Lymphocyte-dependent antibody-mediated cytotoxicity in Hashimoto thyroiditis

In the presence of normal human lymphocytes, decomplemented sera from twentynine out of thirty-nine patients with Hashimoto thyroiditis caused significant lysis of thyroglobulin-coated chicken red blood cells, as estimated by the release of ⁵¹Cr; the mean% specific ⁵¹Cr release being 14·1 ± 1·9 (SEM). Serum from twenty-one control subjects studied concurrently caused no significant lysis of thyroglobulin-coated chicken red blood cells; the mean% specific ⁵¹Cr release being −1·6±0·7 (SEM).     

Circulating lymphocyte subpopulations in Hashimoto thyroiditis

Peripheral blood and T and B lymphocytes and [¹²⁵I]thyroglobulin-binding lymphocytes were investigated in twenty-two euthyroid Hashimoto thyroiditis patients and in twenty-two age- and sex-matched normal subjects. Although the total lymphocyte count in Hashimoto patients (mean±SEM = 1226±187/mm³) was lower than in normal subjects (1603±156/mm³) this difference was not statistically significant. There was, however, a statistically significant reduction in the proportion of circulating T lymphocytes in the Hashimoto patients (mean±SEM = 57·4±2·5%) as assessed by the sheep red-cell rosette method when compared with the normal controls (mean±SEM = 66·7±1·8%). The proportion of B lymphocytes in the peripheral blood as assessed by indirect immunofluorescence, was not significantly different being 21·6±2·1% in the Hashimoto patients and 20·2±1·1% in normal subjects.

[¹²⁵I]thyroglobulin-binding lymphocytes, as assessed by autoradiography were present in the circulation of nineteen Hashimoto patients with a mean frequency of 8·37±1·15/10⁴ lymphocytes and in thirteen normal subjects with a mean of 8·84±0·93/10⁴ lymphocytes. There was no difference in the degree of [¹²⁵I]thyroglobulin binding between the two groups as determined by grain count analysis. There was no apparent correlation between age or thyroglobulin antibody titres and the frequency of [¹²⁵I]thyroglobulin-binding lymphocytes. Thyroglobulin-binding lymphocytes were increased 100-fold in a Hashimoto thyroid biopsy in comparison to the patient's peripheral blood.

     

Haemolytic activity of human blood monocytes. Lysis of human erythrocytes treated with anti-A serum

Purified monocytes and lymphocytes from peripheral blood of healthy human donors were tested in vitro for cytotoxicity against blood group A erythrocytes (RBC) treated with a human hyperimmune anti-A serum. Haemolysis was quantitated by the release of radioactivity from RBC pre-labelled with ⁵¹Cr-chromate.

The antiserum contained high titre antibodies which agglutinated A-RBC. After separation of serum on a Sephadex G-200 column the IgG containing fraction agglutinated A-RBC and precipitated blood group A substance. No or only weak antibody activity was detected in the IgA- and IgM-containing fractions.

Purified monocytes added to a 100-fold excess of RBC in the presence of 0·1% antiserum induced some haemolysis. It was calculated that one monocyte was able to lyse 2–3 RBC within 18 hr incubation. In contrast, lymphocyte suspensions containing more than 97% small lymphocytes had no or only weak haemolytic activity at a lymphocyte-RBC ratio of 25: 1. The effector cells of the monocyte fraction adhered to glass and were eliminated by incubation with carbonyl iron. Phagocytosis by monocytes of antibody-treated RBC was frequently observed. Loading of monocytes by treatment with heat-killed Candida albicans or carbonyl iron particles suppressed their haemolytic action. Cytotoxicity was impaired after treatment of monocytes with 10⁻⁴ M sodium iodo acetate. After separation of serum on Sephadex G-200 column all monocyte induced haemolytic activity was found in the IgG containing fraction. It is assumed that haemolysis is induced by the interaction of monocytes with an IgG anti-A antibody of the antiserum.

     

Quantitative studies on phytohaemagglutinin-induced cytotoxicity by human lymphocytes against homologous cells in tissue culture

The cytotoxic effect of normal human blood lymphocytes on Chang cells in tissue culture was investigated. Cell damage was measured by release of ⁵¹Cr from pre-labelled tissue culture `target' cells. This method was sensitive and rendered highly reproducible results. Released ⁵¹Cr was not re-utilized.

About 25 per cent of the ⁵¹Cr was spontaneously released from labelled Chang cells when incubated for 24 hours at 37°. Lymphocytes at a lymphocyte/Chang cell ratio of 25:1 led to a slight increase of this release. When phytohaemagglutinin was also present, about 50–60 per cent of the isotope appeared in the medium. Under these conditions target cells were significantly damaged within 1 hour. At a lymphocyte/Chang cell ratio of only 1:1, weak cytotoxic effects were also noted after 24 hours of incubation. The results of dose—response experiments suggested that a considerable proportion of the lymphocytes participated in the reaction. Individual variation of the cytotoxic effect of lymphocytes from different donors suggested that it could be related to the degree of histoincompatibility between lymphocytes and Chang cells. Under the present conditions contaminating erythrocytes or granulocytes did not interfere with the cytotoxic action of the lymphocytes.

     

Characterization of human lymphoid cell-mediated antibody-dependent cytotoxicity (LDAC)

Cytotoxicity was studied in a model system using chicken erythrocytes (Ch RBC) labelled with ⁵¹Cr as target cells and human peripheral blood lymphoid cells as effector cells. In vitro human lymphoid cells are highly efficient in destroying target cells coated with anti-target cell antibody, the mean net percentage cytotoxicity of lymphoid cells from fifty-eight control subjects being 64·27±2·06 (SEM). In the absence of antibody the mean net percentage cytotoxicity was 6·90±0·46. As little as 10 ng rabbit anti-Ch RBC IgG was required to cause significant target cell lysis. Studies on the nature of the lymphoid cell-dependent cytotoxic antibody showed that it is localized in the 7S IgG region of whole serum and that an intact Fc region is required; the F(ab')₂ fragment obtained by pepsin digestion of IgG was inactive although able to inhibit the cytotoxic activity of the whole undigested IgG. Investigation of the kinetics of LDAC showed that when antibody was added to the final culture medium target cell lysis progressed rapidly (detectable within 2 hr) and linearly with time up to 8 hr. Thereafter the rate of lysis decreased reaching a maximum after 12 hr culture. With cultures containing target cells which have been pre-incubated with antibody, lysis occurred even more rapidly, detectable within 30 min and reaching a maximum after only 3–4 hr culture. The maximum cytotoxicity in this system was, however, lower than that obtained when antibody was added directly to the culture medium. Cytotoxicity could be inhibited by the addition of aggregated human IgG, as little as 5 μg causing 100% inhibition of target cell lysis. Study of the nature of the effector lymphoid cell showed, first, that viable cells were required, twice frozen/thawed lymphoid cell suspensions being inactive; secondly, that active protein synthesis by the effector cell was not an essential prerequisite, pretreatment of lymphoid cells with mitomycin C having no significan effect on their ability to lyse antibody-coated target cells but significantly reducing their ability to transform in response to the mitogen PHA; thirdly, that the effector cell is non-phagocytic, non-plastic or glass-adherent and does not bear surface immunoglobulin; and fourthly that significant cytotoxicity is detectable even with a total lymphoid cell to target cell ratio of 1:2.     

A kinetic study of antibody producing cells in the spleen of mice immunized intravenously with sheep erythrocytes

The number of antibody producing cells, i.e. rosette forming cells (RFC) has been studied in the spleen of mice injected intravenously with a full immunizing dose of sheep RBC.

The spleen of a non-immunized mouse contains a background of about 70,000 RFC (normal RFC), which do not appear to be the `target cells' for antigens of sheep RBC. Our findings suggest that the spleen of a mouse contains about 4000 `target cells' which initiate the immune response to sheep RBC.

In the primary response, the rise of RFC is exponential for about 96 hours with a doubling time of 13 hours involving seven to eight consecutive doubling periods. The peak value of RFC is 1·6 × 10⁶ per spleen.

In the secondary response, the doubling time of RFC is 6–7 hours. The exponential rise lasts 72 hours and includes nine to eleven doubling periods leading to a peak of 3·5 × 10⁶ RFC/spleen.

Adjuvant in the primary response leaves the doubling time of RFC unaltered but prolongs the exponential rise until the 120th hour leading to a peak of 6·3 × 10⁶ RFC/spleen after ten doubling periods. The effects of priming and adjuvant are not fully additive.

     

Force: velocity pproperties of kitten muscles

1. The characteristics of isometric contractions and force:velocity properties of the extensor digitorum longus (EDL) and soleus (SOL) muscles of neonatal kittens were determined in situ.

2. The mean contraction time is 51 msec for EDL and 70 msec for SOL and the half-relaxation time is 51 msec for EDL and 109 msec for SOL.

3. The average maximum isometric tetanic tension per unit cross-sectional area of muscle is 1·27 kg/cm² for EDL and 1·17 kg/cm² for SOL.

4. The average twitch:tetanus ratio is 0·28 for EDL and 0·119 for SOL; the low value for SOL was found for both indirect and direct stimulation.

5. The average maximum speed of shortening of a sarcomere is 22·8 μ/sec for EDL and 12·7 μ/sec for SOL.

6. These properties of neonatal muscles are compared with those of adult cat muscles and discussed in connexion with differentiation of mammalian muscles into fast and slow types.

     

Cytotoxic lymphocytes in Hashimoto thyroiditis: an in vitro assay system using 51Cr-labelled chicken red blood cells coated with thyroglobulin

An in vitro method is described to detect lymphocytes in patients with Hashimoto thyroiditis that are cytotoxic to thyroglobulin-coated chicken red blood cells. Using this technique, the cytotoxic index of lymphocytes from patients with Hashimoto thyroiditis was 25·46±3·81 (SEM), which is significantly different from that obtained with lymphocytes from control subjects, 6·28±0·80.     

Influence of chronic neostigmine treatment on the number of acetylcholine receptors and the release of acetylcholine from the rat diaphragm

1. Rats were treated twice daily for 7 days with neostigmine and the diaphragm was isolated for study of its acetylcholine content, release upon nerve stimulation and the number of receptors in the end-plate.

2. While the content of total acetylcholine was unchanged, the release of acetylcholine on stimulation with trains of 500 pulses at 100 Hz every 20 sec was reduced by about 50%. Five days after the end of neostigmine treatment the release of acetylcholine recovered to normal.

3. The total number of acetylcholine receptors in the end-plate as measured from the binding of N, O-di[³H]acetyl α-bungarotoxin was reduced from 2·1 × 10⁷ to 1·2 × 10⁷ per end-plate.

4. The above-mentioned changes are not due to acute pharmacological effects of neostigmine, nor to an immediate effect of cholinesterase inhibition but presumably due to chronic accumulation of acetylcholine at the neuromuscular junction.

     

Effects of islet cell surface antibodies and complement on the release of insulin and chromium from perifused β cells

The dynamics of insulin and chromium release from prelabelled rat pancreatic islet cells were studied by perifusion of cells supported in a column of Bio-Gel P-2 polyacrylamide beads. The column-perifused β cells released insulin in a biphasic pattern in response to 30 mmol/l D-glucose and in a monophasic pattern to 20 mmol/l L-arginine. Rat islet cells, first exposed to a rabbit anti-rat islet cell surface serum and complement and then added to the column, were unable to release insulin in response to 30 mmol/l D-glucose and 0·1 mmol/l 3-isobutyl-1-methylxanthine (IBMX). In order to study cytotoxicity by an additional approach, islet cells were prelabelled with radioactive chromium (Na₂⁵¹CrO₄). These cells did not release either insulin or ⁵¹Cr in response to glucose. Furthermore, exposure of the cells to surface antiserum and complement before perifusion did not induce either chromium or insulin release. Similar results were obtained when glucose alone or combined with surface antiserum was added to the perifusate bathing rat islet cells incubated with 0·1 mmol/l non-radioactive Na₂CrO₄ before perifusion. However, a transient, dramatic release of insulin from these cells was induced by adding complement to the perifusion medium (containing surface antibodies). These results indicate that complement-dependent cytotoxicity of islet cell surface antibodies involves different phenomena. Firstly, the cytotoxic reaction results in a transient release of insulin whether the physiological release mechanisms were blocked by chromium or not. Secondly, in cells not treated with chromium the cytotoxic reactions renders the β cells unable to release insulin in response to glucose.     

The cellular transfer of immunity to Trichostrongylus colubriformis in an isogenic strain of guinea-pig: III. The localization and functional activity of immune lymph node cells following syngeneic and allogeneic transfer

Mesenteric lymph node cells obtained from guinea-pigs immunized with Trichostrongylus colubriformis have been transferred allogeneically (McMaster outbred to Heston inbred) and syngeneically (Heston to Heston inbred) by intravenous injection. The immune cells were transferred on days 6, 7 and 8 of infection when the parasite was in the susceptible fourth stage of development. The syngeneically transferred cells caused rejection of the parasite but the allogeneically transferred cells were not effective.

Immune mesenteric lymph node cells were labelled in vitro with ⁵¹Cr before transfer to infected recipients. The percentage of the intravenously injected cells which localized in infected small intestine has been calculated from γ-counts recorded at 1, 6, 16, 24 and 48 hours after injection of the labelled cells. The mean percentage of cell dose per 12 in. of small intestine recorded following syngeneic transfer was 1·29 and following allogeneic transfer 0·68. A significant difference in the localization of allogeneic and syngeneic cells in the small intestine was already apparent at 6 hours following intravenous injection.

     

Cellular cytotoxicity to measles virus during natural measles infection

Little is known about cellular cytotoxicity to measles virus during natural measles infection which is still a major cause of death in many parts of the world. Therefore we measured the ability of peripheral blood mononuclear cells (PBM) from children with measles to kill Hela cells persistently infected with measles virus. In a 6-hr ⁵¹CR-release assay antibody-independent cellular cytotoxicity was shown to be low during the acute stage of measles. This rose to a maximum 1 week after the onset of the rash and fell rapidly on recovery 2 to 3 weeks later. The respective means values for the three periods (expressed as specific immune release of ⁵¹Cr) were 7·9±8·4%, 31·0±16·4% and 6·1±7·7%. Killing in this assay was not effected by T lymphocytes, for concentration of these cells by three different methods failed to increase cytotoxic power. In contrast peripheral blood mononuclear cells depleted of T lymphocytes showed greatly increased antibody-independent cellular cytotoxicity. Antibody-dependent cellular cytotoxicity was not found to vary significantly with the stage of measles. The mean values were 30·0±13·6%, 26·6±11·7% and 23·9±12·1% for the periods 0–2, 3–14 and 15–30 days after the onset of the rash. Both antibody-independent and antibody-dependent cellular cytotoxicity of PBM were lowered by layering these cells on immune complexes fixed to plastic or by incubating them with normal rabbit γ-globulin. Antibody-dependent cellular cytotoxicity was also lowered in the presence of 10% acute-phase autologous plasma. We concluded that antibody-independent cytoxocity was effected either by natural killer cells or by K cells using traces of antibody present in the assay. Antibody-dependent cellular cytotoxicity which is due to K cells may be modulated by circulating immune complexes during the course of disease.     

Mechanisms of corticosteroid action on lymphocyte subpopulations. III. Differential effects of dexamethasone administration on subpopulations of effector cells mediating cellular cytotoxicity in man

The present study investigated the effect of dexamethasone (DEX) administration on different populations of mononuclear cells and neutrophils mediating antibody-dependent cellular cytotoxicity (ADCC) against different target cells.

Mononuclear cells (lymphocytes and monocytes) and neutrophils were obtained from twenty-seven normal volunteers at 0, 4, 24 and 48 hr after oral administration of 21 mg of DEX. ADCC was determined utilizing the following targets: human red blood cells (HRBC), Chang liver cells (Ch) and human heart cells (HHC). The predominant mononuclear effector in HRBC killing was shown to be a monocyte and in Ch and HHC killing, a K cell. As previously shown, DEX produced a profound monocytopenia and lymphocytopenia at 4 hr with a return of lymphocyte counts to normal and monocyte counts to supra-normal at 24 hr. At the point of maximal monocytopenia, monocyte-mediated HRBC killing decreased from a geometric mean of 14 to 4 lytic units per 10⁸ effector cells (P<0·05) and rebounded at 24 hr to a mean of 39 lytic units (P<0·02) with the rebound monocytosis. At the point of absolute lymphopenia (4 hr), there was a relative enrichment in the proportion of lymphocytes bearing an Fc receptor (K cells, P<0·01). Concomitant with this was an increase in ADCC against Ch and HHC from geometric means of 1121 to 7172 lytic units and 939 to 7354 lytic units (P<0·001) respectively. Thus, a major action of DEX administration on mononuclear ADCC was to differentially enrich or deplete different effector cells to and from the circulation, causing changes in cytotoxicity. Since the cytotoxicity paralleled the proportion of effector cells, the cells remaining in the circulation following DEX administration retained normal antibody-dependent cytotoxic capabilities.

Neutrophil-mediated ADCC against HRBC significantly increased at 4 hr from a geometric mean of 3785 to 20142 lytic units (P<0·02) concomitant with the blood neutrophilia and remained elevated for 72 hr. Neutrophil ADCC against Ch and HHC was minimal and was not affected by DEX administration.

This study demonstrates differential effects of corticosteroids on antibody-dependent cell-mediated cytotoxicity, reflecting changes in proportion of circulating effector cell populations. These changes depend on (a) the target cell employed, (b) the effector cell mediating cytotoxicity and (c) the duration of time since the last dose of corticosteroid.

     

An electrophysiological study of the effects of atropine and physostigmine on transmission to the guinea-pig vas deferens

1. The effects of physostigmine and atropine on transmission to the longitudinal musculature of in vitro preparations of the guinea-pig vas deferens have been examined using intracellular micro-electrodes.

2. Atropine (5 × 10^-7 to 10^-6 g/ml.) increased the rate of decay of excitatory junction potentials (EJPs) in response to post-ganglionic stimulation.

3. Physostigmine (5 × 10^-6 g/ml.) reduced the mean resting potential of the muscle cells from -60·5 to -51·5 mV and lowered the voltage of post-ganglionic stimulation necessary for initiation of an action potential in the muscle. In some but not all of the cells studied the time course of the EJP was markedly prolonged.

4. At concentrations which did not alter the response to post-ganglionic stimulation (5 × 10^-7 to 10^-6 g/ml.), physostigmine caused fully facilitated EJPs to appear with the first pulse of a preganglionic train of stimulation.

5. Atropine antagonized all the above effects of physostigmine.

6. Physostigmine (5 × 10^-7 to 10^-6 g/ml.) also lowered the voltage of preganglionic stimulation necessary for initiation of an action potential in the muscle. This effect was not antagonized by atropine.

7. The results are interpreted as being evidence for the existence of separate cholinergic and adrenergic motor fibres to the musculature of the guinea-pig vas deferens.

     

The binding of thyroid hormones to phospholipid membranes

1. Thyroxine and tri-iodothyronine are bound by liposomes (bimolecular phospholipid membranes produced by dispersing egg-yolk lecithin in water). At pH 7·4 the `partition coefficient' between water and the phospholipid is 1·2 × 10⁴ for thyroxine and 2·2 × 10⁴ for tri-iodothyronine. The partition coefficient falls off rapidly in more alkaline solutions as the phenolic group on the hormones becomes ionized.

2. The binding of thyroxine by membranes is very rapid, occurring at rates equivalent to those found in the binding of thyroxine by serum.

3. Lecithin membranes are readily permeable to thyroxine (the oil:water partition coefficient is 1·7).

4. The binding of thyroid hormones by perfused rat hearts and the binding of thyroid hormones by liposomes are similar in their sensitivity to pH. Simple alcoholic extracts of tissues have strong thyroxine-binding activity.

5. These results are discussed in relation to the nature and function of the tissue thyroxine-binding reaction.

     

Assay of anti-hapten antibody with the aid of hapten-coupled bacteriophage

Treatment of bacteriophage with 3-iodo-4-hydroxy-5-nitrophenyl acetic acid chloride (NIP) in aqueous medium killed a proportion of the phage but the survivors were made susceptible to inactivation by rabbit immune sera to NIP-chicken globulin conjugate. The serum factor inactivating NIP-phage (T2) was eluted from a Sephadex G-200 column as 7S γ-globulin and was neutralized by a sheep antiserum against electrophoretically purified rabbit γ-globulin. The inactivation was strongly inhibited by the hapten and its derivatives. As little as 0.001 micromole per millilitre of NIP—ε-amino caproic acid was inhibitory.

Inactivation of NIP-phage was ascribed to anti-NIP antibody.

Inactivation of NIP-phage (T2) in strong anti-NIP sera approximately followed first-order reaction kinetics until 99 per cent of the phage became inactivated. When incubated with phage for 6 hours concentrations of antiserum as low as 10^-7 (1.9×10^-5 μg/ml) or less caused measurable inactivation of the phage. The threshold quantity of antibody was estimated to be less than 10^-5 μg.

     

The incidence of thyroglobulin antibodies and thyroid enlargement in a general practice in north-east England

Antibodies to thyroglobulin in a titre of 1:25 or more were found in 16·2% of women and 4·3% of men between the ages of 21 and 80 years in a random sample of the population from a general practice in the north-east of England. The incidence of antibodies was highest in the seventh decade in both sexes. High antibody titres (1:78125 or more) were found in 4·6% of women and 1·6% of men and it is suggested that this may represent the incidence of diffuse thyroiditis in the population.

Significant thyroid enlargement was found in 12% of women and 0·9% of men, the corrected incidence of goitre obtained by averaging the frequency of goitres in each decade between 21 and 80 years was 8·9% and 0·9% respectively.

     

Membrane dynamics of HL-A–anti-HL-A complexes of human normal and leukaemic lymphocytes

Membrane dynamics of human lymphocytes treated with anti-HL-A antibodies were studied by means of indirect membrane fluorescence. When cells were incubated at 37°C for several hours, a striking diminution of the labelling index was observed. Furthermore, the pattern of the membrane fluorescence changed from a ring-shaped or patchy fluorescence to a polar concentration of the label (`cap formation'). `Cap formation' could be inhibited by incubation at 0°C and by 3 × 10⁻² M sodium azide. Part of the labelled HL-A–anti-HL-A complexes could be detected within intracellular vesicles, which suggests uptake by endocytosis.

Lymphocytes of patients with chronic lymphocytic leukaemia (CLL) did not show a decrease of the labelling index. `Cap formation' was almost completely lacking, and uptake of HL-A–anti-HL-A complexes by endocytosis was not observed.

These findings indicate an impairment of membrane dynamics of CLL lymphocytes.

The importance of these membrane dynamics for the activation of lymphocytes is discussed.

     

Graft reaction in tissue culture: II. Quantification of the lytic action on mouse fibroblasts by rat lymphocytes sensitized on mouse embryo monolayers

A method for assaying the lytic action of large pyroninophilic cells (LPC) in cultures of rat lymphocytes on target fibroblasts in mouse embryo monolayers has been described. The basis of the assay was the labelling of fibroblasts with ⁵¹Cr. The rate of ⁵¹Cr release to the medium was found to be proportional to the number of LPC and expressed precisely the actual number of lysed fibroblasts. The lytic activity in LPC suspension was expressed by the lysis index which is the number of LPC per fibroblast lysed at 50 per cent lysis. A suspension of LPC collected 5–6 days after exposure of rat lymphocytes to the mouse monolayers showed a lysis index ranging from 0.7 to 1.3 after 16–48 hours of incubation. The lytic power of a culture was expressed by a value which was obtained by dividing the total number of LPC in a culture by the index. This determined the total number of fibroblasts lysed if the whole culture content were distributed among plates of test monolayer in such a way as to produce 50 per cent lysis in all the plates. Cultures started with 25 × 10⁶ to 30 × 10⁶ lymphocytes produced on the 5th and 6th day a lytic power of as high as 6 × 10⁶ to 12 × 10⁶ fibroblasts. The study has indicated that rats injected with horse serum and boosted 2 days prior to culturing were completely unable to produce graft reaction cultures. Lymphoid cells from non-boosted rats did generate LPC with lytic ability. A quantitative study of the specificity of the lytic reaction has indicated that LPC originated on C57BL/6 monolayers lysed much more strongly monolayers isologous to originator than C3H monolayers, and vice versa.     

Antiethinyloestradiol antibody activities in oral contraceptive users

Oral contraceptives (OC) can induce in some women, serum immune complexes precipitated in 25% saturated (NH₄)₂SO₄, especially in the case of a thrombotic complication. In this work, the presence of antiethinyloestradiol (anti-EE) antibodies in the complexes was investigated. Binding of ethinyloestradiol-³H (EE-³H), either by (NH₄)₂SO₄ precipitate directly, or in equilibrium dialysis experiments with the anti-EE gamma-globulin isolated through a purification-activation method combining dilution chromatography and affinity chromatography, was measured. The purified proteins were identified by immunoprecipitation and isoelectrophoresis.

Twelve sera from OC users and three from women who had never used them were tested. Eleven of the twelve users had serum immune complexes. Normal human gamma-globulin pools and gamma-globulins prepared from a rabbit immunized with EE, were used as negative and positive controls.

Results demonstrated that serum complexes precipitated in 25% saturated ammonium sulphate contained EE-³H binding gamma-globulins. For seven out of eleven cases, the association constant K_a was calculated and varied from 4 × 10⁵ M⁻¹ to 2·6 × 10⁷ M⁻¹, and the valence from 1·4 to 2, a result consistent with an IgG antibody activity.

These anti-EE antibodies were found in the serum complexes from women on OC with or without thrombosis. However their amount seemed to be greater in the cases with thrombosis.

In the serum of women who had never used the pill, in the only OC user who had no serum complexes in 25% saturated ammonium sulphate, and in the pool of normal human gamma-globulins, no anti-EE activity was detected.