Download Amino acid sequence of an immunogenic corneal stromal protein.

Investigative Ophthalmology & Visual Science, April 1996, Vol. 37, No. 5
944
3 to 30 mm Hg, the Tono-pen XL and the pneumatonometer exhibit similar accuracy, but the variance of
the Tono-pen XL is smaller, making it our tonometer
of choice. Although all three instruments are not as
reliable as one would like, they do play a role in research, particularly when the main objective is to detect a change in IOP rather than an absolute value;
in this case, it would not be necessary to convert to
actual IOP. The high degree of inaccuracy at very high
IOPs demands that caution be used when evaluating
IOP data in rabbits.
Key Words
intraocular pressure, open stopcock, closed stopcock, rabbits, tonometry, variance
References
1. Vareilles P, Conquet P, LeDouarec JC. A method for
the routine intraocular pressure (IOP) measurement
in the rabbit: Range of IOP variations in this species.
ExpEyeRes. 1977; 24:369-375.
2. Katz RS, Henkind P, Weitzman ED. The circadian
rhythm of the intraocular pressure in the New Zealand
White rabbit. Invest Ophthalmol. 1975; 14:775-780.
Amino Acid Sequence of an
Immunogenic Corneal Stromal
Protein
Sammy H. Liu and John D. Gottsch
Purpose. A unique cornea-associated antigen (CO-Ag)
has been purified previously from stromal extracts. The
protein is the target for autoantibodies in patients with
Mooren's ulcer. In this study, the amino acid sequence
of CO-Ag was analyzed and the structure-function properties of CO-Ag was determined.
Methods. Purified CO-Ag was subjected to N-terminal
sequencing by automated Edman degradation. Binding
of calcium (Ca2+) to CO-Ag was measured by a direct
45
Ca2+-binding assay.
Results. The complete amino acid sequence of CO-Ag
has been determined. The protein contains 70 amino
acids in a single chain and lacks cysteine, tryptophan,
and methionine residues. A computerized data base
From the Wilmer Ophthalmobgical Institute, Johns Hopkins University School of
Medicine, Baltimore, Maryland.
Supported in part by the Ginger Gomprecht Research Fund, Baltimore, MD.
Submitted for publication August 28, 1995; revised December 14, 1995; accepted
January 4, 1996.
Proprietary interest category: N.
Reprint requests: John D. Gottsch, Wilmer Eye Institute, Johns Hopkins Hospital,
600 N. Wolfe Street, Baltimore, MD 21205.
3. Neault TR, Cooke D, Brubaker RF. Modification and
calibration of the Bigliano-Webb tonometer for improved accuracy of tonometry in rabbits. CurrEye Res.
1988;8:9-15.
4. Hammond BR, Bhattacherjee P. Calibration of the
Alcon applanation pneumatonograph and Perkins tonometer for use in rabbits and cats. Curr Eye Res.
1984;3:1155-1158.
5. Brubaker RF. Tonometry. In: Duane's Clinical Ophthalmology. Vol. 3. Philadelphia: JB Lippincott; 1993:1-7.
6. Whitacre MM, Emig M, Hassanein K. The effect of
Perkins, Tono-Pen, and Schiotz tonometry on intraocular pressure. Am J Ophthalmol. 1991; 111:59-64.
7. Snedecor GW, Cochran WG. Statistical Methods. 6th
ed. Ames, IA: Iowa State University Press; 1967:195197, 344-348.
8. Moore CG, Milne ST, Morrison JC. Noninvasive measurement of rat intraocular pressure with the Tonopen. Invest Ophthalmol Vis Sci. 1993;34:363-369.
9. Mermoud A, Baerveldt G, Minckler DS, Lee MB, Narsing NA. Intraocular pressure in Lewis rats. Invest Ophthalmol Vis Sci. 1994; 35:2455-2460.
10. Langham ME, Edwards N. A new procedure for the
measurement of the outflow facility in conscious rabbits. ExpEyeRes. 1987; 45:665-672.
search of protein and nucleic acid sequences revealed
strong homology to the Ca2+-binding proteins of the S100 family. The sequence of CO-Ag shows a high homology with calgranulin C (CaG-C) previously purified
from pig granulocytes. The functional Ca2+-binding
sites of CO-Ag and CaG-C were different based on homology with known Ca2+-binding domains and their
Ca2+-binding properties. There are three amino acid
substitutions in the N-terminal Ca2+-binding domain.
Differences were functionally conserved and compatible, with minimum single-base changes in the codon
structures. The greatest difference was located in the Cterminal Ca2+-binding domain. Five consecutive amino
acid changes from Dti3-K-K-G-A67 in CO-Ag to M^-QrDE-Qb7 occurred in CaG-C. These differences alter the
structure of CO-Ag, which no longer can bind Ca2+
ions. The existence of this nonfunctional Ca2+-binding
site was corroborated by its Ca2+-binding properties.
The number of Ca2+-binding sites for the CO-Ag subunit is approximately half that of the CaG-C monomer,
although these two proteins have a similar low binding
constant of approximately 2 X 10~4 M.
Conclusions. These results suggest that CO-Ag is a new
member of the Ca2+-binding protein of the S-100 family
heretofore undescribed in the cornea. Sequence data
provide an important framework to search for sequence
similarity with microbial proteins as possible substrates
for molecular mimicry and for the identification of possible pathogenic epitopes in CO-Ag. Invest Ophthalmol
Vis Sci. 1996; 37:944-948.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 05/08/2017
Reports
.M.ooren's ulcer is a chronic, progressive, painful corneal ulceration. Its cause is unknown, but there is
considerable evidence to suggest that is an autoimmune disease.1"3 Recently, we have purified a corneaassociated antigen (CO-Ag) to study the role of this
antigen in the pathogenesis of Mooren's ulcer.4 Analysis of the CO-Ag amino acid sequence is an essential
prerequisite for understanding its physiological function in the cornea, determining its pathogenic epitopes in the protein molecule, and elucidating the
immunopathogenic mechanisms responsible for
Mooren's ulcer. In this study, we determine the complete amino acid sequence of CO-Ag and analyze the
structure-function properties of this protein.
945
prewashed Centricon 3 microconcentrator (Amicon,
Beverly, MA) with a known amount (5- to 10-fold molar excess) of 45Ca2+ (10 to 14 mCi/mg Ca; Amersham,
Arlington Heights, IL) for 15 minutes at 25°C. The
solution was centrifuged at 10,000 rpm for 25 minutes
in a Beckman ModelJ2-21 centrifuge. Free 45Ca2+ concentration was measured by determining the radioactivity in the filtrate. Each determination was carried
out in triplicate. Binding data were analyzed by the
method of DeH'Angelica et al.8 The amount of Ca2+
bound (mole of Ca2+ bound per mole of CO-Ag) was
plotted against minus logarithm of free Ca2+ concentration where free is the Ca2+ concentration in the
filtrate.
MATERIALS AND METHODS. Protein Sequence
RESULTS. Primary Structure Determination of
Analysis. Automated Edman degradation was performed with an Applied Biosystems (Foster City, CA)
Model 470A gas-phase sequencer connected to an Applied Biosystems Model 120A phenylthiohydantoin
(PTH)-amino acid analyzer and an M900 data system
for on-line analysis of PTH-amino acids. CO-Ag was
purified from bovine corneal stroma as described previously in detail.4 All experiments adhered to the
ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Before application of COAg, filter was treated with polybrene and subjected to
three cycles of treatment with the sequencing reagents
and solvents. CO-Ag (1 nmol) was applied to the
treated filter and then loaded into the cartridge of
the sequencer. The protein sequencing was carried
out by the Johns Hopkins Medical School Peptide/
Protein/DNA Core Facility.
the CO-Ag Protein. When purified CO-Ag was subjected to automated Edman degradation, a single
unique protein of 70 amino acids long was sequenced
(Fig. 1). Quantitation of the PTH-amino acids was
made by peak height measurements and comparison
to corresponding values of PTH-amino acid standards
by an on-line computer system. Plots of the logarithmic yield of PTH-amino acids recovered as a function
of cycle number are demonstrated in Figure 1. The
sequence of CO-Ag was completed after 70 cycles because no amino acid residues were detectable in subsequent steps of Edman degradation. The sequence was
confirmed in two separate determinations. Interesting
features are the absence of cysteine residues, the lack
of intra- and inter-disulfide chains, and the presence
of an excess of 14 basic and 11 acidic residues in the
CO-Ag protein.
Enzymatic Determination of Carboxyl-terminal
Amino Acid Residues. Carboxypeptidase A (Worthington, Freehold, NJ) was used for determination of carboxyl-terminal amino acid sequences. CO-Ag (1 nmol)
was dissolved in 0.1 M, sodium phosphate buffer, pH
8, containing 0.056 M sodium lauryl sulfate,5 and incubated with carboxypeptidase A at 25°C with a substrate
to enzyme ratio of 20:1. Aliquots were removed at time
intervals (e.g., 1, 2, and 3 hours). The reaction was
terminated by adding sufficient acetic acid to lower
the pH to 2.5 to 3. The supernatant was dried in the
vacuum desiccator after centrifugation. Amino acids
present in each aliquot were determined by quantitative amino acid analyzer.
Computerized Homology Search. The program
Blast6 was used to search the European Molecular Biology Organization and GenBank nucleotide data bases
and National Biomedical Research Foundation protein data bases for sequences with similarity to CO-Ag
sequences.
Calcium-Binding Assay. Binding of 45Ca2+ to CO-
Ag was measured according to the method of Mani
and Key.7 CO-Ag (15 to 50 fjM) was incubated in a
Confirmation of Carboxyl-Terminal Sequence As-
signment. Automated Edman degradation sequence
analysis as represented in Figure 1 suggests the entire
length of the CO-Ag protein. The entire protein sequence was confirmed by digestion of the intact protein with carboxypeptidase A. Our previous study4 has
shown that native CO-Ag is a tetramer that can be
dissociated into monomers under denaturing conditions. Carboxypeptidase A is active in sodium lauryl
sulfate solution5 and can thus act under denaturing
conditions in which the CO-Ag monomers are readily
accessible to the enzyme in the carboxyl-terminal sequencing. Addition of carboxypeptidase A to the sample after 1-hour, 2-hour, and 3-hour incubation resulted in the sequential release of phenylalanine, valine, valine, alanine, glycine, and lysine residues (Fig.
2). Results confirm the assignment of the sequence
Lys-Gly-Ala-Val-Val-Phe as the carboxyl terminus of intact CO-Ag protein, thus completing its sequence.
The complete amino acid sequence of CO-Ag, as
established by the current study, yields a calculated
monomeric molecular weight of 8140 Da. This value
is in fair agreement with our previous estimate of 7000
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 05/08/2017
Investigative Ophthalmology 8c Visual Science, April 1996, Vol. 37, No. 5
946
1000 r
FIGURE 1. Quantitative analysis
of CO-Ag sequence data. The
yield of individual phenylthiohydantoin-amino acids was calculated by chromatographic
overlay and peak height measurements. The single-letter
code for the amino acids is: A
= alanine; D = aspartic acid;
E = glutamic acid; F = phenylalanine; G = glycine; H = histidine; I = isoleucine; K = lysine; L = leucine; N = asparagine; P = proline; Q =
glutamine; R = arginine; S =
serine; T = threonine; V = valine; Y = tyrosine.
500
o
100
(p
i 50
el's
22
10
Step Number
Amino Acid
10
20
30
40
50
60
70
TKLEDHLEGIINIFHQYSVRVGHFDTLNKRELKQLITKELPKTLQNTKDQPTIDKIFQDLDADKKGAVVF
Da determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 4% to 12% gel.4 Furthermore, our sequence determination indicates that the
tetrameric native protein must be a homo-tetramer
because there was no evidence of sequence microheterogeneity.
Time (hour)
FIGURE 2. Time course of carboxypeptidase A digestion of
CO-Ag. Aliquots of the carboxypeptidase digest at 1, 2, and
3 hours were subjected to amino acid analysis without acid
hydrolysis. Results indicate the carboxyl-terminal sequence
Lys-Gly-Ala-Val-Val-Phe.
Comparison of Amino Acid Sequences. A search
of computerized data bases of protein and nucleic
acid sequences reveals strong homology of the amino
acid sequence of CO-Ag to the Ca2+-binding proteins
of the S-100 family. The amino acid identity between
CO-Ag and the S-100 proteins ranges from 25% (S100L) to 81% (calgraulin C). The S-100 are a group of
low molecular weight (approximately 10 kDa) acidic
Ca2+-binding proteins. These proteins are expressed
in a cell lineage-specific or tissue-specific manner." To
date, at least 14 proteins of the S-100 family have been
identified and purified. Little is known about their
function, although they are likely to be involved in
cell regulatory processes and intracellular signaling.9
Among the S-100 proteins, CO-Ag shares its greatest degree of protein sequence homology with calgranulin C (CaG-C), which is purified from pig granulocytes.8 It is present at high concentrations in granulocytes but is absent in lymphocytes. Litde is known
about the function of CaG-C other than its Ca2+-binding properties.8 Comparison of the primary structure
of bovine CO-Ag protein with that of pig CaG-C was
shown in Figure 3. Of the 70 residues of CO-Ag compared with diose at corresponding positions in CaGC, identical residues are found at 57 positions, whereas
those infivepositions are both functionally and genetically conserved.
Ca2+ Binding to CO-Ag. Given the sequence relationship between CO-Ag and the Ca2+-binding proteins of the S-100 family, we tested the ability of COAg to bind Ca2+ ions using a direct 45Ca2+-binding
assay. As shown in Figure 4, the CO-Ag protein binds
Ca2+ with a relatively low affinity (R, = 2.4 ± 0.^3 X
10~4 M). Analysis of the binding data according to the
Dell'Angelica et al8 method for CaG-C indicates diat
CO-Ag binds only 0.6 ± 0.05 mole of Ca2+ per subunit.
This value is approximately half that of CaG-C mono-
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 05/08/2017
947
Reports
Ca*+
Ca2+
Hinge
CO-Ag TKLEDHLEGItNIFHQYSVRVGHFDTLNKRELKQLITI
CaG-C TKLEDHLEGIINIFHQYSVRGGHYDTLIKRELKQLITK
10
15
I
20
I
25
1
30
I
35
IDKIFQOLDADKKGAWF
IDKIFQNLDANQDEQVSFKE.
I
40
I
45
50
I
55
I
60
I
65
I
70
FIGURE 3. Comparison of the amino acid sequence of bovine CO-Ag to the amino-terminal
sequence of pig CaG-C. The sequence of CaG-C is taken from Dell'Angelica et al.8 Differences
in amino acids between CO-Ag and CaG-C are highlighted with bold characters. The calciumbinding domains and the hinge region are indicated by shaded boxes.
mer (n = 1.1), although these two proteins have a
similar binding constant of approximately 2 X 10~4 M
(Fig. 4).
DISCUSSION. We present the amino acid sequence of bovine CO-Ag, which was found to be the
target protein for autoantibodies in patients with
Mooren's ulcer.4 The CO-Ag amino acid sequence is
closely related to a family of low molecular weight
Ca2+-binding proteins. The structural relationship of
CO-Ag with these proteins is further corroborated by
its capacity to bind Ca2+. CO-Ag may play a fundamental cellular role in the cornea as a Ca2+ receptor or
mediator of Ca2+-stimulated events, such as cell-cycle
progression, cellular differentiation, cell motility, and
inflammation.9
In spite of strong sequence homology (81%) between CO-Ag and CaG-C, two main differences were
found. First, CO-Ag is a 70-amino acid residue protein
whereas pig CaG-C is a 91-residue protein. Because
p Ca 2+ free
2+
FIGURE 4. Calcium-binding by CO-Ag. The amount of Ca
bound is expressed as mole of Ca2+ per mole of CO-Ag,
using a molecular weight of 8140 Da for the CO-Ag subunit.
The amount of Ca2+ bound is plotted against minus logarithm of free CaH+ concentration.
the N-terminal regions of CO-Ag CaG-C are largely
homologous and can be readily aligned, the size differences must locate in the C-terminal region. It is possible that these two proteins differ in their molecular
size. Alternatively, it is possible that CO-Ag is a product
of posttranslational proteolytic cleavage of a larger
CO-Ag protein. Nucleotide sequence of a full-length
cDNA clone, the size of mRNA code for a CO-Ag protein, and the molecular weight of a recombinant COAg protein will confirm the identity of the sequenced
CO-Ag protein.
Second, it is noteworthy that the sequence homology is striking in die region located outside the Ca2+binding domains and the hinge region. Here, die homology between CO-Ag and CaG-C reaches 100%. The
differences are greatest at the functional sites of COAg and CaG-C, the Ca2+-binding domains, and the
hinge region. At the Ca^-binding domains, there are
three amino acid substitutions at positions 21, 24, and
28 in the N-terminal Ca2+-binding domain (Ser18 to
Glu31). The first two changes in amino acid, at positions 21 and 24, are functionally conserved and can
be explained by minimum single-base changes in the
nucleic acid coding for CO-Ag. The main change residues at position 28 from an isoleucine residue in CaGC to an oxygen-containing asparagine residue in COAg that may serve to coordinate the bound Ca2+ ion.
The most marked difference resides in the C-terminal
Ca2+-binding domain (AspG1 to Glu72). There are five
consecutive amino acid changes from Dh3-K-K-G-A67 in
CO-Ag to M63-Q-D-E-Q67 in CaG-C. These amino acid
changes affect the characteristic feature of the C-terminal Ca2+-binding domain, a region rich in acidic
amino acids in CaG-C, but it has been converted to a
weak basic region in CO-Ag diat may result in an inability to bind Ca2+. The possibility of a nonfunctional
Ca2+-binding domain in CO-Ag is corroborated by its
45
Ca2+-binding properties. Direct calcium-binding
studies suggest that the number of Ca2+-binding sites
for the CO-Ag subunit is approximately half of what
has been reported for the CaG-C monomer.8
In the hinge region, there are three major
changes in amino acids at positions 42, 45, and 51.
The hinge region (Leu40 to Tyr52) is the site for Ca2+-
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 05/08/2017
Investigative Ophthalmology & Visual Science, April 1996, Vol. 37, No. 5
948
dependent interaction with its effector proteins and
is thought to provide specificity to the function of
each S-100 protein. 9 Thus, changes in key amino acids
in the hinge region in the CO-Ag and CaG-C may
result from a unique function of these two proteins
that relates to their specific interactions with effector
proteins in their corresponding tissues, such as CaGC in granulocytes and CO-Ag in corneal stroma.
Epitope mapping and analysis of sequence data
have provided important clues to the cause of certain
autoimmune diseases. For example, a stretch of six
consecutive amino acids of a hepatitis B virus polymerase protein with sequence similarity to the encephalitogenic epitope of myelin basic protein has been
shown to induce central nervous system lesions in rabbits, suggesting that viruses can act as potential initiators of human autoimmune response by molecular
mimicry.10 Our studies of the primary structure of COAg provide the foundation for the identification of
possible pathogenic epitopes in CO-Ag that could be
involved in the initiation and/or progression of
Mooren's ulcer.
Key Words
calcium-binding proteins, calgranulin C, corneal antigen,
sequence homology
The Zebrafish Ultraviolet Cone
Opsin Reported Previously Is
Expressed in Rods
Pamela A. Raymond, Linda K. Barthel, and
Deborah L. Stenkamp
Purpose. To examine expression of the zebrafish ultraviolet cone opsin pigment in goldfish and zebrafish retinas.
Methods. Digoxigenin-labeled cRNA probes were prepared by run-off transcription from plasmids containing
cDNAs for zebrafish ultraviolet opsin, goldfish ultraviolet cone opsin, and goldfish rod opsin. Probes were
From the Department of Anatomy and Cell Biology, University of Michigan Medical
School, Ann Arbor, Michigan.
Supported by grants from the National Science Foundation (IBN9222046) and the
National Institutes of Health (R01 EY04318 and F32 EY06612).
Submitted for publication October 24, 1995; revised December 20, 1995; accepted
January 11, 1996.
Proprietary interest category: N.
Reprint requests: Pamela A. Raymond, Department of Anatomy and Cell Biology,
University of Michigan Medical School, 4610 Medical Sciences II Building, Ann
Arbor, MI 48109-0616.
References
1. Shaap OL, Feltkamp TEW, Breebaart AC. Circulating
antibodies to corneal tissue in a patient suffering from
Mooren's ulcer. Clin Exp Immunol. 1965; 5:365-370.
2. Brown SI, Mondino BJ, Rabin BS. Autoimmune phenomenon in Mooren's ulcer. Am J Ophthalmol.
1976; 82:835-840.
3. Mondino BJ, Brown SI, Rabin BS. Cellular immunity
in Mooren's ulcer. AmJ Ophthalmol. 1978; 85:788-791.
4. Gottsch JD, Liu SH, Minkovitz JB, Goodman DF, Srinivasan M, Stark WJ. Autoimmunity to a cornea-associated stromal antigen in patients with Mooren's ulcer.
Invest Ophthalmol Vis Sri. 1995;36:1541-1547.
5. Ambler RP. Enzymic hydrolysis with carboxypeptidases. Methods Enzymol. 1967; 11:155-166.
6. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ.
Basic level alignment search tool. / Mol Biol.
1990; 215:403-410.
7. Mani RS, Kay CM. Isolation and characterization of a
novel molecular weight 11000 Ca2+-binding protein
from smooth muscle. Biochemistry. 1990;29:13981404.
8. Dell'Angelica EC, Schleicher CH, Santome JA. Primary structure and binding properties of calgranulin
C, a novel SlOO-like calcium-binding protein from pig
granulocytes. J Biol Chem. 1994; 269:28929-28936.
9. Klingman D, Hilt DC. The SI00 protein family. Trends
BiochemSa. 1988; 13:437-443.
10. Oldstone MBA. Molecular mimicry and autoimmune
disease. Cell. 1987; 50:819-820.
hybridized to cryosections of retina and visualized with
immunocytochemistry.
Results. The zebrafish ultraviolet opsin probe hybridized
selectively to rod photoreceptors, but not to ultraviolet
cones or any other cone type, in both zebrafish and
goldfish retinas, and the pattern of expression was identical to that of the goldfish rod opsin probe. The goldfish ultraviolet opsin, in contrast, hybridized to ultraviolet cone photoreceptors in both goldfish and zebrafish.
Conclusions. The cDNA previously identified by Robinson et al1 as zebrafish ultraviolet opsin is not a cone
opsin but is likely to be a rod opsin. Invest Ophthalmol
VisSci. 1996; 37:948-950.
v-lpsins are members of a large and evolutionarily
ancient gene family with several subfamilies that not
share only sequence homology but also tend to exhibit
similar spectral absorption maxima. 1 " 7 A recent article
by Dowling and colleagues1 was the first to identify a
vertebrate ultraviolet opsin gene, which was isolated
from zebrafish (Danio rerio). Sequence analysis of the
zebrafish ultraviolet opsin gene showed that the deduced amino acid sequence was up to 89% identical
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 05/08/2017