BG1 has a major role in MHC-linked resistance to malignant lymphoma in the chicken.

Abstract

Pathogen selection is postulated to drive MHC allelic diversity at loci for antigen presentation. However, readily apparent MHC infectious disease associations are rare in most species. The strong link between MHC-B haplotype and the occurrence of virally induced tumors in the chicken provides a means for defining the relationship between pathogen selection and MHC polymorphism. Here, we verified a significant difference in resistance to gallid herpesvirus-2 (GaHV-2)-induced lymphomas (Marek's disease) conferred by two closely-related recombinant MHC-B haplotypes. We mapped the crossover breakpoints that distinguish these haplotypes to the highly polymorphic BG1 locus. BG1 encodes an Ig-superfamily type I transmembrane receptor-like protein that contains an immunoreceptor tyrosine-based inhibition motif (ITIM), which undergoes phosphorylation and is recognized by Src homology 2 domain-containing protein tyrosine phosphatase (SHP-2). The recombinant haplotypes are identical, except for differences within the BG1 3'-untranslated region (3'-UTR). The 3'-UTR of the BG1 allele associated with increased lymphoma contains a 225-bp insert of retroviral origin and showed greater inhibition of luciferase reporter gene translation compared to the other allele. These findings suggest that BG1 could affect the outcome of GaHV-2 infection through modulation of the lymphoid cell responsiveness to infection, a condition that is critical for GaHV-2 replication and in which the MHC-B haplotype has been previously implicated. This work provides a mechanism by which MHC-B region genetics contributes to the incidence of GaHV-2-induced malignant lymphoma in the chicken and invites consideration of the possibility that similar mechanisms might affect the incidence of lymphomas associated with other oncogenic viral infections.

Full text

Page 1

BG1 has a major role in MHC-linked resistance
to malignant lymphoma in the chicken
Ronald M. Gotoa,1, Yujun Wanga,1, Robert L. Taylor, Jr.b, Patricia S. Wakenellc,2, Kazuyoshi Hosomichid, Takashi Shiinad,
Craig S. Blackmorec, W. Elwood Brilese, and Marcia M. Millera,3
aDepartment of Molecular Biology, Beckman Research Institute, City of Hope, Duarte, CA 91010;bDepartment of Animal and Nutritional Sciences, University
of New Hampshire, Durham, NH 03824;cDepartment of Population Health and Reproduction, University of California Davis, Davis, CA 95616;dDepartment
of Molecular Life Science, Tokai University School of Medicine, Isehara, Kanagawa 259-1143 Japan; andeDepartment of Biological Sciences, Northern
Illinois University, DeKalb, IL 60115
Edited by Richard L. Witter, U.S. Department of Agriculture, East Lansing, MI, and approved August 10, 2009 (received for review June 16, 2009)
Pathogen selection is postulated to drive MHC allelic diversity at loci
for antigen presentation. However, readily apparent MHC infectious
diseaseassociationsarerareinmostspecies.Thestronglinkbetween
MHC-B haplotype and the occurrence of virally induced tumors in the
chicken provides a means for defining the relationship between
pathogen selection and MHC polymorphism. Here, we verified a
significant difference in resistance to gallid herpesvirus-2 (GaHV-2)-
induced lymphomas (Marek’s disease) conferred by two closely-
related recombinant MHC-B haplotypes. We mapped the crossover
breakpoints that distinguish these haplotypes to the highly polymor-
phic BG1 locus. BG1 encodes an Ig-superfamily type I transmembrane
receptor-like protein that contains an immunoreceptor tyrosine-
based inhibition motif (ITIM), which undergoes phosphorylation and
is recognized by Src homology 2 domain-containing protein tyrosine
phosphatase (SHP-2). The recombinant haplotypes are identical, ex-
cept for differences within the BG1 3?-untranslated region (3?-UTR).
The 3?-UTR of the BG1 allele associated with increased lymphoma
contains a 225-bp insert of retroviral origin and showed greater
inhibition of luciferase reporter gene translation compared to the
otherallele.ThesefindingssuggestthatBG1couldaffecttheoutcome
ofGaHV-2infectionthroughmodulationofthelymphoidcellrespon-
sivenesstoinfection,aconditionthatiscriticalforGaHV-2replication
and in which the MHC-B haplotype has been previously implicated.
This work provides a mechanism by which MHC-B region genetics
contributestotheincidenceofGaHV-2-inducedmalignantlymphoma
in the chicken and invites consideration of the possibility that similar
mechanismsmightaffecttheincidenceoflymphomasassociatedwith
other oncogenic viral infections.
disease resistance gene ? GaHV-2 induced lymphoma ? Gallus gallus ?
Marek’s disease
T
vertebrates. Despite the seeming likelihood that pathogen se-
lection has an important role in driving allelic diversity at MHC
class I and class II loci, associations are rarely found between
infectious diseases and particular MHC alleles or haplotypes.
The presence of multiple polymorphic class I and class II gene
family members within MHC haplotypes in many species could
contribute to the difficulty in observing MHC disease associa-
tions. In the chicken, the MHC-B region has an exceptionally
strong role in genetic resistance to Marek’s disease (MD) caused
by gallid herpesvirus-2 (GaHV-2) and to Rous sarcoma virus
(RSV)-induced tumors (1, 2). Resistance to these diseases maps
to the MHC-B subregion marked by a highly expressed classical
MHC class I gene (3–6). It has been suggested that strong
MHC-B disease associations are apparent as the result of alleles
at this single locus for classical class I antigen presentation, BF2,
contributing to immune responses either independently or in
concert with closely-linked transporter associated with antigen-
processing (TAP) genes (6, 7). Antigen presenting molecules
encoded by different BF2 alleles bind quite different classes of
here is an astounding breadth of genetic diversity at the
MHC, a region found within the genomes of all jawed
peptides (7–10) and thereby likely selectively influence adaptive
immune responses to antigen, but the gene or genes that provide
MHC-linked resistance to virally induced tumors in chickens are
not known. Recently, the crossover breakpoint in the recombi-
nant haplotype originally used to map MD resistance to the
MHC-B subregion marked by BF2 was localized, which revealed
that 27 genes lie within the region between the crossover
breakpoint and the BF2 locus (4, 11). Thus, it is now evident that
many genes are candidates for providing the long noted MHC-B
haplotype-linked resistance to MD. Overall, the MHC region in
chicken stands out in contrast to the MHC in mammals. The
chicken MHC is compact and segmented into the MHC-B and
MHC-Y regions in which the MHC class I and class II loci reside,
as well as genes that changed or moved later in evolutionary time
in other species (6, 12–15). Meiotic recombination within
MHC-B is rare (16), but several recombinant haplotypes are
available that are suitable for further investigation of MHC-B-
linked disease resistance.
BR2 and BR4 are two recombinant haplotypes within pedi-
greed matings designed to identify duplicate recombinant hap-
lotypes originating from independent crossover events between
the same parent haplotypes (Fig. 1A). Such recombinant hap-
lotypes were sought as a means for defining which genes within
the MHC-B region confer resistance to MD (16). When tested in
a small challenge trial with GaHV-2 virus at the fourth backcross
generation in the development of congenic lines, birds bearing BR2
andBR4differedsignificantlyintheincidenceofMD(17).BR2and
BR4 also apparently contribute differently to the influence of
MHC-B on the regression of RSV-induced tumors (18). These
observations suggest that, although indistinguishable by the sero-
logically-defined erythroid BG and class I BF antigens used to
isolate them, BR2 and BR4 are likely different as the result of
meiotic recombination breakpoints having occurred at different
locations within the region separating the genes encoding these
serological markers. The breakpoints apparently surround a gene
that influences disease resistance.
Results
A Larger Challenge Trial Confirmed the Difference Between BR2 and
BR4 in Conferring Resistance to Marek’s Lymphoma. We conducted
a GaHV-2 challenge trial using the 003.R2 and 003.R4 lines, now
Author contributions: R.M.G., Y.W., R.L.T., P.S.W., and M.M.M. designed research; R.M.G.,
Y.W.,P.S.W.,K.H.,T.S.,C.S.B.,andM.M.M.performedresearch;R.L.T.,P.S.W.,K.H.,T.S.,and
W.E.B. contributed new reagents/analytic tools; R.M.G., Y.W., R.L.T., P.S.W., K.H., T.S., and
M.M.M. analyzed data; and R.L.T. and M.M.M. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The nucleotide sequences reported in this paper have been deposited in
the GenBank database (Accession Nos. FJ770457 - FJ770460).
1R.M.G. and Y.W. contributed equally to this work.
2Present address: Department of Comparative Pathobiology, School of Veterinary Medi-
cine, Purdue University, West Lafayette, IN.
3To whom correspondence should be addressed. E-mail: MaMiller@coh.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0906776106/DCSupplemental.
16740–16745 ?
PNAS ?
September 29, 2009 ?
vol. 106 ?
no. 39www.pnas.org?cgi?doi?10.1073?pnas.0906776106

Page 2

more inbred than previously as the result of 10 backcross
generations of breeding, to verify the previous observation
suggestingthatBR2andBR4differintheirinfluenceinMD(17).
Pedigree-hatched chicks were inoculated with the highly virulent
RB1B strain of GaHV-2 and observed for the formation of
tumors over 12 weeks. Genotypes were obtained for each bird
and combined with disease data at the conclusion of the trial. A
difference in MD mortality between the lines was evident during
the trial and was consistent with the difference seen in MD
incidence upon gross examination of all birds at completion of
the trial. MD incidence was significantly lower in line 003.R2
birds compared to line 003.R4 birds (Table 1). Inclusion of the
available histopathological findings increased the incidence of
MD overall, but the two lines remained significantly different
(39% in 003.R2 versus 61% in 003.R4, P ? 0.0001). Tumors in
the 003.R2 and 003.R4 lines were present predominantly in
heart, kidney, liver, spleen, and gonad. The difference in tumor
incidence between lines 003.R2 and 003.R4 in this trial, after 10
generations of backcrossing, is essentially identical to that re-
ported for the fourth backcross generation (17). Thus, it appears
that although BR2 and BR4 are identical for MHC-B serological
markers for BG and BF, they differ significantly in the capacity
to confer resistance to MD. This observation supports our hypoth-
esis that the BR2 and BR4 haplotypes have different crossover
breakpoints bounding a gene conferring disease resistance.
The Difference Between MHC-BR2 and MHC-BR4 Maps to BG1. To
define the crossover breakpoints, we initially assayed BR2, BR4,
and the parental haplotypes B2 and BR1 (a recombinant hap-
lotype derived earlier from B23 and B24, Fig. 1A) for BG
restriction fragment patterns, LEI0258 microsatellite PCR-
product lengths, and MHC class I and class II single strand
conformation polymorphism patterns. Alleles in BR2 and BR4
were assigned based on matches with alleles in the parental
haplotypes (Fig. 1B). Consistent with earlier serological typing,
BR2andBR4werefoundtoshareMHCclassIandclassIIalleles
with the B2 parent haplotype (Fig. 1B). The restriction fragment
patterns of the BG genes for BR2 and BR4 resembled those of
the B23 portion of the BR1 parent haplotype, confirming earlier
serological identification that the BG gene cluster originated
from the BR1 parent haplotype. BR2 and BR4 were identical at
the microsatellite marker LEI0258, and matched the BR1 parent
haplotype. Thus, the crossover breakpoints producing BR2 and
BR4 occurred downstream of LEI0258, narrowing further in-
vestigation to seven genes—B-BTN1, B-BTN2, tRNA-leu, BG1 [a
structurally distinct member of the BG gene family, located
about 100-kb distant from the cluster of BG genes encoding BG
antigens (11, 19)], and Blec4, Blec2, and Blec1 (Fig. 1C).
We resolved the BR2 and BR4 crossover breakpoints through
sequencing. Single nucleotide polymorphism (SNP) differences
revealedcloselypositioned,butdistinct,crossoverbreakpointsin
the vicinity of the BG1 gene (encoded on the complementary
strand) (Fig. 1D). In the formation of the BR4 haplotype,
crossover between the BR1 and B2 parent haplotypes occurred
somewhere within a 338-bp interval defined by SNPs 16900 and
C
D
E
AB
Fig. 1.
the genetic differences between them. (A) BR2 and BR4 haplotypes were
recovered in crosses made between animals bearing BR1 (a recombinant
haplotype derived from B23 and B24) and B2 haplotypes when atypical
segregation of antigens encoded by the BG gene cluster and BF (MHC class I)
genes were detected in serological typing of progeny. (B) The crossover
breakpoint regions in BR2 and BR4 were narrowed to a region between
LEI0258 and the BLB loci. The BG gene cluster, about 100 kb upstream from
LEI0258, was typed by restriction fragment patterns; the LEI0258 by PCR
product length; and the BF and BL loci by single strand conformation poly-
morphism. Images for each locus were obtained from single gels and are
duplicated and rearranged as needed here to aid comparisons. (C) Diagram
(not to scale) of the MHC-B, illustrating the genes located within the 27-kb
crossover breakpoint region (arrows indicate direction of transcription). (D)
Map of SNPs within the 27-kb region that define the crossover (CO) break-
points in BR2 and BR4 as distinct, with the BR2 crossover occurring within the
3?-endofBG1,whichisencodedonthecomplementarystrand,asindicatedby
the arrow. Positions are numbered based on GenBank AB268588. E) The BG1
genesequenceispresented5?to3?,illustratingtheconsequencesofcrossover
breakpoints on the structure of the BG1*R2 and BG1*R4 alleles and the
presence of a 225-bp insert in BG1*R4 that provides an alternate polyadenyl-
ation signal sequence (PAS, indicated by *).
Derivation of the BR2 and BR4 MHC-B recombinant haplotypes and
Table 1. Congenic lines 003.R2 and 003.R4 birds differ in the
incidence of Marek’s disease tumors following infection with
GaHV-2
Line
MD mortality
during trial
?2
Total MD
Incidence
?2
003.R2
003.R4
Specific pathogen free
10% (14/140)
39% (51/130)
66% (86/131)
b, c 19% (27/140)
b47% (61/130)
95% (124/131)
b, c
b
TheincidenceofMarek’sdiseasethatoccurredduringthetrial,asindicated
by early mortality, and the total incidence detected upon gross examination
at the conclusion of the trial differed significantly between the 003.R2 and
003.R4 lines. Specific pathogen free (SPF) birds, infected secondarily by inha-
lation of shed virus, served as controls to gauge the success of the initial
infection. b ? P ? 0.0001 vs SPF; c ? P ? 0.0001 vs 003.R4
Goto et al.PNAS ?
September 29, 2009 ?
vol. 106 ?
no. 39 ?
16741
GENETICS

Page 3

17238. These are, respectively, the last SNP matching BR1 and
firstSNPmatchingB2intheBR4haplotype.TheBR2breakpoint
occurred a short distance away, within an 800-bp region between
SNPs 18364 and 19164 (Fig. 1D). The BR4 breakpoint is outside
the BG1 coding region; therefore, the BG1*R4 allele is entirely
identical to the BG1 allele in the B2 parent haplotype. In
contrast, three SNPs (see SNPs 18170, 18184, and 18364 in Fig.
1D) mapped the breakpoint for BR2 to within the 3?-
untranslated region (3?-UTR) of the BG1 gene. Thus, although
identical to the BG1*B4 allele in its coding region, BG1*R2
possesses a 3?-UTR derived from the BR1 parent haplotype.
Sequencing revealed an additional difference between
BG1*R2 and BG1*R4. The 3?-UTR of the BG1*R4 and the BG1
allele in the parent B2 haplotype contain a 225-bp insert that is
absent from BG1*R2. This insert provides an alternate polyad-
enylation signal (PAS) (Fig. 1E, sequences in Fig. S1). BLAST
searches showed that the 225-bp insert is essentially identical to
the 3?-UTR within a retroviral Pro-Pol-dUTPase polyprotein-
encoding sequence that is found at multiple sites throughout the
Gallus gallus genome (Fig. S2). The PAS within the insert is the
dominant signal used in the production of BG1*R4 transcripts,
asshownby3?-RACEanalysisofcDNA(Fig.2A).Lesscommon,
longer BG1*R4 transcripts arise from the use of the endogenous
BG PAS that is displaced downstream as a result of the viral
insertion. The 3?-RACE products from BG1*BR2 transcripts
were of a single size, consistent with the single PAS in this allele.
We found BR2 and BR4 are completely identical across a 61-kb
interval of MHC-B (see GenBank FJ770459 and FJ770460),
aside from the crossover breakpoint region. This includes iden-
tity at the MHC-B loci involved in class I and class II antigen
processing and presentation, as well as Blec2, which encodes an
inhibitory receptor expressed on natural killer cells, and BLec1,
which encodes a putative activating receptor. In addition, typing
of eight individual birds from each line with 2702 SNPs evenly-
distributed across the genome showed that although some
heterozygosity remains in the 003.R2 and 003.R4 lines, there is
no evidence for any other region of the genome becoming fixed
by chance during the production of the 003.R2 and 003.R4
congenic lines (Table S1).
The 3?-UTR Difference Between BG1*R2 and BG1*R4 Had Little Influ-
ence on Transcription but Produced Consistent Differences in Assays
for Translation. To determine whether the 3?-UTR differences
found in BG1*R2 and BG1*R4 affect gene expression, we tested
the relative transcription efficiency of the BG1*R2 and BG1*R4
alleles in quantitative PCR assays. Although BG1 was more
highly expressed in some tissues than others, we found no
significant differences in BG1 ?CTvalues between paired tissue
samples across the 003.R2 and 003.R4 lines (Table S2). These
data indicate that the 3?-UTR difference has little effect on BG1
transcription. We then tested the 3?-UTRs for their effect on
translation in dual luciferase reporter assays. Translation of
firefly luciferase in LMH cells in conjunction with the shorter
and longer BG1*R4 3?-UTRs was consistently repressed as
compared to expression with the BG1*R2 3?-UTR (Fig. 2B),
suggesting that differences in BG1 translation in vivo could be
the basis for the differences in MD between the 003.R2 and
003.R4 lines.
The BG1 Transmembrane Protein Is Phosphorylated on Tyrosine in
Pervanadate-Treated Cells and Co-Precipitates SHP-2. BG1 encodes
a dimerizing 32-kDa type I transmembrane protein bearing an
IgV-like ectodomain and a cytoplasmic tail made up of a
coiled-coil region and a C-terminal domain containing a single
immunoreceptor tyrosine-based inhibition motif (ITIM) (Fig.
3A). Therefore, BG1 is similar to receptors that bear single
IgV-like ectodomains known to attenuate lymphoid cell activa-
tion via cytoplasmic ITIM signaling (20). To determine whether
BG1 is capable of signaling, we expressed a full-length FLAG
epitope-tagged BG1 cDNA clone in LMH cells. At least a
portion of the BG1 expressed reached the cell surface (Fig. 3B).
As suggested by its structure, FLAG-BG1 underwent tyrosine
phosphorylation in the presence of pervanadate, as illustrated in
BG1 immunoblots developed with 4G10, a monoclonal antibody
specific for phosphotyrosine (Fig. 3C). Further, the protein
tyrosine phosphatase SHP-2 (Src homology 2 domain phospha-
tase-1) co-precipitated with phosphorylated BG1 (Fig. 3D).
Although the BG1 protein remains to be fully analyzed, the data
presented here indicate that BG1 reaches the cell surface, can be
phosphorylated on tyrosine, and that, in turn, it associates with
the signaling phosphatase SHP-2. These are all features consis-
tent with BG1 functioning as a surface receptor that limits cell
activation.
Discussion
The selection of rare recombinants originating from the same
parent haplotypes has provided two MHC-B haplotypes, BR2
and BR4, which differ in their contribution to genetic resistance
to lymphomas induced by infection with GaHV-2. When tested
at the fourth backcross generation in the production of congenic
lines, the difference in tumor incidence between BR2 and BR4
line birds was 22% (17). As reported here, when tested in a
substantially larger trial conducted after six additional genera-
B
A
Fig. 2.
and in influence on the translation of mRNA. (A) 3?-RACE reactions revealed
length differences between the BG1*R2 and BG1*R4 cDNAs as a result of
differences in the length of the 3?-UTR. The BG1*R2 allele provides 3?-RACE
productsofasinglelength,consistentwiththeanticipatedsizeof605bp.The
productsformedintheBG1*R43?-RACEreactionsareconsistentwithBG1*R4
transcripts produced primarily through use of the polyadenylation signal
(PAS; indicated by *) introduced within the inserted sequence (701 bp) and
secondarily from the PAS that is inherent in the locus but shifted by the
inserted sequence (825 bp). (B) 3?-UTRs derived from BG1*R2 and BG1*R4
cDNA were cloned downstream of the firefly luciferase gene and tested in
dualluciferaseexpressionassaysinchickenLMHcellsstimulatedwithphorbol
myristate acetate. All three 3?-UTRs inhibited luciferase activity compared to
the vector-only control. The BG1*R4 3?-UTRs were consistently significantly
more inhibitory than the 3?-UTR from the BG1*R2 allele.
The 3?-untranslated regions of BG1*R2 and BG1*R4 differ in length
16742 ?
www.pnas.org?cgi?doi?10.1073?pnas.0906776106Goto et al.

Page 4

tions of backcrossing, the difference in GaHV-2 tumor incidence
between the 003.R2 and 003.R4 lines remained the same. Thus,
the results indicate that genetic variability within the lines in
regions outside MHC-B, which was no doubt greater at the
fourthbackcrossgenerationthanatthetenth,hadlittleinfluence
on the difference in the incidence of lymphomas between the
lines in the two challenge trials. The findings in this study map
thebasisforthedifferenceobservedbetween003.R2and003.R4
chickens in GaHV-2 lymphomas to BG1.
As noted in a recent comprehensive review by Calnek (21), the
pathology that ensues following infection with GaHV-2 is com-
plex, with a variety of cells responding in different ways at
different times to the presence of the virus. In natural infections,
GaHV-2 is transmitted by the inhalation of poultry dander laden
with enveloped GaHV-2. The virus is transported from the
respiratory tract to lymphoid organs. Infection of lymphocytes,
viral proliferation, and subsequent necrosis in the spleen, bursa
and thymus trigger the migration of additional lymphocytes,
granulocytes, and macrophages to these organs and acute in-
flammatoryresponsesensue.Cytolyticinfectionoccursmostlyin
B cells, but also appears in some T cells and in some epithelial
tissues as infected lymphoid cells disseminate. There is a fairly
abrupt shift from cytolytic to latent infection, with latency
developing mostly in T cells. The period of latent infection is
short in genetically susceptible birds, and the virus soon emerges
to begin a second cytolytic phase of infection. Transformation of
activated CD4?T cells soon occurs. Tumors and death typically
follow within weeks. There are a number of points in the course
of infection at which an inhibitory molecule, the role suggested
for BG1, could influence the outcome of GaHV-2 infection.
Numerous investigations into the difference between MHC-
B-resistant (B21-resistant lines N or N2a) and -susceptible
(B19-susceptible lines P or P2a) lines provide intriguing data to
consider in light of identification of BG1 as a candidate gene
affecting MD. There is little evidence that MHC-B genotype-
associated differences exist during the initial phase of infection,
aside from the slightly delayed appearance of splenomegaly in
MHC-B-resistant birds and occasional subtle differences in
cytokine profiles [reviewed by (21); see also (22–24)]. However,
by the end of the first 7–10 days post inoculation (dpi), substan-
tial differences in MD pathogenesis become clearly evident.
There is an abrupt decrease in the viral load in the white cells in
resistant birds (23) that is accompanied by lower levels of
infected T cells (25), relatively higher levels of virus-neutralizing
antibodies (26), fewer AV37? cells (22), and the absence of
induction of the proinflammatory cytokines IL-6 and IL-18,
which in contrast occurs in susceptible birds during the same
interval (23). In the absence of complicating stress or infection,
the genetically resistant birds remain healthy, with the only sign
of active infection being the shedding of virus from the feather
follicle.
Part of the change in MD pathogenesis in genetically resistant
birds that occurs at 7–10 dpi could be due to differential
development of specific cytotoxic T lymphocytes (CTLs) di-
rected against viral antigens. Specific CTL responses in MD-
resistant B21 homozygous birds to the GaHV-2 transcription
regulatory protein ICP4 may be a factor in genetic resistance,
particularly as immunity develops following vaccination (27). At
the same time, there are clearly other genetic factors contrib-
uting in responses to MD. The generally low immune respon-
siveness of MHC-B-resistant birds lead Calnek and colleagues to
suggest that progression of MD tumors is, contrary to logic,
associated with greater immune activation or reactivity (21, 28).
Thisproposalisespeciallyinterestinginlightoftheapparentrole
of BG1 in inhibitory signaling. Many studies provide data
showing greater immune responsiveness in MHC-B genetically
susceptible birds. For example, lymphocytes from MD-
susceptible Line P (MHC-B19) are more responsive to mitogens
(ConAandPHA)thanbirdsfromMHC-B-resistantLineNbirds
(MHC-B21) (29, 30). When infected in vitro lymphocytes from
MHC-B-susceptible lines produce significantly more infected T
cells compared to lymphocytes from MHC-B-resistant lines (31).
Elevated tumor formation in the presence of allogeneic immune
responses suggests, in another way, that immune activation
promotes tumor formation (32). The striking difference in
transcription of the proinflammatory IL-6 and IL-18 cytokine
genes between MHC-B genetically resistant and susceptible lines
suggests these cytokines drive the enhanced immune responses
contributing to increased tumors (23). Although it remains to be
tested, allelic differences at BG1 might contribute to these
differences in activation noted between MHC-B-resistant (B21/
AB
C
D
Fig. 3.
Depiction of the predicted structure of BG1 based on the sequence in Fig. S3.
(B) Flow cytometry of LMH cells stably expressing FLAG-BG1 shows that BG1
reaches the cell surface. Cells were stained with the anti-FLAG monoclonal
antibody M2 and anti-mouse IgG–allophycocyanin (APC) (striped histogram)
and with anti-mouse IgG-APC alone (open histogram) and analyzed by flow
cytometry gating on the DAPI-negative cell population. (C) BG1 is phosphor-
ylatedontyrosine.LMHcellsstablyexpressingFLAG-BG1werelysedfollowing
0-, 5-, and 15-min treatments with pervanadate, a tyrosine phosphatase
inhibitor. BG1 was immunoprecipitated (IP) from total cell lysates using the
anti-FLAG antibody M2 and then immunoblotted (IB) from a 12% polyacryl-
amide gel and probed with either M2 or the monoclonal anti-phosphoty-
rosineantibody4G10.(D)ThetyrosinephosphataseSHP-2co-precipitateswith
BG1.FLAG-BG1wasimmunoprecipitated(IP)fromLMHcellsstablyexpressing
FLAG-BG1withM2usingcultureswithout(–)orwithtreatmentwithpervana-
date for 15 min (?), electrophoresed, transferred from a either a 9% or 12%
gel, and then immunoblotted either with monoclonal M2 (12% gel) or poly-
clonalanti-SHP-2C-18(9%gel)antibodies.IRDye800CWsecondaryantibodies
were used to develop the blots.
BG1 has the structure of a tyrosine-based inhibitory receptor. (A)
Goto et al.PNAS ?
September 29, 2009 ?
vol. 106 ?
no. 39 ?
16743
GENETICS

Page 5

B21) and susceptible (B19/B19) genotypes. BG1 isoforms en-
coded in resistant haplotypes might be more effective at signal-
ing and contribute to the dampened immune responses
associatedwithMDresistance.Itwillbeinterestingtodetermine
thenetworkwithinwhichBG1moleculesfunctionandtoexplore
the possibility that similar mechanisms affect the incidence of
human lymphomas associated with oncogenic viral infection.
BG1*R2 and BG1*R4 represent only two of many BG1 alleles.
BG1*R2, BG1*R4, the BG1 alleles in the parent haplotypes B2
and BR1, and the grandparent haplotype B24 constitute a
subgroup of structurally distinct BG1 alleles. The coding region
in this group is truncated by the presence of an early stop codon
that eliminates 20 amino acids from the C terminus of the final
domain of BG1 found in all other isoforms for which sequences
are published. The early stop codon concomitantly increases the
length of the 3?-UTR (Fig. S1). Truncation of the coding region
could, in itself, alter the function of the BG1 protein compared
to other isoforms (33). We have found that the presence of the
225-bp insert in the 3?-UTR of BG1*R4 allele affects gene
function such that the BG1*R4 allele might best be considered
a mutant allele due to the deleterious effect of the insert. So far,
the 225-bp insert is unique to BR4 and the B2 parent haplotype
among haplotypes for which sequence data are available. It
represents only one of many structural differences found among
BG1 isoforms.
Other BG1 alleles differ in a number of ways (19). In addition
to a few synonymous and non-synonymous nucleotide substitu-
tions, prominent variations in coding sequences appear as result
of duplications of a set of four exons and adjacent introns. These
variations result in different subgroups of BG1 alleles that
possess one, two or four copies of this exon/intron ‘‘quartet’’
encoding a portion of the BG1 cytoplasmic tail. Another group
of BG1 alleles lacks the penultimate BG1 exon that contains the
ITIM-coding sequence (19). While it has not yet been deter-
mined whether the latter structural variation contributes to
disease incidence, it is notable that two haplotypes (B5 and B15)
associated with MD susceptibility and with RSV tumor progres-
sion lack this ITIM encoding exon (34). Additional study of BG1
structural variants is needed to determine whether they contrib-
ute to the disease associations observed among the different
MHC-B haplotypes in which they are found and whether they
reflect the results of pathogen selection.
Even though the BG1*R2 and BG1*R4 alleles have different
3?-UTRs, we found little evidence that these alleles were dif-
ferentially transcribed. In contrast, the dual luciferase assays
indicate that the 3?-UTR differences affect posttranscriptional
regulation (Fig. 2B). Investigations are underway to identify the
mechanism by which the 225-bp insert in the 3?-UTR of BG1*R4
affects translation. The presence of a 3?-UTR derived from a
retroviral sequence could put BG1*R4 under the control of
regulators of retroviral expression that might occur through
binding of sequence-specific small RNAs or proteins. Such
regulators, particularly if they are microRNAs, could result in
BG1*R4 expression being more greatly affected in some tissues
than others. Hence, expression in vivo could vary between tissues,
withthedifferencesbeingeithermoremodestorgreaterthanthose
observed in the LMH cell luciferase assays illustrated in Fig. 2B.
Studies are ongoing to define the mechanism influencing transla-
tionoftheBG1*R4allele,includinginvestigationofseveralmiRNA
target sequences harbored within the 225-bp insert.
Other polymorphic loci within MHC-B, in addition to BG1,
may also contribute to the differences in the incidence of
GaHV-2 lymphoma observed between MHC-B haplotypes. Dif-
ferent MHC class I and class II alleles, some better expressed
than others, likely contribute to the effectiveness of adaptive
immune responses to GaHV-2. An important series of experi-
ments with MHC-B congenic lines has shown that MHC-B
haplotypes differentially influence the effectiveness of vaccinal
(adaptive) immunity (35–37). The MHC-B haplotypes providing
the greatest protective immunity in GaHV-2 challenge trials
following vaccination differ between vaccine serotypes, suggest-
ing that allelic differences in antigen presentation have a critical
role in the development of long-lasting immunity. Importantly,
from the perspective of understanding the contribution of BG1,
haplotypes providing the best protection following vaccination in
this series of experiments were not always the same as those
providing protection in the absence of immunization. Perhaps
allelic differences at BG1 affect activation early in immune re-
sponses, while class I and class II allelic differences contribute
primarily to the effectiveness of memory-based adaptive immunity.
Materials and Methods
OriginofBR2andBR4HaplotypesandtheProductionofCongenicLines.TheBR2
and BR4 recombinant haplotypes were derived from a lineage of birds at
Northern Illinois University originating from a mating between a male of
B23/B24 genotype (derived from a strain of New Hampshire chickens) and
femalesoftheAvianDiseaseandOncologyLaboratoryWhiteLeghorninbred
line72(B2/B2genotype)(16,38,39).Thefirstrecombinanthaplotypethatwas
recovered was designated BR1 and contained elements of B23 and B24 (as
revealed by serological typing reagents) (Fig. 1A). The BR2 and BR4 recombi-
nant haplotypes were recovered in subsequent matings between BR1/B2
males and line 72(B2/B2) females.
Congenic line birds carrying BR2 and BR4 were produced at the University
ofNewHampshirebyintroducingtheBR2andBR4haplotypesintothegenetic
background of the highly inbred line UCD003 and selecting progeny bearing
BR2 and BR4 by serological typing over ten backcross generations. To ensure
that the lines would not be lost, several individuals were selected at each
generation for the next backcross. Hence, alleles at some loci across the
genome continue to segregate at random in lines 003.R2 and 003.R4. The
congenic lines were established by matings among the recombinant progeny
withineachlineandhomozygousbirdsselectedtocarrythelineforward.The
lines are maintained in two to three small breeding colonies, each containing
a single homozygous sire and 10–15 homozygous dams per sire at each
generation.
Viral Challenge Trials of Birds in BR2 and BR4 Congenic Lines. BR2 and BR4
congenic line chicks were pedigree-hatched at University of California, Davis,
double wing-banded, and challenged intra-abdominally with 500 plaque-
forming units (pfu) of the very virulent RB1B strain of GaHV-2 (a serotype 1
virusalsocalledMDV1)(40)attwodays-of-age.MDvirusisclassifiedintothree
serotypes.OnlySerotypeIisoncogenic.SerotypeIstrainsaredividedintofour
pathotypesbasedontheircapacitytocauseMDinvaccinatedbirds.The‘‘very
virulent’’ category is the next to the most virulent category in this scale (41).
Intra-abdominal inoculation broadly mimics natural infection by inhalation
(42), and is the preferred inoculation route to achieve greater uniformity in
GaHV-2viralchallengetrials.TheRB1Binoculawerepropagatedthroughnot
morethantwopassagesonchickkidneyculturecells.Fivereplicatetrialswere
conducted at UC Davis under the supervision of a board-certified avian
pathologist. In all trials, specific pathogen free (SPF) chicks (Charles River
Laboratories) of a similar age were housed with the inoculated birds to serve
as controls for GaHV-2 infectivity. The chicks were checked at least once per
day for clinical signs of disease (paralysis, depression or anorexia) and eutha-
nized if unable to stand and feed. Carcasses of any animals that died in
advanceoftrialterminationwerecollectedandfrozenfornecropsyattheend
of the trial at 12 weeks when all remaining birds were euthanized. At
necropsy, all 270 birds were examined for signs of tumors. Skin, oral cavity,
eyes,thymus,sciaticnerve,vagusnerve,brachialplexus,andallvisceralorgans
were examined for gross evidence of tumors. Tissues were collected from
nearly all birds found negative upon gross examination for microscopic ex-
amination; however, 37 carcasses (23 from 00.3R2 and 14 from 003.R4) were
unavailable for histopathological examination. Histological sections were
examined for evidence of pleomorphic nodules containing lymphocytes with
increased nuclear to cytoplasmic ratios and for the appearance of infiltration
orinvasionofadjacenttissues.Allanimalexperimentswerecarriedoutunder
institutionallyapprovedprotocols.AllBR2andBR4birdsinthechallengetrial
(and their parents) were MHC-B haplotyped by PCR (See SI Materials and
Methods). Disease incidence (number of birds positive versus number of birds
negative) for 003.R2, 003.R4 and SPF controls was compared using the ?2test.
Mapping BR2 and BR4 Crossover Breakpoints and Sequence Differences. A
combination of methods was used narrow the region in BR2 and BR4 con-
16744 ?
www.pnas.org?cgi?doi?10.1073?pnas.0906776106 Goto et al.