PROTEINS IN TEARS FROM HEALTHY AND DISEASED EYES

ARON ZAVARO, ZEMIRA SAMRA, ROBERT BARYISHAK &

DAVID SOMPOLINSKY

(Tserifin, Ramat-Gan, lsrael)

Keywords: Tears; Albumin; IgG; IgA; IgM; Crossed immunoelectrophoresis; Normal values; Conjunctivitis vernalis; Conjunctivitis follicularis; Stevens Johnson's syndrome; Blepharoconjunctivitis; Conjunctivitis sica

ABSTRACT

The levels of total protein in tears from healthy donors, conjunctivitis vernalis patients, and conjunctivitis follicularis patients, were 625, 1370 and 1160 mg% respectively. Serum albumin accounted for 3.3%, 43% and 67% of the total protein of tears from these groups, and the level of proteins probably synthesized by the lacrimal gland, was in tears from conjunctivitis foUicularis patients only half the level in normal tears. By crossed immunoelectrophoresis with intermediate gel, 10 antigenic species could be recognized in normal tears, and of these the following were identified: Lysozyme, IgA, lactoferrin and serum abumin. In tears from patients with conjunti- vifis vernalis three more immunoprecipitates were observed, of which one was due to IgG. No lysozyme could be demonstrated in tears from a case of conjunctivitis sicca by immunoelectrophoresis.

In tears from healthy donors the mean level of IgA was 20 mg%, of IgG 3 rag%, and IgM could not be demonstrated. Rabbit anti-tear immunoglobulin did not precipitate a standard of human IgM in double immunodiffusion. In cases of conjunctivitis ver- nalis and folliculafis the mean levels were increased to 80 and 114 rag% IgG, and 11 and 14 rag% IgM, but IgA was increased only to 32 and 41 rag%. It is assumed that the level of IgA in normal tears is almost entirely due to local synthesis, while serum albumin and other immunoglobulins may have escaped from the circulation by mole- cular sieving. The increased levels of immunoglobulins in inflammatory diseases is probably due to transudation. However, in blepharoconjunetivifis patients several tear samples with a high IgM and a low or zero level of IgG could be demonstrated. Possible explanations for this phenomenon are discussed.

Many investigators have studied the protein components of tears under phy-

siological and pathological conditions, but several problems have remained

unsettled, partly due to diverging reports. Most authors have studied the le-

vels oflysozyme (Ronen et al., 1975; Eylan et al., 1977) or of various immuno-

globulins (McClellan et al., 1973; Sen & Sarin, 1979; Little, Centifanto

& Kaufman, 1969; Allansmith, 1973). It is generally assumed that the IgA

and IgE of tears are, at least in part, synthesized locally by lymphocytes in

the gland tissue, whereas IgG and IgM may have escaped from the blood cir-

culation (Allansmith, 1973), but some authors have indicated the possibility

of local production of IgG as well, in order to explain the ratio between va-

rious immunoglobulins in tears from normal subjects, which often vary from

the ratio found in blood plasma (McClellan et al., 1973).

In the present study we have determined the levels of IgA, IgG and IgM in

Documenta Ophthalmologica 50,185-199 (1980). 0012-4486/80/0501-0185 $ 3.25. �9 Dr. W. Junk B.V. Publishers, The Hague. Printed in The Netherlands.

185

tears of heal thy and externally diseased eyes. We have paid particular atten- t ion to the ratio between these immunoglobulins and serum albumin (HSA).

The concentrat ion of HSA in tears is used as a measure for transudation from the circulation from where it probably originates.

MATERIAL AND METHODS

Tear donors

Tears were obtained from 50 healthy donors without a history of ocular

diseases. The age of the donors varied from 2 to 70 years; 24 were females and 26 were males. Tear samples were also obtained from 60 patients with

well-defined disorders of the external eyes. The tears were drawn into mi-

crotubes by capillarity from the right and the left eye separately, always in this sequence, and no stimulation was employed. In a few cases only one sample could be obtained.

Quantitation of proteins

Total protein of tears was determined by a modification of Lowry's method

(Hartree, 1972). HSA and immunoglobulins were quanti tated by radial im- munodiffusion in 1% agarose (Mancini, Barbonara & Heremans, 1965). Rab-

bit antisera against serum albumin and human heavy chain immunoglobu-

fins (a, 7 and/~, respectively) were used. Four #1 of the tears or of protein standards were introduced into wells in the antiserum containing agarose.

Incubation was 24 hours at 37~ (for HSA, IgG and IgA) and 96 hours (for IgM). For each protein two gels were used, one with 2/A antiserum per cm 2 , the other with 0.2 ~tl per cm 2. In this way linear calibration curves in the intervals ~>0.3 - 100 rag% (HSA, IgG and IgA) or />0.6 - 100 rag% (IgM) were obtained. After washing, the gels were inserted into 1% tannic acid for sharp visualization of the zones of immunoprecipi tat ion. Serum and secre- tory IgA were not differentiated.

Immunization of rabbits

Rabbits were immunized by weekly subcutaneous injections of a pool of equal parts of 24 tear samples from healthy donors. 0.5 ml of the tears was mixed with 0.5 ml incomplete Freund's adjuvant. Rabbit immunoglobulin was obtained from serum by sedimentation with 50% (NH4)2 SO4 and dia- lysis of the redissolved sediment against 10 mM tris (hydroxymethyl ) ami- nomethane, 0.5 M NaC1,0.1% NAN3, pH 7.2.

186

Immunoelectrophoretic methods

Crossed immunoelectrophoresis with intermediate gel and tandem crossed

immunoelectrophoresis in agarose were performed as described by Axelsen (1973) and Bock & Axelsen (1973). The second dimension agarose gel con- tained 20 #l/cm 2 rabbit anti-human tear immunoglobulin. After electro-

phoresis, the plates were washed in 50 mM barbital buffer at pH 8.6 and

thereafter in distilled H20. Staining was with Coomassie Blue R250. For some of the immunoprecipi tates obtained by crossed agarose electropho-

resis the corresponding antigen could be identified, as will be specified un-

der Results. The following techniques were used to identify the antigens, except for lysozyme: 1. Tandem-crossed immunoelectrophoresis (Bock & Axelsen, 1973), with tears in one well and a known protein standard in the second well. By immunological ident i ty the immunoprecipi ta te of the known protein merges with the corresponding precipitate of the tear sample. 2. A known protein is added to the tear sample before crossed agarose electrophoresis. The immunoprecipi ta te corresponding to this protein is

enlarged. 3. A monospecific antiserum to a known protein is incorporated into the intermediate gel of the crossed electrophoresis (Axelsen, 1973).

The corresponding immunoprecipi ta te is drawn into the intermediate gel.

Demonstration o f lysozyme

The electrophoretic mobil i ty of lysozyme was determined by electrophore-

sis on a agarose strip similar to the first dimension of crossed agarose electro- phoresis. The gel was thereafter dried at 40~ and overlayed with a 1.5 mm-

thick agarose layer, which contained 3 mg of dried Mierocoeeus lyseidecti- cus. After incubation for two hours at 37~ a sharp zone of bacteriolysis indicated the lysozyme.

Ouchterlony double immunodiffusion

The antigens and rabbit immunoglobulins were applied in 8/~1 amounts in-

to wells in the 1% agarose gel and the plates were incubated for 48 hours at 37~ thereafter washed and stained as for crossed agarose electrophoresis.

Materials

HSA and rabbit anti-HSA immunoglobulin were from Dakopath AS, Den- mark. Human immunoglobulin standards and rabbit anti-heavy chain immu- noglobulins, transferrin, antitransferrin and antilactoferrin were purchased from Behring AG, Germany.

187

Statistic analyses

Analyses for mean, standard deviation and significance of differences (p)

between the groups by the unpaired t-test were performed on a computer-

ized program.

RESULTS

The source o f tear proteins

Table 1 shows that serum albumin (HSA) accounted for 3.3%, 43% and 67% of total protein in tears from healthy donors, and patients with conjunctivi-

tis vernalis and follicularis, respectively. In cases of conjunctivitis sicca we

did not obtain enough fluid for determination of total protein. We have

therefore washed the conjunctival sac thoroughly and aspirated a part of

the wash fluid. In this case HSA accounted for 77% of total protein. The le- vel of HSA was 20, 590 and 773 rag% in the pools of normal tears, tears

from conjunctivitis vernalis and from conjunctivitis follicularis, respectively. These figures reflect probably an increasing degree of transudation of circu- la tory proteins. The Table shows also that the level of proteins other than HSA and immunoglobulins were similar in tears from healthy persons and patients with conjunctivitis verualis, in both groups about 600 rag%, but it was only 271 rag% in conjunctivitis follicularis.

Table 1. Total protein, serum albumi~ and immunoglobulins in tears.

Total Serum IgA lgG IgM protein 1 albumin

Pool of tears from 24 healthy 625 20.5 23.5 0.93 ~--0.6 donors (3.3%) 2 (3.7%) (0.15%)

Pool of tears from 10 patients 1370 590 26 68 22 with Conjunctivitis vernalis (43%) (2%) (5%) 1.6%)

Pool of tears from 9 patients with 1160 773 42 64 9.9 Conjunctivitis folticularis (67%) (3.6%) (5.5%) (0.8%)

Sample from Conjunctivitis sicca 3 380 292 11 47 5 (77%) (3%) (12%) (1.4%)

1 All values are in mg%. 2The figures in brackets indicate percentage of total protein. 3No tears could be obtained. The right conjunctival sac was thoroughly washed with sterile saline and 40/31 of the fluid was aspirated into a mierocapillar. The figures are relative only.

Proteins in tears f rom healthy eyes

A pool of equal parts of 24 samples of tears from healthy donors was exa-

188

mined by crossed agarose electrophoresis with anti-tear immunoglobulin.

Ten immunoprecipitates could be clearly distinguished, though one of them

(3a) did not always stay clearly out on the photo (see Figs. 1 and 2). Of the-

se, the two most voluminous ones correspond to lysozyme (No. 9) and a

protein with fast anodal mobility (No. 1), probably the low-molecular weight

acidic protein described by Josephson and Weiner (1968) and Bonavida,

Sapse and Sercarz (1969). Precipitate No. 8 was shown to be due to IgA and

No. 3 to be serum albumin (Fig. 2). Precipitate No. 7 was likewise shown to

be due to lactoferrin. In double immunodiffusion the anti-tear immunoglo-

bulin reacted strongly with standards of human IgG, but not with IgM or

transferrin.

In Table 2 are compiled determinations of HSA and immunoglobulins of tear samples from healthy and diseases eyes, and Table 3 shows the signifi- cance of differences between the groups (p) as obtained by a computerized

program of the unpaired t-test. These tables are based on samples from the right eye only. The results of tears from the left eyes have been examined

statistically in the same way and the mean values and degrees of significan-

ce were almost identical.

Fig. 1 (Zavaro, Samra, Baryishak, Sompolinsky). Crossed agarose immunoelectropho- resis with intermediate gel of tear fluid from healthy eyes. The sample was 5 #1 of a pool of equal amounts of tear fhLid from 24 healthy persons. The application slit is indicated by an arrow. 20 #l]cm 2 rabbit anti-tear immunoglobulin was added to the second dimension gel. Immunoprecipitate No. 3 has been identified as due to serum albums, No. 7 as lactoferrin, No. 8 as IgA and No. 9 as lysozyme. Precipitate 3A (see Fig. 2) is not clearly seen on the photo. The anodal electrode to the right and the top in first and second dimension electrophoresis, respectively.

189

Fig. 2 (Zavaro, Samra, Baryishak, Sompolinsky). Identification of serum albumin crossed immunoelectrophoresis with intermediate gel of tears, a. 5//1 of normal tears. b. 5 /21 tears + 5 pl serum albumin, 90 pg/ml. Immunoprecipitate No. 3 is strikingly in- creased, c and d. Tandem-crossed electrophoresis. In c, 5/11 tears were applied into the left and in d, into the right well. Into the other wells was applied 5/11 of 90 pg/ml se- rum albumin. The immunoprecipitate of serum albumin merged with precipitate No. 3. The anode to the right and the top of the figures. T = tears A = serum albumin

In tears from healthy donors HSA levels varied in the range 5-27 mg%, IgA

levels in the range 13-25 rng% and the highest level of IgG observed was 12

mg%, but about 50% of the samples were negative for IgG (<0.3 mg%). IgM

was not demonstrated in any of the samples (<0.6 mg%). The ratio of the

mean values for HSA and for IgG (HSA/G) was 5, about twice the ratio in blood plasma, and this ratio varied 1.5 -- ~ 50. These results might indicate

190

o

0

r~ o

o

0

r~

0

0

r ~

~..~

~.~

~.~

0

+I +1 +1 +1 +1

v = 4 d d d

oh O0 o'~

+1 +1 +1 +1 +1 +1

~ O0 O~

+1 +1 +1 +1 +1 +1 ~- ~ o~ o. ~,

+1 +1 +1 +1 +1 +1

O0

�9 ~ . ~ . ~ : ~ ~ ~ , .~- o o o 0 ~ '

, ~ - = ~ " ~ ' = .~, o~ ~ ~ ~ . ~ 0 o 0 0 0

o~

191

that HSA and IgG in the tears come from the circulation through a blood-

tear barrier which retains high molecular weight proteins most effectively.

The results were analyzed statistically for the correlation with age, sex and

right versus left eye, but no significant correlation was found.

Tears from externally diseased eyes

Fig. 3 shows a crossed immunoelectropherogram of a pool of 10 tear samples

from conjunctivitis vernalis against immunoglobulin to normal tears. All the

immunoprecipitates from Fig. 1 were recognized, and only No. 3 (HSA) was

strongly altered in size. The identity with the corresponding antigens in nor-

mal tears was demonstrated by tandem-crossed immunoelectrophoresis.

Three precipitates, not observed in Fig. 1, occurred on the electrophero-

gram of tears from conjunctivitis vernalis patients, one of which (C in Fig. 3)

was shown to be due to IgG. The lowest concentration of IgG observed in

this group was 19 mg%, which is higher than the highest level in the group of

healthy eyes. The mean value for HSA (600 rag%, see Table 2) was 37 times

Fig. 3. (Zavaro, Samra, Baryishak, Sompolinsky). Crossed immunoelectrophoresis with intermediate gel of 5/dl of a pool of tears from patients with conjunctivitis verna- lis. Immunoprecipitate NO. 3 is strikingly increased. Three immunoprecipitates not observed in Fig. 1 are designated A, B and C. C was shown to be due to IgG. The anode at the right and the top.

192

and the mean IgG (81 mg%) was 26 times as high as in the healthy group,

but the mean IgA was only 1.6 times as high. IgM varied between <0.6 - 3 7

mg%, with a mean of 12 mg%. The mean Alb/G ratio was 7.4 and IgG/IgM

(G/M) ratio was 7.0. Standard deviations of the results were rather high, the

coefficient of variation in aU cases being >0.5; still, the t-test showed that the distribution of concentrations of HSA, IgA, IgG and IgM, was signifi-

cantly different from the group of healthy donors (p <0.005, see Table 3).

Tables 2 and 3 summarize next the levels of proteins in tear samples from

patients which conjunctivitis follicularis. The mean levels for HSA, IgA, IgG

and IgM were all higher than in conjunctivitis vernalis, but the differences

were not statistically significant (Table 3). These proteins accounted for

about 75% of the total tear protein (Table 1). The individual Alb/G ratios

varied in the range of 2.4 - 32.6 and the G/M ratios in the range of 3.6 -

30.8. These results could again indicate that the HSA, IgG and IgM, might

be transported from the blood circulation by molecular sieving.

Table 3. Serum albumin and immunoglobulins in tears of healthy and diseased eyes. Significance of differences (p) according to t-test on basis of the data from Table 2.

Group examined Serum albumin IgA IgG IgM

Healthy versus Co njunctivitis vernalis <0.005 <0.005 <0.005 <0.005 Conjunctivitis foUicularis <0.005 <0.005 <0.005 <0.005 Conjunctivitis sicca <0.005 <0.005 <0.005 <0.005 Stevens Johnson <0.005 <0.005 <0.005 n.s. Blepharo-conjunctivitis <0.005 <0.005 <0.005 <0.005

Conjunctivitis vernalis versus Conjunctivitis foUicularis n.s. 1 n.s. n.s. n.s. Conjunctivitis sicca <0.025 <0.05 n.s. <0.025 Stevens Johnson n.s. n.s. n.s. n.s. Blephar o -conjunctivitis n.s. n.s. < 0.05 n.s.

Conjunctivitis foUicularis versus

Conjunctivitis sicca n.s. n.s. n.s. n.s. Stevens Johnson n.s. n.s. n.s. n.s. Blephar o -conjunctivitis n.s. n.s. <0.005 n.s.

Conjunctivitis sicca versus Stevens Johnson n.s. n.s. n.s. n.s. Blepharo -conjunctivitis <0.005 n.s. <0.005 n.s.

Stevens Johnson versus Blepharo-conjunctivitis n.s. n.s. n.s. n.s.

ln.s. = non-significant (p)'0.05).

193

From five patients with keratoconjunctivitis sicca we obtained tears (Ta- bles 2 and 3). The highest HSA level was 3100 rag%, which is the lower limit of normal plasma concentration. The values for HSA, IgA, IgG and IgM in tears of conjunctivitis sicca patients were significantly higher than in tears

from heal thy donors and from patients with conjunctivitis vernalis (except

for IgG), but not significantly different from the samples of conjunctivitis

follicularis (Table 3). The Alb/G ratio was 14.6 and the G/M ratio 3.7. Crossed immunoelectrophoresis (Fig. 4) demonstrated only three of the ten immu- noprecipitates from the electropherogram of normal tears (Fig. 1), i.e., No. 8 (IgA), No. 3 (HSA) and No. 6, and in addit ion the three precipitates (A, B and C) from conjunctivitis vernalis (Fig. 3).

Three patients suffered from Stevens Johnson's syndrome. HSA, IgG and IgM levels were high, but the mean IgA was only 19.4 rag%, with one sample showing only 4 rag% of IgA, the lowest level observed in any sample in this study.

Fig. 4 (Zavaro, Samra, Baryishak, Sompolinsky). Crossed immunoelectrophoresis with intermediate gel of washing fluid from the eyes of a patient with keratoconjunctivitis sieca. The external eyes and eonjunetiva were thoroughly flushed with sterile saline and a small portion of the fluid was aspirated. Five/11 was applied in the slit indicated by an arrow. Only three of the immunoprecipitates recognized in healthy tears were ob- served, i.e., No. 3 (serum-albumin), No. 8 (IgA), and No. 6. The three precipitates from Fig. 3 (A, B and C) were present. Remark that neither the anodal protein correspond- ing to No. 1 of Fig. 1, lactoferrin (No. 7), or lysozyme (No. 9) could be recognized. The anode to the right and the top.

194

Blepharoconjunctivitis

The study included 17 cases of bilateral blepharoconjunctivitis (Tables 2

and 3). Again, the mean values for IgA and IgM in this series were signifi- cantly higher than in the normal group (p <0.005) and in the same range as

the values for conjunctivitis vernalis and follicularis. The mean HSA value

was 382 mg%, and the mean IgG was only 34 mg%, which is significantly lower than in conjunctivitis vernalis and follicularis (p < 0.005, Table 3).

The Alb/G ratio was 11, and the G/M ratio only 3.4. In several of these tear

samples the G/M ratios were considerably lower than in blood plasma,

and in four of them no IgG, could be demonstrated, neither by radial immu-

nodiffusion, nor by Ouchterlony double diffusion in agarose (Fig. 5). In two

of the samples with no IgG, the IgM values were higher than 26 rag%. It should be added, that in 20 eyes from 11 of the patients conjunctival swabs

were positive for Staphylococcus aureus.

Fig. 5. (Zavaro, Samra, Baryishak, Sompolinsky). Ouchterlony double immunodiffu- sion in agarose of tears from the left eye of a patient with blepharo-conjunctivitis. T = 7 pJ of tear fluid AM = 7/2l of rabbit immunoglobulin to human IgM (anti/~). M = 7 b0 of a~standard of human IgM (20 rag%). AG = 7/21 of rabbit immunoglobulin to human IgG (anti 30. G = 7/J1 of a standard of human IgG (20 rag%). AA = 7 #l of rabbit immunoglobulin to human IgA (anti or). A ---- 7/.ll of a standard of human IgA (20 mg%).

195

The tears of one patient with blepharoconjunctivitis were examined re- peatedly, and the results are compiled in Table 4. This pat ient appeared in

our clinic in July 1978 with a history of recurrent attacks of blepharitis, part icularly pronounced on the left eye. Tear samples from both eyes were

negative for IgG, though IgM levels were relatively high. During the follow- ing year the patient suffered from varying grades of conjunctival inflamma- t ion, always more pronounced on the left eye. At three more occasions tear fluid from the left eye was negative for 'IgG. From swabs of the conjunctiva

S. aureus was grown several t imes from the left eye and S. albus from the right eye. The levels of HSA and immunoglobulin in the blood serum were normal.

DISCUSSION

The concentration of total protein in tears from healthy eyes has been re- por ted to be 620 mg% by West et al. (1961), 670-800 rag% by Moses (1975),

60-260 rag% by Bluestone et al. (1975) and 600-2200 mg% by Josephson

and Weiner (1968). Our results for normal tears were in the range of the

first two of these reports, although our sampling method without stimula-

t ion was comparable to the technique used by Bluestone et al. (1975). How-

ever, the composit ion of tear proteins was strikingly different from the re-

ports of West et al. (1961) and Moses (1975). According to the former, 60% of the tear protein was albumin and 40% globulin, giving an Alb/G ratio si- milar to that in blood plasma. Also, Moses (1975) reported 60% of tear pro-

tein to be albumin and the rest equally divided between globulin and lyso-

Table 4. Serum albumin and immunoglobulins in tears from a patient with blepharo-conjunetivitis.

Serum Date Eye albumin 1 IgA 1 IgG 1 IgM l Bacterial isolate

19.7.1978 Right 980 73.0 02 33.5 Not examined 19.7.1978 Left 810 43.5 0 27.5 Not examined

9.8.1978 Right 178 13.0 30.1 2.0 Not examined 9.8.1978 Left Not examined 19.5 12.0 0 Not examined

28.2.1979 Right 196 45.5 31.0 4.5 Staphylococcus epidermidis

28.2.1979 Left Not examined 21.0 0 18.3 S. aureus 7.3.19793 Right Not examined 65.0 35.0 4.0 S. epidermidis 7.3.1979 Left Not examined 25.5 0 21.0 S. aureus

12.3.1979 Left Not examined 75.5 0 7.8 Not examined

1 All values are in rag%. 2lgG was demonstrable for ~>0.3 rag%, and IgM for ~>0.6 rag%. SAt the same date albumin in serum was 4450 mg%; IgG 1220 mg%; IgA 255 rag% and

IgM 152.5 rag%.

196

zyme. McClellan et al. (1973) reported similar levels of IgA in normal tears as in our study, but in contrast to our data their material showed an IgA/ IgG ratio near 1. Table 1 indicates that HSA + immunoglobulins account for only a minute part of total proteins in normal tears, about 50% in tears from conjunctivitis vernalis, and about 75% of conjunctivitis follicularis. In cases of conjunctivitis sicca, the results probably will vary from patient to patient according to the residual function of the lacrimal gland. The exact figures of the Table should be accepted with due reservation, since the de- termination of total protein was performed with Folin's reagent, whereas the single protein species were determined relative to standards by radial immunodiffusion; however, the results are probably not far from the facts, and the corresponding figures of crossed immunoelectropherograms (Figs. 1, 3, 4) illustrate the differences in the size of precipitates due to HSA and IgG on one hand, and those probably due to the lacrimal gland proteins proper (Gachon et al., 1979) on the other hand. The only three precipitates from healthy tear proteins found in the crossed immunoelectropherogram from conjunctivitis sicca were No. 3 (HSA), No. 8 (IgA) and No. 6 which, remark- ably enough, was of similar size in all three crossed agarose immunoelectro- pherograms. It is possible that precipitate No. 6 is due to an antigen synthe- sized by the conjunctival epithelium rather than by the glandular tissue.

In a crossed immunoelectropherogram from normal tears we could easily demonstrate ten immunoprecipitates and a few more were indicated by very weakly stained precipitates (not seen in the photos). Neither lipid staining (Sudan B) nor polysaccharide staining (Feulgen) outlined these precipi- tates. It is possible that they were due to weak antigens, or antigens not well precipitated by antibody. Josephson and Lockwood (1964) demonstrated 4-8 different proteins in tears by immunoelectrophoresis, whereas Sapse et al. (1967) could identify 10 protein bands by polyacrylamide gel electro- phoresis of tears, and Gachon et al. (1979) have recently demonstrated 60 protein components of tears by two-dimensional polyacrylamide gel electro- phoresis, but only 10 could be developed by immunoelectrophoresis with rabbit human tear antibodies.

Our data show a sharp increase in IgG during inflammation, in contrast to a recent report by Sen & Satin (1979), who found measurable levels of IgG only in the tears from one out of ten cases of conjunctivitis vernalis. These differences are probably not explainable only by the differences in sampling techniques, without stimulation in our study and after exposition to a strong slit lamp by Sen and Satin (1979).

The levels of IgA in conjunctivitis vernalis and foUicularis were increased only up to 1.5 - 2.0 times of the normal level. For technical reasons we have not differentiated between circulatory and secretory IgA. If it is assum- ed that the normal level of IgA is due to local synthesis, and the increase during inflammation is due to circulatory IgA, then the IgG/circulatory IgA ratio would be 81/12 = ~7 in conjunctivitis vernal.is and 145/21 = ~7 in

197

conjunctivitis follicularis or similar to the ratio in blood plasma. The AIb/G ratio in the tears was higher than in blood plasma, both in health

and during inflammation. Also, the G/M ratio was in many of the samples considerably higher than in the circulation. This might be due to a blood- tear barrier which may be most penetrable for small molecules. The barrier

may inhibit the transport from the blood capillaries through the glandular epithelium and through the conjunctival mucosa.

In blepharoconjunctivit is the IgG levels were significantly lower than in tears from the other inf lammatory disorders, a fact that is in concordance with the data presented by Allansmith (1973). The G/M ratio was in many

cases low, and particularly spectacular was the finding of tear samples with high levels of IgM, but no detectable IgG. This might be due to a particular

stimulus of IgM and not IgG synthesis by local lymphoid tissue during ble- pharoconjunctivit is , as is assumed for some mucous membranes (Tomasi, 1976). An alternative explanation might be that IgG was specifically remov-

ed from the tear fluid in some cases of blepharoconjunctivitis. The connec- t ion between this disease and S. aureus infections has been emphasized (Smolin & Okumoto, 1977), and it is tempting to assume that the protein A-coated cocci in some cases remove IgG from the tear fluid, but this is as ye t entirely hypothetical .

ACKNOWLEDGEMENT

Joseph Kedem, P h . D . , and J. Schneider, Ph. D., of Bar41an University,

helped with the statistical analyses. This study was supported by Fisons Ltd., Pharmaceutical Division, England.

REFERENCES

Allansmith, M. Immunology of the tears. Ophthal. Clin. 13:47-72 (1973). Axelsen, N.H. Intermediate gel in crossed and fused rocket immunoelectrophoresis.

Scand. J. lmmunol. 2, Suppl. 1:71-77 (1973). Bluestone, R., Easty, D.L., Goldberg, L.S., Jones, B.R. & Pettit, T.H. Lacrimal immu-

noglobulins and complement quantified by counterimmunoelectrophoresis. Brit. J. Ophthal. 59:279-281 (1975).

Bock, E. & Axelsen, N.H. Comparison of antigens: the reaction of partial identity. Scand. J. lmmunol. 2, Suppl. 1:95-99 (1973).

Bonavida, B., Sapse, A.T. & Sercarz, E.E. Specific tear prealbumin. A unique lachry- mal protein absent from serum and other secretions. Nature (London) 221:375-378 (1969).

Eylan, E., Ronen, D., Romano, A. & Smetana, O. Lysozyme tear levels in patients with herpes simplex virus eye infection. Invest. Ophthal. 16:850-853 (1977).

Gachon, A.M., VerreUe, P., Betail, G. & Dastugue, B. Immunological and electropho- retic studies of human tear proteins. Exp. Eye Res. 29:539-553 (1979).

Hartree, E.F. Determination of protein. A modification of the Lowry method that gives a linear photometric response. Anal, Biochem. 48:422-427 (1972).

Josephson, A.S. & Lockwood, D.W. Immunoelectrophoretic studies of the protein components of normal tears. Z lmmunol. 93: 532-539 (1964).

198

Josephson, A.S. & Weiner, R.S. Studies of the proteins of lacrimal secretions. Immuno- logy 100:1080-1092 (1968).

Little, J.M., Centifanto, Y.M. & Kaufman, H.E. Immunoglobulins in tears. Amer. o r. Ophthal. 68:898-905 (1969).

Mancini, G., Barbonara, A.O. & Heremans J.F. Immunochemical quantitation of anti- gens by single radial immunodfffusion. Immunochemistry 2:235-254 (1965).

Mc Clellan, B .H., Whitney, C.R., Newman, L.P. & Allansmith, M.R. Immunoglobulin in tears.Amer. J. Ophthal. 76:89-101 (1973).

Moses, R.A. Adler's Physiology of the Eye. Clinical Application, 6th ed. St. Louis, Mosby, 1975, p. 20.

Ronen, D., Eylan, E., Romano, A., Stein, R. & Modan, M. A spectrophotometric me- thod for quantitative determination of lysozyme in human tears: description and eva- luation of the method and screening of 60 healthy subjects. Invest. Ophthal. 14: 479- 484 (1975).

Sapse, A.T., Bonavida, B., Stone, W., Jr., Barnett, E.V. & Sercarz, E.E. La d~termina- tion quantitative des prot6ines dans les larmes humaines. Bull. Soc. Fr. Ophtal. 80: 236-248 (1967).

Sen, D.K. & Satin, G.S. Immunoglobulin concentrations in human tears in ocular diseases: Brit. J. Ophthal. 63:297-800 (1979).

Smolin, G. & Okumoto, M. Staphylococcal blepharitis. Arch. Ophthal. 95:812-816 (1977).

Tomasi, T.B., Jr. The Immune System of Secretions. Englewood Cliffs, N.J., 1976, p. 65.

West, E.S., Todd, W.R., Mason, H.S. & Van Bruggen, J.T. Textbook of Biochemistry, 4th ed. New York, Macmillan, 1961, p. 625.

Authors' address: Assaf Harofe Hospital Tserifin, Israel

Requests for reprints to: Dr. Aron Zavaro.

199