GB1580318A - Antibodies active against human hemoglobin a1c - Google Patents

Description

(54) IMPROVEMENTS IN OR RELATING TO ANTIBIODIES ACTIVE AGAINST HUMAN HEMOGLOBIN A,c (71) We, THE ROCKEFELLER UNIVERSITY, of 66th & York Avenues, New York, New York 10021, United States of America, a non profit Corporation organised and existing under the laws of the State of New York, do hereby declare the invention for which we pray that a patent may be granted to us, and the method by which it is to be performed, to be particularly described in and by the following statement:- The present invention relates to an antibody against human hemoglobin (Hb) A,c and the use of an antibody to quantitatively detect hemoglobin A,c in a blood sample.

Hemoglobin A,c is a glycohemoglobin with an amino acid structure which is identical to that of Hb A: the only detectable difference is the presence of 1amino I-deoxyfructose attached in the 2,3-diphosphoglycerate pocket to the N-terminal valine in the beta-chain of Hb A,c. The modification of Hb A to form Hb A,c is a continuous post-translational process, the rate of which is a function of the blood glucose concentration. The level of Hb A,c therefore reflects the status of the individual's carbohydrate metabolism. Normal adults have above 90 /n of their total hemoglobin as Hb A,, 23 /n as Hb A,b and 3-6 wt /n of their total hemoglobin as Hb A,c whereas the range in juvenile and maturity onset diabetics is 6-15 wt% as Hb A,c. A similar increase in Hb A,c concentration has been noted in mice with genetic and chemically induced diabetes and in pancreatectomized dogs.

Diabetes mellitus has been found to occur naturally or can be induced in virtually every species in the animal kingdom. Greater insight into human diabetes can be gained by studying this disease process in animals whose diabetes closely resembles the human condition. Such animal models include the diabetic dog and mouse; see R. Engerman et al. in Diabetes 26: 760--769 (1977); K. P. Hummel et al.

in Science 153: 11271128 (1966); and A. A. Like et al. In Am. J. Pathol. 66:193- 204 (1972). As has been reported by Koenig and Cerami in Proc. Natl. Acad. Sci.

USA 72: 3687-91(1975), both of these species demonstrate increased levels of Hb A,c in the diabetic state. These animal hemoglobins A,c cross-react sufficiently with the human Hb A,c antibody described herein so that their concentrations may be determined accurately by a radioimminoassay (RIA) method. An RIA for animal Hb A,, is very useful because this technique requires only microgram quantities of hemoglobin, compared to milligrams for the widely used prior art column chromatographic method of Tribelli et al. described in New Engl. J. Med. 284: 353-7 (1971). RIA techniques for animal Hb A,c are thus useful in facilitating the evaluation of new drugs and other forms of therapy designed to treat human diabetes and accordingly should facilitate research into the basic pathophysiology of this disease.

Recent studies have indicated that the quantification of Hb A,c concentration is a useful means of assessing carbohydrate intolerance as well as the adequacy of control in patients with diabetes; see Koenig et al. in New England J. Med. 295: 417-420 (1976). One of the difficulties in applying this measurement to clinical studies has been the technical problem of assaying Hb A,c in a large number of samples with the column chromatography method currently available which has been described by Trivelli et al, in New England J. Med. 284:353-357 (1971). In order to circumvent this obstacle, we have studied the immunological properties of Hb Aic with a specific antiserum. The present invention applies these results to the radioimmunoassay of glycohemoglobins.

According to one aspect of the present invention, there is provided an antibody against human hemoglobin A,c which is substantially free of crossreactivity against the human hemoglobins A,, Aia and A,b.

According to another aspect of the present invention there is provided a method of preparing these antibodies which comprises administering purified hemoglobin A,c antigen to an animal capable of producing antibodies thereto and collecting said antibodies from the animal.

Preferably the hemoglobin A,c antigen is administered to an animal whose metabolism does not naturally form hemoglobin A,c.

According to yet another aspect of the invention there is provided a method of quantitatively detecting the presence of hemoglobin A,c in a test antigen comprising a body fluid sample which includes reacting said test antigen with an above mentioned antibody and determining the amount of an antigen-antibody complex formed by radioimmunoassay.

Preferably the antibody is first reacted with a test antigen free of index antigen after which the resultant product is reacted with the index antigen. In a preferred embodiment, the reaction with the index antigen is effected by incubation at a temperature of from ambient temperature to 370C for 0.5 hour.

According to another aspect of the present invention, there is provided a method of diagnosing diabetes mellitus comprising quantitatively detecting the presence of hemoglobin A,c in a blood sample by the above method and determining whether the level of hemoglobin A,c is in excess of 6 wit%.

According to yet another aspect of the invention, there is provided a diagnostic test kit for quantitatively assaying hemoglobin A,c, comprising an above mentioned antibody and a radioactively labelled, substantially pure hemoglobin A.c index antigen.

Because the transformation of hemoglobin A into hemoglobin A,c is a function of glucose concentration in the blood, there is a great need for a quantitative assay method which is rapid and easily conducted without prohibitively expensive equipment. By virtue of certain species cross-reactivity discussed hereinafter, the provision of such a test in accordance with the present invention now provides an additional tool for the study of diabetes and similar diseases causing an elevation in blood glucose concentration in standard laboratory animals.

Highly specific antisera have been prepared that can recognize single amino acid substitutions in human hemoglobin by the radioimmunoassay (RIA) method, e.g. see Javid and Pettis in J. Lab. Clin. Med. 88: 621-626 (1976); Rowley et al. in Blood 43: 607-611(1974); and Garver et al. in Science 196: 1334--36 (1977).

Differences between the structure of the hemoglobin used for immunization and that of the immunized animal's own hemoglobin largely determine the specificities of the antibodies produced. While a wide variety of different species can be employed as the immunized animal for preparing the present antibodies to hemoglobin A,c, it is preferred to immunize an animal whose metabolism does not naturally form hemoglobin A,c e.g. a cat, goat or sheep. Sheep hemoglobin does not react with 2,3-diphosphoglycerate and may lack the configuration of the "DPG pocket" which permits the glycosylation of the beta-chain N-terminus. Indeed, glycosylated hemoglobin cannot be demonstrated in sheep red cell hemolysates.

The sheep, then, apparently recognizes the N-terminus of the human Hb A,c as antigenic. The immunodominant feature of this antigenic determinant must require the spatial conformation provided by l-amino-l-deoxyfructose. Reduction of the keto group strikingly and selectively reduces the reactivity with anti-A,c, while leaving the affinity of other determinants for anti-A unaffected. The ability of the antibody to distinguish between a ketone and an alcohol might in fact reflect its ability to recognize a ring structure, since the keto moiety could readily form a hemiketal with the hydroxyl group of either carbon 5 or 6. The cross-reaction of Hb Aia and Hb A,b with the antibody made to human Hb A,c is of interest since both molecules are probably glycohemoglobins. An exact structure has not been assigned to these hemoglobins.

Dog and mouse Hb A,c react less well than their human counterparts. This observation suggests that the steric fit of the antibody includes more than the sugar molecule and probably extends to surface features of the protein adiacent to it.

The specific immunologic recognition of Hb A,c has an immediate practical value. The quantification of this minor hemoglobin fraction is valuable in monitoring the control of diabetic patients, as has been noted in Koenig et al. In New England Journal of Medicine 295: 417-20 (1976). Currently available assays call for the column chromatographic separation of Hb Aic from the other hemoglobin components of the hemolysate. This is a tedious and not entirely accurate semi-quantitative method since only some 7-80 of the hemoglobin applied to the column is recovered.

The level of Hb A,a+Hb A,b in hemolysates is generally less than 50 wt /" that of Hb A,, and, under assay conditions, their contribution to the blocking of anti A,c is less than 10 wt /n of a comparable amount of Hb A,c. It is therefore unlikely that these two minor glycohemoglobins contribute significantly to the radioimmunoassay value of Hb A,c. In 14 hemolysates in which Hb A,c was assayed by RIA and all three glycohemoglobins were measured by chromatography, correction of the RIA values for the contribution by Hb A,a+b did not significantly affect the correlation between the two methods.

The RIA can be standardized with purified Hb A,c and Hb Ao and is not subject to error due to the selective loss of one of another hemoglobin fractions from the samples. This method offers a number of advantages over available techniques. The assay is specific and reproducible, yet simple enough to permit the processing of 30 or more samples under identical conditions within a working day.

It should find ready application in the study of diabetes in man, as well as in experimental animals.

Hb Ao and Hb A,c are immunologically identical except for a single antigenic determinant. Therefore, following the adsorption of anti-Hb A,c with Hb Ao, the residual specific antibody has a low titer and affinity. This inherent problem may be overcome in a preferred embodiment by modifying the usual RIA method: First, the antibody can be incubated sequentially with the unlabelled or test and radioactively labelled or index antigens, rather than with a mixture of the two, thus maximizing the blocking of the antibody by the test antigen. A further possible modification is to limit the incubation with the index antigen is to 0.5 hour, rather than overnight, so that the displacement of antigen from the low-affinity antibody is minimized.

Specific non-limitative Examples of the present invention, broadly described above, will now be given in more detail. All temperatures are set forth uncorrected in degrees Celsius; unless otherwise indicated, all pressures are ambient and all parts and percentages are by weight.

EXAMPLE I Preparation of Hemoglobin Samples Blood samples were obtained from normal individuals and from patients with diabetes mellitus. Hemolysates were made from saline washed red cells by hypotonic lysis. Stroma was removed by centrifugation. Total hemoglobin concentration was measured as cyanmethemoglobin, using the method of Cartwright described in "Diagnostic Laboratory Hematology", 4th edition, Grune and Stratton, New York (1972). Assay of the Hb A,c fraction was performed as previously described by Trivelli et al.

The main hemoglobin fraction Ao and the minor hemoglobins Aia, A,b and A,c were isolated and purified by column chromatography from the erythrocytes of normal volunteers following the procedures described in New England J. Med. 284: 353-357 (1971) and Proc. National Acad. Sci. USA 72:3687-91(1975). The Hb Ao fraction thus isolated was then stripped of ionically-bound organic phosphates by dialysis in vacuum-expanded bags against 0.5 M NaCl-0.01 M sodium phosphate buffer, pH 7.0. The stripped Hb Ao was then repurified by the above column chromatographic procedure. This process is necessary to prevent small amounts of Hb A,b- like species from forming in the Hb Ao fraction; see V.J.

Stevens et al. in J. Biol. Chem. 252: 2998-3002 (1977). A similar chromatographic method was used for the separation of mouse and dog glycohemoglobins except that the equilibrating and eluting buffers were 0.05 M sodium phosphate, 0.01M KCN, with pH 6.68 for the dog and pH 6.78 for the mouse hemolysates. The components Aia and A Ib were eluted in one peak. The glycoproteins were concentrated by vacuum dialysis. All purified hemoglobins were stored at -200C in the PCG buffer described below.

For reduction with NaBH4, each glycohemoglobin was dialyzed against 0.1M sodium phosphate buffer, pH 7, and was subsequently reacted with a 200-fold molar excess of NaBH4 at room temperature for one hour. The unreacted sodium borohydride was removed by dialysis. The synthesis of glycosylated dipeptides has been previously described by Koenig et al. in J. Biol. Chem. 252: 2992-97 (1977).

EXAMPLE 2 Preparation of Antibodies A cheviot sheep was injected biweekly with 10 mg of purified human Hb A,c in 3 ml water and 2 ml Freund's adjuvant. The first 8 injections utilized complete adjuvant and the five subsequent injections were with incomplete adjuvant. Each 5 ml injection was given into 25 intradermal sites. Following the 8th injection, 500 ml blood was removed from the sheep on each non-immunization week. The blood was left overnight at 400C and the serum was collected by centrifugation and stored at -85 C. This is referred to herein as the parent antiserum. This antiserum showed minimal difference in its reactivity with Hemoglobin A,c and Hemoglobin Ao. This difference was amplified with progressive absorption of the antiserum with Hb Ao at the expense of a considerable fall in antibody titer. Approximately 10% of the initial titer was retained as specific anti-A,c. Anti-A, the fraction of the parent antiserum that was retained on and eluted from the Hb Ao immunoabsorbent, did not discriminate between Hb Aic and Hb Ao.

In the reaction of human hemoglobin A fractions with anti-A,c, the absorbed antiserum clearly distinguishes Hb Ao from its glycosylated derivatives. Of the latter, Hb A,c is most effective in blocking the antiserum; Hb Aia and Hb A,b show about 5% and 10% of its reactivity, respectively. The four hemoglobin A fractions are equally effective in blocking anti-A, which recognizes only the determinants unrelated to the carbohydrate ligand.

Treatment of the hemoglobins with NaBH4 had no effect on their reaction with the non-specific anti-A, but it led to a striking reduction of reactivity with anti A,c, as shown in the following Table 1: TABLE 1 The Specificity of Borohydride Reduction for the Immunodominant Determinant of Glycohemoglobins Maximal blocking of Anti-A1 Anti-A Hb A,a, unmodified 78% 97 /" Hb A1b, unmodified 90 /" 94% Hb A,c, unmodified 98% 91% HbAia, reduced 18 /,, 94% Hb A,b, reduced 24 /" 89% Hb Arcn reduced 19% 94% The common antigenic determinants of hemoglobin (Column 2) are present in all three glycohemoglobins and are not affected by borohydride reduction. The specific glycohemoglobin determinant (Column 1) is most reactive in Hb A,c, least so in Hb Aia, and is significantly altered by borohydride reduction.

EXAMPLE 3 Preparation of Radioimmunoassay Reagents The following reagents were prepared: TBS: (tris-buffered saline) 0.05 tris-hydroxymethyl-aminomethane, 0. I M NaCI, pH 8.3.

PBS: (phosphate buffered saline) 0.05M sodium phosphate, 0.1M NaCI, pH 7.0.

BGG: 2% bovine gamma globulin in TBS. This was used for the final dilution of all antisera for the radioimmunoassay.

PCG: (phosphate-cyanide-glycerol) 0.1M sodium phosphate, 1.5mM KCN in 40 S glycerol, pH 7.0. This buffer was used for storing hemoglobin samples at -200C, at which temperature they did not freeze. The storage stability of the samples was considerably enhanced by this means.

Ammonium sulfate: Saturated solution, adjusted to pH 7.0 with SM NaOH.

Antisera: An aliquot of the parent antiserum was repeatedly absorbed with agarose-linked Hb Ao according to the method of Javid and Liang, described in J.

Lab. Clin. Med. 82: 991-1002(1973) and the specificity of the residual antibodies for Hb A,c was monitored as described below. The final preparation is designated anti-A,c. The cross-reacting antibodies which bound to the immunoabsorbent were eluted with 0.2M glycine, pH 2.8, and dialyzed against PBS. This preparation is referred to as anti-A since it is directed against those antigenic determinants common to Hb A,c and Hb A,.

Each antiserum was calibrated against the index antigen following the procedure of Javid and Yingling described in J. Clin. Invest. 47 2290--2296 (1968).

For the parent antiserum and anti-A, the equivalences (micrograms Hb A,c bound/ml antiserum) were 1120 micrograms/ml and 185 micrograms/ml, respectively. These antisera were diluted in BGG to a final titer of 0.375 micrograms/ml. Equivalent amounts of index antigen were used for these antisera.

The calibration curve for anti-A,c did not show a sharp end point because of the low affinity of the antiserum; the equivalence was estimated at 100 micrograms/ml.

This antiserum was diluted to a titer of 2 micrograms/ml and used in three-fold excess over the index antigen. This yielded lower values for "Antibody Control" (see below) and more reproducible results but did not alter the values obtained.

For an index antigen, purified Hb Aic was iodinated with 125 by the chloramine-T method of McConahey and Dixon, described in the International Archives of Allergy and Immunology 29: 185-189 (1966). The hemoglobin, as obtained by chromatography, is in a buffer containing cyanide. This ion interferes with the oxidation of iodide to iodine. Prior to iodination, the cyanide must be removed from the sample by gel filtration or by dialysis. The specific activity of the preparations was about 1 mCi/mg. The antigen was stored as a 0.3 mg/ml solution in PCG buffer. Immediately prior to use, the antigen was diluted 100-fold in sheep hemoglobin (I mg/ml).

Test hemoglobins: For studies of the primary inhibition of antisera, purified hemoglobins were used in amounts ranging from 0.025 to 25 microgram. A set of mixtures of purified Hb A,c and Hb Ao were used as primary standards for the quantitative assay. These contained 0.0, 1.5, 3.6, 7.5, 9.0, 10.0, 12.0 and 15.0 /n Hb A,c in Hb A,. Stock solutions, 2 mg/ml, were diluted 1:20 in sheep hemoglobin (0.5 mg/ml) and 50 microliters of each mixture was used to construct the standard curve. Assay unknowns were hemolysates from normal and diabetic donors. These were diluted from stock solutions as described for the standards.

EXAMPLE 4 Radioimmunoassay (RIA) Testing: All antisera and hemoglobins were diluted as outlined in the preceding section.

Reactions were carried out in triplicate for the standards, and in duplicate for other test hemoglobins. The assay consists of three stages Stage 1: The reaction mixture which contained 200 microliters antiserum and 50 microliters test hemoglobin was incubated at room temperature for 30 minutes.

Each experiment included two controls. In the "Antibody Control" the test hemoglobin was replaced by the sheep hemoglobin diluent, permitting full reaction between antibody and index antigen in the subsequent stage. In the "Antigen Control" the antibody was omitted (only the BGG diluent was used) to establish the inherent solubility of the free index antigen under the assay conditions.

Stage 2: Freshly diluted index antigen was added to each reaction mixture and further incubated at room temperature for 30 minutes.

Stage 3: TBS, 1.45 ml and ammonium sulfate, 1 ml, were added sequentially to each tube with thorough vortex mixing. After 1/2 hours at 40C the precipitated immune complexes were sedimented at 1,000xG for 10 minutes. Supernatant radioactivity was calculated from the counts in 1 ml and was expressed as the percent of total counts.

Percent Blocking of the Antiserum is defined as: (T-B) - xl00 (A-B) wherein the symbols are the percent supernatant counts for test hemoglobins (T), antigen control (A), and anti-body control (B).

For assay purposes, a standard curve was constructed by plotting the percent blocking for each of the primary standards against its known fractional content of Hb A,c and Hb A,.

EXAMPLE 5 Reaction with glycosyl dipeptides A number of glycosylated derivatives of the N-terminus of the human hemoglobin beta-chain were tested for their ability to inhibit the primary reaction between Hb A,c and its specific antibody. None of the compounds tested showed significant blocking activity, even in up to 1,000-told molar excess, as shown in Table 2: TABLE 2 Maximal Blocking of Anti-A1 Antiserum By Synthetic Glycopeptides Substance Moles Added % Blocking Hb A,c 7x10-" 95 /n Hb A,c, reduced 7x10-" 20% 1-deoxy Mannosyl-valine, reduced 2x 10-9 6% 1 -deoxy Mannosyl-valine, reduced 1x10-9 6% l-deoxy Galactosyl-valine, reduced 2x 10-9 6% l-deoxy Galactosyl-valine, reduced 2xl0-7 8% 1 -deoxy Glucosyl-valyl-histidine, reduced 2x 10-9 9% 1-deoxy Glucosyl-valyl-histidine, reduced 2x 10-9 9% There is no significant blocking of the antiserum by glycopeptides in up to 1,000-fold molar excess over totally blocking amounts of Hb A,c or partially reactive reduced Hb A1c.

EXAMPLE 6 Species Cross-Reactivity The Hb Ao, A Lab and A,c components of dog and mouse hemoglobin were tested for cross reactivity in the RIA system. The reaction of these hemoglobins with anti-A1 is qualitatively analogous to that of their human counterparts. The mouse and dog Hb A,c have about 15% of the reactivity of the human fraction for halfmaximal blocking of anti-A,c. The Ao component in each species is inactive, while the hemoglobins A,a+b have intermediate reactivities.

EXAMPLE 7 Quantitative assay of Hb A,c by RIA: The relation between the percent Hb A,c in the standard mixtures, and the percent blocking of anti-A1 was plotted as a standard curve. Mean and standard deviation for 9 consecutive experiments performed over a 40-day period with the same index antigen were included. There was no progressive shift of the curve in any direction; the range of values for each sample represents the inherent limit of reproducibility for the method. The y-intercept of the curve shows residual cross reactivity of the anti-A1c with Hb Ao.

Assay values for hemolysates with more than 9% Hb A,c fell on the upper, shallower slope of the standard curve. In these instances the hemolysates were diluted with an equal amount of Hb Ao; the percent blocking then fell on the steeper portion of the curve. The corresponding percent Hb A,c was doubled to correct for the two-fold dilution of the hemolysate and more reproducible results were thus obtained. It was found more convenient to include one such dilution with every hemolysate assayed, rather than perform a second assay for those specimens with high Hb Aic.

Thirty-three hemolysates were assayed for Hb Aic by both the RIA and the column method. In general, higher values are obtained by RIA. The regression line for these data is y=1.16x+0.15 with a coefficient of determination r2=0.72. In order to evaluate the source of the discrepancy between the two methods, the following experiment was performed: Two hemolysates, assayed by RIA to have 4.5 and 12.4 /n Hb A1c, respectively, were mixed in 10 different proportions. The original hemolysates and the mixtures were assayed in a blinded fashion by both the RIA and the column methods, and the results were compared with the values of the Hb Ac calculated from the composition of the mixtures. An excellent correlation was observed for the RIA with the regression formula y=x+0.06, and r2=0.99. By contrast, the regression formula for the chromatographic analysis of the same mixtures, using the column values for the original hemolysates for the calculations, was y=1.025x-0.81, and r2=0.87. Thus, the RIA method has a high degree of internal consistency and is linear throughout the range examined.

In view of the fact that the Hb A,c level in diabetics is above 6 wt% as described previously, it follows that the above described method can be used to diagnose diabetes mellitus, a positive diagnosis being obtained if the level of Hb A,c in a patient's blood sample is above 6 wt%.

A diagnostic kit can be provided consisting of an antibody against human hemoglobin A,c as described above and an index antigen comprising a radioactively labelled substantially pure hemoglobin A,c. In a preferred embodiment, the radioactive labelling is with iodine 125.

WHAT WE CLAIM IS: 1. An antibody against huma

Claims (14)

**WARNING** start of CLMS field may overlap end of DESC **. mixtures, using the column values for the original hemolysates for the calculations, was y=1.025x-0.81, and r2=0.87. Thus, the RIA method has a high degree of internal consistency and is linear throughout the range examined. In view of the fact that the Hb A,c level in diabetics is above 6 wt% as described previously, it follows that the above described method can be used to diagnose diabetes mellitus, a positive diagnosis being obtained if the level of Hb A,c in a patient's blood sample is above 6 wt%. A diagnostic kit can be provided consisting of an antibody against human hemoglobin A,c as described above and an index antigen comprising a radioactively labelled substantially pure hemoglobin A,c. In a preferred embodiment, the radioactive labelling is with iodine 125. WHAT WE CLAIM IS:

1. An antibody against human hemoglobin A,c which is substantially free of cross-reactivity against the human hemoglobins Ao, Ala and A,b.

2. A method for preparing an antibody as claimed in claim 1, which comprises administering purified hemoglobin A,c antigen to an animal capable of producing antibodies thereto and collecting said antibodies from the animal.

3. A method according to claim 2, wherein the animal is one whose metabolism does not naturally produce hemoglobin A,,.

4. A method according to claim 3, wherein the animal is a cat, goat or sheep.

5. A method according to claim 2 substantially as herein described or as exemplified in Example 2.

6. Antibodies obtained by a method as described in any one of claims 2 to 5.

7. A method of quantitatively detecting the presence of hemoglobin A,c in a test antigen comprising a body fluid sample, which comprises reacting said test antigen with an antibody as claimed in claim 1 or claim 6 and determining the amount of an antigen-antibody complex formed by radioimmunoassay.

8. A method according to claim 7, wherein the body fluid is blood.

9. A method according to claim 7 or claim 8, wherein said antibody is first reacted with a test antigen free of index antigen after which the resultant product is reacted with index antigen.

10. A method according to claim 9, wherein reaction with the index antigen is effected by incubation at a temperature from room temperature to 370C for 0.5 hour.

II. A method of diagnosing diabetes mellitus comprising quantitatively detecting the presence of hemoglobin A,c in a blood sample by a method according to any one of claims 7 to 10 and determining whether the level of hemoglobin Aic is in excess of 6 wit%.

12. A method according to claim 7 substantially as herein described or exemplified in Example 4.

13. A diagnostic test kit for quantitatively assaying hemoglobin Aic, comprising an antibody as claimed in claim 1 or claim 6 and a radioactively labelled, substantially pure hemoglobin A,c index antigen control.

14. A diagnostic kit according to claim 13, wherein the radioactive labelling is with iodine 125.