Mechanisms Underlying Mammalian Hybrid Sterility in Two Feline Interspecies Models

Abstract

The phenomenon of male sterility in interspecies hybrids has been observed for over a century, however few genes influencing this recurrent phenotype have been identified. Genetic investigations have been primarily limited to a small number of model organisms, thus limiting our understanding of the underlying molecular basis of this well-documented "rule of speciation". We utilized two interspecies hybrid cat breeds in a genome wide association study employing the Illumina 63K SNP array. Collectively, we identified 8 autosomal genes/gene regions underlying associations with hybrid male sterility involved in the function of the blood-testis barrier, gamete structural development, as well as transcriptional regulation. We also identified several candidate hybrid sterility regions on the X chromosome, with most residing in close proximity to complex duplicated regions. Differential gene expression analyses revealed significant chromosome-wide upregulation of X chromosome transcripts in testes of sterile hybrids, which were enriched for genes involved in chromatin regulation of gene expression. Our expression results parallel those reported in Mus hybrids, supporting the 'Large X-Effect' in mammalian hybrid male sterility and the potential epigenetic basis for this phenomenon. These results support the value of the interspecies feline model as a powerful tool for comparison to rodent models of hybrid male sterility, demonstrating unique aspects and potential commonalities that underpin mammalian reproductive isolation.
© The Author 2015. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Full text

Article
Mechanisms Underlying Mammalian Hybrid Sterility in Two
Feline Interspecies Models
Brian W. Davis,
1,2
Christopher M. Seabury,
1,2
Wesley A. Brashear,
1,2
Gang Li,
1
Melody Roelke-Parker,
1,3
and William J. Murphy*
,1,2
1
College of Veterinary Medicine, Texas A&M University
2
Interdisciplinary Program in Genetics, Texas A&M University
3
National Cancer Institute, National Institutes of Health, Bethesda, MD
*Corresponding author: E-mail: wmurphy@cvm.tamu.edu.
Associate editor: Beth Shapiro
Abstract
The phenomenon of male sterility in interspecies hybrids has been observed for over a century, however, few genes
influencing this recurrent phenotype have been identified. Genetic investigationshavebeenprimarilylimitedtoasmall
number of model organisms, thus limiting our understanding of the underlying molecular basis of this well-documented
“rule of speciation.” We utilized two interspecies hybrid cat breeds in a genome-wide association study employing the
Illumina 63 K single-nucleotide polymorphism array. Collectively, we identified eight autosomal genes/gene regions
underlying associations with hybrid male sterility (HMS) involved in the function of the blood-testis barrier, gamete
structural development, and transcriptional regulation. We also identified several candidate hybrid sterility regions on
the X chromosome, with most residing in close proximity to complex duplicated regions. Differential gene expression
analyses revealed significant chromosome-wide upregulation of X chromosome transcripts in testes of sterile hybrids,
whichwereenrichedforgenesinvolvedinchromatinregulation of gene expression. Our expression results parallel those
reported in Mus hybrids, supporting the “Large X-Effect” in mammalian HMS and the potential epigenetic basis for this
phenomenon. These results support the value of the interspecies feline model as a powerful tool for comparison to rodent
models of HMS, demonstrating unique aspects and potential commonalities that underpin mammalian reproductive
isolation.
Key words: hybrid sterility, Haldane’s rule, large X-effect, speciation, feline.
Introduction
Reproductive isolation is a primary barrier supporting the
establishment of new species by restricting gene flow be-
tween related taxa. Darwin recognized reproductive isolation
as essential to the formation of nascent species. While exam-
ining hybridization, he observed a gradient of sterility and
hypothesized a precursory causation of hybrid incompatibil-
ities, stating that sterility was not a result of direct adaptation
but was “incidental on other acquired differences” (Darwin
1859). In 1922, J.B.S. Haldane formalized hybridization obser-
vations from diverse organisms into what is now referred to as
Haldane’s Rule, which states that the heterogametic sex (XY
males in mammals) is preferentially rare, absent, or sterile in
the offspring of two different species (Haldane 1922). This
phenomenon is observed across animal phyla, yet only a
few genes that regulate hybrid sterility have been identified
in two model organisms (i.e., Drosophila and mouse) (Sun
et al. 2004;Mihola et al. 2009;Phadnis and Orr 2009).
Dobzhansky (1937) and Muller (1942) proposed a simple
model explaining hybrid male sterility (HMS) as the result
of incompatible epistatic interactions (known as
Dobzhansky–Muller incompatibilities) between divergent ge-
nomes. Few species have been formally utilized in genetic
analyses to examine this principle of evolutionary biology
and suitable vertebrate models with which to genetically dis-
sect HMS are very rare. In mammals, a single mouse HMS
locus has been identified: TheautosomalH3K4trimethyl-
transferase gene Prdm9 (Mihola et al. 2009). Additionally,
multiple regions of the mouse X chromosome have also
been implicated (Storchov
a et al. 2004;Good et al. 2008;
White et al. 2012), and while fine mapping has uncovered
several candidate genes, none have been shown to be causal
(Bhattacharyya et al. 2014).
The observation that the X chromosome harbors many
loci underlying HMS underpins one of the rules of speciation
(Coyne and Orr 2004), the “Large X-Effect,” based on its dis-
proportionately large influence on postzygotic reproductive
isolation. In Drosophila, genomic incompatibilities are en-
riched on the X and heterospecific introgression of the X
induces sterility more readily than autosomes (Turelli and
Orr 1995;Masly and Presgraves 2007). In natural populations,
gene flow is drastically reduced on the X chromosome com-
paredwithautosomes(Machol
an et al. 2007). Multiple mech-
anisms have been proposed to explain this phenomenon,
including “Dominance theory,” which is based on the expo-
sure of recessive epistatic X-linked incompatibilities in hemi-
zygous males (Turelli and Orr 1995), and “Faster-male theory,”
which suggests male-limited reproductive traits accumulate
ßThe Author 2015. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please
e-mail: journals.permissions@oup.com
Mol. Biol. Evol. doi:10.1093/molbev/msv124 Advance Access publication May 25, 2015 1
MBE Advance Access published June 27, 2015
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

incompatibilities faster than do female-limited traits, due
either to the sensitivity of spermatogenesis to molecular per-
turbation and/or to sexual selection (Wu and Davis 1993).
More generally, the X may simply evolve at a faster rate than
the autosomes (“faster-X theory”) and accumulate incompat-
ibilities more readily (Charlesworth et al. 1987). Mouse
models support the Large X-Effect based on results from nat-
ural hybrid zones (Machol
an et al. 2007) and experimental
crosses (Good et al. 2008,2010;White et al. 2012;
Bhattacharyya et al. 2014). Given the rapid evolution of re-
productive genes and pathways, it currently remains unclear,
however, whether the same or even convergent genetic
mechanisms or pathways drive HMS in divergent taxa.
Reproductive isolation via HMS is proposed to be poly-
genic, involving genes with both large and small influence
(Turelli and Orr 1995;Maside and Naveira 1996). Interspecies
crosses are powerful resources for mapping complex traits
and examining the mechanisms of speciation, requiring
fewer individuals to map genes with moderate to large phe-
notypic effects (L’H^
ote et al. 2010). To explore the genetic
basis of Haldane’s Rule and the potential for a Large X-Effect in
a novel, complimentary mammalian model, we exploited two
interspecies cat hybrids that are among the most common
cat breeds worldwide. Despite a 10 My evolutionary
divergence time (Li G, Davis BW, Eizirik E, Murphy WJ, sub-
mitted), domestic cats will naturally interbreed with African
servals (Profelis serval) in captivity to produce the Savannah
breed, with over 10,500 registered members worldwide
(Wheeler 2011). A second intercross between domestic cats
and Asian leopard cats (Prionailurus bengalensis)produces
the Bengal, one of the most popular cat breeds in the
world (Johnson 1991). These latter two species diverged ap-
proximately 7.2 Ma (Li G, Davis BW, Eizirik E, Murphy WJ,
submitted).
The foundation lines for each breed almost exclusively
originate through mating female domestic cats to males of
the wild species. Fertilization may be accomplished by natural
mating, or more rarely, via artificial insemination. Lines are
normally perpetuated via unidirectional crossing of fertile F1
females to male domestic cats. Hybrid sterility manifests sim-
ilarly in both breeds with F1 and early backcross generations
exhibiting azoospermia and severe seminiferous tubule de-
generation. Later backcross generations show meiotic pro-
gression but still may produce low amounts of sperm with
a high proportion of abnormalities (fig. 1Band D). Later gen-
eration hybrid cats show fertility characteristics analogous to
normospermic domestic cats (fig. 1Aand C); however, on
average, Bengal males regain fertility in earlier backcross
μ
AB
CD
FIG.1. H&E-stained testis from fertile and sterile hybrids. (A) Fertile fifth-generation Savannah backcross hybrid testis with mature, normal sperm. (B)
Sterile fourth generation Savannah backcross hybrid testis showing hypospermatogenesis, with a high incidence of pachytene spermatocytes and the
presence of vacuoles. (C) Fertile fourth-generation Bengal backcross hybrid testis with mature, normal sperm. (D) Sterile fourth-generation Bengal
backcross hybrid testis exhibiting globozoospermia.
2
Davis et al. .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

generations than do Savannahs (supplementary fig. S1,
Supplementary Material online). This is complicated as late
generation, fertile Bengal and Savannah males are often used
to perpetuate breeding lines via intercrossing with fertile F1 or
backcross females from early generations.
Here, we demonstrate the utility of these hybrid cat
models for detecting loci that contribute to HMS by genotyp-
ing cohorts of each hybrid breed on the Illumina 63 K domes-
tic cat single-nucleotide polymorphism (SNP) array. Following
the Dobzhansky–Muller model, we hypothesize that sterile
individuals will have an enrichment for SNP alleles putatively
descending from the wild progenitor species rather than of
domestic origin. We also present genome-wide expression
results based on testis RNAseq data from the parent species
and fertile and sterile hybrid backcross cats, to investigate
patterns of genome-wide and locus-specific misexpression
associated with HMS.
Results
Genome-Wide Association Study Identifies Multiple
Candidate Hybrid Sterility Genes in Savannahs and
Bengals
We analyzed cohorts of sterile and fertile Bengals and
Savannahs in independent genome-wide association studies
(GWAS) to identify SNPs associated with the sterility pheno-
type. We leveraged the domestic cat SNP array to detect
markers fixed in wild progenitor species but variable in the
domestic cat. The results of our two GWAS produced biolog-
ically relevant associations for five chromosomal regions in
Savannahs and three in Bengals (fig. 2). For seven of the eight
loci associated with sterility, there was an enrichment of 1)
putative ancestral alleles that were monomorphic in the wild
parent species and 2) homozygous wild genotypes in sterile
hybrids (supplementary fig. S2,Supplementary Material
online). This enrichment for wild alleles within a predomi-
nantly domestic cat genomic background follows the
Dobzhansky–Muller model and may drive reproductive iso-
lation by genomic incompatibility, thereby corroborating the
theoretical predictions of our model system.
In the savannah GWAS, two top ranking SNP markers
mapped within a pair of genes (CADM1 and AKAP9,fig.
3Aand B) that promote proper organization and function
of the blood-testis barrier (BTB), which is vital for germ
cell development and sperm quality by influencing the
ultrastructural composition of developing sperm
(Cavicchia et al. 1996). Both genes are essential to sper-
matogenesis in mouse and are upregulated in testes of
sterile Savannahs (fig. 3Aand B).CADM1servesasan
adhesion molecule at the Sertoli-spermatid interface of
the BTB, and Cadm1
/
mouse mutants show oli-
goastheno-teratozoospermia and BTB disruption
(Maekawa et al. 2011). Like Cadm1,Akap9 null mice dem-
onstrate failure of spermatogenic progression due to mis-
localization of developing germ cells (Schimenti et al.
2013). Similar phenotypes are observed in some sterile
savannahs (fig. 1B,supplementary fig. S3,Supplementary
Material online). AKAP9 is unique among all the identified
candidate HMS genes in that it is the only gene to display
significant nonsynonymous evolution between the do-
mestic cat and wild species (supplementary table S1,
Supplementary Material online). All other genes show
few or no amino acid altering substitutions that would
be predicted under the faster-male hypothesis, thus im-
plying a major role for interspecific regulatory divergence
in feline HMS (Good et al. 2010).
Thetop-rankingmarkerintheSavannahGWAS(additive
model) lies within intron 4 of GRM8 (glutamate receptor,
metabotropic 8). This is the only testis-expressed gene
within one megabase of the associated marker and is signif-
icantly downregulated in testes of sterile Savannahs (fig. 3C).
In mice, Grm8 is present on the surface of spermatozoa and
neurons (Lee et al. 2011). Although its precise function within
spermatogenesis is currently unknown, GRM8 is one of the
most highly expressed metabotropic glutamate receptors in
the human testis (Julio-Pieper et al. 2013). The fourth signif-
icant savannah SNP resides within LATS2, which encodes a
novel multifunctional serine/threonine kinase whose roles in-
clude control of cell proliferation and ensuring accurate
ploidy during cell division (McPherson et al. 2004). This
gene induces G2/M arrest and apoptosis via its kinase activity
(Kamikubo et al. 2003) and is significantly downregulated in
sterile Savannah testes (fig. 3D). The fifth SNP localizes to a
gene-dense region on chromosome E3 that contains a zinc-
finger gene cluster and resides within ZSCAN25,agenethat
encodes a SCAN domain-containing transcription factor of
unknown function (fig. 3E).
In the Bengal cohort analysis, the top sterility-associated
SNP marker resides within an intergenic interval between two
genes (fig. 3F). Immediately upstream of the SNP lies DNA
methyltransferase 3-like (DNMT3L), which encodes a product
that regulates the establishment of DNA methylation at dis-
persed and paternally inherited sites in the early stages of
spermatogenesis (Bourc’his et al. 2001). Dnmt3l-/- male
mice are sterile due to failure of meiosis as a result of hypo-
methylation (Webster et al. 2005) and exhibit phenotypes
similar to those of many sterile Bengal hybrids, including
loss of spermatocytes due to sloughing, vacuolarization, and
a Sertoli-cell-only like appearance (fig. 1D,supplementary fig.
S3,Supplementary Material online). DNMT3L normal expres-
sion and relative misexpression was the highest of any gene in
the candidate gene region. Moreover, misregulation patterns
were in opposing directions depending on the parent species
under comparison (i.e., domestic vs. wild). DNMT3L was upre-
gulated in sterile hybrids relative to domestic cats and fertile
hybrids but downregulated in comparisons with the Asian
leopard cat, suggesting divergent levels of transcription that
may be incompatible in a hybrid background due to dosage
effects (fig. 3F). Downstream of the same top ranking SNP lies
C21orf33, a mitochondrial homolog of HES1. Although no
specific function has been ascribed to this gene, the protein
product is decreased in sperm samples from men with unex-
plained infertility issues (McReynolds et al. 2014)andmay
warrant future consideration.
Three remaining Bengal GWAS markers reside within or
near a pair of genes whose products putatively interact
3
Mechanisms Underlying Mammalian Hybrid Sterility .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

and are involved in intracellular trafficking via the
Golgi-associated retrograde protein (GARP)/SNARE complex,
a process that is essential for proper sperm development. The
product of the first gene, VPS53, cooperates with VPS52 and
VPS54 to form the GARP complex (Hickey et al. 2013).
Defects in GARP lead to acrosome biogenesis failure and
globozoospermia (Jockusch et al. 2014), a phenotype also
observed within some sterile Bengals (fig. 1D). The second
gene, SNAP25, possessed two significant markers approxi-
mately 25–50 kb upstream of the transcriptional start site.
SNAP25 is a component of the SNARE complex, which di-
rectly interacts with the GARP complex (Fasshauer and
Margittai 2004). SNAP25 was moderately upregulated in ster-
ile versus fertile hybrids (fig. 3G). The VPS53 amino acid se-
quence is conserved between the domestic cat and Asian
leopard cat, and testis transcription showed only modest
changes in expression between sterile hybrids and domestic
cats but more pronounced differences with the Asian leopard
cat (fig. 3H). Taken together, these observations lead us to
hypothesize that misregulation of both genes may alter crit-
ical GARP-SNARE interactions that support proper acrosomal
formation.
X Chromosome-Wide Overexpression in Hybrid
Testes
Whole-testis transcriptome RNAseq data showed very similar
expression patterns between Bengal and Savannah hybrids:
70% of all misregulated transcripts were shared between both
hybrid breeds when sterile hybrids were compared with do-
mestic cat or fertile hybrids (supplementary fig. S4,
Supplementary Material online) and measures of misregula-
tion were robust to different normalization methods (supple-
mentary fig. S5,Supplementary Material online). We observed
a very strong and statistically significant skew toward upre-
gulation of the X chromosome relative to that of autosomes,
in both sterile Savannahs and Bengals (fig. 4Band C)
(Komolgorov–Smirnov P<0.01). Furthermore, only the X
A
B
C
FIG.2. Manhattan plots and marker details for GWAS in two hybrid feline breeds. (A) Five markers (SAV1-5) exceeding Wellcome Trust recommen-
dations for genome-wide significance (P
uncorrected
<510
5
;log
10
P= 4.30, red line) (Wellcome Trust Case Control Consortium 2007) based on
analysis of the savannah cohort (n= 103). The Manhattan plot shown represents the full results under the dominant model of inheritance, but for
brevity, we inserted the –log10 Pvalue of SAV1 based on the additive model, for comparison. The full Manhattan plot under the additive model is
shown in supplementary figure S9,Supplementary Material online. (B) Manhattan plot under a dominant model of inheritance for the Bengal cohort
(n= 101), showing three markers (the two markers for BEN3 are in LD) exceeding genome-wide significance (red line) (Wellcome Trust Case Control
Consortium 2007). (C) Table of markers, Pvalues, coordinates in the FelCat5 assembly and most proximal gene to each marker. P-P plots for each
analysis are shown in supplementary figure S10,Supplementary Material online.
4
Davis et al. .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

chromosome was found to possess significant chromosome-
wide misexpression between sterile and fertile individuals
(Bengals: P=5.010
35
,Savannahs:P=1.510
22
,supple-
mentary table S2,Supplementary Material online). This same
X-upregulation bias was observed in laboratory mouse intra-
species hybrids (Good et al. 2010;Campbell et al. 2013), and
we identified a suite of shared genes misregulated in both
rodent and feline hybrids (supplementary table S3,
Supplementary Material online). An analysis of functional
gene enrichment of the upregulated feline X chromosome
genes revealed “chromatin regulation” as the top enriched
category for both Bengal and Savannah hybrids (supplemen-
tary table S4,Supplementary Material online).
One complication in assessing gene expression differences
between the developmentally different testes of sterile versus
fertile animals is the “cellular composition” bias (Good et al.
2010). Specifically, postmeiotic cells are found in lower pro-
portions in sterile testes, therefore their transcripts are ex-
pected to show lower expression levels compared with
normal testes, even if transcript levels were equivalent per
cell in the two samples. As a result, it has been proposed that
in sterile testes, postmeiotic genes will appear to have reduced
expression levels, whereas mitotic cells will appear overex-
pressed, and therefore differential expression may not actually
reflect true misregulation. To examine whether the proposed
tissue composition bias impacted our differential expression
results, we examined the misregulation patterns of three
genes that were shown to be robust to tissue composition
in comparisons between normal human adult testis and mul-
tiple testicular germ cell tumor pathologies (Svingen et al.
2014) and have been validated to be reliable for normalization
experiments. Two of these putative “housekeeping” genes,
RPS20 and SRSF4, were not misregulated in any of the sterile
hybrid to fertile comparisons (supplementary table S5,
Supplementary Material online). The third gene, RPS29,was
not examined due to the presence of multiple processed
pseudogenes and numerous high identity BLAT hits else-
where in the domestic cat genome, making it a poor
F
DW
F
DWF
DW
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
W
D
D
D
D
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
F
DWF
DW
F
DWF
DW
F
D
W
D
D
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
F
DW
W
D
W
F
D
F
D
W
D
F
DWF
DW
F
DWF
DWF
DW
F
DWF
DWF
DWF
DWF
DW
D
W
D
D
F
D
W
AE
BF
CG
DH
FIG.3. (A–H) Genomic regions surrounding each top ranked marker and testis expression data for the region. A 500-kb window surrounding each
significant marker is indicated by vertical lines on the chromosome ideogram. –log10 Pvalues (yaxis) for both additive and dominant inheritance
models are plotted along each chromosome (xaxis), with significance threshold indicated by horizontal red line. Log
2
fold testis expression change in
sterile hybrids when compared with domestic cat (D), fertile hybrids (F), and wild species (W) are shown in red (significant upregulation), green
(significant downregulation), and gray (nonsignificant misregulation) (supplementary table S8,Supplementary Material online).
5
Mechanisms Underlying Mammalian Hybrid Sterility .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

candidate for RNAseq normalization. The stability of RPS20
and SRSF4 across fertile and sterile testes suggests that tissue
composition is unlikely to have strongly biased our results,
specifically the highly significant chromosome-wide upregu-
lation of genes on the X chromosome and not on any auto-
some (supplementary table S2,Supplementary Material
online). Future experiments using flow-sorted testis cell-
type preparations will further test the validity of these results.
X-chromosome overexpression in hybrid mice is hypothe-
sized to be due in part to failed trans-acting loci on the X
chromosome (Good et al. 2008;White et al. 2012;Turner et al.
2014). GWAS results for the X chromosome identified sev-
eral suggestive markers (P
uncorrected
<510
-4
)that
approached but did not exceed our imposed significance
threshold (see Materials and Methods; supplementary table
S6,Supplementary Material online). We attribute this to 1)
lower X chromosome SNP density on the array (supplemen-
tary fig. S6,Supplementary Material online) and 2) the obser-
vation in less divergent mouse intraspecific hybrids that the
Large X-Effect is polygenic (Good et al. 2008). Given the
greater evolutionary divergence observed in the feline inter-
specific hybrids, we hypothesized that the X-linked signal
would be even more polygenic, with multiple X-linked loci
modulating small to moderate individual effects.
Nevertheless, in Savannahs, 42% of the top 50 SNP markers
were X-linked, thereby affirming an X-linked contribution to
log210
FMR1
A
B
C
D
FIG.4. X chromosome copy number, GWAS significance, and RNA-seq gene misregulation in sterile feline hybrids. (A) Number of duplicated copies
detected by CNVator across the X chromosome. Blue lines on the ideogram indicate position along the X chromosome (xaxis). Purple lines indicate
number of detected copies (yaxis). (B)–Log
10
Pvalues of SNP markers (only values above 3.3 are shown, for brevity) that approached significance on
the X chromosome of each breed. Orange vertical lines indicate Bengal and red indicates Savannah. (C) Gene misexpression on the X chromosome in
each sterile hybrid when compared with the domestic cat. Red indicates upregulation and green downregulation. The xaxis denotes the position along
the X chromosome, and the yaxis indicates the log
2
(fold change) in expression for each gene when compared with domestic cat. (D) Contrasting
misregulation patterns on the X chromosome and autosomes in sterile hybrids when compared with domestic cat, fertile hybrids, and wild species. x
axes list the integer fold-change categories (from negative 10-fold change to positive 10-fold change) for which each gene is classified based on the
magnitude of misexpression. yaxes show the frequency of genes falling into these categories. Each X-autosome distribution was significantly different for
all six comparisons (Komolgorov–Smirnov P<0.01).
6
Davis et al. .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

sterility (supplementary table S7,Supplementary Material
online). When we compared the highest X-linked GWAS re-
gions defined by SNPs in Savannahs and Bengals (fig. 4B,
supplementary table S6,Supplementary Material online) to
regional duplication profiles generated by structural variation
analysis, we observed that most of the top markers mapped
near regions containing duplications (fig. 4A).
In the Bengal GWAS, the top four X-linked markers shared
identical support values (log
10
P= 3.57), and one marker
maps 70 kb upstream of fragile X mental retardation syn-
drome 1 homolog (FMR1), a highly expressed gene in
mouse spermatogonia and spermatocytes (Hickey et al.
2013;Alpatov et al. 2014). Notably, Fmr1 resides within the
mouse Hstx2 (hybrid sterility X 2) critical interval
(Bhattacharyya et al. 2014). A cluster of three additional
SNP markers localize to distal Xq within another structurally
complex region (fig. 4A). In humans, naturally occurring du-
plications within the syntenic Xq27.3–q28 region induce fer-
tility problems and reduce testis weight, underpinning the
role of this region in eutherian mammal testis function.
Discussion
As an increasing number of intra and interspecific genome
comparisons become available, an emerging theme from
these studies is that postspeciation gene flow often extends
for several million years following an initial divergence (Toews
and Brelsford 2012;Kutschera et al. 2014;LiG,DavisBW,
Eizirik E, Murphy WJ, submitted). Although our examination
of divergent interspecies hybrids cannot guarantee identifica-
tion of all genomic regions that may have initiated the diver-
gence of these felid species, it does highlight several that are
essential to the reproductive success of males by their disrup-
tion in hybrids and hence, those potentially restricting post-
speciation gene flow. Our combined GWAS plus RNAseq
strategy identified multiple candidate hybrid sterility genes
and a strong X-linked expression bias in sterile hybrids.
Collectively, our results implicate several molecular/cellular
pathways that appear to play a major role in felid hybrid
sterility, as well as some intriguing commonalities with
mouse models of hybrid sterility that may point to conserved
mammalian pathways that are sensitive to hybrid
incompatibilities.
BTB Compartmentalization Failure
Regulation of cellular signaling between developing germ cells
and supporting Sertoli cells is necessary for accurate sperma-
togenesis in mammals. Proper organization of the BTB is es-
sential to this process (Mital et al. 2011). The primary function
of adult Sertoli cells is to provide structure to the seminiferous
tubule, nurture maturing germ cells, and compose specialized
junction complexes that comprise the BTB (Mruk and Cheng
2004). One of these specialized junctionsistheectoplasmic
specialization (ES). Assembly of the ES anchors elongating
spermatids to Sertoli cells via the forming acrosome and dis-
assembly releases sperm into the seminiferous tubule lumen
(Vogl et al. 2000). CADM1 is a Sertoli-spermatid adhesion
molecule in the apical ES, and Cadm1 deficiency causes
round and elongating spermatids to detach from Sertoli
cells, resulting in sloughing of germ cells prior to the comple-
tion of maturation (Wakayama et al. 2007;Wakayama and
Iseki 2009). The resulting severe sperm head and midpiece
abnormalities cause poor binding to the zona pellucida
(Mueller et al. 2003). These phenotypes are similar to those
observed in hybrid cats (fig. 1Band D,supplementary figs. S2
and S7,Supplementary Material online).
AKAP9 also plays a role in the BTB, and when ablated in
mice, the BTB junction organization is disrupted, leading to
compartmentalization failure of developing spermatozoa by
Sertoli cells (Cavicchia et al. 1996), which is characterized by
mislocalization of developing germ cells, few round sperma-
tids, and Sertoli cells with characteristics resembling prepu-
bertal testes (Sehrawat et al. 2011). The large sequence
divergence between domestic cat and serval AKAP9 alleles
in sterile Savannah hybrids, including putative frameshift and
numerous nonsynonymous substitutions, may facilitate the
severe testis compartmentalization and failure of sperm pro-
duction present in the Savannah HMS phenotype (fig. 1B,
supplementary fig. S3,Supplementary Material online).
AKAP9 also functions as a scaffolding protein that is required
for assembling phosphatases and protein kinases on the cen-
trosome and Golgi apparatus. It is therefore plausible that
AKAP9 interacts with another gene product identified in
our GWAS, the serine/threonine kinase LATS2, which local-
izes to centrosomes and regulates the cell cycle and apoptosis
(Yabuta et al. 2014).
Acrosomal Development and GARP/SNARE
Intracellular Trafficking
The importance of acrosome development in felid HMS phe-
notypes is highlighted by three of the top eight genes in the
two separate GWAS, including the top ranked Savannah
GWAS marker that resides within GRM8. Multiple lines of
evidence point to a role for GRM8 in sperm acrosome func-
tion (Lee et al. 2011;Julio-Pieper et al. 2013) and knockout of a
similar gene, Gmr7, produces mice with reduced numbers of
motile sperm (Marciniak et al. 2014).Onecluetotheroleof
GRM8 in spermatogenesis and hybrid sterility is the cAMP
regulatory task played by GRM8 in other tissues. Because
spermatozoa control the acrosome reaction via cAMP-de-
pendent processes (Lin et al. 2006), the sperm-specific pres-
ence of this mGlu receptor in the mammalian acrosome may
repress the production of cAMP, and in turn prevent acro-
some exocytosis (Baker et al. 2013). Therefore, the observed
downregulation of this gene in sterile hybrid testes is consis-
tent with a failed interaction within a hybrid background.
Cargo transport between endosomal-lysosome mem-
branes is essential to the proper development of the sperm
acrosome (Vassilieva and Nusrat 2008). Vesicular membrane
fusion is a vital component of sperm maturation that requires
interaction of the SNARE motif and the N-terminus of VPS53/
VPS54 within the GARP complex (P
erez-Victoria and
Bonifacino 2009). A naturally occurring missense mutation
in VPS54, the co-interacting partner of our candidate gene
VPS53, defines the “wobbler” mouse mutant, and changes the
7
Mechanisms Underlying Mammalian Hybrid Sterility .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

dynamics of retrograde endocytosis between the sperm head
and the Golgi, impairing acrosome formation and inducing
globozoospermia (Paiardi et al. 2011). Furthermore, the weak-
ening of Sertoli-germ cell contacts in “wobbler” mutants im-
pedes the promotion of germ cell survival and maturation in
a manner similar to BTB failure, resulting in poor motility and
overall reduced germ cell numbers. Both phenotypes are ob-
served in sterile Bengal hybrids (fig. 1B, supplementary fig. S3,
Supplementary Material online). SNAP25 is a primary com-
ponent of the SNARE complex and is essential to acrosome
exocytosis (Tomes et al. 2002). Given the demonstrated link
between GARP-SNARE complexes in the formation of the
acrosome, we hypothesize that VPS53 and SNAP25 represent
partners in a Dobzhansky–Muller incompatibility in the
Bengal interspecies model and warrant further study within
the contexts of normal spermatogenesis and reproductive
isolation.
Transcriptional Regulation and Methylation
Oneemergingfeaturesharedbyproteinsencodedby
Drosophila and mouse HMS genes, as well as genes found
within chromosomal regions of high intraspecific divergence,
is that they encode binding proteins involved in transcrip-
tional regulation (Carneiro et al. 2014).Theputativefunctions
of several candidate felid HMS genes identified in this study as
well as the observed patterns of gene misregulation also sug-
gest an important role of transcriptional regulation in feline
HMS (Good et al. 2010). For example, the second ranked
Savannah HMS SNP resides within ZSCAN25, which encodes
a zinc-finger protein with SCAN domains important for DNA
binding and transcriptional regulation. ZSCAN25 is one of
many mammalian transcription factors with unknown func-
tion, and our results warrant further functional investigation
via knockdown strategies to better understand its role in
normal spermatogenesis. Notably, a mouse hybrid sterility
QTL localizes to this same gene-dense interval on the distal
q-arm of mouse chromosome 5 and also overlaps a strong
expression QTL (White et al. 2012;Turner et al. 2014). Further
fine mapping of both feline and mouse loci would address the
intriguing possibility of an HMS gene shared across mamma-
lian orders.
The top ranked Bengal GWAS SNP lies between two genes,
DNMT3L and C21orf33, neither of which could be excluded
based on our existing mapping resolution or RNAseq analy-
ses. However, the most compelling HMS candidate gene is
DNMT3L, based on several key observations. DNMT3L regu-
lates the functional activity of two DNA methyltransferases,
DNMT3A/B, which are required during early meiotic stages of
development, and its absence leads to a myriad of meiotic
defects and male infertility (Bourc’his et al. 2001;Webster
et al. 2005). Many of these meiotic defects were also observed
in testes of sterile Bengal hybrids (supplementary fig. S3,
Supplementary Material online). Dnmt3l also resides within
a mouse hybrid sterility QTL/eQTL region on chromosome 10
(Turner et al. 2014), and its downstream regulatory target
Dnmt3a is misregulated in mouse hybrids (Good et al.
2010). DNMT3L regulates chromatin changes essential for
formation of the sex body during the pachytene stage of
meiosis that silences sex chromosome gene expression, a pro-
cess termed meiotic sex chromosome inactivation (Burgoyne
et al. 2009). Dnmt3L heterozygous mutant male mice show
extended sex bodies and misregulation of a large number of
meiotic transcripts, many from the X chromosome (Zamudio
et al. 2011). This known connection between Dnmt3L func-
tion and X chromosome epigenetic modifications is strongly
suggestive of a connection between this candidate hybrid
sterility locus and the overexpression observed in X chromo-
somes of sterile Bengal hybrids. Finally, the concordant pat-
tern of testis misregulation observed in the two feline
interspecies hybrids (supplementary fig. S4,Supplementary
Material online) indicates a similar mechanism influencing
male sterility within both felid hybrids.
Conservation of the Large X-Effect in Mammalian
Hybrid Gene Expression
TheroleoftheXchromosomeinhybriddysfunctioniswell
documented across animal phyla, but the molecular basis of
the Large X-Effect in different organisms is not fully under-
stood. Any number of properties of the X chromosome, in-
dividually or in combination, may contribute to the Large X-
Effect, including faster male evolution (Wu and Davis 1993;
Torgerson and Singh 2003), adaptive evolution of male-biased
genes (Ellegren and Parsch 2007), and gene movement
(Moyle et al. 2010). Moreover, patterns in X chromosome-
wide overexpression in sterile intraspecific mouse hybrids
(Good et al. 2010;Turner et al. 2014) allude to the influence
of both cis- and trans-acting factors on the Large X-Effect. The
enrichment of X-upregulated genes involved in chromatin
modification in felid hybrids coupled with the disruption of
meiotic sex chromosome inactivation in infertile hybrid mice
(Good et al. 2010;Campbell et al. 2013;Bhattacharyya et al.
2014) indicates the plausibility of either conserved or conver-
gent regulatory mechanisms shared between felid and rodent
hybrids.
One of four top ranked Bengal X-linked SNP markers
(log
10
P= 3.57) lies immediately upstream of FMR1,which
encodes fragile X mental retardation protein FMRP. FMRP has
been shown to interact with pachytene-stage meiotic chro-
mosomes and DNA damage response components during
mouse spermatogenesis, and Fmr1 knockout mice show ap-
optosis and failure of spermatogenesis (Alpatov et al. 2014).
FMRP’s direct chromatin interactions with components nec-
essary for sex body formation suggest a potential role in
hybrid X misregulation, especially when considered within
the context of the misregulated genes associated with chro-
matin binding in sterile cat hybrids. Given that the top ranked
Bengal autosomal candidate gene DNMT3L directly regulates
methylation and epigenetic changes in early spermatogenesis,
and histone H3K79 methylation aids in recruitment of FMRP
to meiotic chromatin, we speculate that failed heterospecific
interactions involving DNMT3L and FMRP may contribute to
the X chromosome-wide upregulation that induces apoptosis
and sterility.
8
Davis et al. .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

In addition to the region upstream of FMR1,weidentified
several other candidate X chromosome intervals potentially
associated with the sterility phenotype in both hybrid crosses.
Interestingly, many of the top X chromosome SNPs reside
adjacent to or within structurally complex chromosomal re-
gions that contain duplications (fig. 4Aand B). This observa-
tion is relevant based on the previously described association
between mapped mouse hybrid sterility QTL and ampliconic
gene clusters on the mouse X chromosome (Mueller et al.
2013). In mice, ampliconic gene clusters are enriched for
postmeiotic genes and comprise 18% of all protein-coding
genes on the mouse X (Mueller et al. 2008). These highly
repetitive regions are typically collapsed and/or unassigned
in the majority of draft mammalian genome assemblies, like
cat, but leave signatures of high copy number.
The recent resolution of mouse X and Y chromosome
amplicon sequences has led to the recognition of several
gene family members that are shared between the two sex
chromosomes and whose relative copy number has evolved
under the pressures of sex-linked meiotic drive (Soh et al.
2014), an additional mechanism that has been proposed to
underlie the Large X-Effect. Two particular genes, Sly on the Y
chromosome and Slx on the X chromosome, are among the
largest genes families, with greater than 130 and 25 functional
copies in the Mus musculus C57BL/6 strain, respectively. Mice
lacking adequate Sly expression, either via chromosomal de-
letion or knockdown show sperm abnormalities and infertility
(Tour
e et al. 2004;Ellis et al. 2005;Cocquet et al. 2009). In both
scenarios, the paralogous Slx gene(s)andotherpostmeiotic
genes are upregulated. This has led to the hypothesis that Slx
and Sly recently evolved in the genus Mus as a meiotic drive
system and that the acquisition and massive amplification of
these ampliconic gene families has been driven by X-Y inter-
chromosomal conflict (Cocquet et al. 2009; Ellis et al. 2011;
Soh et al. 2014).
X-Y conflict may be widespread in mammals, as evidenced
by lineage-specific X-Y shared gene families in several species,
including VCX-VCY gene families in simian primates
(Skaletsky et al. 2003)andCUL4BX-CUL4BY in the cat
family (Murphy et al. 2006;Li et al. 2013). Felid ampliconic
Y chromosome gene families show considerable size and
sequence variation between different cat species and there-
fore may have evolved as a parallel genomic manifestation of
sex-linked meiotic drive (Li et al. 2013). We hypothesize that
copy number variation between sex-linked amplicons of dif-
ferent species may provide a mechanism for reproductive
isolation and the “islands of speciation” that are commonly
observedinmammalianhybridzones(e.g.,Carneiro et al.
2014;Phifer-Rixey et al. 2014). Future fine mapping using
unbiased, sequence-based approaches in larger cohorts will
aid in the further elucidation of specific X-linked genes/gene
families that contribute to feline HMS and allow us to test this
hypothesis.
In summary, we identified eight autosomal HMS candidate
loci and several X chromosome candidate regions within hy-
brids of two divergent feline interspecific crosses. Most HMS
loci relate to a single compelling hybrid sterility candidate
gene based on mapping resolution, expression data, and
published functional data for mouse and human. At least
three candidate gene regions are syntenic with mouse
hybrid sterility QTL/eQTL on Chromosomes 5, 10, and X,
raising the possibility that conserved hybrid sterility network
components might be disrupted in mouse and cat hybrids.
Felid X chromosome genes were strongly overexpressed in
sterile hybrids relative to fertile controls and were functionally
enriched for chromatin-modification-related biological pro-
cesses, suggesting a similar epigenetic manifestation of the
Large X-Effect in both felids and mice. Divergence of noncod-
ing regulatory sequences has long been suspected to play an
important role in interspecies divergence (King and Wilson
1975;Haerty and Singh 2006), and this point is emphasized by
the lack of amino acid divergence, yet significant expression
divergence and hybrid misregulation observed for most felid
HMS genes. Future resequencing and functional/epigenetic
annotation of candidate intervals defined by these two
hybrid crosses will shed light on the mechanistic basis of
noncoding incompatibilities that lead to HMS. Finally, similar
genetic mapping studies in less-diverged feline interspecific
breeds, such as the Chausie (a hybrid between the Jungle cat
and domestic cat, which diverged ~3–4 Ma) will expand op-
portunities to examine the evolutionary origins and accumu-
lation of hybrid incompatibilities within this recent species
radiation.
Materials and Methods
Creation of Interspecies Domestic Cat Hybrids
To produce F1 Savannah and Bengal hybrids, breeders se-
lected males from African serval and Asian leopard cat spe-
cies, respectively. These were crossed to female domestic cats
by either cohabitation and natural mating or more rarely,
artificial insemination. Testes of F1 males from both crosses
show Leydig cell hyperplasia with seminiferous tubules char-
acterized by Sertoli-cell-only phenotype and an absence of a
defined lumen (supplementary fig. S8,Supplementary
Material online). First-generation backcross males generally
show a similar, though often milder phenotype, with the oc-
casional presence of spermatogonia and early spermatocytes.
F1 hybrid females from both breeds are fertile, thus they are
used in subsequent, unidirectional backcross matings to fer-
tile domestic cat males, until fertile backcross hybrid males are
produced. The backcross generation in which each hybrid
breed regains fertility varies with evolutionary distance be-
tween the pair of parent cat species (over 10 and 7 My for
the parents of the Savannah and Bengal, respectively) (Li G,
Davis BW, Eizirik E, Murphy WJ, submitted). Within our study
population, some Bengals produced viable sperm as early as
the second backcross generation, and Savannahs as early as
the third backcross generation, although these are considered
rare in the breeding communities, who regularly backcross
early generation females to late generation, fertile hybrid
males.
Fertility Evaluation
Determination of fertility was performed using one or more of
three methods: Breeding records, semen evaluation, and
9
Mechanisms Underlying Mammalian Hybrid Sterility .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

histopathology. Sterility was defined as repeated, confirmed
matings with multiple proven females over 1 or more years
with no conception, whereas fertile individuals were defined
by documented breeding with validation via pedigree records.
When available, electroejaculation was used to evaluate the
presence of sperm and ultrastructural abnormalities (Howard
et al. 1990). Cats with ejaculates possessing greater than 70%
abnormal sperm (i.e., teratospermic) or exhibiting complete
azoospermia were classified as sterile (Pukazhenthi et al.
2006), when breeding information was not available.
Histopathological evaluation was performed on testes and
epididymis from sexually mature males that underwent orch-
idectomy. Testes were laterally bisected and stored in Bouin’s
fixative. Testis and epididymis were embedded in paraffin,
sectioned, and stained with H & E. Histology was evaluated
to determine the presence/absence of germ cells, stage of
meiotic progression, and the presence/absence of normal
sperm. Fertile individuals possessed seminiferous tubules
and caput epididymis with large numbers of sperm with
normal morphology. Histological and semen evaluation
data were available for all individuals utilized in RNA-Seq
analyses.
Feline SNP Array
DNAs from 101 Bengals, 103 Savannahs, 20 African servals,
and 13 Asian leopard cats were genotyped using the Illumina
63 k Feline SNP array, which assays 62,897 biallelic markers.
Hybrid cats possessing a genotype call rate <0.90 were re-
moved (99% of all study cats possessed call rates 495%).
Since the creation of the 63 K array, revisions to the feline
genome required reassessment of the physical marker loca-
tions. The probe sequences were obtained and compared
with FelCat5 using a local BLAST implementation (Altschul
1997) that identified 1,101 SNPs that could not be remapped
to the genome assembly.
GWAS
Hybrid cats used for each binary case-control style GWAS
were as follows: Bengals (n= 60 fertile; n=41 sterile) and
Savannahs (n=56 fertile; n=47 sterile) (supplementary fig.
S1,Supplementary Material online). We searched for marker-
based associations meeting or exceeding the Wellcome Trust
recommendations (P
uncorrected
510
-5
;log
10
P= 4.30)
(Wellcome Trust Case Control Consortium 2007). Notably,
this significance threshold is conservative considering the
polygenic nature of hybrid sterility and the modest SNP den-
sity of the Illumina feline SNP array, with the Wellcome Trust
recommendations developed for a much higher density SNP
array (Human Affymetrix 500 K GeneChip; see Wellcome
Trust Case Control Consortium 2007). All marker-based as-
sociation analyses were carried out using a mixed linear
model, as described and implemented in EMMAX (Kang
et al. 2010;Segura et al. 2012), and were executed in the
SVS environment (Golden Helix, Version 7.7.6) as described
(Neibergs et al. 2014). Briefly, the mixed model can be gener-
ally specified as: y¼XþZu þ,whereyis a n1vector
of the observed phenotypes, Xis a nqmatrix of fixed
effects, is a q1 vector representing the coefficients of
the fixed effects, and Zis a ntmatrix relating the instances
of the random effect to the phenotypes of interest (Neibergs
et al. 2014;http://doc.goldenhelix.com/SVS/8.2.1/mixed_
models.html, last accessed June 10, 2015). Notably, we must
assume that varðuÞ¼2
gKand varðÞ¼2
eI,suchthat
varðyÞ¼2
gZKZ0þ2
eI, but in this study, Zis simply the
identity matrix I,andKis a kinship matrix among all samples.
To solve the mixed model equation using the generalized
least squares solution, the variance components (2
gand
2
e) must first be estimated as described previously (Kang
et al. 2010;Segura et al. 2012;Neibergs et al. 2014;http://
doc.goldenhelix.com/SVS/8.2.1/mixed_models.html,last
accessed June 10, 2015). We used the REML-based EMMA
approach to estimate the variance components (Kang et al.
2008), with stratification among hybrid cats accounted for
and controlled using an IBS kinship matrix computed from
the Illumina 63 K genotypes (Kang et al. 2010;Segura et al.
2012). Given precedence for dominance effects between the
parental genomes of interspecific hybrids (Payseur and
Hoekstra 2005;Chang and Noor 2007;White et al. 2012),
we evaluated both dominant and additive models (fig. 2A,
supplementary fig. S9,Supplementary Material online) using
EMMAX, with subsequent comparison of the results. We also
estimated the impact of including backcross generation as a
covariate in our GWAS analyses and found that most of the
top ranking markers described in this study remained, indi-
cating that the kinship matrix corrects for most of the gen-
eration-based stratification. For example, in at least one
model (i.e., Additive or Dominant with inclusion of backcross
generation), markers SAV1–SAV3 ranked among the top 3
markers, whereas SAV5 and SAV4 were ranked among the
top 6 or 7 markers, respectively. Likewise, BEN1 and BEN2
were ranked among the top 4 markers, whereas BEN3 (com-
prised two adjacent markers) ranked lower (i.e., among the
top 31-32 markers). P-P plots are shown in supplementary
figure S10,Supplementary Material online.
RNA-Seq and Expression Analysis
Whole-testis RNA was extracted from 25 mg of tissue using
the Purelink RNA extraction kit (Life Technologies) for three
fertile domestic cats, three fertile Bengals (assessed via histo-
pathology and electroejaculation), three azoospermic (sterile)
Bengals (two first and one third generation backcross hy-
brids), three fertile Savannahs (multiply proven breeders),
and three azoospermic (sterile) Savannahs (one second-
and two fourth-generation backcross hybrids). All animals
were over the age of two and sexually mature. The integrity
and concentration of total RNA samples was quantified using
a Bioanalyzer 2100 system (Agilent). Fifty base-pair, single-end
Illumina sequencing was executed for Bengal and domestic
cat samples, and 100-bp single-end sequencing for Savannah,
Asian leopard cat, and African serval samples using the
Illumina HiSeq2000. RNAseq reads have been deposited
under SRA accession numbers SRS913307–SRS913310,
SRS913312, SRS913313, SRS913318, SRS913319, and
SRS913323–SRS913330. Sequence reads were aligned to the
10
Davis et al. .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

domestic cat genome (FelCat5) using a two-pass alignment
strategy in the program STAR (Dobin et al. 2013). The align-
ments were sorted with SAMtools (Li et al. 2009). Polymerase
chain reaction duplicates marked with PicardTools
(http://broadinstitute.github.io/picard/, last accessed June
10, 2015) for the purpose of indel and SNP detection only.
The GATK walker SplitNCigarReads was used to split reads
into exons and hard-clip intronic sequences (McKenna et al.
2010). Mapping quality reassignment (MAPQ60) was applied
to the alignments via GATK ReassignOneMappingQuality.
Base quality score recalibration and indel realignment was
performed on each individual separately, with SNP/indel dis-
covery and genotyping across all 17 samples implemented
simultaneously using variant quality score recalibration ac-
cording to GATK Best Practices recommendations (Van der
Auwera et al. 2013). Read count tables were generated using
HTSeq-count (Liao et al. 2014). Reads were counted per-gene
and as a union of all feature sets. Every feature with an
Ensembl identifier was included in the differential expression
set. Multiply mapped reads of equal quality were randomly
assigned primary status and reads that mapped to more than
five locations were discarded. We implemented several nor-
malization approaches implemented in the EdgeR
Bioconductor package (Robinson et al. 2010), including quar-
tile normalization, trimmed mean of M (TMM) values, and
relative log expression. To be conservative, we utilized the
intersection of these approaches (supplementary fig. S5,
Supplementary Material online) for downstream statistical
analyses of chromosome expression bias and gene ontology
enrichment. Relative expression differences between pairwise
comparisons for the candidate genes were based on the TMM
values. Chromosome-wide misexpression was tested for sig-
nificance using a chi-squared test.
X Chromosome Duplication Analysis
We identified putative genomic duplications in the domestic
cat version 6.2 assembly (felCat5) by remapping all Illumina
fragment reads (Montague et al. 2014) and estimating copy
number with CNVnator (Abyzov et al. 2011).
Supplementary Material
Supplementary figures S1–S10 and tables S1–S8 are available
at Molecular Biology and Evolution online (http://www.mbe.
oxfordjournals.org/).
Acknowledgments
We would like to thank Marilyn Raymond, Eduardo Eizirik,
Lynn Colenda, and Buhdan Pukazhenthi for assistance with
the generation of early generation Bengal hybrids at NIH. We
thank Alison Pearks Wilkerson, Breanne Harty, and Nicholas
Kennedy for technical assistance. We thank Lynn Bustamante,
TAMU Veterinary Histopathology Laboratory, for testis slide
preparations, Dr Brad Weeks for slide scanning and imaging,
and Dr Beverly Purswell for savannah histopathology reports.
We also thank the numerous Bengal and Savannah cat bree-
ders for generously contributing samples, pedigrees, and
breeding data for this project, especially Dr Kristin Filseth
and Kathrin Stucki. This work was supported, in part, by
Morris Animal Foundation (MAF) Grant D12FE-019 to
W.J.M., Cat Health Network/MAF grant D12FE-502 to
W.J.M. and C.M.S., and Winn Feline Foundation Grant 07-
034 to W.J.M. B.W.D. acknowledges the support of NIH LRP
award 1L50HD070410-01 from the NICHD and the Texas
A&M College of Veterinary Medicine & Biomedical
Sciences. C.M.S. acknowledges support from the Texas
A&M College of Veterinary Medicine & Biomedical Sciences
and Texas AgriLife Research. B.W.D., C.M.S., M.R.-P., and
W.J.M. designed research; B.W.D., C.M.S., W.B., G.L., M.R.-P.,
and W.J.M. performed research; C.M.S. and M.R.-P. contrib-
uted new reagents or analytic tools; B.W.D., C.M.S., W.B., G.L.,
and W.J.M. analyzed data; and B.W.D., C.M.S., and W.J.M.
wrote the paper with input from all authors.
References
Abyzov A, Urban AE, Snyder M, Gerstein M. 2011. CNVnator: an ap-
proach to discover, genotype, and characterize typical and atypical
CNVs from family and population genome sequencing. Genome Res.
21:974–984.
Alpatov R, Lesch BJ, Nakamoto-Kinoshita M, Blanco A, Chen S, St €
utzer
A, Armache KJ, Simon MD, Xu C, Ali M, et al. 2014. A chromatin-
dependent role of the fragile X mental retardation protein FMRP in
the DNA damage response. Cell 157:869–881.
Altschul S. 1997. Gapped BLAST and PSI-BLAST: a new generation of
protein database search programs. Nucleic Acids Res. 25:3389–3402.
Baker MA, Naumovski N, Hetherington L, Weinberg A, Velkov T, Aitken
RJ. 2013. Head and flagella subcompartmental proteomic analysis of
human spermatozoa. Proteomics 13:61–74.
Bhattacharyya T, Reifova R, Gregorova S, Simecek P, Gergelits V, Mistrik
M, Martincova I, Pialek J, Forejt J. 2014. X chromosome control of
meiotic chromosome synapsis in mouse inter-subspecific hybrids.
PLoS Genet. 10:e1004088.
Bourc’his D, Xu GL, Lin CS, Bollman B, Bestor TH. 2001. Dnmt3L and the
establishment of maternal genomic imprints. Science 294:2536–
2539.
Burgoyne PS, Mahadevaiah SK, Turner JM. 2009. The consequences of
asynapsis for mammalian meiosis. Nat Rev Genet. 10:207–216.
Campbell P, Good JM, Nachman MW. 2013. Meiotic sex chromosome
inactivation is disrupted in sterile hybrid male house mice. Genetics
193:819–828.
Carneiro M, Albert FW, Afonso S, Pereira RJ, Burbano H, Campos R,
Melo-Ferreira J, Blanco-Aguiar JA, Villafuerte R, Nachman MW, et al.
2014. The genomic architecture of population divergence between
subspecies of the European rabbit. PLoS Genet. 10:e1003519.
Cavicchia JC, Sacerdote FL, Ortiz L. 1996. The human blood-testis barrier
in impaired spermatogenesis. Ultrastruct Pathol. 20:211–218.
Chang AS, Noor MAF. 2007. The genetics of hybrid male sterility be-
tween the allopatric species pair Drosophila persimilis and D.pseu-
doobscura bogotana: dominant sterility alleles in collinear autosomal
regions. Genetics 176:343–349.
Charlesworth B, Coyne JA, Barton NH. 1987. The relative rates of evo-
lution of sex chromosomes and autosomes. Am Nat. 130:113–146.
Cocquet J, Ellis PJ, Yamauchi Y, Mahadevaiah SK, Affara NA, Ward MA,
Burgoyne PS. 2009. The multicopy gene Sly represses the sex chro-
mosomes in the male mouse germline after meiosis. PLoS Biol.
7:e1000244.
Coyne JA, Orr HA. 2004. Speciation. Sunderland (MA): Sinauer
Associates.
Darwin C. 1859. On the origin of species by means of natural selection,
or the preservation of favoured races in the struggle for life. London:
John Murray.
Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P,
Chaisson M, Gingeras TR. 2013. STAR: ultrafast universal RNA-seq
aligner. Bioinformatics 29:15–21.
11
Mechanisms Underlying Mammalian Hybrid Sterility .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

Dobzhansky TG. 1937. Genetics and the origin of species. New York:
Columbia University Press.
Ellegren H, Parsch J. 2007. The evolution of sex-biased genes and sex-
biased gene expression. Nat Rev Genet. 8:689–698.
Ellis PJ, Clemente EJ, Ball P, Tour
e A, Ferguson L, Turner JM, Loveland KL,
Affara NA, Burgoyne PS. 2005. Deletions on mouse Yq lead to
upregulation of multiple X- and Y-linked transcripts in spermatids.
Hum Mol Genet. 14:2705–2715.
Fasshauer D, Margittai M. 2004. A transient N-terminal interaction of
SNAP-25 and syntaxin nucleates SNARE assembly. JBiolChem.
279:7613–7621.
Good JM, Dean MD, Nachman MW. 2008. A complex genetic basis to X-
linked hybrid male sterility between two species of house mice.
Genetics 179:2213–2228.
Good JM, Giger T, Dean MD, Nachman MW. 2010. Widespread over-
expression of the X chromosome in sterile F1 hybrid mice. PLoS
Genet. 6:e1001148.
Haerty W, Singh RS. 2006. Gene regulation divergence is a major con-
tributor to the evolution of Dobzhansky-Muller incompatibilities
between species of Drosophila.MolBiolEvol.23:1707–1714.
Haldane JBS. 1922. Sex ratio and unisexual sterility in hybrid animals.
J Genet. 12:101–109.
Hickey SE, Walters-Sen L, Mosher TM, Pfau RB, Pyatt R, Snyder PJ, Sotos
JF, Prior TW. 2013. Duplication of the Xq27.3-q28 region, including
the FMR1 gene, in an X-linked hypogonadism, gynecomastia, intel-
lectual disability, short stature, and obesity syndrome. Am J Med
Genet A. 161 A:2294–2299.
Howard J, Brown JL, Bush M, Wildt DE. 1990. Teratospermic and nor-
mospermic domestic cats: ejaculate traits, pituitary—gonadal hor-
mones, and improvement of spermatozoal motility and
morphology after swim—up processing. JAndrol.11:204–215.
Jockusch H, Holland A, Staunton L, Schmitt-John T, Heimann P, Dowling
P, Ohlendieck K. 2014. Pathoproteomics of testicular tissue deficient
in the GARP component VPS54: the wobbler mouse model of
globozoospermia. Proteomics 14:839–852.
Johnson G. 1991. The bengal cat. Greenwell Springs (LA): Gogees Cattery.
Julio-Pieper M, O’Connor RM, Dinan TG, Cryan JF. 2013. Regulation of
the brain-gut axis by group III metabotropic glutamate receptors.
Eur J Pharmacol. 698:19–30.
Kamikubo Y, Takaori-Kondo A, Uchiyama T, Hori T. 2003. Inhibition of
cell growth by conditional expression of kpm, a human homologue
of Drosophila warts/lats tumor suppressor. JBiolChem.278:17609–
17614.
Kang HM, Sul JH, Service SK, Zaitlen NA, Kong S-Y, Freimer NB, Sabatti
C, Eskin E. 2010. Variance component model to account for sample
structure in genome-wide association studies. Nat Genet. 42:348–
354.
Kang HM, Zaitlen NA, Wade CM, Kirby A, Heckerman D, Daly MJ, Eskin
E. 2008. Efficient control of population structure in model organism
association mapping. Genetics 178:1709–1723.
King M-C, Wilson AC. 1975. Evolution at two levels in humans and
chimpanzees. Science 188: 107-116.
Kutschera VE, Bidon T, Hailer F, Rodi JL, Fain SR, Janke A. 2014. Bears in a
forest of gene trees: phylogenetic inference is complicated by in-
complete lineage sorting and gene flow. MolBiolEvol.31: 2004–
2017.
L’H^
ote D, Laissue P, Serres C, Montagutelli X, Veitia RA, Vaiman D. 2010.
Interspecific resources: a major tool for quantitative trait locus clon-
ing and speciation research. Bioessays 32:132–142.
Lee A, Anderson AR, Barnett AC, Chan A, Pow DV. 2011. Expression of
multiple glutamate transporter splice variants in the rodent testis.
Asian J Androl. 13:254–265.
Li G, Davis BW, Raudsepp T, Wilkerson AJP, Mason VC, Ferguson-Smith
M, O’Brien PC, Waters PD, Murphy WJ. 2013. Comparative analysis
of mammalian Y chromosomes illuminates ancestral structure and
lineage-specific evolution. Genome Res. 23:1486–1495.
Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G,
Abecasis G, Durbin R. 2009. The Sequence Alignment/Map format
and SAMtools. Bioinformatics 25:2078–2079.
Liao Y, Smyth GK, Shi W. 2014. featureCounts: an efficient general pur-
pose program for assigning sequence reads to genomic features.
Bioinformatics 30:923–930.
Lin M, Lee YH, Xu W, Baker MA, Aitken RJ. 2006. Ontogeny of tyrosine
phosphorylation-signaling pathways during spermatogenesis and
epididymal maturation in the mouse. Biol Reprod. 75:588–597.
Machol
an M, Munclinger P, Sugerkov
aM,Dufkov

aP,B

ımov
aB,
Boz
ıkov
aE,ZimaJ,Pi

alek J. 2007. Genetic analysis of autosomal
and X-linked markers across a mouse hybrid zone. Evolution
61:746–771.
Maekawa M, Ito C, Toyama Y, Suzuki-Toyota F, Fujita E, Momoi T,
Toshimori K. 2011. Localisation of RA175 (Cadm1), a cell adhesion
molecule of the immunoglobulin superfamily, in the mouse testis,
and analysis of male infertility in the RA175-deficient mouse.
Andrologia 43:180–188.
Marciniak M, Chruscicka B, Lech T, Burnat G, Pilc A. 2014. Expression of
group III metabotropic glutamate receptors in the reproductive
system of male mice. Reprod Fertil Devdx. doi.org/10.1071/RD14132.
Maside XR, Naveira HF. 1996. A polygenic basis of hybrid sterility may
give rise to spurious localizations of major sterility factors. Heredity
77:488–492.
Masly JP, Presgraves DC. 2007. High-resolution genome-wide dissection
of the two rules of speciation in Drosophila.PLoS Biol. 5:e243.
McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A,
Garimella K, Altshuler D, Gabriel S, Daly M, et al. 2010. The Genome
Analysis Toolkit: a MapReduce framework for analyzing next-gen-
eration DNA sequencing data. Genome Res. 20:1297–1303.
McPherson JP, Tamblyn L, Elia A, Migon E, Shehabeldin A, Matysiak-
Zablocki E, Lemmers B, Salmena L, Hakem A, Fish J, et al. 2004. Lats2/
Kpm is required for embryonic development, proliferation control
and genomic integrity. EMBO J. 23:3677–3688.
McReynolds S, Dzieciatkowska M, Stevens J, Hansen KC, Schoolcraft WB,
Katz-Jaffe MG. 2014. Toward the identification of a subset of unex-
plained infertility: a sperm proteomic approach. Fertil Steril.
102:692–699.
Mihola O, Trachtulec Z, Vlcek C, Schimenti JC, Forejt J. 2009. A mouse
speciation gene encodes a meiotic histone H3 methyltransferase.
Science 323:373–375.
Mital P, Hinton BT, Dufour JM. 2011. The blood-testis and blood-epi-
didymis barriers are more than just their tight junctions. Biol Reprod.
84:851–858.
Montague M, Li G, Gandolfi B, Khan R, Aken B, Searle S, Minx P, Hillier L,
Kolboldt D, Davis BW, et al. 2014. Comparative analysis of the
domestic cat genome reveals genetic signatures underlying feline
biology and domestication. Proc Natl Acad Sci U S A. 111:17230–
17235.
Moyle LC, Muir CD, Han M V, Hahn MW. 2010. The contribution of
gene movement to the “two rules of speciation”. Evolution 64:1541–
1557.
Mruk DD, Cheng CY. 2004. Sertoli-Sertoli and Sertoli-germ cell interac-
tions and their significance in germ cell movement in the seminif-
erous epithelium during spermatogenesis. Endocr Rev. 25:747–806.
Mueller JL, Mahadevaiah SK, Park PJ, Warburton PE, Page DC, Turner
JMA. 2008. The mouse X chromosome is enriched for multicopy
testis genes showing postmeiotic expression. Nat Genet. 40:794–799.
Mueller JL, Skaletsky H, Brown LG, Zaghlul S, Rock S, Graves T, Auger K,
Warren WC, Wilson RK, Page DC. 2013. Independent specialization
of the human and mouse X chromosomes for the male germ line.
Nat Genet. 45:1083–1087.
Mueller S, Rosenquist TA, Takai Y, Bronson RA, Wimmer E. 2003. Loss of
nectin-2 at Sertoli-spermatid junctions leads to male infertility and
correlates with severe spermatozoan head and midpiece malforma-
tion, impaired binding to the zona pellucida, and oocyte penetra-
tion. Biol Reprod. 69:1330–1340.
Muller HJ. 1942. Isolating mechanisms, evolution, and temperature. Biol
Symp. 6:71–125.
Murphy WJ, Pearks Wilkerson AJ, Raudsepp T, Agarwala R, Sch€
affer AA,
Stanyon R, Chowdhary BP. 2006. Novel gene acquisition on carni-
vore Y chromosomes. PLoS Genet. 2:e43.
12
Davis et al. .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

Neibergs HL, Seabury CM, Wojtowicz AJ, Wang Z, Scraggs E, Kiser J,
Neupane M, Womack JE, Van Eenennaam A, Hagevoort GR, et al.
2014. Susceptibility loci revealed for bovine respiratory disease com-
plex in pre-weaned holstein calves. BMC Genomics 15:1164.
Paiardi C, Pasini ME, Gioria M, Berruti G. 2011. Failure of acrosome
formation and globozoospermia in the wobbler mouse, a Vps54
spontaneous recessive mutant. Spermatogenesis 1:52–62.
Payseur BA, Hoekstra HE. 2005. Signatures of reproductive isolation in
patterns of single nucleotide diversity across inbred strains of mice.
Genetics 171:1905–1916.
P
erez-Victoria FJ, Bonifacino JS. 2009. Dual roles of the mammalian
GARP complex in tethering and SNARE complex assembly at the
trans-golgi network. MolCellBiol.29:5251–5263.
Phadnis N, Orr HA. 2009. A single gene causes both male sterility
and segregation distortion in Drosophila hybrids. Science
323:376–379.
Phifer-Rixey M, Bomhoff M, Nachman MW. 2014. Genome-wide pat-
terns of differentiation among house mouse subspecies. Genetics
198:283–297.
Pukazhenthi BS, Neubauer K, Jewgenow K, Howard J, Wildt DE. 2006.
The impact and potential etiology of teratospermia in the domestic
cat and its wild relatives. Theriogenology 66:112–121.
Robinson MD, McCarthy DJ, Smyth GK. 2010. edgeR: a Bioconductor
package for differential expression analysis of digital gene expression
data. Bioinformatics 26:139–140.
Schimenti KJ, Feuer SK, Griffin LB, Graham NR, Bovet CA, Hartford S,
Pendola J, Lessard C, Schimenti JC, Ward JO. 2013. AKAP9 is essential
for spermatogenesis and sertoli cell maturation in mice. Genetics
194:447–457.
Segura V, Vilhj
almsson BJ, Platt A, Korte A, Seren €
U, Long Q, Nordborg
M. 2012. An efficient multi-locus mixed-model approach for
genome-wide association studies in structured populations. Nat
Genet. 44:825–830.
Sehrawat S, Ernandez T, Cullere X, Takahashi M, Ono Y, Komarova Y,
Mayadas TN. 2011. AKAP9 regulation of microtubule dynamics
promotes Epac1-induced endothelial barrier properties. Blood
117:708–718.
Skaletsky H, Kuroda-Kawaguchi T, Minx PJ, Cordum HS, Hillier L, Brown
LG, Repping S, Pyntikova T, Ali J, Bieri T, et al. 2003. The male-specific
region of the human Y chromosome is a mosaic of discrete se-
quence classes. Nature 423:825–837.
SohYWS,AlfoldiJ,PyntikovaT,BrownLG,GravesT,MinxPJ,FultonRS,
Kremitzki C, Koutseva N, Mueller JL, et al. 2014. Sequencing the
mouse Y chromosome reveals convergent gene acquisition and
amplification on both sex chromosomes. Cell 159:800-813.
Storchov
a R, Gregorov
a S, Buckiov
aD,Kyselov

aV,DivinaP,ForejtJ.
2004. Genetic analysis of X-linked hybrid sterility in the house
mouse. Mamm Genome. 15: 515-524.
Sun S, Ting C-T, Wu C-I. 2004. The normal function of a speciation gene,
Odysseus, and its hybrid sterility effect. Science 305:81–83.
Svingen T, Jørgensen A, Rajpert-De Meyts E. 2014. Validation of endog-
enous normalizing genes for expression analyses in adult human
testis and germ cell neoplasms. Mol Hum Reprod. 20:709–718.
Toews DPL, Brelsford A. 2012. The biogeography of mitochondrial and
nuclear discordance in animals. Mol Ecol. 21:3907–3930.
Tomes CN, Michaut M, De Blas G, Visconti P, Matti U, Mayorga LS. 2002.
SNARE complex assembly is required for human sperm acrosome
reaction. Dev Biol. 243:326–338.
Torgerson DG, Singh RS. 2003. Sex-linked mammalian sperm proteins
evolve faster than autosomal ones. MolBiolEvol.20:1705–1709.
Tour
e A, Szot M, Mahadevaiah SK, Rattigan A, Ojarikre OA, Burgoyne
PS. 2004. A new deletion of the mouse Y chromosome long arm
associated with the loss of Ssty expression, abnormal sperm devel-
opment and sterility. Genetics 166:901–912.
Turelli M, Orr HA. 1995. The dominance theory of Haldane’s rule.
Genetics 140:389–402.
Turner LM, White MA, Tautz D, Payseur BA. 2014. Genomic networks of
hybrid sterility. PLoS Genet. 10:e1004162.
Van der Auwera GA, Carneiro MO, Hartl C, Poplin R, Del Angel G, Levy-
Moonshine A, Jordan T, Shakir K, Roazen D, Thibault J, et al. 2013.
Curr Protoc Bioinform. 11:11.10.1–11.10.33.
Vassilieva E V, Nusrat A. 2008. Vesicular trafficking: molecular tools and
targets. Methods Mol Biol. 440:3–14.
Vogl AW, Pfeiffer DC, Mulholland D, Kimel G, Guttman J. 2000. Unique
and multifunctional adhesion junctions in the testis. Ectoplasmic
specializations. Arch Histol Cytol. 63:1–15.
Wakayama T, Iseki S. 2009. Role of the spermatogenic-Sertoli cell inter-
action through cell adhesion molecule-1 (CADM1) in spermatogen-
esis. Anat Sci Int. 84:112–121.
Wakayama T, Sai Y, Ito A, Kato Y, Kurobo M, Murakami Y, Nakashima E,
Tsuji A, Kitamura Y, Iseki S. 2007. Heterophilic binding of the adhe-
sion molecules poliovirus receptor and immunoglobulin superfamily
4 A in the interaction between mouse spermatogenic and Sertoli
cells. Biol Reprod. 76:1081–1090.
Webster KE, O’Bryan MK, Fletcher S, Crewther PE, Aapola U, Craig J,
Harrison DK, Aung H, Phutikanit N, Lyle R, et al. 2005. Meiotic and
epigenetic defects in Dnmt3L-knockout mouse spermatogenesis.
Proc Natl Acad Sci U S A. 102:4068–4073.
Wellcome Trust Case Control Consortium. 2007. Genome-wide associ-
ation study of 14,000 cases of seven common diseases and 3,000
shared controls. Nature 447:661–678.
Wheeler JC. 2011. Savannah cats. Edina (MN): ABDO Publishing
Company.
White MA, Stubbings M, Dumont BL, Payseur BA. 2012. Genetics and
evolution of hybrid male sterility in house mice. Genetics 191:917–
934.
Wu CI, Davis AW. 1993. Evolution of postmating reproductive isolation:
the composite nature of Haldane’s rule and its genetic bases. Am
Nat. 142:187–212.
Yabuta N, Mukai S, Okada N, Aylon Y, Nojima H. 2014. The tumor
suppressor Lats2 is pivotal in Aurora A and Aurora B signaling
during mitosis. Cell Cycle 10:2724–2736.
Zamudio NM, Scott HS, Wolski K, Lo C-Y, Law C, Leong D, Kinkel SA,
Chong S, Jolley D, Smyth GK, et al. 2011. DNMT3L is a regulator of X
chromosome compaction and post-meiotic gene transcription.
PLoS One 6:e18276..
13
Mechanisms Underlying Mammalian Hybrid Sterility .doi:10.1093/molbev/msv124 MBE
at Medical Sciences Library on June 27, 2015http://mbe.oxfordjournals.org/Downloaded from

!"#$%&'
(%)%##%*'
+,$-."'(/0'(%12&"'(,3"4'56'(7".,&"'%#8'+".9&"':;<.,840''!"#$%&'()*+,-"./0&*$1,*02)3,'#*%4*#$,'"5,*(0+*4,'/5,*
163'"+#*2/5"7,+*"0*$1,*89:;<**=,+*+,0%$,#*#$,'"5,*(0+*&',,0*+,0%$,#*4,'/5,>*?"$1*.%20$#*(5%0&*$1,*!@(A"#*(0+*
,(.1*3(.B.'%##*&,0,'(/%0*(5%0&*$1,*"@(A"#<**;,C,0*#$,'"5,*D,0&(5#*?,',*%4*20B0%?0*&,0,'(/%0<*
EF*
EG*
HF*
HG*
IF*
IG*
F*
G*
EF*
EG*
HF*
HG*
IF*
IG*
F*
G*

G*
IG*
HG*
EG*
JG*
FG*
KG*
LG*
;:MI* ;:MH* ;:ME* ;:MJ* ;:MF* DNOI* DNOH* DNOE*
G*
HG*
JG*
KG*
PG*
IGG*
IHG*
IJG*
;:MI* ;:MH* ;:ME* ;:MJ* ;:MF* DNOI* DNOH* DNOE*
G*
F*
IG*
IF*
HG*
HF*
EG*
EF*
JG*
;:MI* ;:MH* ;:ME* ;:MJ* ;:MF* DNOI* DNOH* DNOE*
G*
IG*
HG*
EG*
JG*
FG*
KG*
LG*
PG*
;:MI* ;:MH* ;:ME* ;:MJ* ;:MF* DNOI* DNOH* DNOE*
Q,'/5,*!63'"+#R*8,0%$6-,#*
+,$-."'(=0''8,0%$6-,*(0+*$%$(5*(55,5,*.%20$#*4%'*,(.1*#"&0"S.(0$*89:;*)('B,'*T;:MI@F>*DNOI@EU<**Q%'*,C,'6*.5(##*
%4*163'"+>*$1,*02)3,'#*%4*1%)%76&%2#*(0+*1,$,'%76&%2#*&,0%$6-,#*(',*#1%?0<**V5(##"S.(/%0*"#*3(#,+*%0*
-2$(/C,*W?"5+X*(0+*W+%),#/.X*(55,5,#*(#.,'$("0,+*4'%)*&,0%$6-"0&*(*.%1%'$*%4*,(.1*?"5+*#-,.",#<**;$,'"5,*
;(C(00(1#*1(C,*,0'".1),0$*%4*?"5+*(55,5,#*4%'*,C,'6*)('B,'*,A.,-$*;:MF<**Y1"#*)('B,'*1(#*(0*(320+(0.,*%4*
1%)%76&%2#@?"5+*&,0%$6-,#*"0*4,'/5,*C,'#2#*#$,'"5,*;(C(00(1#>*+,#-"$,*.%)-('(35,*02)3,'#*%4*-2$(/C,*?"5+Z
+%),#/.*(55,5,#<**;$,'"5,*D,0&(5#*1(C,*(0*,0'".1),0$*%4*?"5+*(55,5,#*4%'*DNOI*(0+*DNOH>*32$*(0*,0'".1),0$*%4*
+%),#/.*(55,5,#*4%'*DNOE>*?1".1*(',*(3#,0$*"0*4,'/5,*163'"+#<'
;$,'"5,*!63'"+#R*8,0%$6-,#* ;$,'"5,*!63'"+#R*:55,5,#*
Q,'/5,*!63'"+#R*:55,5,#*
[%),#/.*
9"5+*
[%),#/.*
9"5+*
!%)%76&%2#*[%),#/.*
!,$,'%76&%2#*
*
!%)%76&%2#*9"5+*
!%)%76&%2#*[%),#/.*
!,$,'%76&%2#*
*
!%)%76&%2#*9"5+*

+U*F$1*8,0,'(/%0*
D(.B.'%##*
;(C(00(1*
+,$-."'(>0':,4752%7*5&5$;'56'(7".,&"'?35542".1,@'!%@A@.544':;<.,840**;,)"0"4,'%2#*$2325,#*#1%?*(*'(0&,*%4*(3,''(0$*
-1,0%$6-,#<*Y?%*%4*$1,*)%#$*.%))%0*"0*3%$1*163'"+*3',,+#*(',*16-%#-,')($%&,0".*$2325,#*?"$1*-%%'56*+,S0,+*52),0*T(>*
+@4U>*(0+*$2325,#*?"$1*('',#$,+*),"%/.*-'%&',##"%0>*-',+%)"0(0$56*($*$1,*-(.16$,0,*#$(&,*T3>*.>*&>*1U>*$1($*%\,0*.%0$("0*
5('&,*C(.2%5,#<*
.U*J$1*8,0,'(/%0*
D(.B.'%##*
;(C(00(1*
3U*E'+*8,0,'(/%0*
D(.B.'%##*
;(C(00(1*
(U*E'+*8,0,'(/%0*
D(.B.'%##*
;(C(00(1*

,U*I#$*8,0,'(/%0*
D(.B.'%##*
D,0&(5*
4U*J$1*8,0,'(/%0*
D(.B.'%##*
D,0&(5*
1U*F$1*8,0,'(/%0*
D(.B.'%##*
D,0&(5*
&U*F$1*8,0,'(/%0*
D(.B.'%##*
D,0&(5*
+,$-."'(>'B@5#9#-"8C0':,4752%7*5&5$;'56'(7".,&"':;<.,840***

+,$-."'(D0'E512%.,45#'56'7"494'$"#"'1,4."$-&%95#'<"7F""#'47".,&"'%#8'6".9&"'$.5-240'M,00*+"(&'()#*+"#-5(6"0&*$1,*
02)3,'*%4*20"]2,*(0+*%C,'5(--"0&*)"#',&25($,+*&,0,#*TQ[=*G<GIU*"+,0/S,+*"0*+"^,',0$*163'"+@163'"+*%'*-(',0$@
163'"+*.%)-('"#%0#<**Y1,*5,\*.%52)0*+,-".$#*(55*&,0,#>*(0+*$1,*'"&1$*.%52)0*+,-".$#*$1,*_*.1'%)%#%),<**Y1,*3%`%)*
M,00*+"(&'()*+,-".$#*$1,*02)3,'*%4*&,0,#*)"#',&25($,+*"0*#$,'"5,*163'"+#*T(0+*$1,"'*%C,'5(-*3,$?,,0*3',,+#U*
.%)-(',+*$%*(55*$1',,*&'%2-#*%4*4,'/5,*(0")(5#*
;$,'"5,*!63'"+*Z*[%),#/.>*Q,'/5,*!63'"+>*(0+*9"5+*;-,.",#*

+,$-."'(G0'E512%.,45#'56'HI?J("K'I5.1%&,3%95#'L"7*5840**M,00*+"(&'()#*+"#-5(6"0&*%C,'5(-*"0*&,0,*"+,0/$6*
3,$?,,0*+"^,',0/(5*,A-',##"%0*(0(56#,#*2#"0&*$1',,*0%')(5"7(/%0*),$1%+#R*Yaa>*=bN>*(0+*cd<*
!"#$%&#'!()(**(+','-./#012'3("'
''''
4567'
89'
:;<'
=>>'
?@' A'4A5'
6' ?B@'
6C'
!"#$%&#'!()(**(+','D#$1&#'!()(**(+'
''''
4BB@'
89'
:;<'
=>>'
B' ?E'A@B'
57' 7?5'
6'
!"#$%&#'F#*G(&','-./#012'3("'
''''
46BC'
89'
:;<'
=>>'
6CC' E'4C4'
E' @6B'
677'
!"#$%&#'F#*G(&','D#$1&#'F#*G(&'
''''
4@6E'
89'
:;<'
=>>'
B?C' E'???'
E' 77?'
6?A'

M'
GM'
=G'
?/'
+,$-."'(N0'O&&-1,#%'+"&,#"'N>P'?..%;'(IQ'L%.A".'R"#4,7;0'';Oe*+,0#"$6*(.'%##*(55*.1'%)%#%),#*?"$1"0*$1,*
+%),#/.*.($*&,0%),*(##,)356*C,'#"%0*K<H*T4,5V($FU<*Y1,*!@(A"#*+,0%$,#*$1,*02)3,'*%4*;Oe#*-,'*I@a3*?"0+%?>*
.(5.25($,+*(.'%##*,(.1*',#-,./C,*.1'%)%#%),*T.1'cO*0%$*#1%?0U<**Y1,*_*.1'%)%#%),*#1%?#*(*)('B,+*
',+2./%0*"0*;Oe*+,0#"$6*.%)-(',+*$%*(55*(2$%#%),#<*
?='
M'
GM'
=G'
?>'
M'
GM'
=G'
!/'
M'
GM'
=G'
!='
M'
GM'
=G'
!>'
M'
GM'
=G'
!D'
M'
GM'
=G'
E/'
M'
GM'
=G'
E='
M'
GM'
=G'
M'
GM'
=G'
R/'
R>'
M'
GM'
=G'
R='
M'
GM'
=G'
RD'
M'
GM'
=G'
S/'
M'
GM'
=G'
S>'
M'
GM'
=G'
S='
M'
GM'
=G'
M'
GM'
=G'
+/'
M'
GM'
=G'
+='
M'
GM'
=G'
T'

+,$-."'(U0'!"#7'L,82,"@"'?<#5.1%&,9"4',#'(7".,&"':;<.,840**Nf(.25($,*4'%)*(*$,'($%#-,')".*D,0&(5*163'"+*,A()"0,+*
2#"0&*5"&1$*)".'%#.%-6<***:*1"&1*-'%-%'/%0*TgKGhU*%4*#-,')*+,)%0#$'($,+*-%%'*4%'?('+*)%/5"$6>*+2,*"0*-('$*$%*3,0$@
)"+-",.,*+,4,.$#*T',+*."'.5,#U<**
'

+,$-."'(V0'+/'!"#$%&'W"494':,4752%7*5&5$;0**!iN*#$("0,+*$,#/#*#,./%0*%4*(*#$,'"5,*QI*&,0,'(/%0*163'"+*
%^#-'"0&*%4*(*)(5,*:#"(0*5,%-('+*.($*(0+*(*4,)(5,*+%),#/.*.($<**;,)"0"4,'%2#*$2325,#*.%0$("0*-%%'56*+,S0,+*
52),0*(0+*%056*;,'$%5"*.,55#<*

+,$-."'(X0'YZ?('H"4-&74'!%4"8'5#'%#'?88,9)"'L58"&0**a(01(`(0*-5%$#*+"#-5(6"0&*#"&0"S.(0.,*%4*89:;*
',#25$#*4%'*$1,*:U*;(C(00(1*.%1%'$>*(0+*DU*D,0&(5*.%1%'$>*-,'4%'),+*20+,'*(0*(++"/C,*)%+,5*%4*"01,'"$(0.,<*
*
?'
!'

+,$-."'(/M0'QJQ'2&574'65.'SLL?T'%#%&;4"4'

W%<&"'(/0''?##57%7"8')%.,%#74'<"7F""#'42"@,"40''M('"(0$#*+,$,.$,+*?"$1"0*=O:#,]*.%+"0&*',&"%0#*%4*
#"&0"S.(0$56*(##%."($,+*89:;*&,0,#<*
'
(?[?II?:('
Y"#"'
E*.0'
Q54,95#'
H"6'
?&7'
(-<497-95#'W;2"'
??'(-<470'
E585#'(-<470'
R51"49@' +".9&"':;<.,84' (7".,&"':;<.,84'
Z,&8'
#$%&'
O%*)"##,0#,*)2$(/%0#*
'' '' '' '' '' '' '' '' '' ''
()*%+'
[I*
IEIjLKIP*
:* 8*
)"##,0#,kC('"(0$*
Q*l*b*
Y`*l*V`*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjEFGKI*
8* 8:*
4'(),#1"\kC('"(0$*Z*4,($2',k,5%0&(/%0*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjGjEEL*
:8* :*
4'(),#1"\kC('"(0$*Z*#-5".,kC('"(0$*Z*$'20.(/%0*
'' '' '' '' '' '' T' T' T' T'
),)-.'
:H*
jFPKFKEj*
Y* V*
)"##,0#,kC('"(0$*
b*l*e*
.Y&*l*.V&*
'' '' '' '' '' '' '' T' '' ''
),)-.'
:H*
jFPKKILP*
Y* V*
)"##,0#,kC('"(0$*
Q*l*b*
Y$.*l*V$.*
'' '' '' '' '' '' '' T' '' ''
),)-.'
:H*
jFPLLJjK*
V* :*
)"##,0#,kC('"(0$*
e*l*!*
.V$*l*.:$*
'' '' '' '' '' '' T' '' '' ''
),)-.'
:H*
jFPPHPGH*
:* Y*
)"##,0#,kC('"(0$*
b*l*Q*
`:*l*`Y*
T' T' '' '' '' T' T' T' '' ''
),)-.'
:H*
jFPPHPEJ*
:* 8*
)"##,0#,kC('"(0$*
d*l*=*
.:(*l*.8(*
T' T' '' '' '' '' T' T' T' ''
),)-.'
:H*
jFPjKIJP*
8* V*
)"##,0#,kC('"(0$*
8*l*=*
8&$*l*V&$*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFPjKIKJ*
V* Y*
)"##,0#,kC('"(0$*
Y*l*a*
(V&*l*(Y&*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFPjKHJP*
:* 8*
)"##,0#,kC('"(0$*
N*l*8*
&:(*l*&8(*
'' '' '' '' '' '' '' '' T' ''
),)-.'
:H*
jFPjKHFG*
:* 8*
)"##,0#,kC('"(0$*
O*l*[*
:($*l*8($*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFPjKHLF*
V* :*
)"##,0#,kC('"(0$*
Y*l*O*
(V.*l*(:.*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFPjKHLL*
8* :*
)"##,0#,kC('"(0$*
[*l*O*
8($*l*:($*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFPjKEGL*
8* :*
)"##,0#,kC('"(0$*
N*l*m*
8((*l*:((*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFPjKEIJ*
:* 8*
)"##,0#,kC('"(0$*
!*l*=*
.:$*l*.8$*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjGjFIE*
:* Y*
)"##,0#,kC('"(0$*
;*l*V*
:&$*l*Y&$*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJIKJE*
Y* V*
)"##,0#,kC('"(0$*
n*l*Y*
(Y.*l*(V.*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJHLPL*
8* V*
)"##,0#,kC('"(0$*
M*l*b*
8$&*l*V$&*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJHPJj*
8* :*
)"##,0#,kC('"(0$*
a*l*n*
($8*l*($:*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJEFII*
8* :*
)"##,0#,kC('"(0$*
;*l*O*
(8.*l*(:.*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJEFFH*
:* V*
)"##,0#,kC('"(0$*
m*l*d*
:((*l*V((*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJEFKK*
Y* 8*
)"##,0#,kC('"(0$*
n*l*a*
($Y*l*($8*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJFLKE*
V* 8*
)"##,0#,kC('"(0$*
d*l*N*
V((*l*8((*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjJPJJG*
V* 8*
)"##,0#,kC('"(0$*
=*l*8*
V&$*l*8&$*
'' '' '' '' '' '' T' T' T' T'
),)-.'
:H*
jFjFJjPG*
Y* V*
)"##,0#,kC('"(0$*
M*l*:*
&Y$*l*&V$*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjFLjPj*
8* V*
)"##,0#,kC('"(0$*
N*l*d*
8(&*l*V(&*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjFjKjF*
V* :*
)"##,0#,kC('"(0$*
e*l*Y*
V.(*l*:.(*
'' '' '' '' '' '' '' '' '' T'
),)-.'
:H*
jFjGjEPK*
VY8*
V*
#-5".,k+%0%'kC('"(0$*Z*4,($2',k$'20.(/%0*
'' '' '' '' '' '' '' '' '' T'
/0()123'
O%*)"##,0#,*)2$(/%0#*
'' '' '' '' '' '' '' '' '' ''
4)502'
O%*)"##,0#,*)2$(/%0#*
'' '' '' '' '' '' '' '' '' ''
!SIY?\('
Y"#"'
E*.0'
Q54,95#'
H"6'
?&7'
(-<497-95#'W;2"'
??'(-<470'
E585#'(-<470'
R51"49@' +".9&"':;<.,84' (7".,&"':;<.,84'
Z,&8'
*1%564'
O%*)"##,0#,*)2$(/%0#*
**
**
**
**
**
**
**
**
**
**
7-036'
O%*)"##,0#,*)2$(/%0#*
**
**
**
**
**
**
**
**
**
**
01)-23'
O%*)"##,0#,*)2$(/%0#*
**
**
**
**
**
**
**
**
**
**

W%<&"'(=0*V1'%)%#%),@#-,."S.*)"#',&25(/%0*%4*$,#/#*$'(0#.'"-$#*"0*#$,'"5,*C,'#2#*4,'/5,*163'"+#*36*.1"@#]2(',*$,#$>*
+4oI<*
''
E*.515451"'
?/' ?=' ?>' !/' !=' !>' !D' E/' E=' R/' R=' R>' RD' S/' S=' S>' +/' +=' T'
!"#$%&'
c-*
HGJ*
HPH*
IPP*
HGK*
IFG*
HIJ*
HHH*
EGP*
IKK*
HIK*
IHE*
IEK*
IJE*
HIG*
ILE*
IHI*
IEF*
LI*
HFF*
[%?0*
HEj*
EIj*
HJE*
HGH*
HGJ*
HFE*
HEP*
EEL*
IFj*
HHP*
IHL*
IEL*
ILJ*
HHG*
IKF*
IHF*
IHL*
LG*
EF*
;$,'"5,*!63'"+*Z*[%),#/.*-8C(52,*
G<GjKEE*
G<IEIHE*
G<GGPGL*
G<PJEGH*
G<GGJI*
G<GLIIH*
G<JFFKL*
G<HFEFI*
G<KjLP*
G<FKjGH*
G<PGGHP*
G<jFILJ*
G<GPIKK*
G<KHjKE*
G<KKEJK*
G<LjPL*
G<KHIIJ*
G<jEHPj*
E<FN@EP*
c-*
HGG*
HPG*
HGI*
HGG*
IFF*
HIH*
HEK*
EHK*
ILE*
HHI*
IHF*
IJJ*
IJj*
HIF*
IPH*
IEJ*
IJH*
PI*
HKI*
[%?0*
HJj*
EEP*
HJL*
HIH*
HHH*
HFP*
HFF*
EFJ*
ILj*
HEI*
IEE*
IEL*
ILI*
HHJ*
IPI*
IHK*
IEJ*
LJ*
JF*
;$,'"5,*Z*Q,'/5,*!63'"+**-@C(52,*
G<GHGLF*
G<GIjKJ*
G<GHjLK*
G<FFJEj*
G<GGGFK*
G<GEEPF*
G<EjIIj*
G<HPHjE*
G<LJjIH*
G<KEPI*
G<KIPJJ*
G<KLKHF*
G<HIPLK*
G<KKLFE*
G<jFPIJ*
G<KIjP*
G<KEGIE*
G<FLEjJ*
FN@EF*
''
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
(%)%##%*'
c-*
HHP*
EHF*
HEG*
HHP*
IjE*
HEI*
HJF*
EKL*
ILL*
HJG*
IEJ*
IJP*
IFj*
HJH*
Ijj*
IEH*
IJJ*
PE*
HFJ*
[%?0*
HPF*
EEj*
HKG*
HHI*
HIH*
HLK*
HLF*
EPE*
IjP*
HJH*
IHP*
IFj*
ILj*
HIL*
IPI*
IEj*
IJH*
Lj*
KG*
;$,'"5,*!63'"+*Z*[%),#/.*-@C(52,*
G<GIIPF*
G<FPKjH*
G<ILFEE*
G<LJIIJ*
G<EJFII*
G<GJFKK*
G<IPPEI*
G<FFjGK*
G<HLPIL*
G<jHLJH*
G<LIGPL*
G<FEGIE*
G<HLKKK*
G<HJEHF*
G<EFFPI*
G<KLGKP*
G<jGFPK*
G<LFEEH*
K<PN@HP*
c-*
HGI*
HLL*
IPI*
IjI*
IFj*
Ijj*
HIK*
EIF*
IFP*
IjJ*
IGj*
IIj*
IEF*
IPL*
IKH*
IGL*
IIK*
LI*
HHF*
[%?0*
HFF*
EEJ*
HFE*
HIJ*
HGI*
HKI*
HKL*
EKE*
IjH*
HEK*
IHG*
IFI*
ILK*
HHI*
ILE*
IEL*
IEK*
PG*
KG*
;$,'"5,*Z*Q,'/5,*!63'"+**-@C(52,*
G<GIIJF*
G<GHIII*
G<GGGFF*
G<HFEGj*
G<GHKPK*
G<GGEPJ*
G<GHGEI*
G<GKFHL*
G<GKjIK*
G<GJHPH*
G<JKLHj*
G<GFIJP*
G<GHGGP*
G<GjHEE*
G<FJLPJ*
G<GFJLj*
G<HGLLI*
G<JKEjH*
I<FN@HH*

W%<&"'(>0'TJ&,#A"8'L,4."$-&%95#'!"7F""#'E%7'%#8'L5-4"0''
V%)-('"#%0*%4*)"#',&25($,+*_@5"0B,+*&,0,#*#1(',+*3,$?,,0*
"0$'(#-,.",#*%9:*163'"+#*T8%%+*,$*(5<*HGIGU*(0+*"0$,'#-,.",#*
4,5"+*163'"+#<*
8,0,*
a2#* D_[*
D_[kQ[=*
;_[*
;_[kQ[=*
(*,+;'
2-* 2-*
G<GGGLLJ*
2-*
I<PPN@GF*
(<4=>?'
2-* 2-*
J<jPN@GK*
2-*
G<GGJIE*
*@1456'
2-* 2-*
F<GJN@GP*
2-*
J<jLN@IE*
A0*+B>+C'
2-* 2-*
F<EHN@IG*
2-*
I<HLN@IE*
4)%-2'
2-* 2-*
F<KFN@IF*
2-*
L<EEN@GK*
%)#D*2'
2-* 2-*
E<LGN@HJ*
2-*
H<KGN@IP*
%5%$+'
2-* 2-*
G<GGGHFP*
2-*
G<GGEEHE*
1*<E>++'
2-* 2-*
H<FFN@GP*
2-*
H<jJN@IE*
-#$%(+'
2-* 2-*
I<LFN@HI*
2-*
j<JGN@IJ*
-406'
2-* 2-*
I<FKN@IG*
2-*
P<FFN@GP*
-0%*+C'
2-* 2-*
I<jIN@Gj*
2-*
I<KLN@Gj*
05)#2'
2-* 2-*
P<FHN@IE*
2-*
K<GIN@Gj*
0@)-+'
2-* 2-*
G<GGIFJF*
2-*
I<jEN@GF*
50-@42'
2-* 2-*
I<KLN@IF*
2-*
I<EKN@GL*
50$2'
2-* 2-*
P<GFN@GK*
2-*
E<GPN@GP*
F*$+6'
2-* 2-*
E<IGN@II*
2-*
P<EFN@IJ*
50-)1;'
2-*
+%?0*
I<GKN@Ij*
+%?0*
E<KEN@IH*
E%$+1>'
2-*
+%?0*
G<GGJIIF*
+%?0*
J<PFN@EE*
(@4(+'
2-*
+%?0*
j<KJN@EF*
+%?0*
E<GLN@HF*
-$*?='
2-*
+%?0*
G<GGGIjJ*
+%?0*
G<GGGIKK*
$G>(+'
2-*
+%?0*
H<PIN@FK*
+%?0*
E<PFN@FF*
0(%42'
2-*
+%?0*
G<GGJLPK*
+%?0*
E<HJN@IL*
-G#)'
2-* 2-*
G<GGIGKP*
2-*
G<JLEPGK*
0%-?'
2-* 2-*
F<FHN@GK*
2-*
G<EFFKHL*
0D-5='
)"+*
+%?0*
G<JJjGKH*
+%?0*
G<GGKFE*
DEA(2'
2-*
+%?0*
G<GKKGJL*
+%?0*
H<jLN@IF*
5D?++'
2-*
+%?0*
G<IHIIFK*
+%?0*
E<LIN@IE*
5,54+'
2-*
+%?0*
G<PIGGI*
+%?0*
I<FFN@IG*
,4A4+3'
)"+* 2-*
I<IJN@GF*
2-*
G<GGIFLF*
D>-'
)"+* 2-*
J<GFN@II*
2-*
G<GGHKjj*
A*A*+'
+%?0* 2-*
K<jIN@GF*
2-*
K<EIN@GK*
)(H5.'
)"+* 2-*
G<GGGFKL*
2-*
G<GGGIIE*
(05E2'
2-* 2-*
G<GJJLLE*
2-*
G<ELEjLJ*
-*A)+'
2-* 2-*
G<GLHjFH*
+%?0*
G<jPjELj*
$-0;,);'
2-*
+%?0*
G<GFHJJF*
+%?0*
G<GPLPJE*
/>5>66'
2-* 2-*
G<IIHjHK*
+%?0*
G<GjEIJE*

W%<&"'(D0'Y"#"'S#.,@*1"#7'?#%&;4,40''[:Mn[*',#25$#*#1%?"0&*,0'".1),0$*%4*8,0,*p0$%5%&6*T8pU*$,')#*?"$1*
2-',&25($,+*&,0,@,A-',##"%0*%0*$1,*_*.1'%)%#%),*"0*D,0&(5*(0+*;(C(00(1*163'"+#<*
'
:;<.,8'
E%7"$5.;'
QJ)%&-"'
+5&8J"#.,@*1"#7'
!"#$%&'
V1'%)(/0*',&25($%'*
F<FK*A*IG@J*
E<K*
''
V1'%)(/0*)%+"S.(/%0*
G<GGIH*
E*
''
O%0@),)3'(0,@3%20+,+*%'&(0,55,*
G<GGK*
I<J*
(%)%##%*'
a:8N*-'%$,"0*
I<GE*A*IG@F*
IE<L*
''
V1'%)(/0*',&25($%'*
j<jH*A*IG@J*
E<K*
''
V1'%)(/0*)%+"S.(/%0*
G<GGF*
H<j*
''
O%0@),)3'(0,@3%20+,+*%'&(0,55,*
G<GGF*
I<J*

W%<&"'(G0'+%&4"'8,4@5)".;'.%7"4'%#8'&5$=65&8'@*%#$"')%&-"4'65.'$"#"4'8"7".1,#"8'75'<"'47%<&;'"]2."44"8',#'
7"49@-&%.'$".1'@"&&'#"52&%41'7-15.^#5.1%&'KHWJQEH'"]2".,1"#740''O,"$1,'*5%.2#*?(#*#"&0"S.(0$56*)"#',&25($,+*
TQ[=*G<GIU<*
**
Q[=*;$,'"5,*;(C(00(1* Q[=*;$,'"5,*D,0&(5*
** [%),#/.* Q,'/5,* 9"5+*;-,.",#* [%),#/.* Q,'/5,* 9"5+*;-,.",#*
$-02C'
G<GHjHFjHLJ*
G<KGHFEEKKH*
G<GLGEJPjJJ*
G<PKILPKJL*
G<LILJPJK*
G<PIHILGIFI*
0$0E='
G<GIEEILKJK*
G<GEFLJFjJI*
G<GPJjGJJJL*
G<EIPGIFJjP*
G<GJPPELGLj*
G<EHIJELEIL*
** b%&*QV*;$,'"5,*;(C(00(1* b%&*QV*;$,'"5,*D,0&(5*
** [%),#/.* Q,'/5,* 9"5+*;-,.",#* [%),#/.* Q,'/5,* 9"5+*;-,.",#*
$-02C'
G<PHKJHjJJI*
G<IEPPEEFjI*
G<KJJHEJHLE*
@G<GFGjHGKHL*
@G<GjFFjjPJJ*
@G<GjLHPGHLH*
0$0E='
G<FKLLPJHjL*
G<JjHGHKLKL*
G<KGLFLLIGI*
G<HJHGKjKL*
G<JFJLLjPFL*
G<EPJGjFHJj*

W%<&"'(N0'T'E*.515451"'YZ?('."$,5#4'BQ_G']/MJD'C'65.'<57*'*;<.,8'<.""84'
D',,+*
a('B,'*
e%#"/%0**
e@M(52,*
@5%&IGe*
T(++"/C,U*
@5%&IGe*
T+%)"0(0$U*
;(C(00(1*
.1'_<IPIGJFLI**
IJ>JGH>PLH*
G<GGGJKIFKJ*
E<EEFLKLPPP*
E<EEFLKLPPP*
.1'_<IjEIGPJj**
IF>FEj>HFK*
G<GGGJKIFKJ*
E<EEFLKLPPP*
E<EEFLKLPPP*
.1'_<HGHHJIHL**
IK>HjJ>LEL*
G<GGGJKIFKJ*
E<EEFLKLPPP*
E<EEFLKLPPP*
.1'_<HGLFFLFP**
IK>LPG>FFK*
G<GGGJKIFKJ*
E<EEFLKLPPP*
E<EEFLKLPPP*
.1'_<HHGLJEIL**
IL>PLj>HGj*
G<GGGEIKJjI*
E<JjjKEPIIj*
E<JjjKEPIIj*
**
.1'_<HPFFEjEG**
HE>IGP>IFj*
G<GGGHPHJJL*
E<FJjGKHjIF*
E<FJjGKHjIF*
.1'_<HjEHHEIG**
HE>LEK>PFK*
G<GGGHEjKHJ*
E<KHGJKjHjI*
E<KHGJKjHjI*
.1'_<HjKEIKJj**
HE>jjJ>JjL*
G<GGGHEjKHJ*
E<KHGJKjHjI*
q*E<E*
.1'_<HjKLFPGJ**
HJ>GEH>HJJ*
G<GGGHEjKHJ*
E<KHGJKjHjI*
E<KHGJKjHjI*
.1'_<HjLHLHHE**
HJ>GLP>JPE*
G<GGGHEjKHJ*
E<KHGJKjHjI*
E<KHGJKjHjI*
**
.1'_<JIHjPFPj**
EE>GFF>GPF*
G<GGGIFLILL*
E<PGEKIGEGH*
E<PGEKIGEGH*
**
.1'_<JIjEHIHj**
EE>FFH>LjL*
G<GGGEFFPHL*
qE<E*
E<JJPLKIGEJ*
.1'_<IHIKLLHjG**
IGG>HEJ>JJJ*
G<GGGJJIFPL*
E<EFJjPEFFP*
E<EFJjPEFFP*
.1'_<IHIjPGjIP**
IGG>JjH>LHI*
G<GGGJJIFPL*
E<EFJjPEFFP*
E<EFJjPEFFP*
.1'_<IHHGJEKEF**
IGG>FJL>IFP*
G<GGGJJIFPL*
E<EFJjPEFFP*
E<EFJjPEFFP*
.1'_<IHHGjLFjI**
IGG>FjI>FIL*
G<GGGJJIFPL*
E<EFJjPEFFP*
E<EFJjPEFFP*
.1'_<IHHIHHFGL**
IGG>KGP>LFP*
G<GGGJJIFPL*
E<EFJjPEFFP*
E<EFJjPEFFP*
.1'_<IHHEHFIEj**
IGG>LKK>HEL*
G<GGGJJIFPL*
E<EFJjPEFFP*
E<EFJjPEFFP*
.1'_<IHHKLLHFL**
IGI>GFE>EPH*
G<GGGJJIFPL*
E<EFJjPEFFP*
E<EFJjPEFFP*
**
.1'_<IEJHFKFEF**
IIG>IGK>FjF*
j<LJIKJN@GF*
J<GIIEKPIHI*
E<KjHKKHHKH*
D,0&(5*
.1'_<IJPEGIKjH**
IIj>jEH>FFG*
G<GGGHLGGJP*
E<FKPFFjEEL*
G<GGGHLGGJP*
.1'_<IFJILGKEL**
IHJ>IKL>ILK*
G<GGGHLGGJP*
E<FKPFFjEEL*
G<GGGHLGGJP*
.1'_<IFFGjPEGP**
IHJ>PHG>PPF*
G<GGGHLGGJP*
E<FKPFFjEEL*
G<GGGHLGGJP*
.1'_<IFKFJHEFj**
IHF>LGK>FLE*
G<GGGHLGGJP*
E<FKPFFjEEL*
G<GGGHLGGJP*

W%<&"'(U0'W52'GM'YZ?('1%.A".4'65.'7*"'(%)%##%*'@5*5.70''
;(C(00(1*V%1%'$>*:++"/C,*a%+,5*TNaa:_U*
**
;(C(00(1*V%1%'$>*[%)"0(0$*a%+,5*TNaa:_U*
a('B,'*
Q,5V($F*
e%#/%0* e@M(52,* @5%&IGe* **
a('B,'*
Q,5V($F*
e%#/%0* e@M(52,* @5%&IGe*
.1':H<HHKFEIjPE*
IJEKIEGKK*
I<FHN@GK*
F<PIjJPLEHE*
**
.1'NE<LjJPJLj*
KFFLEKL*
H<GEKKN@GK*
F<KjIGPjjEE*
.1'NE<LjJPJLj*
KFFLEKL*
I<LGN@GF*
J<LKPELKEPj*
**
.1':H<IFEFKKGHL*
jFjJPPFj*
H<FjN@GF*
J<FPKHKjIIF*
.1'[I<IFjJGLPG*
IHjPjHjP*
F<LLN@GF*
J<HEPKFFJI*
**
.1'[I<IFjJGLPG*
IHjPjHjP*
H<KJN@GF*
J<FLLPLLLGH*
.1'DH<IFLIPFHHK*
IELIGIKFP*
K<EFN@GF*
J<IjLGLLLjH*
**
.1':I<HGLJJPj*
IjFjJLj*
H<jEN@GF*
J<FEEJPEFF*
.1':H<IFJEILLKj*
jKFPHELH*
K<FEN@GF*
J<IP