6;191)5)%:):-'51<-891:?6;191)5)%:):-'51<-891:?
%'1/1:)3644659%'1/1:)3644659
)+;3:?#;*31+):1659 -7)8:4-5:6.1636/1+)3%+1-5+-9
-A515/)3152=1:0)9:04)1541+-+65/-51:)33?,-A+1-5:15-A515/)3152=1:0)9:04)1541+-+65/-51:)33?,-A+1-5:15
-6915670139-6915670139
)4-9--
)?63151+%+6::9,)3-#06-51>81@65)
)=51415)
)?63151+%+6::9,)3-#06-51>81@65)
1 1# )+1)9
)?63151+%+6::9,)3-#06-51>81@65)
%-8/-1"+02;8
)?63151+%+6::9,)3-#06-51>81@65)
1+0)-3# +)88?
)?63151+%+6::9,)3-#06-51>81@65)
%--5->:7)/-.68),,1:165)3);:0689
6336=:019)5,),,1:165)3=6829):0::79,1/1:)3+64465939;-,;*169+1(7;*9
$-+644-5,-,1:):165$-+644-5,-,1:):165
--1415) )+1)9 "+02;8% +)88? "!-133#86:0-86-#-86$!/;?-5&
6841-8%-521-=1+@63*-8:$15)3,1+2-84)5%8<15--!-A515/)3152
=1:0)9:04)1541+-+65/-51:)33?,-A+1-5:15-6915670139
%+1-5+-
0::79,6168/
9+1-5+-
&0198:1+3-19*86;/0::6?6;.68.8--)5,67-5)++-99*?:0--7)8:4-5:6.1636/1+)3%+1-5+-9):%'1/1:)3
644659:0)9*--5)++-7:-,.6815+3;916515)+;3:?#;*31+):1659*?)5);:0681@-,),41519:8):686.%'1/1:)3
64465968468-15.684):16573-)9-+65:)+:1839;-,;
;:0689;:0689
)4-9--)=51415) 1 1# )+1)9%-8/-1"+02;8 1+0)-3# +)88?):1-$"!-1330-8?3
#86:0-86-$)370#-86&0)50!/;?-5%:-70)51)6841-831@)*-:0-521-=1+@)5)63*-8:19)
$15)3,1%:-<-5+2-84)50)83-98<15)5,!)5+?--
&019)8:1+3-19)<)13)*3-):%'1/1:)36446590::79,1/1:)3+64465939;-,;*169+1(7;*9
References and Notes
1. A. S. Greenberg et al., Nature 374, 168 (1995).
2. K. H. Roux et al., Proc. Natl. Acad. Sci. U.S.A. 95,
11804 (1998).
3. M. H. Richards, J. L. Nelson, Mol. Biol. Evol. 17, 146 (2000).
4. S. Muyldermans, T. Atarhouch, J. Saldanha, J. A. Bar-
bosa, R. Hamers, Protein Eng. 7, 1129 (1994).
5. M. Diaz, R. L. Stanfield, A. S. Greenberg, M. F. Flajnik,
Immunogenetics 54, 501 (2002).
6. K. R. Hinds, G. W. Litman, Nature 320, 546 (1986).
7. F. Kokubu, R. Litman, M. J. Shamblott, K. Hinds, G. W.
Litman, EMBO J. 7, 3413 (1988).
8. S. S. Lee, D. Fitch, M. F. Flajnik, E. Hsu, J. Exp. Med.
191, 1637 (2000).
9. M. J. Flajnik, H. Dooley, unpublished data.
10. M. Diaz, M. F. Flajnik, Immunol. Rev. 162, 13 (1998).
11. C. Hamers-Casterman et al., Nature 363, 446 (1993).
12. A. Desmyter et al., Nature Struct. Biol. 3, 803 (1996).
13. S. Spinelli et al., Nature Struct. Biol. 3, 752 (1996).
14. K. Decanniere et al., Struct. Fold. Des. 7, 361 (1999).
15. K. Decanniere, S. Muyldermans, L. Wyns, J. Mol. Biol.
300, 83 (2000).
16. S. Spinelli et al., Biochemistry 39, 1217 (2000).
17. S. Spinelli, M. Tegoni, L. Frenken, C. van Vliet, C.
Cambillau, J. Mol. Biol. 311, 123 (2001).
18. A. Desmyter, K. Decanniere, S. Muyldermans, L. Wyns,
J. Biol. Chem. 276, 26285 (2001).
19. K. Decanniere et al., J. Mol. Biol. 313, 473 (2001).
20. A. Desmyter et al., J. Biol. Chem. 277, 23645 (2002).
21. W. Vranken et al., Biochemistry 41, 8570 (2002).
22. R. Loris et al., J. Biol. Chem. 278, 28252 (2003).
23. M. Dumoulin et al., Nature 424, 783 (2003).
24. Camel and llama V
H
H domains share about 65%
sequence identity with mouse and human V
H
do-
mains; however, several mutations in the canonical
V
H
interface region render the camelid V
H
H domains
more soluble than isolated human or murine V
H
domains. Additionally, the long CDR3s in several of
the camelid V
H
H domains are folded back across this
interface region, partially rescuing it from solvent
exposure. The camelid CDR3 is sometimes anchored
to the immunoglobulin core by formation of a disul-
fide bridge between CDR3 and CDR1 or FR2. Further-
more, some camelid V
H
H domains can access re-
cessed cavities, such as the HEL active-site cleft, by
means of long CDR3 loops (12).
25. J. Davies, L. Riechmann, FEBS Lett. 339, 285 (1994).
26. H. Dooley, M. F. Flajnik, A. J. Porter, Mol. Immunol.
40, 25 (2003).
27. Materials and methods are available as supporting
material on Science Online.
28. The IgNAR CDR3 is 20 residues in length, spanning
N84 to N103 (Gly-Leu-Gly-Val-Ala-Gly-Gly-Tyr-Cys-
Asp-Tyr-Ala-Leu-Cys-Ser-Ser-Arg-Tyr-Ala-Glu).
29. Comparison of the IgNAR V domain structure with
other antibody x-ray structures with the programs Dali
and SSM (27) reveals similar root mean square devia-
tions for the IgNAR V domain from V
␣
,V
L
,andV
H
domains of around 0.7 to 1.0 Å for 45 core C␣ atoms.
30. C. Chothia, A. M. Lesk, J. Mol. Biol. 196, 901 (1987).
31. The IgNAR CDR1 has root mean square deviations
from the cAb-Lys3 and cAb-RN05 CDR1s of 1.2 and
1.5 Å, respectively, for 15 C␣ atoms (IgNAR residues
N22 to N36).
32. Of the seven residues that differ between HEL and
turkey egg-white lysozyme, two (Arg
73
3 Lys
73
and
Asp
101
3 Gly
101
) are found in the IgNAR-HEL interface.
Arg
L73
makes six van der Waals contacts, one hydrogen
bond, and one salt bridge, and Asp
L101
makes 17 van der
Waals contacts and five hydrogen bonds in the interface
with IgNAR. The camelid V
H
H cAb-Lys3 also binds an
epitope encompassing the recessed HEL active site,
similarly using its CDR3 for access (12). Several murine
antibodies to lysozyme (HyHEL10, HyHEL8, HyHEL26,
and HyHEL63) recognize a common epitope around the
active site, but do not penetrate this site as deeply as do
cAb-Lys3 and IgNAR.
33. The camelid anti–ribonuclease A V
H
H domain cAb-
RN05 uses only CDR1 and CDR3 for binding antigen,
whereas the camelid anticarbonic anhydrase V
H
H
domain cAb-CA05 uses CDR3 almost exclusively with
only two van der Waals contacts to the antigen from
CDR1. Similarly, the camelid anti–␣-amylase V
H
H
domains AMB7 and AMD10 use all three CDR loops,
but CDR1 contributes only 5% of the total buried
surface area on the antibody upon antigen binding.
34. The IgNAR-lysozyme interface is comparable in size
to those seen in lysozyme-Fab (Fv) complexes that
range from 538 to 829 Å
2
for the Fab and 540 to 831
Å
2
for the lysozyme (table S3). Similar interface sizes
have also been observed for lysozyme complexes
with camel V
H
H domains. A total of 122 van der
Waals contacts (table S2), eight hydrogen bonds, and
three charged interactions are made between HEL
and the IgNAR (table S4). The majority of the buried
surface on the IgNAR V domain is contributed by
CDR3 residues N85 to N89, N91, N93, N95 to N96,
and N98 to N103 (75%), with the remainder by
CDR1 N26 to N33 (Fig. 4). The HEL-IgNAR V region
interface has a good shape-complementarity index of
0.70 (0.72 and 0.70 for crystal form 2) (27), with
waters filling several cavities in the interface (fig. S4).
A total of 14 water molecules contact both the
IgNAR V domain and HEL, with 7 of these (waters 2,
4, 7, 8, 60, 114, and 266) sequestered from contact
with external solvent.
35. Type I and type II sequences are 90% identical.
IgNAR type II V domains have only four conserved
cysteines, rather than the six or more found in type
I, and tend toward smaller CDR3 lengths [9 to 18
amino acids (26)].
36. The structurally equivalent positions Lys
N84
(IgNAR)
or Lys
96
(AMD10) correspond to positions Leu
89
or
His
93
of conventional V domains.
37. IgNAR residues Asn
N45
, Glu
N46
, Ser
N48
, Ser
N50
,
Lys
N51
, Gly
N62
, Ser
N63
, and Lys
N64
are under strong
positive selection.
38. S. D. Nuttall et al., Mol. Immunol. 38, 313 (2001).
39. S. D. Nuttall et al., FEBS Lett. 516, 80 (2002).
40. Unexpectedly, a set of receptors (variable lympho-
cyte receptors) that may modulate immune recogni-
tion in lamprey has recently been identified. These
receptors are arranged from leucine-rich repeats and
may constitute a component of a primitive immune
system in lampreys (41).
41. Z. Pancer et al., Nature 430, 174 (2004).
42. V. A. Streltsov et al., Proc. Natl. Acad. Sci. U.S.A. 101,
12444 (2004).
43. We thank P. Horton for technical assistance. I. Hol-
ton, G. Meigs, the staff of Advanced Light Source
Beamline 8.3.1 for support during data collection, and
The National Aquarium in Baltimore for meticulous
care of the sharks. Supported by NIH grants GM38273
(R.L.S. and I.A.W.) and RR06603 (H.D. and M.F.F.). This
is manuscript no. 16602-MB from The Scripps Re-
search Institute. The coordinates and structure fac-
tors have been deposited in the Protein Data Bank
(PDB) with accession codes 1SQ2 (crystal form 1)
and 1T6V (crystal form 2).
Supporting Online Material
www.sciencemag.org/cgi/content/full/1101148/DC1
Materials and Methods
SOM Text
Figs. S1 to S4
Tables S1 to S4
References and Notes
7 June 2004; accepted 4 August 2004
Published 19 August 2004;
10.1126/science.1101148
Include this information when citing this paper.
Defining a Link with Asthma
in Mice Congenitally Deficient
in Eosinophils
James J. Lee,
1,2
* Dawn Dimina,
1,2
MiMi P. Macias,
1,2
†
Sergei I. Ochkur,
1,2
Michael P. McGarry,
1,2
Katie R. O’Neill,
2,3
Cheryl Protheroe,
2,3
Ralph Pero,
2,3
Thanh Nguyen,
1,2
Stephania A. Cormier,
1,2
‡ Elizabeth Lenkiewicz,
1,2
Dana Colbert,
2,3
Lisa Rinaldi,
4
Steven J. Ackerman,
5
Charles G. Irvin,
4
Nancy A. Lee
2,3
*
Eosinophils are often dominant inflammatory cells present in the lungs of
asthma patients. Nonetheless, the role of these leukocytes remains poorly
understood. We have created a transgenic line of mice (PHIL) that are specif-
ically devoid of eosinophils, but otherwise have a full complement of hema-
topoietically derived cells. Allergen challenge of PHIL mice demonstrated that
eosinophils were required for pulmonary mucus accumulation and the airway
hyperresponsiveness associated with asthma. The development of an eosino-
phil-less mouse now permits an unambiguous assessment of a number of
human diseases that have been linked to this granulocyte, including allergic
diseases, parasite infections, and tumorigenesis.
The underlying features of asthma display a
marked heterogeneity (1, 2), yet the presence of
eosinophils in the airway lumen and lung tissue
has been recognized even in the earliest studies
(3) and is often regarded as a defining feature of
this disease (4, 5). Moreover, the recruitment of
eosinophils occurs in animal models of aller-
gen-mediated respiratory inflammation; in par-
ticular, mouse models have offered unique op-
portunities with which to examine detailed
pathologic features of this disease. However,
the availability of clinical studies and numerous
mouse models of asthma have not led to an
unambiguous description of eosinophil ef-
fector functions in asthma, and questions
remain as to the specific role(s), if any, of
these leukocytes (6).
A line of mice devoid of eosinophils was
created to test hypotheses that link eosinophils
and asthma-related pathogenesis. Transgenic
mice devoid of eosinophils were created by lin-
eage-specific expression of a cytocidal protein
R EPORTS
www.sciencemag.org SCIENCE VOL 305 17 SEPTEMBER 2004 1773
Downloaded from https://www.science.org at Louisiana State University on October 06, 2021
with a promoter fragment identified from studies
of secondary granule protein genes expressed in
mouse eosinophils (7–11). A candidate promoter
from the gene for eosinophil peroxidase (EPO)
was selected on the basis of transfection studies
with EPO promoter–luciferase reporter con-
structs in the eosinophilic cell line AML14.3D10
(12). These studies revealed that upstream se-
quences from the mouse EPO gene were capable
of supporting high-level expression that was
unique to eosinophil lineage–committed cells
(fig. S1). In addition to mouse EPO-derived
sequences, the transgenic construct developed
included the diphtheria toxin A (DTA) chain
open reading frame (13). The cytocidal charac-
ter of diphtheria toxin is mediated by the DTA
chain (the B chain provides entry into eukary-
otic cells) through the catalytic degradation of
elongation factor-2 and the subsequent col-
lapse of protein synthesis (14).
Assessment of circulating leukocytes (15)in
the resulting EPO-DTA transgenic (PHIL) mice
demonstrated that these animals were devoid of
eosinophils but otherwise have a full comple-
ment of hematopoietically derived cells (Fig.
1A). An examination of splenic lymphoid cells
(15) revealed normal numbers of B cells, T cells,
and the T lymphocyte CD4
⫹
and CD8
⫹
sub-
types (Fig. 1B). Assessments of lung sections
and peritoneal cavity exudates from PHIL mice
revealed wild-type levels of mast cells (fig. S2,
A and B). Moreover, circulating basophils were
identified in peripheral blood from PHIL mice
(fig. S2C), demonstrating that even a leukocyte
lineage sharing a direct common precursor with
the eosinophil lineage was unaffected. The spe-
cific ablation of eosinophils in PHIL mice also
occurred with no effects on either erythropoiesis
or the production of platelets (Fig. 1C). A nom-
inal elevation of total white blood cell counts
was consistently observed in PHIL mice relative
to negative littermates. This increase, however,
was not specific to any one cell type and did not
elevate circulating cell numbers beyond the nor-
mal observable range in wild-type mice.
The loss of eosinophils in PHIL mice was
nearly absolute, with only an occasional eosino-
phil identified in surveys of blood films from 1
of 20 animals examined. This eosinophil defi-
ciency is lifelong and a Mendelian inheritable
trait of the line. The specificity of the eosinophil
deficiency in PHIL mice was achieved through a
cross with interleukin (IL)–5 transgenic animals
(16). These IL-5 transgenic mice have circulat-
ing eosinophil levels that, in some cases, exceed
100,000 per mm
3
of blood, representing ⬃50%
of all white blood cells. Analyses of blood from
double transgenic animals (i.e., mice carrying
both the DTA and IL-5 transgenes) again revealed
a complete absence of eosinophils (Fig. 1D).
The eosinophil-deficient character of
PHIL mice was extended further by immu-
nohistochemistry with antibodies specific for
Major Basic Protein (MBP) (15, 17, 18).
Tissues with abundant resident populations of
eosinophils (i.e., bone marrow, uterus, small
1
Division of Pulmonary Medicine,
2
Department of
Biochemistry and Molecular Biology,
3
Division of
Hematology/Oncology, Mayo Clinic Arizona,
Scottsdale, AZ 85259, USA.
4
Vermont Lung Center,
Department of Medicine, University of Vermont,
Burlington, VT 05405, USA.
5
Department of Bio-
chemistry and Molecular Biology, University of Il-
linois, College of Medicine, Chicago, IL 60612, USA.
*To whom correspondence should be addressed. E-mail:
jjlee@mayo.edu (J.J.L.) and nlee@mayo.edu (N.A.L.)
†Present address: The EAR Foundation of Arizona,
Phoenix, AZ 85008, USA.
‡Present address: Department of Biology, Louisiana
State University, Baton Rouge, LA 70803, USA.
Fig. 1. The eosinophil deficiency of PHIL mice is specific and definitive. (A) Peripheral blood of PHIL
mice is devoid of eosinophils without effects on the composition of other leukocytes (mean ⫾ SE,
n ⫽ 17 animals per group). (B) The targeted loss of eosinophils had no effects on lymphocyte
subtypes (n ⫽ 5 animals per group). (C) The specific ablation of eosinophil lineage–committed cells
had no additional effects on other hematopoietic parameters, although a nonspecific marginal
increase in the steady-state levels of total circulating white blood cells was observed. (D)
Fluorescence-activated cell sorting analyses demonstrated that the marked blood eosinophilia (i.e.,
the presence of CCR3
⫹
cells) of the IL-5 transgenic line NJ.1726 (16) was completely abolished in
NJ.1726/PHIL double transgenic mice. PE, phycoerythrin. (E) Immunohistochemistry (dark purple–
stained cells) with eosinophil-specific rabbit polyclonal antisera to MBP demonstrates that tissues
or organs with prominent resident populations of eosinophils at baseline in wild-type mice were
devoid of these granulocytes in PHIL mice. Scale bar, 100 m.
R EPORTS
17 SEPTEMBER 2004 VOL 305 SCIENCE www.sciencemag.org1774
Downloaded from https://www.science.org at Louisiana State University on October 06, 2021
intestines, and thymus) in wild-type animals
were shown to be devoid of these granulo-
cytes in PHIL mice (Fig. 1E).
PHIL mice were subjected to an acute aller-
gen sensitization/aerosol challenge model of
asthma (15) to determine if the presence of
eosinophils was causatively linked to the devel-
opment of disease symptoms. Whereas wild-
type mice sensitized/aerosol challenged with
chicken ovalbumin (OVA) developed a signifi-
cant airway eosinophilia [⬃50% of bronchoal-
veolar lavage fluid (BAL) cells], PHIL mice
were essentially devoid of eosinophils, with only
trace numbers (⬍0.5%) of eosinophils identifi-
able in the BAL of one of four OVA-treated
animals (Fig. 2A). The loss of eosinophils from
the lungs of OVA-treated PHIL mice also ex-
tended to tissue-infiltrating cells. Specifically,
the lungs of OVA-treated PHIL mice were de-
void of eosinophils, unlike the significant tissue
eosinophilia that occurred in the areas surround-
ing the central airways (peribronchial) and the
vasculature (perivascular) of OVA-treated wild-
type animals (Fig. 2, B to D, and fig. S3).
Examination of blood films and bone marrow
smears from OVA-treated PHIL animals again
revealed only an occasional eosinophil in a
fraction of the animals examined.
The targeted ablation of eosinophils had sig-
nificant effects on allergen-induced pulmonary
pathology, suggesting a causative role for these
granulocytes. Overall, OVA-induced histopa-
thology in PHIL mice was attenuated relative to
OVA-treated wild-type littermates. This lack of
pathology was manifested by the reduced airway
epithelial hypertrophy in OVA-treated PHIL
mice (Fig. 2, B and C). In addition, assessment
of airway mucins by periodic acid–Schiff (PAS)
staining (15) demonstrated that OVA-induced
goblet cell metaplasia/mucus accumulation
(GM/MA) in PHIL mice was significantly re-
duced (Fig. 3, A to C). A quantitative assessment
of the staining in these tissue sections revealed a
68% reduction of PAS staining in PHIL mice
relative to OVA-treated wild-type animals (Fig.
3D). However, the GM/MA observed in OVA-
treated PHIL mice was still significant when
compared to allergen-naive animals. This obser-
vation suggests that eosinophils contribute to,
and are necessary for, the levels of pathology
observed in wild-type mice, but that they are not
alone sufficient to account for these wild-type
levels. That is, both eosinophil-dependent and
-independent mechanisms exist in the lung that
elicit GM/MA after allergen challenge.
The association between allergen-induced
pulmonary eosinophilia and the development of
lung dysfunction in both asthma patients and
mouse models has been, at best, a collection of
confusing and often contradictory observations.
The lack of symptom improvement in asthma
patients after administration of antibodies to IL-5
exemplifies the ambiguous character of clinical
studies that attempt to ablate eosinophils (6).
Mouse models purporting to ablate eosinophils
Fig. 2. The pulmonary eosinophilia associated with OVA sensitization/aerosol challenge was abolished
in PHIL mice. (A) OVA-induced eosinophilia of the airway lumen was lost (decreased by ⬎99%) in PHIL
mice (mean ⫾ SE, n ⫽ 5 animals per group). *, P ⬍ 0.001. (B and C) Assessments of infiltrating
eosinophils in (B) wild-type and (C) PHIL mice by immunohistochemistry (dark purple–stained cells)
with rabbit polyclonal antisera to mouse MBP revealed that OVA-induced accumulation of eosinophils
was also extinguished in PHIL mice. Scale bar, 100 m. (D) Quantitative assessments of the number of
eosinophils infiltrating peribronchial areas (i.e., eosinophils per mm
2
) demonstrated that OVA-treated
PHIL mice were devoid of tissue eosinophils (mean ⫾ SE, n ⫽ 5 animals per group). Ø indicates the
absence of eosinophils in any of the sections of any of the mice in the cohort examined.
Fig. 3. The specific loss of eosinophils in PHIL mice resulted in a significant reduction in OVA-induced
GM/MA. Representative lung sections after PAS staining are shown for (A) saline control and (B) OVA
sensitized/OVA aerosol challenged wild-type mice in comparison to (C) OVA sensitized/OVA aerosol
challenged PHIL mice. The sections of each panel show early-branching central conducting airways,
whereas the insets show smaller, more distal bronchioles. Scale bars, 100 m. (D) Quantitative
assessments of airway epithelial mucus content showed a marked decrease (relative to wild type) in
PHIL mice (mean ⫾ SE, 5 to 10 animals per group). All evaluations of histopathology were performed
in duplicate as independent observer-blinded assessments. *, P ⬍ 0.05.
R EPORTS
www.sciencemag.org SCIENCE VOL 305 17 SEPTEMBER 2004 1775
Downloaded from https://www.science.org at Louisiana State University on October 06, 2021
