Stimulation index obtained with tears from healthy nonsmokers (HNS), smokers who are otherwise healthy (HS), and patients with Graves disease (GD) and Graves ophthalmopathy (GO) measured on cells expressing (lulu) and not expressing (zulu) the thyrotropin receptor. *Stimulation index = 21.1. The bold dashed line shows the cutoff (97th percentile of values from HNS) for the lulu cell line; the light dashed line, the cutoff for the zulu cell line.
Relative light units obtained when treating cells expressing (lulu) and not expressing (zulu) the thyrotropin receptor when treated with 0.01% trypsin at time 0 and at 10 and 30 minutes of incubation.
Silver-stained sodium dodecyl sulfate–polyacrylamide gel electrophoresis of tear samples. Lane 1 contains molecular weight standards; lane 2, tears from healthy nonsmokers; lane 3, tears from patients with Graves ophthalmopathy; lane 4, tears from smokers who are otherwise healthy; lane 5, lysozyme; lane 6, lysozyme; lane 7, blank; and lane 8, mast cell tryptase. The arrows indicate the proteins excised and analyzed by mass spectrometry.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Baker GRC, Morton M, Rajapaska RS, et al. Altered Tear Composition in Smokers and Patients With Graves Ophthalmopathy. Arch Ophthalmol. 2006;124(10):1451–1456. doi:10.1001/archopht.124.10.1451
To analyze, compare, and contrast tear composition in healthy nonsmokers, smokers who were otherwise healthy, and patients with Graves disease (GD) accompanied by Graves ophthalmopathy (GO) of varying severity.
Reflex tears were collected using Schirmer strips from 37 healthy nonsmokers, 33 otherwise healthy smokers, 51 patients with GD, and 85 patients with GO. Thyrotropin receptor–stimulating activity and serum thyroid-stimulating antibodies were measured. Pooled tear samples from healthy nonsmokers, healthy smokers, and patients with GO were separated by means of electrophoresis. Proteins expressed in healthy smokers, patients with GO, and healthy nonsmokers were separated by means of electrophoresis and analyzed by mass spectrometry.
Based on the 97th percentile of findings from healthy nonsmokers, specific thyrotropin receptor–stimulating activity was detected in 25% of the tear samples from healthy smokers, 32% of those from patients with GD, and 41% of those from patients with GO. Clinical activity scores correlated with serum thyroid-stimulating antibody levels but not tear thyrotropin receptor–stimulating activity. Electrophoresis revealed additional proteins of 30 to 41 kDa in the tear samples from patients with GO and healthy smokers compared with samples from healthy nonsmokers. These proteins were identified as zinc-α2-glycoprotein and lactoferrin but have no thyrotropin receptor–stimulating activity.
We demonstrate similar changes in tear composition in patients with GO and healthy smokers compared with healthy nonsmokers. Expression of zinc-α2-glycoprotein and lactoferrin is increased and their molecular weights are modified, suggesting degradation and/or changes during glycosylation, which may affect the bioactivities of zinc-α2-glycoprotein and lactoferrin.
Smoking, which is a significant risk factor for the development of GO, modifies tear composition.
Thyroid-stimulating antibodies (TSAbs) bind the thyrotropin receptor, mimic the action of thyrotropin, and cause Graves disease (GD). Some patients with GD develop ophthalmopathy (GO) in which expansion of orbital contents occurs by adipogenesis, edema, and overproduction of extracellular matrix.1 These mechanisms can produce a range of symptoms and may require medical or surgical intervention. Progress in understanding pathogenesis is hindered by limited availability of orbital tissue samples from patients with GO and the restriction to samples obtained during surgery—often when disease is no longer active. We sought a noninvasive method to evaluate immune and endocrine mediators present in active orbital disease and opted for tear collection to assess potential changes.
Previous evidence supports lacrimal gland involvement in GO. High-performance liquid chromatography of normal tears separated constituent proteins into 5 peaks. An increase in 1 or more peaks was obtained in 28% of samples from patients with GO2 and an increased IgA/lysozyme ratio in 33%.3 Computed tomograms demonstrated enlarged lacrimal glands in 22% of patients with GO.4 Octreotide scintigraphy findings revealed tracer uptake in the vicinity of the lacrimal gland.5 Eckstein et al6 found significantly reduced tear production in patients with GO, correlating with ocular surface damage. Eckstein et al6 also found that the acinar cells from lacrimal glands were positive for thyrotropin receptor immunoreactivity and thus susceptible to activation. Because TSAbs have been detected in saliva,7 it seemed reasonable to seek them in tears. Our initial experiments revealed that thyrotropin receptor–stimulating activity (TSHRSA) was higher in normal tears from smokers than from nonsmokers (data not shown).
Smoking is a significant risk factor for the development of GO.8 The effects are immediate and direct. Several mechanisms have been investigated, including hypoxia, in which orbital fibroblasts produced increased levels of tumor necrosis factor α, interferon γ, interleukin (IL) 1, and glycosaminoglycans in vitro compared with normal aeration.9 Another study examined the effect of cigarette smoke constituents on orbital fibroblasts and found HLA-DR expression only in the presence of interferon γ.10 When mast cells were exposed to cigarette smoke in vitro, release of IL-4, IL-5, IL-10, IL-13, and tumor necrosis factor α was induced.11
Mast cells are present in human GO biopsy specimens and in murine models of GO.12,13 Circulating levels of IgE, stem cell factor (c-Kit ligand), and c-Kit, the last 2 central to mast cell activation, are increased in patients with GD.1 Our group demonstrated IgE antibodies binding directly to the thyrotropin receptor in patients with GO,14 indicating some degree of antigen specificity, although it is unlikely to activate the receptor. However, apart from thyrotropin and TSAbs, the thyrotropin receptor can be activated by proteases such as trypsin,15 and a related protease, tryptase, is a major product of activated mast cells.
We hypothesized that smoking could stimulate the production of mast cell protease, leading to activation of the acinar cells (via the thyrotropin receptor) and to subsequent modification in tear composition. The TSAbs in tears could have a similar effect.
Patients were recruited and provided informed consent, after ethical approval by the Bro Taf Local Research Ethics Committee of Wales and following the tenets of the Declaration of Helsinki. Reflex tears were collected with the use of Schirmer strips, together with serum samples, clinical ophthalmologic data, and information on smoking status. Patients were divided into 2 groups on the basis of their clinical activity score.16 Patients with GD were recruited from an endocrine clinic, were not being treated by an ophthalmologist, and were considered to have minimal GO activity (n = 51; 44 female and 7 male; age range, 17-60 years). The remainder had GO (n = 85; 67 female and 18 male; age range, 28-64 years) and had eye disease of sufficient severity to require management by a clinical ophthalmologist.
Reflex tears and serum samples were also obtained from smokers who were otherwise healthy (n = 25; 13 female and 12 male; age range, 22-48 years) and from healthy nonsmokers (n = 28; 15 female and 13 male; age range, 21-55 years) who were free of autoimmune thyroid disease.
A second cohort consisting of 9 healthy nonsmokers (7 female and 2 male; age range, 24-50 years), 8 healthy smokers (7 female and 1 male; age range, 24-60 years), and 38 patients (32 female and 6 male; age range, 18-73 years) with a spectrum of clinical activity scores (range, 0-9; mean ± SD, 2.78 ± 2.28; median, 3) was also examined to find correlations between TSAb level, TSHRSA, and clinical activity score and to investigate whether TSHRSA is a TSAb.
Individual samples were analyzed for TSHRSA with a previously described bioassay for TSAbs17 using Chinese hamster ovary cells stably expressing the human thyrotropin receptor and a cyclic adenosine monophosphate–responsive luciferase (lulu) or the reporter gene alone (zulu). We used the following 2 protocols: (1) tears (10% vol/vol) in serum-free medium and (2) tears (10% vol/vol) in salt-free buffer.18 In each case, the total volume was 100 μL/well with incubation for 4 hours at 37°C. Individual serum samples (10% vol/vol) were also assayed in both conditions. All measures were performed in triplicate and repeated if tear volume permitted. After incubation, the cells were lysed, and luciferase activity was measured on a plate-reading luminometer with firefly luciferin (Tropix; Promega Corporation, Southhampton, England).
The results are expressed as a stimulation index, which is calculated from the ratio of light units in the presence of tears or serum to light units in the equivalent assay buffer alone (serum-free medium or salt-free buffer). The stimulation index was considered positive if it was greater than the 97th percentile for the stimulation index of normal tears or of serum assayed in the same conditions. Activity in the lulu and zulu cell lines was measured because Chinese hamster ovary cells may have endogenous cyclic adenosine monophosphate–coupled receptors capable of responding to tear components.
Apart from the biological fluids, the bioassay was applied to investigate the TSHRSAs of polyacrylic acid liquid gel (Viscotears; Novartis, Zurich, Switzerland) (10% vol/vol), lysozyme (0.002μM to 20μM), lactoferrin (LF) (0.001μM to 10μM; recombinant and biochemically purified from Sigma-Aldrich Corp, St Louis, Mo), zinc-α2-glycoprotein (ZAG) (2% to 20% vol/vol conditioned medium or affinity-purified protein), mast cell tryptase (0.002μM to 20μM), and trypsin (0.01% vol/vol). All bioassays were performed in serum-free medium for varying incubation times.
Tears from healthy nonsmokers, healthy smokers, patients with GD, and patients with GO were pooled and boiled with an equal volume of 2× loading buffer containing 10% β-mercaptoethanol,19 and 50 μL/lane was loaded onto 10% sodium dodecyl sulfate–polyacrylamide gels. Gels were stained with silver (Amersham Biosciences, Uppsala, Sweden) or were transferred to polyvinyl difluoride membranes for Western blot analysis using standard protocols.19
The blots were probed with monoclonal antibodies to mast cell tryptase (1:10 000) or monoclonal antibodies to ZAG (1:500) (Santa Cruz Biochemicals, Santa Cruz, Calif). Detection was achieved by using an antimouse IgG–horseradish peroxidase conjugate (1:5000) (Amersham Biosciences) and visualization was achieved by means of enhanced chemoluminescence (ECL Plus; Amersham Biosciences).
This same method was also used to separate lysozyme (Sigma-Aldrich Corp), purified ZAG, mast cell tryptase, and Viscotears.
After silver staining, several proteins were cut from the gel, destained, reduced, and alkylated before digestion with sequencing-grade trypsin. The resulting peptides were extracted, dried in a vacuum centrifuge, redissolved in formic acid, and concentrated on C18 chromatography resin (u-C18 ZipTips; Millipore Corp, Bedford, Mass). The peptides were eluted directly onto the matrix-assisted laser desorption ionization (MALDI) target, and mass fingerprinting was performed using a MALDI–time-of-flight (TOF) mass spectrometer (Bruker Reflex III; Bruker Daltonics Inc, Coventry, England) in the reflectron mode. Mascot software (Matrix Science, London, England) was used for searching the NCBInr database, and protein identification was based on a MOWSE score with a significance value of P<.05.
The ZAG-conditioned culture supernatants from A293 cells expressing full-length His-tagged human ZAG were centrifuged and underwent dialysis against 100 volumes of native binding buffer for purification on resin (ProBond; Invitrogen Corp, Carlsbad, Calif). An unrelated His-tagged protein was used as a negative control.
Using tear samples from 28 healthy nonsmokers in serum-free medium (mean ± SD stimulation index, 1.75 ± 0.57) as the basis for reference, we considered stimulation index values of more than 2.7 to be positive. (All results are expressed as mean ± SD unless otherwise indicated.)
Figure 1 illustrates the stimulation index values obtained by using thyrotropin receptor–expressing (lulu) cells; a positive response was obtained in tear samples from 9 of 25 healthy smokers (2.41 ± 0.99), 7 of 51 patients with GD (2.20 ± 1.19), and 13 of 47 patients with GO (2.67 ± 3.06). Differences in the proportion of stimulation index values between healthy smokers and healthy nonsmokers (P = .01) and between healthy smokers and patients with GO were significant (χ2 test, P = .03) but not when we compared the stimulation index values of healthy smokers and patients with GD.
To determine whether the activity was mediated via the thyrotropin receptor, tear samples of sufficient volume were also tested using zulu cells. There was no significant difference (unpaired, 2-tailed t test) between the stimulation index values obtained using lulu cells and those obtained using zulu cells for the healthy nonsmokers' tears (P = .1); in contrast, the stimulation index values were significantly higher in the lulu cells for tear samples from healthy smokers (P = .02), patients with GD (P = .02), and patients with GO (P = .01). The stimulation index value of the healthy nonsmokers was 2.53 ± 0.57, and a positive response (cutoff value, 3.4) was obtained in tear samples from 1 of 20 healthy smokers (2.18 ± 0.7), 0 of 19 patients with GD (1.9 ± 0.63), and 2 of 17 patients with GO (2.45 ± 0.85).
To calculate specific TSHRSA, we subtracted the stimulation index value of the zulu cell line from that of the lulu cell line and obtained TSHRSA values for the healthy nonsmokers (−0.15 ± 0.52), healthy smokers (0.35 ± 1.09), patients with GD (0.71 ± 1.83), and patients with GO (0.73 ± 1.29). The 97th percentile for the lulu-zulu stimulation index difference in the healthy nonsmokers was 0.85. From these data, we concluded that specific TSHRSA is present in the tear samples from approximately 25% of the healthy smokers, 32% of the patients with GD, and 41% of the patients with GO.
Tear and serum samples from a second cohort consisting of 9 healthy nonsmokers, 8 healthy smokers, and 38 patients with GD (clinical activity score range, 0-9) underwent testing for TSHRSA in serum-free medium and salt-free buffer.
Based on the 97th percentile of the samples from 9 healthy nonsmokers, we found TSHRSA in the tear samples (in serum-free medium) from 4 of 8 healthy smokers and 13 of 38 patients with GD. The clinical activity score in the 13 patients with GD and positive TSHRSA findings was 3.54 ± 2.37 (median, 3). In contrast, the 25 GD patients with negative TSHRSA findings had a clinical activity score of 2.50 ± 2.18 (median, 2). The differences between the tears from healthy smokers and those from patients with GD, when compared with those from healthy nonsmokers, are significant (P = .02 and P = .045, respectively), but the difference between the tears from healthy smokers and those from patients with GD is not. The number of tear samples positive for TSHRSA was decreased (although not significantly) in salt-free buffer (Table 1). The patients with GO included smokers and nonsmokers, and there was no difference in the responses of these 2 groups of patients with GO.
Serum TSAbs were not detected in the tears of healthy smokers in either condition. In the patients with GD, serum TSAbs were found in 8 of 38 samples in serum-free medium; this result increased to 31 of 38 samples in salt-free buffer (Table 1) (P<.001). The 31 patients with GD who were positive for TSAbs in salt-free buffer had a clinical activity score of 3.10 ± 2.31 (median, 3); in the 7 who were negative for TSAbs, the score was 1.00 ± 0.89 (median, 1).
Serum TSAb values determined when using salt-free buffer correlated with those determined when using serum-free medium (ρ = 0.59; P<.001) but did not correlate with tear TSHRSA in salt-free buffer or serum-free medium. The clinical activity score was positively correlated with serum TSAb findings in salt-free buffer (ρ = 0.44; P = .005) but not in serum-free medium. The clinical activity score did not correlate with tear TSHRSA in either condition.
Tear samples from 5 TSHRSA-positive patients with GD were pooled and assayed on thyrotropin receptor–expressing cells, in serum-free medium and in the presence of thyroid-blocking antibody (TBAb)–containing or normal human serum (both at 5%, 10%, and 20% vol/vol). There was no difference in TSHRSA in the pooled tears in the presence of the TBAbs compared with those in the presence of the normal human serum.
Bioassay findings excluded Viscotears and lysozyme as possible explanations for TSHRSA (data not shown). Thus, we investigated whether trypsin activation of the thyrotropin receptor18 was detectable using the bioassay. We used 0.01% vol/vol trypsin and, as shown in Figure 2, detected a modest response (a 4-fold increase) after 10 minutes, which persisted after 30 minutes of incubation. A smaller response (a 2-fold increase) was also obtained using the zulu cell line (in which unstimulated light levels were similar to those of the lulu cell line). The difference in magnitude is due to the activation of the thyrotropin receptor in the lulu cell line. In a second experiment, we obtained similar results and demonstrated that the effects disappeared when the cells were incubated with trypsin for 60 or 120 minutes (data not shown).
We consistently observed an increased volume-per-volume protein content of tears from patients with GO and from healthy smokers. This finding was accompanied by additional bands in the 30- to 40-kDa molecular weight range. Similar results were obtained when we used at least 3 different batches of pooled tears. A representative example is shown in Figure 3.
We excised 4 bands ranging from 30 to 40 kDa from each of the 3 tear-containing lanes and mast cell tryptase from the adjacent lane. After MALDI-TOF analysis, we were able to identify the 4 bands as being ZAG and LF (the 2 higher and lower molecular weight bands, respectively). The results are summarized in Table 2.
Biochemically purified and recombinant LF did not display any TSHRSA. After purification of the ZAG-conditioned medium, similar quantities (assessed by silver staining) of an approximately 40-kDa protein were present in all fractions. The purified ZAG and the conditioned medium did not display any TSHRSA.
We have identified similar differences in the bioactivity and composition of tears from patients with GO and those from healthy smokers, compared with tears from healthy nonsmokers.
Several indications suggest that TSHRSA does not represent an antibody. The activity was detected in almost one third of the healthy smokers; however, in the small group examined, none had serum TSAbs. The TSHRSA was not affected by the presence of human serum samples containing TBAbs. Furthermore, the optimal conditions for TSHRSA are physiological, unlike the salt-free buffer preferred by TSAbs.18 The bioassay is able to demonstrate thyrotropin receptor activation by trypsin. However, the kinetics of this response differ from those of TSHRSA, possibly because of differences in concentration. A proportion of the activity we detected in tears is due to activation of Gsα-coupled receptors other than the thyrotropin receptor endogenously expressed by the Chinese hamster ovary cell line. The stimulation index values obtained using the zulu cell line ranged from 1.1 to 4.5, contrasting with those of the thyrotropin receptor–expressing cells, which ranged from 1.1 to 21.1.
Attempts to characterize the TSHRSA demonstrated differences in protein composition, including increased protein content in tears from patients with GO. This difference could be attributed to the dry-eye syndrome that can accompany GO, although more commonly these patients will have reflex tearing that tends to dilute the protein concentration. We also detected a number of additional bands corresponding to proteins of approximately 30 to 40 kDa in the tears from the patients with GO and in those from the healthy smokers. This finding agrees with a recent publication demonstrating increased protein in smokers' tears, in which the region between 25 and 40 kDa was the most discriminatory.20 We have identified 2 proteins from this region as LF and ZAG, which are both constituents of normal tears.21
Lactoferrin is an iron-binding protein with pleiotropic effects, including anti-inflammatory and antimicrobial effects and the ability to bind glycosaminoglycans.22 The molecular weight of LF is approximately 78 kDa; however, we detected the protein at a lower molecular weight. Lactoferrin is highly resistant to degradation, but cigarette smoke components23 and cathepsins24 increase LF proteolysis, thereby reducing its antimicrobial properties. The loss of LF bioactivity is potentially detrimental and may occur at the ocular surface in smokers and patients with GO.
Zinc-α2-glycoprotein is a 41-kDa protein component of plasma, saliva, and tears. It stimulates lipid breakdown in adipocytes and produces the extreme weight loss that occurs in some cancers.25 It is postulated to increase cyclic adenosine monophospate levels in adipocytes (which secrete ZAG) via β3-adrenoreceptors.26 We have demonstrated that the tears of patients with GO and smokers contain an additional ZAG of smaller molecular weight, possibly the result of incomplete glycosylation. The precise function of ZAG in tears and other biological fluids is not known.
Lactoferrin and ZAG do not have TSHRSA under the assay conditions we used. We postulate that the TSHRSA constitutes a protease enzyme, which could activate the thyrotropin receptor expressed on acinar cells and could increase cyclic adenosine monophosphate levels with multiple effects.27 In support of this hypothesis, a recent study28 reported increased in vitro production of mast cell proteinases when exposed to cigarette smoke condensate.
In conclusion, smoking induces changes in the bioactivity and composition of tears that are similar to the changes seen in tears of patients with GO. These changes may contribute to the development of GO in patients with existing GD.
Correspondence: Marian Ludgate, PhD, Centre for Endocrine and Diabetes Sciences, School of Medicine, Cardiff University, Main Building, Heath Park, Cardiff CF14 4XN, Wales (email@example.com).
Submitted for Publication: October 4, 2005; final revision received January 12, 2006; accepted March 27, 2006.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grants from the Wales Office of Research and Development and The Jules Thorn Trust.
Acknowledgment: We thank Carol Lane, PhD, and John Lazarus, MD, for allowing us to study patients in their care, and Laura Hale, PhD, of Duke University, Durham, NC, for providing ZAG-expressing cells.
Create a personal account or sign in to: