Professional Documents
Culture Documents
Keng-Chang Tsai, Chen-Wei Chiang, Yan-Ni Lo, Fu-Ling Chang, Tsai-Yu Lin,
Chang-Yu Chang, Wang-Chuan Chen & Yu-Ching Lee
To cite this article: Keng-Chang Tsai, Chen-Wei Chiang, Yan-Ni Lo, Fu-Ling Chang, Tsai-Yu Lin,
Chang-Yu Chang, Wang-Chuan Chen & Yu-Ching Lee (2018): Generation and characterization
of avian-derived anti-human CD19 single chain fragment antibodies, Animal Biotechnology, DOI:
10.1080/10495398.2018.1486323
Article views: 4
ORIGINAL ARTICLE
ABSTRACT KEYWORDS
The human cluster of differentiation 19 (CD19) is highly expressed in most leukemia, rendering Human cluster of
is a promising therapeutic target. In this study, we generated anti-CD19 single-chain variable differentiation 19; leukemia;
fragments (scFv) from immunized chickens by phage display technology. After constructing a single chain variable
fragment; phage display
scFv antibody library with 2.5 108 compositional diversity for panning, one representative technology; molecu-
scFv clone S2 which can specifically recognize to the CD19 protein was isolated and character- lar modeling
ized. The binding reactivity of the scFv S2 to the endogenous CD19 protein of the ARH-77
leukemia cancer cell was verified through flow cytometry and the binding affinity of scFv S2
is 6.9 108 M determined by the surface plasmon resonance system. Compared with the
chicken germline, hyper mutation in the complementarity-determining regions (CDRs) sug-
gested that scFv S2 could be generated through an antigen-driven humoral response. By
molecular modeling, the possible CDR configurations of scFv S2 were constructed rationally.
Furthermore, the characteristics of chicken antibodies of a protein database were investigated.
The findings in this study contribute to antibody development and engineering because they
reveal the geometric structures and properties of the CDRs in chicken antibodies.
CONTACT Yu-Ching Lee ycl@tmu.edu.tw Research Center of Translational Medicine, Taipei Medical University, #250 Wu-Hsing Street, Taipei
110, Taiwan.
Supplemental data for this article can be accessed here.
These authors contributed equally to this work.
ß 2018 Taylor & Francis Group, LLC.
2 K.-C. TSAI ET AL.
through phage display technology for generating University. The chicken was immunized with 10 lg
monoclonal antibodies.11 Numerous antibodies gener- recombinant CD19 protein (Sino Biological Inc.,
ated through this novel technique are currently China.) with an equal volume of Freund’s complete
being developed for target therapy.12,13 Traditional adjuvant through intramuscular injection. Three add-
approaches (such as hybridoma) for generating mono- itional immunizations were carried out at intervals of 7
clonal antibodies are complicated and time consuming days. After final immunization, IgY antibodies in egg
and have numerous difficulties.14 Alternatively, the yolks were purified and titrated through an enzyme-
phage display system is a safe and effective in vitro linked immunosorbent assay (ELISA) to determine the
procedure for enriching specific antibodies from com- presence of a humoral immune response. The purifica-
plicated phage antibody libraries.15 In general, small tion of IgY from the egg yolks was in accordance with
mammals such as mouse and rabbit are often used a published protocol.20,21 The purified IgY antibodies
because of their convenient operation. However, it is from each egg were dissolved in 5 mL of Tris-buffered
frequently to develop tolerance with highly conserved saline containing 0.05% sodium azide and stored
proteins of mammals during immunization due to at –20 C.
phylogenetic relationship. This issue limits experimen-
tal options and hinders the progress of research. To
solve this problem, the chicken is a suitable host choice Construction of chicken scFv library
because it can be used to induce stronger humoral and biopanning
responses based on species differences. The chicken A phage library displaying the scFv antibody was con-
has only a lambda light-chain isotype, compared with structed according to published protocols with minor
two light-chain isotypes (kappa and lambda) in mam- modifications.16,19 Briefly, chicken spleens were har-
mals, facilitating the construction of an antibody vested 7 days after the final immunization. After hom-
library for efficient binder screening.16 Moreover, sin- ogenization, total RNA was extracted and reverse
gle-chain variable fragment (scFv) antibodies based on transcribed into the first-strand cDNA by using a
chicken Ig frameworks often exhibit high expression SuperScript RT kit (Invitrogen, USA). Chicken scFv
levels in simple bacterial cultures.17 Therefore, lots of primers and PCR conditions were provided in the
studies have examined the various applications of supplementary files. After amplification using chicken-
chicken-derived scFv antibodies.18,19 Although the specific primers, gene products of heavy-chain and
humanization of chicken scFv antibodies is still light-chain variable (VH and VL, respectively) regions
required for potential therapeutic use in the future, this were subjected to a second round of polymerase chain
obstacle could be overcome through antibody reaction (PCR) with a linker to form full-length scFv
engineering. molecules, which were cloned further into the
We conducted this study to generate specific anti- pComb3X vector.16 The recombinant DNA was trans-
CD19 chicken-derived scFv antibodies, with a focus formed into the Escherichia coli ER2738 strain through
on investigating the complementarity-determining electroporation, and the VCS-M13 (Stratagene, USA)
region (CDR) structure within the most acceptable helper phage was then added to initiate recombinant
scaffold through molecule modeling. Moreover, the phage production. The total phage library was precipi-
characteristics of antibody sequences by the published tated with 4% polyethylglycol 8000 and 3% NaCl (w/v)
chicken antibody structure were explored. The results and resuspended in phosphate-buffered saline (PBS)
from this study can provide an improved understand- containing 1% bovine serum albumin (BSA) and stored
ing of the chicken scFv antibodies, the possible appli- at 4 C.
cations of which can be explored in future research. For biopanning, 1011 recombinant phage particles
from the constructed scFv antibody library were
Materials and methods added to a microtiter plate precoated with a recom-
binant CD19 protein (1 lg/well) and incubated at
Chicken immunization room temperature for 2 h. Next, unbound phage par-
Female white leghorn (Gallus domesticus) chickens ticles were removed and the well was washed with
were maintained carefully in the animal facility of the PBST (PBS with 0.05% Tween 20) for 10 times.
Taipei Medical University. All of the animal experi- Bound phage particles were eluted with 0.1 M HCl/gly-
ments followed ethical standards, with the protocols cine (pH 2.2)/0.1% BSA elution buffer, neutralized
having been reviewed and approved by the Animal with 2 M Tris base buffer, and immediately used to
Use and Management Committee of Taipei Medical infect the E. coli ER2738 strain for recombinant phage
ANIMAL BIOTECHNOLOGY 3
Antibody Framework module from Discovery Studio gene fragments was performed using two consecutive
v. 2017 (BIOVIA Inc., USA). The lowest PDF total PCR steps (Supplemental Fig. 1). After the M13 helper
energy (probability density function energy) was phage was used for rescue, a high-quality phage dis-
employed to select a reliable scFv S2 model among playing scFv library was constructed; the library com-
200 models. plexity was determined to be 2 108. To evaluate the
ratio of clones with correct inserts in the constructed
Results library, 30 clones of the library were selected to con-
firm the presence of a 750 bp fragment containing
Characterization of humoral IgY response in both heavy and light chain genes by using colony
the chicken PCR. The result revealed that 93% (28/30) of the
The recombinant CD19 protein used for immuniza- clones selected from the library had correct gene
tion was analyzed through SDS-PAGE, revealing an inserts (data not shown). Next, we performed four
approximate molecular weight of 47 kDa due to glyco- rounds of panning to eliminate unwanted clones and
sylation (Fig. 1a, lane CD19). The purified IgY anti- to enrich specific binders against CD19. As shown in
bodies from the egg yolks were tested for their Fig. 2a, eluted phage titer of the final round was over
binding activity to CD19 immobilized on nitrocellu- one order higher than that of the first round of pan-
lose membranes or ELISA plate wells. Through ning. The composition of the antibodies in the ampli-
Western blotting, polyclonal anti-CD19 IgY antibodies fied library was varied after panning. Moreover, we
obtained after the fourth immunization substantially confirmed the binding reactivity against CD19 of the
bound to CD19. By contrast, the IgY antibodies amplified phage form in each round of panning
obtained from the pre-immunized chicken egg showed through phage ELISA (Fig. 2b). Notably, compared
no reaction (Fig. 1a). In the ELISA, the IgY antibodies with the original library, the binding signal increased
prepared from the egg yolks after the fourth immun- with the number of panning cycles.
ization specifically bound to CD19 but not to BSA,
even when titered at a high concentration (Fig. 1b). Characterization of anti-CD19 chicken scFv clones
The result suggests that a strong humoral antibody
response was elicited in the chicken host. After panning, we randomly selected 14 clones for
single colony analysis. The expression levels of IPTG-
induced scFv clones were tested for binding activity to
Construction of scFv antibody library the CD19 recombinant protein. These selected scFv
and biopanning
clones showed significant binding activity to CD19
The immunized chicken was sacrificed and total RNA but not to BSA (Fig. 3a). Through sequencing, we
was extracted from the spleen cells to construct the determined the nucleotide sequences of the variable
antibody library. The amplification of full-length scFv regions of heavy and light chain genes of the 14 clones.
Figure 1. Humoral antibody response in chicken after immunization. (a) Recombinant CD19 protein was visualized using
Coomassie blue staining or transferred onto a nitrocellulose membrane, which was subsequently probed with polyclonal IgY puri-
fied from chicken egg before immunization and after forth immunization. Arrows indicate the recombinant CD19 molecular weight
at 47 kDa. (b) Humoral IgY responses in chicken egg after the fourth immunization analyzed through ELISA. A series of diluted IgY
antibodies were examined for their binding activity to CD19 (solid bars) or BSA (open bars).
ANIMAL BIOTECHNOLOGY 5
The results revealed that 13 clones shared identical than the expected monomer form (25 kDa). One rep-
heavy and light chain genes, except for clone S4. resentative scFv S2 was expressed and purified for fur-
Corresponding to the sequencing results, 13 identical ther experiments. The purified scFv S2 was analyzed
scFv clones were clearly detected through Western by SDS-PAGE to confirm the purity as well
blotting, showing consistent patterns (Fig. 3b). (Supplemental Fig. 2).
Because the scFv linker was designed to seven residues We also examined the recognition of scFv S2 to
(GGSSRSS), a short linker caused the scFv expression recombinant CD19. The results showed that scFv S2
trend to be a dimer form (50 kDa), which was more could clearly recognize the CD19 protein at an
Figure 2. Evaluation of panning process. (a) Eluted phage numbers were monitored after each round of panning against CD19.
Wild-type M13 phage was used in the panning process as a negative control. (b) The amplified phage library after each round of
panning was used to confirm the binding reactivity by using phage ELISA. Ori denotes original phage library before the pan-
ning process.
Figure 3. Characterization of isolated anti-CD19 scFvs. (a) Binding activity of randomly selected scFvs expressed in the cell lysate
to CD19 was examined using ELISA. (b) Western blotting of selected scFv expression. Arrows indicate the scFv molecular weight of
monomer or dimer forms. (c) Recombinant CD19 protein indicated with arrows were visualized using Coomassie blue staining or
detected using a representative scFv S2 through Western blotting. (d) Binding activity of scFv S2 to endogenous CD19 expressed
on the ARH77 leukemia cell membrane was analyzed through flow cytometry. PC denotes commercial anti-CD19 polyclonal anti-
bodies used in the test as a positive control. NC denotes an irrelevant scFv without a binding signal.
6 K.-C. TSAI ET AL.
expected molecular weight (Fig. 3c). By using flow scFv S2. The extra amino acids in the CDR-L1 loop
cytometry to test the binding reactivity of scFv to are one glycine, one arginine, and two tyrosines,
endogenous human CD19 intrinsically expressed on whereas the CDR-L3 loop containe one aspartic
the membrane of ARH77 leukemia cells, we observed amino acid. The lengths of the other CDR loops are
that scFv S2 showed significant binding signals to the same and the CDR-H1 sequence is identical to
ARH77 tumor cells; we noted no binding activity scFv S2. Comparing the constructed structure of scFv
when we applied an irreverent scFv (Fig. 3d). A sur- S2 with scFv 4P48 reveal a shorter CDR-L1 (Fig. 5b
face plasmon resonance (SPR) system was used for and c), and the variability of CDR-L1 is similar to
the assay to measure the binding affinity of scFv S2 to that of CDR-H3 having high tolerance to display on
the CD19 molecules (Supplemental Fig. 3). The KD the chicken antibody.
value of scFv S2 was 6.9 108 M, and the affinity
constants are presented in the Table 1. By nature rep-
ertoire after immunization, the affinity at pM level of Discussion
isolated scFv is generally acceptable. The result is Different from the traditional hybridoma approach,
similar to previous studies.23,24 the phage display system provides a molecular bio-
logical method for rapidly and conveniently preparing
ScFv gene sequence analysis monoclonal antibodies. However, when the scFv gene
is assembled in a library, the combination of heavy
The nucleotide sequences of scFv S2 were aligned to chains and light chains is random, which may affect
the germline gene sequences of chicken Ig. As shown the binding activity of the displayed scFv because of
in Fig. 4, the overall mutation rates as compared with the unsuitable structure. Nevertheless, such problems
the germline gene sequences were 18.3% and 23.1% in
can be ameliorated through affinity maturation by
VL and VH regions, respectively. The mutation rate
using antibody engineering to enhance antibody avid-
in the CDRs was higher than that in the framework
ity, which has been widely used in the development of
regions (FRs). Insertion mutations occurred in the
many clinical antibody drugs. In this study, we used a
heavy chain CDR3.
short linker (7 aa) in the structure of scFv, which
caused the scFv expression to become a dimer form
Homology modeling of scFv S2 structure (Fig. 3b) to improve the selection efficiency based on
The structure of CDR loops of the chicken antibody the binding affinity of the panning process. The bind-
scFv S2 was constructed using homology modeling ing affinity of scFv S2 identified through the SPR sys-
strategies. Based on three x-ray crystallography tem was 69 pM. The structural change caused by the
chicken antibody structures that have been published, linker contribute to enhance the binding ability of
4P48 (PDB code) is a chicken anticardiac Troponin I scFv. In Fig. 3b, scFv clone S4 has a molecular weight
scFv that is most similar to the scFv S2 framework. larger than other clones in dimer form. After sequenc-
As shown in Fig. 5a, the lengths of the CDR-L1 and ing, the length of VH and VL genes of scFv S4 was
CDR-L2 loops of scFv 4P48 are greater than those of confirmed to be general. The aggregation of scFv is
often occurring via the domain-swapping manner.
Table 1. kon and koff rate constants of chicken scFv S2 target- Different antibody sequence associated with the vari-
ing to CD19 protein. able domain interface is one of the factors leading to
Ligand Analyte (scFv) kon (102 M1 S1) koff (105 S1) KD (108 M) aggregation or multimer expression.25 As scFv S2 and
CD19 S2 6.32 ± 0.00317 4.39 ± 0.00623 6.94 ± 0.0718
S4 have been heated before SDS-PAGE analysis, the
Figure 4. Sequence analysis of VL and VH genes of the scFv S2 antibody. VH and VL nucleotide sequences were determined and
translated into amino acid sequences to be aligned with the chicken germline gene. Sequence gaps were introduced to maximize
the alignment; these are indicated by blank spaces. Dots indicate the consensus sequences. FR and CDR boundaries are indicated
above the germline sequences.
ANIMAL BIOTECHNOLOGY 7
aggregation of scFvs were dissociated to monomer To understand the differences in sequence between
(Supplemental Fig. 4). the isolated chicken antibody (scFv S2) and human
Compared with the result obtained from the first antibodies, we conducted a preference analysis using
panning process (using the original library as an input three published chicken antibody structures (4P48,
phage), the eluted phage number obtained from the 4P49 and 4GLR) and 10 representative human anti-
fourth panning process increased by 40 times, demon- body structures (3G6A, 3NH7, 4QHU, 4QCI, 4OD2,
strating the enrichment of the specific antibody clone 4QF1, 8FAB, 4HPY, 5BV7 and 1W72) from the PDB,
in the library. Moreover, the eluted phage number which all have high similarity to the framework
from the second to the fourth panning showed an sequence of scFv S2 (Fig. 6a and b). The most signifi-
obvious upward trend, also suggesting successful selec- cant difference between the chicken and human anti-
tion. Similar results have been reported in our previ- body sequences was determined to be in the variable
ous studies and other studies.26,27 Based on the region of the light chain. The length of the CDR-L1
differences of species, the chicken tends to have more of chickens was noted to be highly variable compared
somatic mutations after immunization with proteins with human CDR-L1, which has a fixed and con-
of mammalians. Notably, the somatic mutation of served length. This may be explained by this feature
scFv clones is mainly from amino acid insertion (Fig. of chicken antibodies to be able to have more effective
4) compared with the germline of chickens. This abilities. Notably, the chicken framework was observed
means that somatic mutation causes high variability. to have one more amino acid of serine or alanine
The greater length of CDR3-H3 is the basis for the than the human sequence has at the position of 45A
variable CDR, which has the ability to enter deeper (based on the numbering in the international
antigenic epitope regions. ImMunoGeneTics information system). The difference
Figure 5. Homology modeling of scFv S2 structure. (a) Multiple sequence alignment of variable domains of the light and heavy
chains of 4G48, 4P49 and 4GLR with the scFv S2 antibody. The background of the residues is colored according to sequence simi-
larity. The deep color shows the conserved residues in all sequences. The color scheme is from deep to light, corresponding to
identity, highly conserved residues, and low-conserved residues, respectively. The residue backgrounds colored in white are not
similar. Residues of the CDRs are indicated by bold letters with arrow marks. (b) PDB id: 4P48 was used as the major structural
template with the highest sequence similarity to construct the (c) scFv S2 scaffold model.
8 K.-C. TSAI ET AL.
Figure 6. Sequence preference analysis of published chicken and human antibody with scFv S2. Multiple sequence alignment of
variable domains of the light chains (a) and the heavy chains (b) of chicken antibodies (PDB id: 4G48, 4P49 and 4GLR) and the
representative nine human antibodies (PDB id: 3G6A, 3NH7, 4QHU, 4QCI, 40D2, 4QF1, 8FAB, 4HPY and 1W72) with the scFv S2
chicken antibody. All amino acids of sequences are renumbered using IMGT schemes. The background of the residues is colored
according to sequence similarity. The deep color shows the conserved residues in all sequences. The color scheme is from deep to
light, corresponding to identity, highly conserved residues, and low-conserved residues, respectively. The residue backgrounds col-
ored in white are not similar. Residues of the CDRs are indicated by bold letters with arrow marks.
in the position of chicken antibodies may be related CDRL1 and CDRH3. Overall, this information affords
to the complexity tolerance of CDR-L1. The chicken a clearer understanding of the characteristics of
and human characteristics of the VH region were chicken antibodies and can thus assist us by this way
determined to be similar, with all of them exhibiting a to develop antibody drugs in the future. Such an
high degree of CDR-H3 diversity. Therefore, chicken immune strategy can be used to induce highly specific
antibodies can induce effective somatic mutations on antibodies, and through antibody engineering, can
ANIMAL BIOTECHNOLOGY 9
improve and develop the possibility of clinical 12. Rahbarnia L, Farajnia S, Babaei H, et al. Evolution of
applications. phage display technology: from discovery to applica-
tion. J Drug Target. 2017;25:216–224.
13. Shim H. Therapeutic antibodies by phage display.
Disclosure statement Curr Pharm Des. 2016;22:6538–6559.
14. Groves DJ, Morris BA. Veterinary sources of nonro-
No potential conflict of interest was reported by dent monoclonal antibodies: interspecific and intra-
the authors. specific hybridomas. Hybridoma. 2000;19:201–214.
15. Barbas 3rd CF, Kang AS, Lerner RA, Benkovic SJ.
Assembly of combinatorial antibody libraries on
Funding phage surfaces: the gene III site. Proc Natl Acad Sci
This work was financially supported by the “TMU Research USA. 1991;88:7978–7982.
Center of Cancer Translational Medicine” from The 16. Andris-Widhopf J, Rader C, Steinberger P, Fuller R,
Featured Areas Research Center Program within the frame- Barbas 3rd CF,. Methods for the generation of
work of the Higher Education Sprout Project by the chicken monoclonal antibody fragments by phage dis-
Ministry of Education (MOE) in Taiwan; Ministry of play. J Immunol Methods. 2000;242:159–181.
Science and Technology in Taiwan under Grant [MOST 17. Finlay WJ, Shaw I, Reilly JP, Kane M. Generation of
105-2628-B-038-007-MY3]. high-affinity chicken single-chain Fv antibody frag-
ments for measurement of the Pseudonitzschia pun-
gens toxin domoic acid. Appl Environ Microbiol.
References 2006;72:3343–3349.
18. Li C, He J, Ren H, Zhang X, Du E, Li X. Preparation
1. Carter RH, Barrington RA. Signaling by the CD19/ of a chicken scFv to analyze gentamicin residue in
CD21 complex on B cells. Curr Dir Autoimmun. animal derived food products. Anal Chem.
2004;7:4–32. 2016;88:4092–4098.
2. Tedder TF. CD19: a promising B cell target for 19. Lee YC, Chen WC, Liang MH, et al. Single chain
rheumatoid arthritis. Nat Rev Rheumatol. 2009;5: antibody fragment with serine protease inhibitory
572–577. property capable of neutralizing toxicity of
3. Cooper LJ, Al-Kadhimi Z, DiGiusto D, et al. Trimeresurus mucrosquamatus venom. Biochem
Development and application of CD19-specific T cells Biophys Res Commun. 2015;460:170–176.
for adoptive immunotherapy of B cell malignancies. 20. Akita EM, Nakai S. Production and purification of
Blood Cells Mol Dis. 2004;33:83–89. Fab’ fragments from chicken egg yolk immunoglobu-
4. Ginaldi L, De Martinis M, Matutes E, Farahat N, lin Y (IgY). J Immunol Methods. 1993;162:155–164.
Morilla R, Catovsky D. Levels of expression of CD19 21. Akita EM, Nakai S. Comparison of four purification
and CD20 in chronic B cell leukaemias. J Clin Pathol. methods for the production of immunoglobulins from
1998;51:364–369. eggs laid by hens immunized with an enterotoxigenic
5. Yang W, Agrawal N, Patel J, et al. Diminished expres- E. coli strain. J Immunol Methods. 1993;160:207–214.
sion of CD19 in B-cell lymphomas. Cytometry B Clin 22. Fairhead M, Howarth M. Site-specific biotinylation of
Cytom. 2005;63:28–35. purified proteins using BirA. Methods Mol Biol.
6. Mills DM, Stolpa JC, Cambier JC. Modulation of 2015;1266:171–184.
MHC class II signal transduction by CD19. Adv Exp 23. Pelat T, Hust M, Hale M, Lefranc MP, Dubel S,
Med Biol. 2007;596:139–148. Thullier P. Isolation of a human-like antibody frag-
7. Wu J, Fu J, Zhang M, Liu D. Blinatumomab: a bispe- ment (scFv) that neutralizes ricin biological activity.
cific T cell engager (BiTE) antibody against CD19/ BMC Biotechnol. 2009;9:60.
CD3 for refractory acute lymphoid leukemia. J 24. Huang L, Su X, Federoff HJ. Single-chain fragment
Hematol Oncol. 2015;8:104. variable passive immunotherapies for neurodegenera-
8. Zhu M, Wu B, Brandl C, et al. Blinatumomab, a bis- tive diseases. Int J Mol Sci. 2013;14:19109–19127.
pecific T-cell Engager (BiTE(V R )) for CD-19 targeted 25. Arndt KM, Muller KM, Pluckthun A. Factors influ-
cancer immunotherapy: clinical pharmacology and its encing the dimer to monomer transition of an anti-
implications. Clin Pharmacokinet. 2016;55:1271–1288. body single-chain Fv fragment. Biochemistry.
9. Sommermeyer D, Hill T, Shamah SM, et al. Fully 1998;37(37):12918–12926.
human CD19-specific chimeric antigen receptors for 26. Lee CH, Lee YC, Leu SJ, et al. Production and charac-
T-cell therapy. Leukemia. 2017;31:2191–2199. terization of neutralizing antibodies against Bungarus
10. Turtle CJ, Hanafi LA, Berger C, et al. CD19 CAR-T multicinctus snake venom. Appl Environ Microbiol.
cells of defined CD4þ:CD8þ composition in adult B 2016;82(23):6973–6982.
cell ALL patients. J Clin Invest. 2016;126:2123–2138. 27. Lee YC, Tsai KC, Leu SJ, Wang TJ, Liu CY, Yang YY.
11. Huse WD, Sastry L, Iverson SA, et al. Generation of a Isolation, characterization, and molecular modeling of
large combinatorial library of the immunoglobulin a rheumatoid factor from a Hepatitis C virus infected
repertoire in phage lambda. Science. 1989;246: patient with Sjogren’s syndrome. ScientificWorld
1275–1281. Journal. 2013;2013:516516.