Selective IgA deficiency in the dog

Selective IgA deficiency in the dog

CLINICAL IMMUNOLOGY AND IMMUNOPATHOLOGY Selective PETER J. FELSBURG,* 36, 297-305 (1985) IgA Deficiency LAWRENCE in the Dog’ T. GLICKMAN,~ A...

624KB Sizes 10 Downloads 68 Views

CLINICAL

IMMUNOLOGY

AND

IMMUNOPATHOLOGY

Selective PETER

J. FELSBURG,*

36, 297-305 (1985)

IgA Deficiency LAWRENCE

in the Dog’

T. GLICKMAN,~

AND PETER F. JEZYKt

*Department of Veterinary Pathobiology, College of Veterinary Medicine, i7niversit.y of Illinois, Urbanu, Illinois 61801, and tDepartment of Clinical Studies, School of Veterinary Medicine. University of Pennsylvania, Philadelphia. Pennsylvania 19104 This study documents the occurrence of selective IgA deficiency in the dog. This is a unique spontaneous animal model with clinical and immunologic findings similar to that of selective IgA deficiency in humans, the most common human primary immunodeficiency. The disease in the dog is characterized by chronic, recurrent respiratory infections and dermatitis, low concentrations of serum IgA, normal concentrations of serum IgG and IgM, normal T-cell function as measured by lymphocyte transformation tests, the presence of autoantibodies, and a defect in the maturation or terminal differentiation of IgA B cells into IgA-secreting plasma cells. 0 1985 Academic Presb. hc.

INTRODUCTION

Since the original cases reported by West et al. (l), selective IgA deficiency has become recognized as the most common primary immunodeficiency disease in humans. Recent studies using automated screening techniques have reported the prevalence of selective IgA deficiency to range between I :252 and I:3000 (2). IgA-deficient individuals are characterized by absent or markedly reduced levels of serum, and usually secretory, IgA with normal or elevated levels of serum IgG and IgM. In general, the cell-mediated immune response is normal in these individuals. Although many of these IgA-deficient individuals may appear asymptomatic, studies of individual patients as well as extensive studies of large numbers of patients suggest the absence of IgA predisposes to a variety of diseases. Infectious diseases, primarily of the respiratory and gastrointestinal tracts, are common in selective IgA deficiency. Mild or moderate recurrent upper respiratory (URT) infections, including otitis, or viral and bacterial origin beginning in early childhood are a common finding in these patients (3, 4). In addition to the URT and gastrointestinal infections, Buckley (4) reported 7% of selective IgA deliciency pateints also exhibit chronic or recurrent skin infections. Seizures are observed in approximately 5% of IgA-deficient patients (3, 4). Individuals with selective IgA deficiency have an increased incidence of allergies, autoantibodies, and autoimmune disease (3-8). An interesting finding in selective IgA-deficient atopic patients is the lack of seasonality and the beginning of symptoms in infancy and early childhood. Up to 50% of IgA-deficient atopic children also have a history of nonspecific dermatitis or eczema (4,9, 10). Rheumatoid factors (35%) and antinuclear antibodies (7%) are two of the more common autoantibodies found in IgA-deficient patients (3, 6, 7). The three most common ’ Supported by USPHS Grant AI 17791 and a grant from the AVMA Foundation. 297 0090-1229/85 $1.50 Copyright All rights

8 1985 by Academic Press, Inc. of reproduction in any form reserved.

298

FELSBURG.

GLICKMAN,

AND

JEZYK

autoimmune diseases associated with selective IgA deficiency are rheumatoid arthritis, systemic lupus erythematosus, and autoimmune thyroiditis (3. 6, 8). Although much is known concerning the clinical manifestations of selective IgA deficiency, the mode of inheritance and the basic immunologic defect still remains unclear. Part of the problem resides in the fact that, unlike other immunodeficiency diseases, there is no appropriate animal model for the disease. This study describes a selective IgA deficiency in the dog which resembles the clinical presentation and laboratory findings of selective IgA deficiency in humans. This unique spontaneous animal mode1 of selective IgA deficiency promises an exciting opportunity to further delineate the genetics and pathogenesis of this immunodeficiency disease. MATERIALS

AND METHODS

Study population. The source of dogs in this study is a large breeding kennel of approximately 6000 to 7000 beagle dogs. While there is a rigorous preventive medical program in effect at this kennel, several infectious diseases remain endemic. Upper respiratory infections caused by Bordetella bronchisepticu and canine parainfluenza virus occur in epizootics despite intranasal vaccination with an effective bivalent vaccine beginning at 3 weeks of age. Recurrent otitis is very common in these dogs as well. Similarly, canine parvovirus enteritis remains endemic even though a commercially available modified-live canine parvovirus vaccine is given parenterally to all puppies. Since similar preventive programs have been effective in controlling these infectious diseases in other kennel populations, there was some concern of an underlying immunologic and/or genetic susceptibility to these infections. Over the past 5 years, there has also been an increase in the prevalence of young dogs with a chronic, recurrent dermatitis which is partially responsive to antibiotic therapy. At any one time, approximately 2% of the dogs are affected with this condition. Over 50% of these dogs also have a chronic, recurrent otitis. Due to the concern of a potential underlying immunologic defect as a cause for these infectious diseases, two nonconsanguineous dogs (one male and one female) with dermatitis were sent to us for an immunologic workup. During these studies, the dogs also developed recurrent upper respiratory infections. Quantitation of serum immunoglobulins. Serum IgG, IgM, and IgA concentrations were measured by single radial immunodiffusion (11, 12) using heavy-chainspecific antisera to each class of immunoglobulin. Immunoglobulin determinations were repeated on several different occasions and also confirmed by an independent laboratory. Normal serum immunoglobulin in levels were determined in age- and sex-matched beagles from another kennel with no excess frequency of infectious diseases and no reported immune defects. Quantitation of B and T lymphocytes. Peripheral blood mononuclear cells (PBMC) were separated from heparinized whole blood by density gradient centrifugation on Hypaque-Ficoll according to a modification of the technique of Boyum (13). Monocytes were identified by latex ingestion. Identification and enumeration of peripheral B lymphocytes was performed by direct immunofluorescence (14- 16) using F(ab’), fluorescein isothiocyanate (FITC)-conjugated goat

CANINE

I&4 DEFICIENCY

299

anti-dog IgG (heavy and light chain specific; Cappell Laboratories, Cochranville. Pa.) reagents. Peripheral T cells were identified and quantitated by direct immunofluorescence using an anti-rat thymocyte serum which has been shown to cross-react by greater than 95% with canine T cells (17, 18). A total of 200 lymphocytes were counted in each of these assays and the percentage of cells stained with the respective reagents was calculated. Lymphocyte transformation. The lymphocyte transformation tests were performed in quadruplicate using a micro whole-blood technique developed for the dog (19). The mitogens used were phytohemagglutinin (PHA), concanavalin A (Con A), and pokeweed mitogen (PWM). The results are expressed as the net counts per minute (cpm) per 5 ~1 blood used in each culture. Tetanus toxoid immunization. Dogs were immunized intramuscularly with 0.5 cc tetanus toxoid. A booster immunization was given 2 weeks following the initial immunization. Serum samples were collected prior to immunization and then at weekly intervals for 4 weeks. The primary and secondary humoral response was evaluated by a passive hemagglutination test (20). Purified tetanus toxoid for use in the passive hemagglutination test was kindly supplied by Wyeth Laboratories (Marietta, Pa.). Polyclonal activation of canine lymphocytes. Polyclonal activation of canine PBMC and the determination of immunoglobulin-secreting plasma cells by a staphylococcal protein A-reverse hemolytic plaque assay was performed according to the technique of Felsburg et al. (21). Essentially, 1 x lo6 lymphocytes were cultured with PWM for 5 days; the cells were washed and added to protein A-coated sheep red blood cells. A monolayer was created in a liquid phase, and the class-specific immunoglobulin-secreting plaque-forming cells (PFC) were determined using individual heavy-chain-specific antisera. Detection ofautoantibodies. Rheumatoid factor (RF) and antinuclear antibody (ANA) were determined using routine procedures (22). Rheumatoid factor was determined by the Rose-Waaler passive hemaggluntination test and ANA was detected by indirect immunofluorescence using monkey kidney cells as a substrate. Skin testing. Intradermal skin testing with a battery of approximately 30 allergens were performed according to procedures used in the Small Animal Clinic. Positive and negative controls consisting of histamine and saline were incorporated in all skin testing. RESULTS

The preliminary studies on the two principal cases included our normal immunologic screening tests for potential immunodeficiencies-immunoglobulin quantitation, quantitation of B and T lymphocytes, and lymphocyte transformation. Table 1 summarizes the results of these screening tests. Both dogs demonstrated normal levels of serum IgG and IgM, but a marked deficiency in serum IgA (the geometric mean concentration for normal age-matched controls is 36 mg/dl with a 95% confidence interval of 18 to 56 mg/dl). Quantitation of peripheral B and T lymphocytes showed that both dogs possessed normal levels of circu-

FELSBURG,

300

GLICKMAN, TABLE

AND JEZYK I

SERUM IMMUNOGLOBULIN

CONCENTRATIONS. B AND T LYMPHOCYTE MARKERS. AND LYMPHOWW BLASTOCENIC RESPONSE IN PATIENTS AND AGE-MATCHED CONTROLS

Patients Male Serum immunoglobulins W kM &A

Female

Controls”

(mg/dl) 1450 185 <5

1300 138 <5

1235 k 261 130 k 55 35 2 8

31.6 61.2

24.8 61.3

26.4 -+ 6.1 64.9 k 5.6

65 11890 11389 8310

72 I7881 6642 9917

B and T lymphocytes (%) B Cells T Cells Lymphocyte transformation (cpm) Control PHA Con A PWM

58 15840 8415 5060

t + e 2

22 8030 4235 2681

u Normal values (mean r SD) for dogs 4 to 6 months of age.

lating B and T cells. Lymphocyte transformation studies revealed a normal in vitro lymphocyte blastogenic response to PHA, Con A, and PWM. The only immunologic defect revealed by the preceding screening immunologic studies was a marked deficiency of serum IgA in both dogs. To evaluate the functional capacity of the humoral immune system we immunized the two dogs with tetanus toxoid and determined their serologic response. Both dogs exhibited a primary and secondary humoral response following immunization (Table 2). We next examined the ability of both dogs to generate IgG-, IgM-, and IgAsecreting plasma cells following in vitro polyclonal activation with PWM. Both dogs demonstrated the ability to generate normal numbers of IgG- and IgMsecreting plasma cells, however, no IgA-secreting plasma cells were detected in the female and the male demonstrated a marked reduction in the number of IgAsecreting plasma cells (Table 3). The lower limit of sensitivity of our reverse hemolytic plaque assay is 10 PFC/106 cells. TABLE HUMORAL

ANTIBODY

2

RESPONSE FOLLOWING PRIMARY AND SECONDARY WITH TETANUS TOXOID

IMMUNIZATION

Time” 0

Week 1

Week 2

Week 3

Week 4

Patients Male Female

110 110


20 40

160 320

160 320

Controls*


5

36

295

410

’ Dogs were vaccinated at Time 0 and at Week 2. Results are the recipricals of antibody titers. b Geometric mean titer of 15 control dogs.

CANINE

IgA DEFICIENCY

301

TABLE 3 GENERATION OF IMMUN~GLOBULIN-SECRETING PLAQUE-FORMING CELLS (PFC) FOLLOWING POLYCLONALACTIVATION WITH PWM Patients Male

Female

1260 780 10 1940

1480 580
[email protected] W kA Total

Controls (means k SD; n = 10) 1276 575 176 2158

r k t 2

240 248 64 511

Note. Results are given as PFC/106 cells.

In order to determine whether the IgA deficiency observed in these two dogs was an isolated event or representative of the affected dogs as a group, we obtained coded samples for immunoglobulin quantitation and lymphocyte transformation from ten other affected dogs and from ten unrelated age- and sex-matched clinically normal dogs (controls). The results of the lymphocyte transformation tests were normal in all 20 dogs. Seven of the ten dogs in the affected group had decreased concentrations of serum IgA similar to the first two dogs evaluated (Table 4). The serum IgG and IgM concentrations were within the normal range for dogs of this age. Since a familial inheritance of IgA deficiency has been shown to exist in humans, we evaluated the serum immunoglobulin concentrations of clinically normal littermates of the affected dogs. Unfortunately, most of the littermates had already been sold. Table 5 illustrates that 4 of the 5 remaining littermates had serum IgA concentrations below the 95% confidence interval of age-matched control dogs. One of these dogs also appeared to be deficient in IgG. Because of the common finding of autoantibodies in human patients with selective IgA deficiency, we examined all the preceding dogs for the presence of TABLE 4 SERUM IMMUNOGLOBULIN CONCENTRATIONSIN AFFECTED DOGS AND AGE-AND SEX-MATCHEDCONTROLS Affected

1200 800 1100 890 1250 960 1600 1300 1450 1400

Controls

kM

Id

kG

I&f

[email protected]

130 200 98 125 130 150 81 115 98 91

19 <5 15 <5 17 5 22 6 <5 23

1100 1200 1250 1500 1500 loo0 1200 1100 1250 1250

170 71 130 115 150 91 170 100 170 130

34 27 23 28 46 29 36 40 30 44

Note. Data are given as mg/dl.

302

FELSBURG,

SERUM

Note.

Results

IMMIJNOGI.OBUI.IN

GLICKMAN.

AND

TABLE 5 CONCENTRATIONS IN CUNICALLY ‘I HE AFFECTED DOGS

JEZYK

NORMAL.

LITIERMATES

w

kM

kA

1750 1250 1000 1450 490

250 76 130 115 86

6 i.s 23 5 16

are given

OI

as mgidl.

rheumatoid factor and antinuclear antibody. All samples from the age- and sexmatched control group were negative for RF, <1:8, and ANA, ~1: 10. However, in the other dogs, five had RF titers of I:8 or greater (1:s for three dogs; 1:32 for one dog; and 1:64 for one dog), None of the dogs demonstrated the presence of antinuclear antibodies. The initial two dogs are now approximately 2 years of age. The serum IgA concentration in the male is 22 mg/dl and in the female is 12 mg/dl. Although both dogs now have detectable IgA, these values fall below the normal values for beagles 2 years of age (geometric mean of 58 mg/dl with a 95% confidence interval of 34 to 147 mg/dl). Both dogs still experience mild recurrent respiratory infections and the female still has her dermatitis. In addition, the female has experienced several episodes of seizures of unexplained origin. These two dogs were bred when they were approximately 10 months old. This mating resulted in a litter of five puppies-two males and three females. At three months of age, one male had detectable levels of serum IgA, whereas all the other puppies had undetectable levels (less than 5 mg/dl) of IgA. The male with no detectable serum IgA died of canine parvovirus enteritis even though he had been vaccinated. The rest of the pups are now approximately 1 year old and their serum IgA concentrations are: male, 28 mg/dl; female 1, 16 mg/dl; female 2, 11 mg/dl; and female 3, <5 mg/ dl. The geometric mean serum IgA concentration in normal beagles 1 year of age is 63 mg/dl with a 95% confidence interval of 24 to 166 mg/dl. The females have all experienced mild recurrent respiratory infections and female 3 has also developed a chronic dermatitis very similar to that observed in her mother. In addition, female 3 has also experienced several episodes of seizures during the past few months. DISCUSSION

This study has documented the presence of a selective IgA deficiency in young dogs suffering from recurrent upper respiratory infections and chronic dermatitis. The immunologic findings are similar to those observed in human IgA deficient patients-either an absence, or very low concentrations, of serum IgA. Selective IgA deficiency is the most common primary immunodeficiency disease in humans. Many cases are healthy when diagnosed as adults, but it is not known whether these individuals had increased frequencies of infections during

CANINE

IgA DEFICIENCY

303

childhood. In this regard, screening of 830 “healthy” adult dogs which are part of the breeding stock of the kennel revealed undetectable IgA (less than 5 mg/dl) in 1% of the dogs, and low IgA (below the 95% confidence interval for agematched controls) in another 8% of the dogs (Felsburg et al., manuscript in preparation). Serum IgA concentrations of 100 age- and sex-matched beagles from a kennel with no history of an increased frequency of infectious diseases fell within the normal range reported for normal adult dogs (23, 24). The clinical manifestations of selective IgA deficiency in the young are varied, although, in most instances, are related either directly or indirectly to the mucosal surfaces of the body. This is not surprising since the major immunologic defense mechanism of these surfaces is secretory IgA. Stimulation of secretory IgA production is of primary importance in preventing viral infection of the respiratory and gastrointestinal tract (25, 26), inhibiting bacterial adherence to mucosal surfaces and colonization (27), and inhibiting the absorption of macromolecular antigens in the gastrointestinal tract (28). The URT infections in the dogs in this study were associated with B. bronchiseptica and canine parainfluenza virus, both which produce a tracheobronchitis. B. bronchiseptica colonizes the mucosal surface of the trachea and canine parainfluenza virus replicates locally in the epithelium of the URT. Intranasal vaccination with a modified-live bivalent vaccine stimulates the local production of IgA which, in normal dogs, prevents, not only clinical disease, but also infection with these agents (29). More importantly, dogs that recover from a natural infection are immune to reinfection for at least 12 months (30). It is evident that these young IgA-deficient dogs do not mount a local immune response to these agents either through vaccination or natural infection. A pilot case-control study revealed that puppies born of “healthy” IgA-deficient dams had a relative risk of developing upper respiratory infections of 3.5 (95% confidence limits = 2.4 to 5.1) when compared to puppies born of dams with normal serum concentrations of IgA (Payton et al., unpublished data). The dermatitis observed in these dogs is similar to the nonspecific dermatitis or eczema which has been reported in children with IgA deficiencies. The convulsive episodes of unexplained etiology observed in the mother and daughter described under Results is another clinical similarity with the human disease. One of the intruiging aspects of selective IgA deficiency in humans is the association with allergic and autoimmune diseases. Although none of our dogs appeared to be atopic at this time as determined by skin testing, there was a high incidence of autoantibodies, namely RF, in these young dogs. A definite familial inheritance of selective IgA deficiency has been well documented in humans (31-36). The mode of inheritance is still to be resolved. Some families demonstrate an autosomal dominant pattern, whereas others exhibit an autosomal recessive pattern (36, 37). Although our initial breeding study does not shed much light on the mode of inheritance, it does indicate a familial occurrence and ability to reproduce both the immunologic and clinical abnormalities. The basic immunologic defect(s) in the maturation and differentiation of the IgA system which leads to selective IgA deficiency is/are still to be resolved. In vitro studies have demonstrated a heterogeneity of defects involving both the B-

304

FELSBUKG.

CLICKMAN,

AND JEZY K

and T-cell populations in IgA-deficient individuals; studies have shown that some patients have a B-cell defect (38, 39), some patients have increased suppressorT-cell function (38, 40-44), and some patients have decreased helper-T-cell function (38. 42). In vitro studies of these dogs have also demonstrated a defect in the maturation or terminal differentiation of IgA B cells into IgA-secreting plasma cells. This unique spontaneous animal model of selective IgA deficiency may prove useful in addressing some of the important questions which remain to be answered concerning human selective IgA deficiency. The underlying immunologic defect is not fully defined nor has the mode of inheritance been well defined. Another point of controversy is the role of a “true” IgA deficiency and low IgA concentrations in the manifestation of clinical disease. This animal model will also be useful in evaluating various modes of immunotherapy. Last, these dogs deficient in IgA will provide an experimental model to study and confirm the role of IgA in the biologic function of the mucosal immune system. REFERENCES 1. West, C. D., Hong, R., and Holland, N. H., J. C/in. Invest. 41, 2054, 1962. 2. Ropars, C.. Muller, A.. Paint, N.. Beige. D.. and Avenard. G.. J. Immunol. Methods 54, 838. 1982. 3. Ammann, A. J., and Hong, R., Medicine 50, 223. 1971, 4. Buckley, R. H.. Birth Defects 11, 134, 1975. 5. Buckley, R. H., and Dees, S. C.. N. Engl. J. Med. 281, 465, 1969. 6. Ammann, A. J., and Hong, R., Clin. Exp. Immunol. 7, 833, 1970. 7. Koistinen, J., and Sarna, S., VOX Sang. 29, 203, 1975. 8. Wells, J. V.. Michaelli, D., and Fudenberg, H. H.. Birth Defects 11, 144. 1975. 9. Ostergard, P. -A., C/in. Exp. Immunol. 40, 561. 1980. 10. Taylor, B., Norman, A. P.. Orge], H. A.. Turner, M. W., Stokes, C. R., and Soothi]], J. F., Lance1 2, 111, 1978. I]. Mancini, G., Cargonara, A. 0.. and Heremans, J. F.. Immunochemistry 2, 235, 1965. 12. Fahey, J. L., and McKelvey, E. M., J. Zmmunol. 94, 84. 1965. 13. Boyum. A., Scund. J. Lab. Invest. 21, 77, 1968. 14. Miller, C. H.. Carbonell. A. R.. Peng. R.. MacKenzie, M. R., and Shifrine, M., Amer J. Vet. Res. 39, 1191, 1978. 15. Ho, C. K., and Babiuk, C. A., Immunology 35, 733, 1978. 16. Chandler, J. P., and Yang, T. J., Int. Arch. Allergy Appl. Immunol. 65, 62, 1981. 17. Balch, C. M., Dagg, M. K., and Cooper, M. D.. J. Immunol. 117, 447, 1976. 18. Gillman, C. F., Emeson, E. E., Veith, F. J., and Norin, A. J., Transplantation 29, 454, 1980. 19. Felsburg, P. J., Reilley, M. T.. and Sinnigen, J. K., Vet. Immunol. Immunopathol. 1, 251. 1980. 20. Stavitsky, A. B.. J. Immunol. 72, 368, 1954. 21. Felsburg, P. J.. Serra. D. A., Mandato. V. N., and Jezyk, P. F., Vet. Immunol. Immunopathol. 6, 353, 1984. 22. Halliwell, R. W., Adv. Vet. Sci. Camp. Med. 22, 221, 1978. 23. Heddle, R. J.. and Rowley, D., Immunology 29, 185, 1975. 24. Reynolds, H. Y., and Johnson, J. S., J. Immunol. 105, 698, 1970. 25. Ogra, P. L., and Karyon, D. T., J. Immunol. 102, 1423, 1969. 26. Perkins, J. C., Tucker, D. N., Knopf, H. L. S., Wenzel, R. P., Hornick, R. B., Kapikian, A. Z., and Chanock, R. M., Amer. J. Epidemiol. 90, 319, 1969. 27. Williams, W. A., and Gibbons, R. J., Science (Washington, D.C.) 177, 697. 1972. 28. Walker, W. A., Isselbacher, K. J., and Block, K. J., Science (Washington, D.C.) 177, 608, 1972. 29. Bey. R. F., Shade. F. J., Goodnow, R. A., and Johnson, R. C., Amer. J. Vet. Res. 42, 1130, 1981. 30. Bernis, D. A., Greisen, H. A., and Appel, M. J. G.. J. Infect. Dis. 135, 753, 1977.

CANINE 31. 32. 33. 34. 35. 36. 31. 38. 39. 40. 41. 42. 43. 44.

IgA DEFICIENCY

305

Hobbs, J. R., Lancet 1, 110, 1968. Kirkpatrick, C. H., and Ruth, W. E., Amer. J. Med. 41, 427, 1966. Goldberg, L. S., Barnett, E. V., and Fudenberg, H. H., J. Lab. C/in. Med. 72, 204, 1968. Stocker, R., Ammann, P., and Rossi, E., Arch. Dis. Child. 43, 585, 1968. Schwartz, R. M., Schenk, E. A., Berman, J., and Ellis, B. A., J. Lab. C/in. Med. 74, 203, 1969. Nell, P. A., Ammann, A. J., Hong, R., and Steihm. E. R.. Pediatrics 41, 71, 1972. Van Loghem, E., Eur. J. Immunol. 4, 57, 1974. Inoue, T., Okubo, H., Kudo, J., Ikuta, T.. Hachimine, K., Shibata. R.. Yoskinari, 0.. Fukada, K., and Yanase, T., J. Clin. Immunol. 4, 235, 1984. Cassidy, J. T., Oldham, G., and Platts-Mills, T. A. E., C/in. Exp. Immunol. 35, 296, 1979. Waldman, T. A., Broder, S., Krakauer. R., Durn, M., Meade, B.. and Goldman, C.. Trans. Assoc. Amer. Physicians 89, 215, 1976. King, M. A., Wells, J. V., and Nelson, D. S., Clin. Exp. fmmunol. 38, 306, 1979. Atwater, J. S., and Tomasi, T. B., C/in. Immanol. Immunopathol. 9, 379, 1978. Levitt, D., and Cooper, M. D., J. Pediat. 98, 52, 1981. Delespesse, G., Gausset, P., Cauchie, C., and Govaeits. A., C/in. Exp. Immunol. 24, 273, 1976.

Received December 10, 1984; accepted with revision March 25. 1985