0013-7227/91/1293-1559$03.00/0 Endocrinology Copyright © 1991 by The Endocrine Society
Vol. 129, No. 3 Printed in U.S.A.
The Heterogeneity of Human Chorionic Gonadotropin (hCG). III. The Occurrence and Biological and Immunological Activities of Nicked hCG* LAURENCE A. COLE, ANDREW KARDANA, PATRICIA ANDRADE-GORDON, MARY-ANN GAWINOWICZ, JOHN C. MORRIS, ELIZABETH R. BERGERT, JOHN O'CONNOR, AND STEVEN BIRKEN Yale University (L.A.C., A.K., P.A.-G.), New Haven, Connecticut 06510; Columbia University College of Physicians and Surgeons (M.-A.G., J.O., S.B.), New York, New York 10032; and Mayo Medical School (J.C.M., E.R.B.), Rochester, Minnesota 55905
ABSTRACT. Nicks, or missing peptide linkages, have been found in hCG /J-subunit between residues 44 and 45 and between residues 47 and 48. We examined the occurrence and biological and immunological activities of nicked hCG. As shown by sequence analysis, CR127 standard hCG is approximately 20% nicked, half at 044-45 and half at 047-48. Treatment with human leukocyte elastase increased the extent of nicking of CR127 standard hCG. The longer the incubation of CR127 standard with human leukocyte elastase (0, 2, and 21 h), the greater the extent of nicked hCG (20%, 46%, and 89%). As the extent of nicking increased, the receptor-binding ability diminished, as did the ability to stimulate progesterone production by rat corpus luteal cells in vitro (0.9, 0.74, and 0.29 fig/fig hCG, respectively). In a regression analysis, a linear relationship was indicated between the extent of nicking and receptor binding values (97% correlation) and between the extent of nicking and steroidogenic activity in vitro (99% correlation). From the intercepts of the regression lines, it was estimated that nicks reduced receptor binding by 11-fold and reduced the steroidogenic activity of hCG by 5-fold. We examined eight individual hCG preparations, three purified from pregnancy urine, three from urine from patients with hydatidiform mole, and two from urine from women with choriocarcinoma. In descending order, the eight individual hCG preparations were 100%, 100%, 85%, 76%, 42%, 41%, 0%, and 0% intact. Although no correlation was observed between the percent intact and the ability of the eight individual samples to displace 50% [125I]hCG in binding CG/LH receptor (r < 0.5), a close correlation was noted between the percent intact and the steroidogenic activity in vitro (98% correlation). This separated the effects of nicking on receptor binding and steroidogenic
h
CG IS a glycoprotein hormone composed of two dissimilar subunits, a (92 residues) and |8 (145 residues), joined noncovalently. While the a-subunit of hCG Received March 12,1991. Address all correspondence and requests for reprints to: Laurence A. Cole, Ph.D., Department of Obstetrics and Gynecology, Yale University School of Medicine, 333 Cedar Street, New Haven, Connecticut 06510. • This work was supported by NIH Grants CA-44131 (to L.A.C.), CA-46828 (to L.A.C.), HD-15454 (to S.B.), and HD-22970 (to F. Naftolin).
activities and indicated that while multiple factors influence receptor binding, only nicking suppresses the steroidogenic activity of bound hCG. We examined the recognition of nicked hCG molecules by different hCG immunoassays. The Hybritech Tandem assay measured total hCG and did not distinguish nicked and intact hCG molecules (in a regression analysis, immunoactivity vs. percent intact hCG, r < 0.5). In contrast, the immunometric assay using B109 hCG dimer-specific monoclonal antibody and anti-/3-peroxidase only detected the intact component of hCG (in a regression analysis, immunoreactivity vs. percent intact hCG, 98% correlation). We used these assays together to estimate the percentage of intact hCG and to deduce the extent of nicking. In 38 parallel sets of serum and urine samples, 20 from pregnancy and 18 from patients with trophoblast disease, wide variation was detected in the percentage of intact hCG (range, 6-100%). Mean values for the percentage of intact hCG in serum and urine were 72 ± 22% and 76 ± 35%, respectively (mean ± SD; P > 0.5). The percentage of intact hCG was not significantly different in pregnancy (71 ± 26%) and trophoblast disease (77 ± 33%) samples (P > 0.3). We concluded that nicked hCG accounts, on the average, for approximately one quarter of the total hCG in serum and urine. The finding of a major hCG component with only minimal steroidogenic activity is novel and raises issues concerning the physiological or regulatory function of nicking and the medical significance of variations thereof. Also, the finding that a hCG assay using a dimer-specific monoclonal antibody only measures intact hCG calls into question the results from many assays in current use that employ this type of antibody. (Endocrinology 129: 1559-1567,1991)
has the same peptide sequence as the a-subunits of LH, FSH, and TSH, the jS-subunit is unique and distinguishes the functions of hCG from those of the other glycoprotein hormones. hCG is normally produced in the placenta and is present at similar levels in urine and blood (serum/ plasma) in women with pregnancy and trophoblast disease. In 1973, Shome and Parlow (1) discovered missing peptide linkages at different sites on the LH /3-subunit between residues 46 and 49. More recent studies have confirmed this (2, 3) and have located the nicks (missing linkages) at 044-45, 047^48, and 048^49. Compari-
1559
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG
1560
sons of nicked and intact LH preparations have indicated that these nicks cause a 40% loss of biological activity (2). In 1988, Nishimura and colleagues (4) showed that nicks at undetermined sites were present on hCG from individuals with trophoblast disease. Since then, several groups, including ours, have confirmed this finding and located nicks at sites analogous to those found on LH (044—»45 and 047—48; see Fig. 1) in pregnancy, trophoblast disease, or standard hCG preparations (5-9). These studies, however, were preliminary, in that they were limited to one or two sequence analyses and to hCG purified from pools of urine. The origin and occurrence of nicked hCG and the effects of nicking on hCG actions still need to be addressed. The preceding articles are comprehensive studies of nicking, examining the origin, sites, and occurrences of hCG peptide cleavages in detail (10, 11). In the first paper, hCG was purified from 13 individual urine samples, and the peptide sequences were examined (11). Nicking was demonstrated in 11 of the 13 purified preparations at 047-48 and, with 3 exceptions, at 044-45 or 046-47. The proportion of hCG molecules nicked is shown to vary widely (0-100% of the total). In the second paper, the structures of 7 different reference preparations of hCG were examined (10) in which variable nicking (10-20% of the molecules) at 047^48 and 044-45 was demonstrated. Human leukocyte elastase was found to specifically nick hCG at sites in the 044-49 region in vitro and could be a cause of nicking in vivo. This paper complements the preceding articles, examining the biological effect of nicking and the development of an immunological approach to measure the extent of nicking in serum and urine samples. The receptor-binding and steroidogenic and immunological activities of 8 pure hCG preparations (0-100% nicked) and of elastase-nicked Asp Arg 6 0 Tyr Asn Cys 57 Tyr CyS'*'*' V al G| y 38 Pro Val
Ala
Thr
40
Gin
Met Thr Arg 44 Val
Pro Leu Ala Pro 50 Leu
45 Leu Gin
\Val 47 \
FIG. 1. Sequencing of hCG /3-subunit residue 55-61, showing major sites of missing peptide linkages or nicks (10).
Endo • 1991 Voll29«No3
hCG standard are measured. For the first time, specific immunoassays are developed for the measurement of total hCG and intact hCG (that not nicked) and applied to determine the extent of nicking in pregnancy and trophoblast disease samples.
Materials and Methods Samples Pure hCG samples, coded PI, P3, and P5 (from pregnancy urine), Ml, M2, and M4 (from individuals with hydatidiform mole), and C3 and C5 (from individuals with choriocarcinoma), were prepared, as described previously (11). Parallel serum and urine samples were collected from 21 individuals with normal pregnancy, 13 individuals with hydatidiform mole, and 12 individuals with choriocarcinoma or invasive trophoblast disease. Urine samples were frozen (—20 C) within 15 min of collection; serum samples were frozen immediately after centrifugation. Electrophoresis and immunoblotting Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was carried out on 12% polyacrylamide gels in a Bio-Rad Mini Protean II slab gel apparatus (Bio-Rad, Richmond, CA), using the discontinuous buffer system of Laemmli (12). Reduced samples were heated to 80 C for 10 min, then applied to gels. Nonreduced samples were applied directly to gels without heating. hCG standard (NIH batch CR127) and mol wt markers (Bio-Rad) in sample buffer were also applied to gels. The gels were subjected to a constant 200 V for 1 h at ambient temperature. Immunoblotting was carried out in a Bio-Rad Mini Trans-Blot apparatus, using the methods of Burnette (13). Briefly, gels were placed on a 0.2-/im nitrocellulose membrane, fitted into the transfer apparatus, and subjected to 100 V for 1 h at ambient temperature. Proteins transferred onto the membrane were washed free of sodium dodecyl sulfate, and the membranes were blocked from nonspecific binding by soaking in buffer containing 5% BSA. Nitrocellulose membranes were then incubated 4 h at ambient temperature with antiserum raised to hCG /3-subunit C-terminal peptide, residues 111-131 (CC11, kindly donated by Dr. H. C. Chen at NIH), and bands were visualized by further incubation with protein-A-gold conjugate (Bio-Rad Laboratories). Added sensitivity was achieved by a gold enhancement procedure (Bio-Rad Laboratories). This consisted of soaking the washed membranes with 0.2 M citrate buffer (pH 3.7) and then incubating with a solution of 0.85% hydroquinone plus 0.11% silver lactate for 15 min in the dark. The reaction was stopped using the supplied fixing solution. Gels were scanned and uniformly intensified using a HewlettPackard ScanJet Plus scanner (Palo Alto, CA; 256 Gray scale; 90 x 109 dots/in, resolution) and printed on a Hewlett Packard LaserJet Series II printer. Immunological activities Fresh solutions of CR127 hCG standard and PI, P3, P5, Ml, M2, M4, C3, and C5 hCG (pure) were prepared. Absolute concentrations of hCG were determined by amino acid analysis, as described in the preceding paper (11). Levels of immunoreactive hCG were determined by the Hybritech Tandem assay and the Bl09:anti-j8-peroxidase assay using PI hCG (0%
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG nicked) as standard. Levels were also determined using the Columbia anti-hCG immunoradiometric assay, which uses CR127 hCG (17% nicked) as standard. Levels were expressed as micrograms per ml. Immunoactivity was expressed as micrograms per ml by immunoassay divided by micrograms per ml by amino acid analysis, or micrograms per /xg hCG. The Bl09:anti-/3-peroxidase assay is a two-stage immunoenzymometric assay using a hCG dimer-specific monoclonal antibody, B109 (generated by Drs. A. Krichevsky and E. Armstrong of Columbia University) (14) to capture hCG and antij8 antiserum-peroxidase (Bios Pacific, Inc., Emeryville, CA) to label or detect it. Microtiter plates were coated with antibody B109 (200 /xl; 2 Mg/ml; in 0.25 M NaHCO3-0.1 M NaCl buffer, pH 9.5) and plates were incubated overnight at 4 C. Plates were washed five times with water and aspirated before use. In triplicate, 100 ix\ sample or standard (0-25 ng/ml) were added to coated wells together with 100 /ul buffer-carrier protein mix [0.05 M NaH2PO3/Na2HPO3-0.14 M NaCl (pH 7.5) plus 0.1% (wt/vol) ovalbumin]. Plates were shaken on a plate rotator overnight at ambient temperature, then washed five times with water and aspirated. Goat anti-/3 antiserum-peroxidase [200 /xl; 1:3500; in 0.1 M Tris-HCl-0.025 M CaCl2 (pH 7.5) plus 0.1% (wt/vol) ovalbumin] was then added to wells, and plates were shaken for a further 2 h at ambient temperature. After another five washes with water, 200 /xl substrate mix (5-mg tablet of orthophenylenediamine and 4 /xl 30% H2O2 in 25 ml 0.01 M sodium citrate, pH 4.9) were added, and plates were shaken in the dark for 30 min at ambient temperature. HC1 (50 /xl; 4 M) was added to stop reactions, and the absorbance of the wells was determined in a Flow Laboratories Titertek Multiscan NCC-340 plate reader (McClean, VA) at 492 nm. Data were sent to a Zenith 80286 computer, and standard curves were plotted and levels determined using Titersoft software (Flow Laboratories). All values were determined in triplicate. This assay is specific for hCG, with less than 1% cross-reactivity with hCG free /3-subunit, free a-subunit, or human (h) LH. The in-house anti-a:anti-j3 test was carried out by procedures similar to those used for the B109:anti-j8-peroxidase assay. The exception was the use of Unipath 2119:12 monoclonal anti-a (4 /xg/ml) in place of B109 anti-hCG. PI hCG was used as standard. All values were determined in triplicate. This assay is specific for hCG, with less than 0.1% cross-reactivity with hCG free j8-subunit, free a-subunit, or hLH. The C-terminal peptide assay (CCll anti-/3-subunit peptide:[125I]anti-/3) was described previously (10). This assay detects equimolar levels of hCG and free /3-subunit, with less than 0.1% cross-reactivity with free a-subunit and hLH. Tandem assay kits were provided by Hybritech (San Diego, CA). A two-step modification of the assay was used, as previously described (15). Briefly, methods were carried out according to the instructions, except two 2-h incubations were used in place of a single 1-h protocol. Beads (coated with antia antibody) were first incubated with sample, washed, then separately incubated with tracer antibody ([125I]anti-/3). After further washes, bead-bound radioactivity was determined, and concentrations were calculated. Values were determined in triplicate. This assay is also specific for hCG, with less than 0.1% cross-reactivity with hCG free /3-subunit, free a-subunit, or hLH. The Columbia University anti-hCG assay also uses monoclo-
1561
nal antibody B109, but at different coating concentrations and with a different tracer antibody ([125I]monoclonal antibody B108) than the B109:anti-/?-peroxidase test. The methods were reported by O'Connor et al. (16) and are widely used by research laboratories measuring low levels of hCG, as in early fetal loss studies (17). Biological activity
Immature female Sprague-Dawley rats, 26 days old, were injected sc with 50 IU hMG, followed 56 h later by 25 IU hCG (Sigma Chemical Co., St. Louis, MO). Ovaries were collected 6 days post-hCG injection, and luteal cells were prepared by a modification of the method of Behrman and colleagues (18). Luteinized ovaries were finely minced with a razor and dispersed in Ca2+-free Minimum Essential Medium 1380 (Gibco, Grand Island, NY) containing 0.1% (wt/vol) BSA, 2 mM glutamine, 2000 IU collagenase (Worthington Biochemical Co., Freehold, NJ), and 3000 IU deoxyribonuclease (Worthington; 5 ml/g ovarian tissue) and incubated for 1 h at 37 C in 95% air-5% CO2 in a culture cabinet. The population of luteal cells was enriched by application to a discontinuous Percoll density gradient (Pharmacia Fine Chemicals, Uppsala, Sweden). Luteal cells present in the density layers 1.018 and 1.033 g/ml Percoll were removed by aspiration, washed by 3-fold dilution with medium, collected by centrifugation (100 X g; 5 min.), and resuspended in Minimum Essential Medium 2360 (Gibco) containing 0.1% (wt/vol) BSA and 2 mM glutamine. To study hCG steroidogenic activity, cells were distributed into 12 x 75-mm glass tubes (2.5 x 106 cells/tube). Fresh solutions of CR127 hCG standard and PI, P3, P5, Ml, M2, M4, and C3 and C5 preparations of pure hCG were prepared. Absolute concentrations of hCG were determined by amino acid analysis. Two different concentrations of hCG samples (or six concentrations of standards) in the range 0.55 ng/ml were applied in triplicate to tubes containing luteal cells. Minimum Essential Medium 2360 was added to bring the volume to 1 ml, and tubes were incubated 90 min in a 37 C 95% air-5% CO2 culture cabinet. Conditioned medium was removed, and progesterone levels were determined by RIA (R1A kit, Serono-Baker Diagnostics, Inc., Allentown, PA). Pi hCG, which is 0% nicked (no missing linkages or Nterminal sequence deficiency in a- or 0-subunit), was used as the standard for biological activity studies. The absolute biological activity of PI hCG is 1.1 times that of NIH CR127 standard hCG (14,900 vs. 16,600 IU/mg). Progesterone values were plotted against concentrations of PI hCG, and activities relative to PI hCG were determined. Assay values were expressed as micrograms per ml, and biological activity as micrograms per ng (/xg/ml by bioassay -s- /xg/ml by amino acid analysis). Receptor-binding activity Receptor-binding activity was examined by competitive RRA, using [125I]hCG and homogenates of luteinized rat ovaries. hCG was radioiodinated by the chloramine-T method using the method of Greenwood and colleagues (19), as modified by Lee and Ryan (20). Homogenates of luteinized rat ovaries were prepared, and competitive binding assay was performed as described by Lee and Ryan (21). Serial dilutions of hCG samples were prepared (concentration determined by amino
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
Endo • 1991 Voll29«No3
OCCURRENCE AND ACTIVITIES OF NICKED hCG
1562
acid analysis), and displacement of [125I]hCG was determined. Values were linearized by the log-logit method, and the concentration necessary to displace 50% of [125I]hCG in binding the receptor (EC50) was determined.
Results Biological activity of nicked hCG Standard hCG (NIH batch CR127) is approximately 20% nicked at 044-45 or 047-48 (Fig. 1). Treatment of standard hCG with human leukocyte elastase increased the extent of nicking from 20-89% (10). We examined the effect of this elastase digestion on the receptor binding and steroidogenic activities of standard hCG. As shown in Table 1, the longer the incubation with human leukocyte elastase (0, 2, and 21 h), the smaller the percentage of intact (nonnicked) hCG (80%, 54%, and 11%) and the greater the concentration required to displace 50% of 125I-labeled standard hCG (EC50) in binding the CG/LH receptor (1.0, 2.2, and 10 /xg). A linear relationship was indicated between the percent intact hCG and receptor-binding activity (97% correlation). From the intercept of the regression line an 11-fold difference was suggested in the receptor-binding activity of nicked and intact hCG. A linear relationship (99% correlation) was also demonstrated between the percent intact hCG and steroidogenic activity in vitro (0.90, 0.74, and 0.29 /xg/ixg hCG). From the intercept of the regression line, nicked molecules were calculated to have approximately 20% the steroidogenic activity of intact hCG.
We also examined the receptor binding and steroidogenic activities of eight individual hCG preparations, three purified from pregnancy urine, three from urine from women with hydatidiform mole, and two from urine from patients with choriocarcinoma (Table 2). In ascending order, the eight individual preparations were 0%, 0%, 15%, 24%, 58%, 59%, 100%, and 100% nicked (or 100%, 100%, 85%, 76%, 42%, 41%, 0%, and 0% intact). No relationship was observed between the percent intact and the ability of the eight individual hCG samples to displace 50% of [125I]hCG in binding the CG/LH receptor (r < 0.5). In contrast, a linear relationship was evident between the percent intact and the steroidogenic activity of the eight individual samples in vitro (98% correlation). Western blots of nicked hCG Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of reduced hCG, followed by Western blotting and staining of the blot with antibodies to the 0-subunit Cterminal peptide region, permits visualization of nicked hCG (4-8). An example of this method is shown in Fig. 2. As observed, when Western blots were made of serum and urine samples containing nicked hCG, two bands were detected. The first, mol wt (Mr) 35,000, corresponded to the intact 0-subunit of hCG. The second, Mr 24,000, corresponded to the peptide 048-145 released from nicked hCG after reduction of disulfide bonds. The presence of this band indicated nicking. The example in Fig. 2 is the starting urine of sample C3 and a parallel
TABLE 1. The biological, receptor binding, and immunological activities of CR127 standard hCG, nicked by the action of human leukocyte elastase (HLE) in vitro Sequence analysis (% intact)0
Sample
CR127 hCG after 0 h with HLE 2 h with HLE 21 h with HLE
80 54 11
Amino acid analysis (/ig/ml)
850 850 850
Immunoassay Hybritech Tandem
Immunoassay BlO9:anti-/8peroxidase
Bioassay steroidogenic activity in vitro
RRA
(Mg/ml)
840d 950 890
jig/ml
Atg/Mg hCG c
fig/'ml
Mg/^g hCG c
760 630 140
0.89e 0.74 0.16
770 630 250
0.90' 0.74 0.29
1.0* 2.2 10
The protocol is that described in the preceding paper (10). Briefly, hCG is digested with enzyme at 37 C in pH 8.5 buffer (0.03 U enzyme/nmol hCG). Aliquots are taken, and the reaction halted by the addition of phenylmethylsulfonylfluoride for reverse phase HPLC and sequence analysis or ovalbumin for immuno- and bioassays. All assays used Pi hCG, which is 0% nicked, as standard. 0 Percent intact is calculated as 100% minus percent nicked, determined by reverse phase HPLC and N-terminal sequence analysis (10). Values are estimates plus and minus five percentage points. 6 Values are the concentration (nanograms per ml by amino acid analysis) necessary to displace 50% of [126I]hCG in binding receptor; values are normalized to results for CR127 hCG after 0 h with HLE. c hCG determined by amino acid analysis. d Values in Hybritech Tandem assay were 105 ± 6.0% (mean ± SD) of amino acid analysis values. No correlation was observed between Hybritech Tandem assay values (micrograms per ng hCG) and percent intact (r > 0.6). " In a regression analysis, immunoactivity (micrograms per ng hCG) at 0, 2, and 21 h vs. percent intact, slope = 0.011, intercept = 0.071, and r = 0.98 (98% correlation). 1 In a regression analysis, bioactivity (micrograms per ng hCG) at 0, 2, and 21 h vs. percent intact, slope = 0.0090, intercept = 0.20, and r = 0.99 (99% correlation). * In a regression analysis, receptor binding EC50 values at 0, 2, and 21 h vs. percent intact, slope = 0.14, intercept = 11, and r = 0.97 (97% correlation).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG TABLE 2. The biological and receptor-binding activities of eight individual hCG preparations
Sample
Sequence Amino acid analysis analysis (% intact) 0
Bioassay steroidogenic activity in vitro
RRA (EC 60 ) 6
Mg/ml Mg/Mg hCG'
1. Pregnancy Pi
100
41 P3 P5 42 2. Trophoblast disease Ml
M2 M4
C3 C5
620 520
1460 1090 1490 3400
85 100 0 76 0
1.0e
620 250 730
1.0" 0.48 0.50
950
0.87
1.3
1.2
0.82 0.60 0.82
0.20
1.6
320
1760 ND' ND /
2100
420
1.6 1.1
The in vitro biological activity of each hCG preparation was determined in the rat luteal cell progesterone test. Values were compared to the biological activity of P i hCG, which is 100% intact. ° Percent intact is calculated as 100% minus percent nicked, determined by N-terminal sequence analysis (10). Values are estimates plus and minus five percentage points. b Values are the concentration (nanograms per ml by amino acid analysis) necessary to displace 50% of [125I]hCG in binding receptor, values are normalized to results for P i hCG. c hCG determined by amino acid analysis. d In a regression analysis, bioactivity (micrograms per ng hCG) of P1-C5 vs. percent intact, slope = 0.0091, intercept = 0.15, and r = 0.98 (98% correlation). ' In a regression analysis, receptor binding EC50 values vs. percent intact, no correlation was observed {P < 0.5). ' Not determined.
75,000* 50,000*
f
34,000* 24,000*
M24,000
Reduced C3 Serum
C3 Urine
FIG. 2. Scan of nitrocellulose blot of sodium dodecyl sulfate-polyacrylamide gel electrophoresis of reduced (+) and nonreduced (-) C3 patient serum and a parallel urine sample.
serum sample. As shown in a preceding paper (11), after purification and N-terminal sequence analysis, sample C3 was found to be 24% nicked at 047-48. Using this Western blot method we examined 24 urine and 12 parallel serum samples from individuals with pregnancy or trophoblast disease. As shown in Table 3, regardless of origin (pregnancy or trophoblast disease), 79% of urine and a similar proportion of serum samples (83%) gave a Mr 24,000 band, indicating that approximately 80% of serum and urine samples were nicked. Studies with different hCG standards (10-20% nicked; see preceding paper) indicated that 10% nicking was
1563
TABLE 3. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting of pregnancy and trophoblast disease serum and urine (paired) hCG samples Source Pregnancy Trophoblast disease Total
No. with 24,000 Mr band" Serum 4 of 5 7 of 7 10 of 12 (83%)
Matching urine 5 of 7 14 of 17 19 of 24 (79%)
hCG nicking can be detected in serum and urine samples by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting. Intact hCG /3-component (from intact hCG) reduced with (3mercaptoethanol migrates at 35,000-36,000 daltons on these gels; the C-terminal part of nicked hCG ^-component (peptide 048-145 of nicked hCG), however, migrates faster, at 24,000 daltons (6). " Studies with hCG standards (up to 12% nicked) indicate that samples need to be more than 10% nicked for unequivocal visual detection of the 24,000 Mr band.
required for consistent observation of a 24,000 Mr band. Thus, the approximately 20% of samples without an obvious 24,000 Mr band may represent samples with small amounts of nicking ( 0.5). The mean percent intact hCG was not significantly different in pregnancy (71 ± 26%) and trophoblast disease (77 ± 33%) samples (P > 0.3). We inferred that nicked hCG accounts, on the average, for approximately 26% of the total hCG in serum
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG and urine samples, regardless of origin (pregnancy or trophoblast disease).
Discussion Here we show that nicks at 044-45 or 047-48 reduce the receptor binding and steroidogenic activities of hCG. When standard hCG is nicked by the action of human leukocyte elastase in vitro, both the ability to displace [125I]hCG in binding the receptor and steroidogenic activity are reduced. It was estimated that molecules nicked this way have 9% of the receptor binding and 20% of the steroidogenic activity of intact hCG. Recently, Sakakibara and colleagues (8) purified the nicked component of hCG 0-subunit and combined it with w-subunit to form a recombinant nicked hCG. Steroidogenic activity was measured and was found to be less than 1% that of natural hCG. This is much lower activity for nicked hCG than reported here. The lower activity, however, might be due to a poor or abnormal combination of a- and nicked 0-subunit or to denaturation or molecular damage of nicked 0-subunit during reverse phase HPLC procedures using trifluoroacetic acid. Studies by Bousfield and Ward (22) show that ovine LH 0-subunit, nicked at 043-44 in vitro by endoproteinase Arg C and recombined with ovine LH a-subunit, had approximately 2% the receptor-binding activity of intact oLH. This is approximately one quarter of the result reported here with elastase-cleaved hCG. The difference again could be due to the use of recombined, rather than natural, dimers. We considered how nicks in the 044-50 region of standard hCG could diminish the receptor-binding activity of hCG 11-fold and steroidogenic activities of hCG 5fold. As shown in Fig. 1, nicking occurs in a hydrophobic loop in the hCG 0-subunit, which is held in place by a disulfide link between Cys38 and Cys57. Keutmann and colleagues (23) showed that this loop may be part of the receptor-binding domain of hCG. Cleavages at 044-45 or 047-48 would open the loop, which could change the tertiary structure and thus ablate the receptor-binding properties of hCG. Keutmann and colleagues (24) made analogs missing the Cys38-Cys57 disulfide linkage. The analogs retained 20% of receptor-binding activity. Consistent with this is our finding that molecules nicked in this same region retain 9% of receptor binding and 20% of steroidogenic activity. Our results confirm those of Keutmann and colleagues (23, 24) that the 038-57 region is important for biological activity and that minor cleavages in this region result in an 80% loss of hormone function. Significant variation has been reported in the in vitro biological activity of hCG, whether from a single source, such as first trimester pregnancy, or from multiple origins, such as pregnancy and trophoblast disease (reviews in Refs. 25-27). Much of this variation has been
1565
attributed to minor differences in hCG oligosaccharide structure (26-31). We purified hCG from urine from eight individuals, three with pregnancy, three with hydatidiform mole, and two with choriocarcinoma, and investigated the source of variations in hCG biological activities. The receptor-binding activity of the eight individual samples varied (0.82-1.6 ng/ng), regardless of source. Studies with CR127 standard hCG indicated that nicking can reduce the receptor-binding activity of hCG by 11-fold. We compared the extent of nicking and receptor-binding activity. No correlation was evident (linear regression, r < 0.5), indicating that nicking was not the primary cause of variation in in vitro receptorbinding activity of individual hCG samples. a-Subunit N-terminal heterogeneity was the only other peptide variable in these eight hCG preparations (11). Regression analysis, however, showed no correlation between receptor-binding activity and the extent of a-subunit N-terminal heterogeneity (0%, 0%, 0%, 27%, 18%, 12%, 0%, and 19%, respectively, in samples P1-C5; r < 0.5). We concluded that variations in receptor-binding activity were not primarily due to peptide variations, but, rather, as suggested by others (26-31), were due to carbohydrate microheterogeneity. The in vitro steroidogenic activity of the eight individual hCG preparations also varied widely (0.20-1.0 ng/ng hCG). The variation, however, whether in pregnancy or trophoblast disease samples, correlated with the percentage of intact hCG (98% correlation). We conclude that variations in the extent of nicking, rather than differences in carbohydrate content, are the principal cause of disparity in the steroidogenic activity of hCG. These findings indicated a dissociation of the effects of nicking on receptor binding and steroidogenic activities, and that although carbohydrate structure and extent of nicking affect receptor binding, only nicking suppresses the steroidogenic activity of bound hCG. The difference in results for receptor-binding activity and steroidogenic activity between individual hCG samples nicked in vivo and hCG standards nicked by human leukocyte elastase in vitro could be due to the different cleavage sites. While the former are only cleaved at 04748, the latter are cleaved at 044-45 and 047-48 and, in addition, at 048-49 and a70-71 depending on the incubation time with elastase. It is possibile that only the cleavage at 047-48 affects steroidogenic activity. This, however, seems unlikely, considering that all three cleavage sites in the 044-49 region are likely to open up the 038-57 intercysteine loop. We cannot, however, rule out the possibility that cleavages at sites other than 047-48 lead to the loss of receptor-binding activity observed only with elastase-cleaved preparations. Studies by Catt and Dufau (32) may provide another explanation for the different effects of nicked hCG on receptor binding and steroidogenesis. They show that steroidogenesis is max-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
1566
OCCURRENCE AND ACTIVITIES OF NICKED hCG
imal when only a small proportion of gonadotropin receptors are occupied. Thus, the differential effect of nicked hCG may be due in part to the greater sensitivity of steroidogenesis. We examined the specificity of different immunoassays for nicked hCG. Individual hCG samples were tested in three types of assays, anti-a:anti-/5 assays, anti-/? Cterminal segment assays, and an assay using hCG dimerspecific antibody. Two anti-a:anti-/3 immunometric tests were evaluated: the Hybritech Tandem assay and our in-house test. Both reflected amino acid analysis results and were not significantly affected by the percentage of nicking (r < 0.5). Values in the Hybritech Tandem assay were 94 ± 20% (±SD), and those in the in-house anti-a:anti-/? assay were 108 ± 22% of amino acid analysis values. A hCG (3subunit C-terminal peptide assay was also evaluated. As with the anti-a:anti-/3 assays, values represented total hCG and were not significanty affected by the percentage of nicking (r < 0.5). The results in a third type of immunoassay, using a hCG dimer-specific monoclonal antibody, were, however, affected by the percentage of nicking. In a regression analysis, values from the Columbia anti-hCG assay, which is used at numerous centers in the U.S. for the detection of low levels of hCG (16, 17), correlated with the percentage of intact (nonnicked) hCG (91% correlation). A limitation of this nick-sensitive assay was the use of CR127 hCG standard (20% nicked), which could falsely elevate hCG results. An improved version of this assay (B109:anti-/3 peroxidase), using P i hCG (0% nicked) as standard and a lower B109 coating concentration, gave results more closely correlating with the percentage of intact hCG (94% correlation). Similar results were observed in an assay using B107, another hCG dimer-specific antibody. This indicated that the differential recognition of intact and nicked hCG may be a property of most antibodies specific for hCG dimer. Varying results have been reported with the same sample in different hCG immunoassays and by different clinical laboratories. The problems and confusion created by these differences is exemplified by a recent article by Painter, "Discordant hCG results in pregnancy: a method in crisis" (33), and in several other recent articles (3436). In general, the competitive nature of RIAs, such as the hCG/? RIA, renders them more sensitive to small peptide or carbohydrate structural differences. As shown here, results from different types of immunometric assays, such as the anti-a:anti-/? (measures total hCG) and the anti-hCG dimer type (distinguished nicked and intact molecules), can also conflict and be confusing. Furthermore, the international hCG standards (First International Reference Preparation, Third International Standard, and CR series) are 10-20% nicked (9) and can themselves confuse results in different competitive and
Endo • 1991 Voll29«No3
immunometric assays. We concluded that variable nicking in individual serum and urine samples and in standard hCG preparations is a major contributor to the microheterogeneity of hCG and may be a cause of discordant results in different hCG immunoassays. Clearly, some assays, such as our in-house anti-a:anti-/? test, the Hybritech Tandem kit, and possibly numerous other similar assays, are unaffected by nicking and thus measure total hCG. Other assays, such as the Columbia antihCG test, the B109:anti-/3 peroxidase assay, and test kits that use this common type of antibody, are specific for intact or nonnicked hCG. While the anti-a:anti-/3 test, which measures total hCG, may be optimal for following hCG production, the latter, the intact hCG assay, may be preferable for measuring functional or biologically active hormone. We considered methods other than purification and sequence analysis to investigate the occurrence of nicking in hCG samples. Nishimura et al. (4), Bidart and colleagues (5), Sakakibara et al. (8), and Puisieux and colleagues (7) all used electrophoresis of reduced hCG and Western blotting with C-terminal peptide antisera to detect nicked molecules. Using this method we were able to demonstrate the broad occurrence of nicked molecules in pregnancy and trophoblast disease fluids, detecting nicks in 19 of 24 urine and 10 of 12 serum hCG samples. Quantitative methods were needed for investigating the amount of nicking in serum and urine samples. Using the Hybritech Tandem assay to measure total hCG and the B109:anti-/?-peroxidase assay to detect intact molecules, we were able to deduce the amount of nicked hCG (total minus intact) in serum and urine samples. In 38 parallel sets of serum and urine samples, 20 from pregnancy and 18 from patients with trophoblast disease, a similar percentage of intact hCG was detected in serum and urine (mean ± SD, 72 ± 22% and 76 ± 35%, respectively; P > 0.5). The percentage of intact hCG was not significantly different in pregnancy (71 ± 26%) and trophoblast disease (77 ± 33%) samples (P > 0.3). We deduced from these findings that approximately onequarter of hCG molecules, whether from urine or serum, pregnancy or trophoblast disease, are nicked. The finding of a major hCG component in pregnancy and trophoblast disease serum and urine with only minimal biological activity is novel. It raises questions concerning the physiology of hCG and the possibility that nicking is a specific regulatory process. Questions are also raised about the clinical significance of nicked hCG and whether the extent of nicking could mark gestational disorders.
Acknowledgments The authors wish to thank Dr. H. C. Chen at NIH for the gift of hCG/3 C-terminal peptide antisera (CC 11), and Dr. Fu
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG Chai-ying and the medical staff in the Department of Obstetrics and Gynecology at Yale University for providing the parallel urine and serum samples used in this paper. Thanks also go to Kay Shirey at Columbia University for proofreading this manuscript before submission.
17. 18.
References 1. Shome B, Parlow AF 1973 The primary structure of the hormonespecific, beta subunit of human pituitary luteinizing hormone (hLH). J Clin Endocrinol Metab 36:618-624 2. Ward DN, Glenn SD, Nahm HS, Wen T 1986 Characterization of cleavage products in selected human lutropin preparations. Int J Peptide Protein Res 27:70-78 3. Hartree AS, Lester JB, Shownkeen RC 1985 Studies of the heterogeneity of human pituitary LH by fast protein liquid chromatography. J Endocrinol 105:405-414 4. Nishimura R, Ide K, Utsunomiya T, Kitajima T, Yuki Y, Mochizuki M 1988 Fragmentation of the /9-subunit of human chorionic gonadotropin produced by choriocarcinoma. Endocrinology 123:420425 5. Bidart J-M, Puisieux A, Troalen F, Foglietti MJ, Bohuon C, Bellet D 1988 Characterization of the cleavage product in the human choriogonadotropin /8-subunit. Biochem Biophys Res Commun 154:626-632 6. Cole LA, Kardana A, Birken S 1989 The isomers, subunits and fragments of hCG. Serono Symp 65:59-78 7. Puisieux A, Bellet D, Troalen F, Razafindratsita A, Lhomme C, Bohoun C, Bidart J-M 1990 Occurrence of fragmentation of free and combined forms of the 0 subunit of human chorionic gonadotropin. Endocrinology 126:687-694 8. Sakakibara R, Miyazaki S, Ishiguro M 1990 A nicked /3-subunit of human chorionic gonadotropin purified from pregnancy urine. J Biochem 107:858-862 9. Lustbader JW, Birken S, Pileggi NF, Kolks MAG, Pollak S, Cuff ME, Yang W, Hendrickson WA, Canfield RE 1989 Crystallization and characterization of human chorionic gonadotropin in chemically deglycosylated and enzymatically desialylated states. Biochemistry 28:9239-9243 10. Birken S, Gawinowicz MA, Kardana A, Cole LA 1991 The heterogeneity of hCG. II. Characteristics and origins of nicks in hCG reference standards. Endocrinology 129:1551-1558 11. Kardana A, Elliott ME, Gawinowicz MA, Birken S, Cole LA 1991 The heterogeneity of hCG. I. Characterization of peptide variations in 13 individual preparations of hCG. Endocrinology 129:15411550 12. Laemmli UK 1970 Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680-685 13. Burnette WN 1981 "Western blotting:" electrophoretic transfer of proteins from sodium dodecyl sulfate-polyacrylamide gels to unmodified nitrocellulose and radiometric detection with antibody and radioiodinate protein A. Anal Biochem 112:195-201 14. Krichevsky A, Armstrong EG, Schlatterer J, Birken S, O'Connor J, Bikel K, Silverberg S, Lustbader JW, Carfield R 1988 Preparation and characterization of antibodies to the urinary fragment of the human chorionic gonadotropin /3-subunit. Endocrinology 123:584-593 15. Cole LA, Birken S 1988 Origin and occurrence of human chorionic gonadotropin /3-subunit core fragment. Mol Endocrinol 2:825-830 16. O'Connor J, Schlatterer JP, Birken S, Krichevsky A, Armstrong
19. 20. 21. 22. 23.
24.
25.
26. 27. 28. 29. 30.
31. 32. 33. 34. 35. 36.
1567
EG, McMahon D, Canfield R 1988 Development of highly sensitive immunoassays to measure human chorionic gonadotropin, its fisubunit, and /3 core fragment in the urine: application of malignancies. Cancer Res 48:1361-1366 Wilcox AJ, Weinberg CR, O'Connor JF 1988 Incidence of early pregnancy loss. N Engl J Med 319:189-194 Behrman HR, Preston SL, Hall AK 1980 Cellular mechanism of the antigonadotropic action of luteinizing hormone-releasing hormone in the corpus luteum. Endocrinology 107:656-664 Greenwood FC, Hunter WM, Glover JS 1963 The preparation of 12S I-labeled human growth hormone of high specific radioactivity. Biochemistry 89:114-123 Lee CY, Ryan RJ 1973 Interactions of ovarian receptors with luteinizing hormone and human chorionic gonadotropin. Biochemistry 12:4609-4615 Lee CY, Ryan RJ 1972 Luteinizing hormone receptors: specific binding of human luteinizing hormone to homogenates of luteinized rat ovaries. Proc Natl Acad Sci USA 69:3520 Bousfield GR, Ward DN 1988 Selective proteolysis of ovine lutropin or its j8-subunit by endoproteinase arg-C. J Biol Chem 263:12602-12607 Keutmann HT, Charlesworth MC, Mason KA, Ostrea T, Johnson L, Ryan RJ 1987 A receptor-binding region in human choriogonadotropin/lutropin beta subunit. Proc Natl Acad Sci USA 84:2038-2042 Keutmann HT, Charlesworth MC, Mason KA, Johnson L, Ryan RJ 1988 Primary and secondary structural determinants in the receptor binding sequence beta (38-57) from human luteinizing hormone. Biochemistry 27:8939-8944 Vaitukaitis JL 1978 Glycoprotein hormones and their subunitsimmunological and biological characterization. In: McKerns KW (ed) Structure and Function of the Gonadotropins. Plenum Press, New York, pp 339-360 Fauser BCJ M, Hsueh AJW 1989 Bioassays of gonadotropins: background and clinical applications. Serono Symp 65:123-136 Grotjan HE, Cole LA 1988 Human chorionic gonadotropin microheterogeneity. In: Keel B, Grotjan H (eds) Microheterogeneity of Glycoprotein Hormones. CRC Press, Boca Raton, pp 219-237 Kalyan NK, Lippes HA, Bahl OP 1982 Role of carbohydrate in human chorionic gonadotropin. J Biol Chem 257:12624-12631 Moyle WR, Bahl OP, Marz L 1975 Role of carbohydrate of human chorionic gonadotropin in the mechanism of hormone action. J Biol Chem 23:9163-9169 Birken S, Agosto G, Amr S, Nisula B, Cole L, Lewis J, Canfield R 1988 Characterization of antisera distinguishing carbohydrate structures in the /3-carboxyl-terminal region of human chorionic gonadotropin. Endocrinology 123:572-583 Amir SM, Kasagi K, Ingbar SH 1987 The role of subunit sialic acid in the thyrotropic and gonadotropic activities of human chorionic gonadotropin. Endocrinology 121:160-166 Catt KJ, Dufau ML 1973 Spare gonadotropin receptors in rat testis. Nature [New Biol] 244:219-221 Painter PC 1989 Discordant hCG results in pregnancy: a method in crisis. Clin Casebook 27:20-24 Hussa RO 1987 The Clinical Marker hCG. Praeger Press, New York Chen HC, Matsuura S, Ohashi M 1980 Limitations and problems of hCG-specific antisera. In: Segal SJ (ed) Chorionic Gonadotropin. Plenum Press, New York, pp 231-252 Hussa RO, Rinke ML, Schweitzer PG 1985 Discordant human chorionic gonadotropin results: causes and solutions. Obstet Gynecol 65:211-219
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
Vol. 129, No. 3 Printed in U.S.A.
The Heterogeneity of Human Chorionic Gonadotropin (hCG). III. The Occurrence and Biological and Immunological Activities of Nicked hCG* LAURENCE A. COLE, ANDREW KARDANA, PATRICIA ANDRADE-GORDON, MARY-ANN GAWINOWICZ, JOHN C. MORRIS, ELIZABETH R. BERGERT, JOHN O'CONNOR, AND STEVEN BIRKEN Yale University (L.A.C., A.K., P.A.-G.), New Haven, Connecticut 06510; Columbia University College of Physicians and Surgeons (M.-A.G., J.O., S.B.), New York, New York 10032; and Mayo Medical School (J.C.M., E.R.B.), Rochester, Minnesota 55905
ABSTRACT. Nicks, or missing peptide linkages, have been found in hCG /J-subunit between residues 44 and 45 and between residues 47 and 48. We examined the occurrence and biological and immunological activities of nicked hCG. As shown by sequence analysis, CR127 standard hCG is approximately 20% nicked, half at 044-45 and half at 047-48. Treatment with human leukocyte elastase increased the extent of nicking of CR127 standard hCG. The longer the incubation of CR127 standard with human leukocyte elastase (0, 2, and 21 h), the greater the extent of nicked hCG (20%, 46%, and 89%). As the extent of nicking increased, the receptor-binding ability diminished, as did the ability to stimulate progesterone production by rat corpus luteal cells in vitro (0.9, 0.74, and 0.29 fig/fig hCG, respectively). In a regression analysis, a linear relationship was indicated between the extent of nicking and receptor binding values (97% correlation) and between the extent of nicking and steroidogenic activity in vitro (99% correlation). From the intercepts of the regression lines, it was estimated that nicks reduced receptor binding by 11-fold and reduced the steroidogenic activity of hCG by 5-fold. We examined eight individual hCG preparations, three purified from pregnancy urine, three from urine from patients with hydatidiform mole, and two from urine from women with choriocarcinoma. In descending order, the eight individual hCG preparations were 100%, 100%, 85%, 76%, 42%, 41%, 0%, and 0% intact. Although no correlation was observed between the percent intact and the ability of the eight individual samples to displace 50% [125I]hCG in binding CG/LH receptor (r < 0.5), a close correlation was noted between the percent intact and the steroidogenic activity in vitro (98% correlation). This separated the effects of nicking on receptor binding and steroidogenic
h
CG IS a glycoprotein hormone composed of two dissimilar subunits, a (92 residues) and |8 (145 residues), joined noncovalently. While the a-subunit of hCG Received March 12,1991. Address all correspondence and requests for reprints to: Laurence A. Cole, Ph.D., Department of Obstetrics and Gynecology, Yale University School of Medicine, 333 Cedar Street, New Haven, Connecticut 06510. • This work was supported by NIH Grants CA-44131 (to L.A.C.), CA-46828 (to L.A.C.), HD-15454 (to S.B.), and HD-22970 (to F. Naftolin).
activities and indicated that while multiple factors influence receptor binding, only nicking suppresses the steroidogenic activity of bound hCG. We examined the recognition of nicked hCG molecules by different hCG immunoassays. The Hybritech Tandem assay measured total hCG and did not distinguish nicked and intact hCG molecules (in a regression analysis, immunoactivity vs. percent intact hCG, r < 0.5). In contrast, the immunometric assay using B109 hCG dimer-specific monoclonal antibody and anti-/3-peroxidase only detected the intact component of hCG (in a regression analysis, immunoreactivity vs. percent intact hCG, 98% correlation). We used these assays together to estimate the percentage of intact hCG and to deduce the extent of nicking. In 38 parallel sets of serum and urine samples, 20 from pregnancy and 18 from patients with trophoblast disease, wide variation was detected in the percentage of intact hCG (range, 6-100%). Mean values for the percentage of intact hCG in serum and urine were 72 ± 22% and 76 ± 35%, respectively (mean ± SD; P > 0.5). The percentage of intact hCG was not significantly different in pregnancy (71 ± 26%) and trophoblast disease (77 ± 33%) samples (P > 0.3). We concluded that nicked hCG accounts, on the average, for approximately one quarter of the total hCG in serum and urine. The finding of a major hCG component with only minimal steroidogenic activity is novel and raises issues concerning the physiological or regulatory function of nicking and the medical significance of variations thereof. Also, the finding that a hCG assay using a dimer-specific monoclonal antibody only measures intact hCG calls into question the results from many assays in current use that employ this type of antibody. (Endocrinology 129: 1559-1567,1991)
has the same peptide sequence as the a-subunits of LH, FSH, and TSH, the jS-subunit is unique and distinguishes the functions of hCG from those of the other glycoprotein hormones. hCG is normally produced in the placenta and is present at similar levels in urine and blood (serum/ plasma) in women with pregnancy and trophoblast disease. In 1973, Shome and Parlow (1) discovered missing peptide linkages at different sites on the LH /3-subunit between residues 46 and 49. More recent studies have confirmed this (2, 3) and have located the nicks (missing linkages) at 044-45, 047^48, and 048^49. Compari-
1559
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG
1560
sons of nicked and intact LH preparations have indicated that these nicks cause a 40% loss of biological activity (2). In 1988, Nishimura and colleagues (4) showed that nicks at undetermined sites were present on hCG from individuals with trophoblast disease. Since then, several groups, including ours, have confirmed this finding and located nicks at sites analogous to those found on LH (044—»45 and 047—48; see Fig. 1) in pregnancy, trophoblast disease, or standard hCG preparations (5-9). These studies, however, were preliminary, in that they were limited to one or two sequence analyses and to hCG purified from pools of urine. The origin and occurrence of nicked hCG and the effects of nicking on hCG actions still need to be addressed. The preceding articles are comprehensive studies of nicking, examining the origin, sites, and occurrences of hCG peptide cleavages in detail (10, 11). In the first paper, hCG was purified from 13 individual urine samples, and the peptide sequences were examined (11). Nicking was demonstrated in 11 of the 13 purified preparations at 047-48 and, with 3 exceptions, at 044-45 or 046-47. The proportion of hCG molecules nicked is shown to vary widely (0-100% of the total). In the second paper, the structures of 7 different reference preparations of hCG were examined (10) in which variable nicking (10-20% of the molecules) at 047^48 and 044-45 was demonstrated. Human leukocyte elastase was found to specifically nick hCG at sites in the 044-49 region in vitro and could be a cause of nicking in vivo. This paper complements the preceding articles, examining the biological effect of nicking and the development of an immunological approach to measure the extent of nicking in serum and urine samples. The receptor-binding and steroidogenic and immunological activities of 8 pure hCG preparations (0-100% nicked) and of elastase-nicked Asp Arg 6 0 Tyr Asn Cys 57 Tyr CyS'*'*' V al G| y 38 Pro Val
Ala
Thr
40
Gin
Met Thr Arg 44 Val
Pro Leu Ala Pro 50 Leu
45 Leu Gin
\Val 47 \
FIG. 1. Sequencing of hCG /3-subunit residue 55-61, showing major sites of missing peptide linkages or nicks (10).
Endo • 1991 Voll29«No3
hCG standard are measured. For the first time, specific immunoassays are developed for the measurement of total hCG and intact hCG (that not nicked) and applied to determine the extent of nicking in pregnancy and trophoblast disease samples.
Materials and Methods Samples Pure hCG samples, coded PI, P3, and P5 (from pregnancy urine), Ml, M2, and M4 (from individuals with hydatidiform mole), and C3 and C5 (from individuals with choriocarcinoma), were prepared, as described previously (11). Parallel serum and urine samples were collected from 21 individuals with normal pregnancy, 13 individuals with hydatidiform mole, and 12 individuals with choriocarcinoma or invasive trophoblast disease. Urine samples were frozen (—20 C) within 15 min of collection; serum samples were frozen immediately after centrifugation. Electrophoresis and immunoblotting Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was carried out on 12% polyacrylamide gels in a Bio-Rad Mini Protean II slab gel apparatus (Bio-Rad, Richmond, CA), using the discontinuous buffer system of Laemmli (12). Reduced samples were heated to 80 C for 10 min, then applied to gels. Nonreduced samples were applied directly to gels without heating. hCG standard (NIH batch CR127) and mol wt markers (Bio-Rad) in sample buffer were also applied to gels. The gels were subjected to a constant 200 V for 1 h at ambient temperature. Immunoblotting was carried out in a Bio-Rad Mini Trans-Blot apparatus, using the methods of Burnette (13). Briefly, gels were placed on a 0.2-/im nitrocellulose membrane, fitted into the transfer apparatus, and subjected to 100 V for 1 h at ambient temperature. Proteins transferred onto the membrane were washed free of sodium dodecyl sulfate, and the membranes were blocked from nonspecific binding by soaking in buffer containing 5% BSA. Nitrocellulose membranes were then incubated 4 h at ambient temperature with antiserum raised to hCG /3-subunit C-terminal peptide, residues 111-131 (CC11, kindly donated by Dr. H. C. Chen at NIH), and bands were visualized by further incubation with protein-A-gold conjugate (Bio-Rad Laboratories). Added sensitivity was achieved by a gold enhancement procedure (Bio-Rad Laboratories). This consisted of soaking the washed membranes with 0.2 M citrate buffer (pH 3.7) and then incubating with a solution of 0.85% hydroquinone plus 0.11% silver lactate for 15 min in the dark. The reaction was stopped using the supplied fixing solution. Gels were scanned and uniformly intensified using a HewlettPackard ScanJet Plus scanner (Palo Alto, CA; 256 Gray scale; 90 x 109 dots/in, resolution) and printed on a Hewlett Packard LaserJet Series II printer. Immunological activities Fresh solutions of CR127 hCG standard and PI, P3, P5, Ml, M2, M4, C3, and C5 hCG (pure) were prepared. Absolute concentrations of hCG were determined by amino acid analysis, as described in the preceding paper (11). Levels of immunoreactive hCG were determined by the Hybritech Tandem assay and the Bl09:anti-j8-peroxidase assay using PI hCG (0%
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG nicked) as standard. Levels were also determined using the Columbia anti-hCG immunoradiometric assay, which uses CR127 hCG (17% nicked) as standard. Levels were expressed as micrograms per ml. Immunoactivity was expressed as micrograms per ml by immunoassay divided by micrograms per ml by amino acid analysis, or micrograms per /xg hCG. The Bl09:anti-/3-peroxidase assay is a two-stage immunoenzymometric assay using a hCG dimer-specific monoclonal antibody, B109 (generated by Drs. A. Krichevsky and E. Armstrong of Columbia University) (14) to capture hCG and antij8 antiserum-peroxidase (Bios Pacific, Inc., Emeryville, CA) to label or detect it. Microtiter plates were coated with antibody B109 (200 /xl; 2 Mg/ml; in 0.25 M NaHCO3-0.1 M NaCl buffer, pH 9.5) and plates were incubated overnight at 4 C. Plates were washed five times with water and aspirated before use. In triplicate, 100 ix\ sample or standard (0-25 ng/ml) were added to coated wells together with 100 /ul buffer-carrier protein mix [0.05 M NaH2PO3/Na2HPO3-0.14 M NaCl (pH 7.5) plus 0.1% (wt/vol) ovalbumin]. Plates were shaken on a plate rotator overnight at ambient temperature, then washed five times with water and aspirated. Goat anti-/3 antiserum-peroxidase [200 /xl; 1:3500; in 0.1 M Tris-HCl-0.025 M CaCl2 (pH 7.5) plus 0.1% (wt/vol) ovalbumin] was then added to wells, and plates were shaken for a further 2 h at ambient temperature. After another five washes with water, 200 /xl substrate mix (5-mg tablet of orthophenylenediamine and 4 /xl 30% H2O2 in 25 ml 0.01 M sodium citrate, pH 4.9) were added, and plates were shaken in the dark for 30 min at ambient temperature. HC1 (50 /xl; 4 M) was added to stop reactions, and the absorbance of the wells was determined in a Flow Laboratories Titertek Multiscan NCC-340 plate reader (McClean, VA) at 492 nm. Data were sent to a Zenith 80286 computer, and standard curves were plotted and levels determined using Titersoft software (Flow Laboratories). All values were determined in triplicate. This assay is specific for hCG, with less than 1% cross-reactivity with hCG free /3-subunit, free a-subunit, or human (h) LH. The in-house anti-a:anti-j3 test was carried out by procedures similar to those used for the B109:anti-j8-peroxidase assay. The exception was the use of Unipath 2119:12 monoclonal anti-a (4 /xg/ml) in place of B109 anti-hCG. PI hCG was used as standard. All values were determined in triplicate. This assay is specific for hCG, with less than 0.1% cross-reactivity with hCG free j8-subunit, free a-subunit, or hLH. The C-terminal peptide assay (CCll anti-/3-subunit peptide:[125I]anti-/3) was described previously (10). This assay detects equimolar levels of hCG and free /3-subunit, with less than 0.1% cross-reactivity with free a-subunit and hLH. Tandem assay kits were provided by Hybritech (San Diego, CA). A two-step modification of the assay was used, as previously described (15). Briefly, methods were carried out according to the instructions, except two 2-h incubations were used in place of a single 1-h protocol. Beads (coated with antia antibody) were first incubated with sample, washed, then separately incubated with tracer antibody ([125I]anti-/3). After further washes, bead-bound radioactivity was determined, and concentrations were calculated. Values were determined in triplicate. This assay is also specific for hCG, with less than 0.1% cross-reactivity with hCG free /3-subunit, free a-subunit, or hLH. The Columbia University anti-hCG assay also uses monoclo-
1561
nal antibody B109, but at different coating concentrations and with a different tracer antibody ([125I]monoclonal antibody B108) than the B109:anti-/?-peroxidase test. The methods were reported by O'Connor et al. (16) and are widely used by research laboratories measuring low levels of hCG, as in early fetal loss studies (17). Biological activity
Immature female Sprague-Dawley rats, 26 days old, were injected sc with 50 IU hMG, followed 56 h later by 25 IU hCG (Sigma Chemical Co., St. Louis, MO). Ovaries were collected 6 days post-hCG injection, and luteal cells were prepared by a modification of the method of Behrman and colleagues (18). Luteinized ovaries were finely minced with a razor and dispersed in Ca2+-free Minimum Essential Medium 1380 (Gibco, Grand Island, NY) containing 0.1% (wt/vol) BSA, 2 mM glutamine, 2000 IU collagenase (Worthington Biochemical Co., Freehold, NJ), and 3000 IU deoxyribonuclease (Worthington; 5 ml/g ovarian tissue) and incubated for 1 h at 37 C in 95% air-5% CO2 in a culture cabinet. The population of luteal cells was enriched by application to a discontinuous Percoll density gradient (Pharmacia Fine Chemicals, Uppsala, Sweden). Luteal cells present in the density layers 1.018 and 1.033 g/ml Percoll were removed by aspiration, washed by 3-fold dilution with medium, collected by centrifugation (100 X g; 5 min.), and resuspended in Minimum Essential Medium 2360 (Gibco) containing 0.1% (wt/vol) BSA and 2 mM glutamine. To study hCG steroidogenic activity, cells were distributed into 12 x 75-mm glass tubes (2.5 x 106 cells/tube). Fresh solutions of CR127 hCG standard and PI, P3, P5, Ml, M2, M4, and C3 and C5 preparations of pure hCG were prepared. Absolute concentrations of hCG were determined by amino acid analysis. Two different concentrations of hCG samples (or six concentrations of standards) in the range 0.55 ng/ml were applied in triplicate to tubes containing luteal cells. Minimum Essential Medium 2360 was added to bring the volume to 1 ml, and tubes were incubated 90 min in a 37 C 95% air-5% CO2 culture cabinet. Conditioned medium was removed, and progesterone levels were determined by RIA (R1A kit, Serono-Baker Diagnostics, Inc., Allentown, PA). Pi hCG, which is 0% nicked (no missing linkages or Nterminal sequence deficiency in a- or 0-subunit), was used as the standard for biological activity studies. The absolute biological activity of PI hCG is 1.1 times that of NIH CR127 standard hCG (14,900 vs. 16,600 IU/mg). Progesterone values were plotted against concentrations of PI hCG, and activities relative to PI hCG were determined. Assay values were expressed as micrograms per ml, and biological activity as micrograms per ng (/xg/ml by bioassay -s- /xg/ml by amino acid analysis). Receptor-binding activity Receptor-binding activity was examined by competitive RRA, using [125I]hCG and homogenates of luteinized rat ovaries. hCG was radioiodinated by the chloramine-T method using the method of Greenwood and colleagues (19), as modified by Lee and Ryan (20). Homogenates of luteinized rat ovaries were prepared, and competitive binding assay was performed as described by Lee and Ryan (21). Serial dilutions of hCG samples were prepared (concentration determined by amino
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
Endo • 1991 Voll29«No3
OCCURRENCE AND ACTIVITIES OF NICKED hCG
1562
acid analysis), and displacement of [125I]hCG was determined. Values were linearized by the log-logit method, and the concentration necessary to displace 50% of [125I]hCG in binding the receptor (EC50) was determined.
Results Biological activity of nicked hCG Standard hCG (NIH batch CR127) is approximately 20% nicked at 044-45 or 047-48 (Fig. 1). Treatment of standard hCG with human leukocyte elastase increased the extent of nicking from 20-89% (10). We examined the effect of this elastase digestion on the receptor binding and steroidogenic activities of standard hCG. As shown in Table 1, the longer the incubation with human leukocyte elastase (0, 2, and 21 h), the smaller the percentage of intact (nonnicked) hCG (80%, 54%, and 11%) and the greater the concentration required to displace 50% of 125I-labeled standard hCG (EC50) in binding the CG/LH receptor (1.0, 2.2, and 10 /xg). A linear relationship was indicated between the percent intact hCG and receptor-binding activity (97% correlation). From the intercept of the regression line an 11-fold difference was suggested in the receptor-binding activity of nicked and intact hCG. A linear relationship (99% correlation) was also demonstrated between the percent intact hCG and steroidogenic activity in vitro (0.90, 0.74, and 0.29 /xg/ixg hCG). From the intercept of the regression line, nicked molecules were calculated to have approximately 20% the steroidogenic activity of intact hCG.
We also examined the receptor binding and steroidogenic activities of eight individual hCG preparations, three purified from pregnancy urine, three from urine from women with hydatidiform mole, and two from urine from patients with choriocarcinoma (Table 2). In ascending order, the eight individual preparations were 0%, 0%, 15%, 24%, 58%, 59%, 100%, and 100% nicked (or 100%, 100%, 85%, 76%, 42%, 41%, 0%, and 0% intact). No relationship was observed between the percent intact and the ability of the eight individual hCG samples to displace 50% of [125I]hCG in binding the CG/LH receptor (r < 0.5). In contrast, a linear relationship was evident between the percent intact and the steroidogenic activity of the eight individual samples in vitro (98% correlation). Western blots of nicked hCG Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of reduced hCG, followed by Western blotting and staining of the blot with antibodies to the 0-subunit Cterminal peptide region, permits visualization of nicked hCG (4-8). An example of this method is shown in Fig. 2. As observed, when Western blots were made of serum and urine samples containing nicked hCG, two bands were detected. The first, mol wt (Mr) 35,000, corresponded to the intact 0-subunit of hCG. The second, Mr 24,000, corresponded to the peptide 048-145 released from nicked hCG after reduction of disulfide bonds. The presence of this band indicated nicking. The example in Fig. 2 is the starting urine of sample C3 and a parallel
TABLE 1. The biological, receptor binding, and immunological activities of CR127 standard hCG, nicked by the action of human leukocyte elastase (HLE) in vitro Sequence analysis (% intact)0
Sample
CR127 hCG after 0 h with HLE 2 h with HLE 21 h with HLE
80 54 11
Amino acid analysis (/ig/ml)
850 850 850
Immunoassay Hybritech Tandem
Immunoassay BlO9:anti-/8peroxidase
Bioassay steroidogenic activity in vitro
RRA
(Mg/ml)
840d 950 890
jig/ml
Atg/Mg hCG c
fig/'ml
Mg/^g hCG c
760 630 140
0.89e 0.74 0.16
770 630 250
0.90' 0.74 0.29
1.0* 2.2 10
The protocol is that described in the preceding paper (10). Briefly, hCG is digested with enzyme at 37 C in pH 8.5 buffer (0.03 U enzyme/nmol hCG). Aliquots are taken, and the reaction halted by the addition of phenylmethylsulfonylfluoride for reverse phase HPLC and sequence analysis or ovalbumin for immuno- and bioassays. All assays used Pi hCG, which is 0% nicked, as standard. 0 Percent intact is calculated as 100% minus percent nicked, determined by reverse phase HPLC and N-terminal sequence analysis (10). Values are estimates plus and minus five percentage points. 6 Values are the concentration (nanograms per ml by amino acid analysis) necessary to displace 50% of [126I]hCG in binding receptor; values are normalized to results for CR127 hCG after 0 h with HLE. c hCG determined by amino acid analysis. d Values in Hybritech Tandem assay were 105 ± 6.0% (mean ± SD) of amino acid analysis values. No correlation was observed between Hybritech Tandem assay values (micrograms per ng hCG) and percent intact (r > 0.6). " In a regression analysis, immunoactivity (micrograms per ng hCG) at 0, 2, and 21 h vs. percent intact, slope = 0.011, intercept = 0.071, and r = 0.98 (98% correlation). 1 In a regression analysis, bioactivity (micrograms per ng hCG) at 0, 2, and 21 h vs. percent intact, slope = 0.0090, intercept = 0.20, and r = 0.99 (99% correlation). * In a regression analysis, receptor binding EC50 values at 0, 2, and 21 h vs. percent intact, slope = 0.14, intercept = 11, and r = 0.97 (97% correlation).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG TABLE 2. The biological and receptor-binding activities of eight individual hCG preparations
Sample
Sequence Amino acid analysis analysis (% intact) 0
Bioassay steroidogenic activity in vitro
RRA (EC 60 ) 6
Mg/ml Mg/Mg hCG'
1. Pregnancy Pi
100
41 P3 P5 42 2. Trophoblast disease Ml
M2 M4
C3 C5
620 520
1460 1090 1490 3400
85 100 0 76 0
1.0e
620 250 730
1.0" 0.48 0.50
950
0.87
1.3
1.2
0.82 0.60 0.82
0.20
1.6
320
1760 ND' ND /
2100
420
1.6 1.1
The in vitro biological activity of each hCG preparation was determined in the rat luteal cell progesterone test. Values were compared to the biological activity of P i hCG, which is 100% intact. ° Percent intact is calculated as 100% minus percent nicked, determined by N-terminal sequence analysis (10). Values are estimates plus and minus five percentage points. b Values are the concentration (nanograms per ml by amino acid analysis) necessary to displace 50% of [125I]hCG in binding receptor, values are normalized to results for P i hCG. c hCG determined by amino acid analysis. d In a regression analysis, bioactivity (micrograms per ng hCG) of P1-C5 vs. percent intact, slope = 0.0091, intercept = 0.15, and r = 0.98 (98% correlation). ' In a regression analysis, receptor binding EC50 values vs. percent intact, no correlation was observed {P < 0.5). ' Not determined.
75,000* 50,000*
f
34,000* 24,000*
M24,000
Reduced C3 Serum
C3 Urine
FIG. 2. Scan of nitrocellulose blot of sodium dodecyl sulfate-polyacrylamide gel electrophoresis of reduced (+) and nonreduced (-) C3 patient serum and a parallel urine sample.
serum sample. As shown in a preceding paper (11), after purification and N-terminal sequence analysis, sample C3 was found to be 24% nicked at 047-48. Using this Western blot method we examined 24 urine and 12 parallel serum samples from individuals with pregnancy or trophoblast disease. As shown in Table 3, regardless of origin (pregnancy or trophoblast disease), 79% of urine and a similar proportion of serum samples (83%) gave a Mr 24,000 band, indicating that approximately 80% of serum and urine samples were nicked. Studies with different hCG standards (10-20% nicked; see preceding paper) indicated that 10% nicking was
1563
TABLE 3. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting of pregnancy and trophoblast disease serum and urine (paired) hCG samples Source Pregnancy Trophoblast disease Total
No. with 24,000 Mr band" Serum 4 of 5 7 of 7 10 of 12 (83%)
Matching urine 5 of 7 14 of 17 19 of 24 (79%)
hCG nicking can be detected in serum and urine samples by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting. Intact hCG /3-component (from intact hCG) reduced with (3mercaptoethanol migrates at 35,000-36,000 daltons on these gels; the C-terminal part of nicked hCG ^-component (peptide 048-145 of nicked hCG), however, migrates faster, at 24,000 daltons (6). " Studies with hCG standards (up to 12% nicked) indicate that samples need to be more than 10% nicked for unequivocal visual detection of the 24,000 Mr band.
required for consistent observation of a 24,000 Mr band. Thus, the approximately 20% of samples without an obvious 24,000 Mr band may represent samples with small amounts of nicking ( 0.5). The mean percent intact hCG was not significantly different in pregnancy (71 ± 26%) and trophoblast disease (77 ± 33%) samples (P > 0.3). We inferred that nicked hCG accounts, on the average, for approximately 26% of the total hCG in serum
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG and urine samples, regardless of origin (pregnancy or trophoblast disease).
Discussion Here we show that nicks at 044-45 or 047-48 reduce the receptor binding and steroidogenic activities of hCG. When standard hCG is nicked by the action of human leukocyte elastase in vitro, both the ability to displace [125I]hCG in binding the receptor and steroidogenic activity are reduced. It was estimated that molecules nicked this way have 9% of the receptor binding and 20% of the steroidogenic activity of intact hCG. Recently, Sakakibara and colleagues (8) purified the nicked component of hCG 0-subunit and combined it with w-subunit to form a recombinant nicked hCG. Steroidogenic activity was measured and was found to be less than 1% that of natural hCG. This is much lower activity for nicked hCG than reported here. The lower activity, however, might be due to a poor or abnormal combination of a- and nicked 0-subunit or to denaturation or molecular damage of nicked 0-subunit during reverse phase HPLC procedures using trifluoroacetic acid. Studies by Bousfield and Ward (22) show that ovine LH 0-subunit, nicked at 043-44 in vitro by endoproteinase Arg C and recombined with ovine LH a-subunit, had approximately 2% the receptor-binding activity of intact oLH. This is approximately one quarter of the result reported here with elastase-cleaved hCG. The difference again could be due to the use of recombined, rather than natural, dimers. We considered how nicks in the 044-50 region of standard hCG could diminish the receptor-binding activity of hCG 11-fold and steroidogenic activities of hCG 5fold. As shown in Fig. 1, nicking occurs in a hydrophobic loop in the hCG 0-subunit, which is held in place by a disulfide link between Cys38 and Cys57. Keutmann and colleagues (23) showed that this loop may be part of the receptor-binding domain of hCG. Cleavages at 044-45 or 047-48 would open the loop, which could change the tertiary structure and thus ablate the receptor-binding properties of hCG. Keutmann and colleagues (24) made analogs missing the Cys38-Cys57 disulfide linkage. The analogs retained 20% of receptor-binding activity. Consistent with this is our finding that molecules nicked in this same region retain 9% of receptor binding and 20% of steroidogenic activity. Our results confirm those of Keutmann and colleagues (23, 24) that the 038-57 region is important for biological activity and that minor cleavages in this region result in an 80% loss of hormone function. Significant variation has been reported in the in vitro biological activity of hCG, whether from a single source, such as first trimester pregnancy, or from multiple origins, such as pregnancy and trophoblast disease (reviews in Refs. 25-27). Much of this variation has been
1565
attributed to minor differences in hCG oligosaccharide structure (26-31). We purified hCG from urine from eight individuals, three with pregnancy, three with hydatidiform mole, and two with choriocarcinoma, and investigated the source of variations in hCG biological activities. The receptor-binding activity of the eight individual samples varied (0.82-1.6 ng/ng), regardless of source. Studies with CR127 standard hCG indicated that nicking can reduce the receptor-binding activity of hCG by 11-fold. We compared the extent of nicking and receptor-binding activity. No correlation was evident (linear regression, r < 0.5), indicating that nicking was not the primary cause of variation in in vitro receptorbinding activity of individual hCG samples. a-Subunit N-terminal heterogeneity was the only other peptide variable in these eight hCG preparations (11). Regression analysis, however, showed no correlation between receptor-binding activity and the extent of a-subunit N-terminal heterogeneity (0%, 0%, 0%, 27%, 18%, 12%, 0%, and 19%, respectively, in samples P1-C5; r < 0.5). We concluded that variations in receptor-binding activity were not primarily due to peptide variations, but, rather, as suggested by others (26-31), were due to carbohydrate microheterogeneity. The in vitro steroidogenic activity of the eight individual hCG preparations also varied widely (0.20-1.0 ng/ng hCG). The variation, however, whether in pregnancy or trophoblast disease samples, correlated with the percentage of intact hCG (98% correlation). We conclude that variations in the extent of nicking, rather than differences in carbohydrate content, are the principal cause of disparity in the steroidogenic activity of hCG. These findings indicated a dissociation of the effects of nicking on receptor binding and steroidogenic activities, and that although carbohydrate structure and extent of nicking affect receptor binding, only nicking suppresses the steroidogenic activity of bound hCG. The difference in results for receptor-binding activity and steroidogenic activity between individual hCG samples nicked in vivo and hCG standards nicked by human leukocyte elastase in vitro could be due to the different cleavage sites. While the former are only cleaved at 04748, the latter are cleaved at 044-45 and 047-48 and, in addition, at 048-49 and a70-71 depending on the incubation time with elastase. It is possibile that only the cleavage at 047-48 affects steroidogenic activity. This, however, seems unlikely, considering that all three cleavage sites in the 044-49 region are likely to open up the 038-57 intercysteine loop. We cannot, however, rule out the possibility that cleavages at sites other than 047-48 lead to the loss of receptor-binding activity observed only with elastase-cleaved preparations. Studies by Catt and Dufau (32) may provide another explanation for the different effects of nicked hCG on receptor binding and steroidogenesis. They show that steroidogenesis is max-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
1566
OCCURRENCE AND ACTIVITIES OF NICKED hCG
imal when only a small proportion of gonadotropin receptors are occupied. Thus, the differential effect of nicked hCG may be due in part to the greater sensitivity of steroidogenesis. We examined the specificity of different immunoassays for nicked hCG. Individual hCG samples were tested in three types of assays, anti-a:anti-/5 assays, anti-/? Cterminal segment assays, and an assay using hCG dimerspecific antibody. Two anti-a:anti-/3 immunometric tests were evaluated: the Hybritech Tandem assay and our in-house test. Both reflected amino acid analysis results and were not significantly affected by the percentage of nicking (r < 0.5). Values in the Hybritech Tandem assay were 94 ± 20% (±SD), and those in the in-house anti-a:anti-/? assay were 108 ± 22% of amino acid analysis values. A hCG (3subunit C-terminal peptide assay was also evaluated. As with the anti-a:anti-/3 assays, values represented total hCG and were not significanty affected by the percentage of nicking (r < 0.5). The results in a third type of immunoassay, using a hCG dimer-specific monoclonal antibody, were, however, affected by the percentage of nicking. In a regression analysis, values from the Columbia anti-hCG assay, which is used at numerous centers in the U.S. for the detection of low levels of hCG (16, 17), correlated with the percentage of intact (nonnicked) hCG (91% correlation). A limitation of this nick-sensitive assay was the use of CR127 hCG standard (20% nicked), which could falsely elevate hCG results. An improved version of this assay (B109:anti-/3 peroxidase), using P i hCG (0% nicked) as standard and a lower B109 coating concentration, gave results more closely correlating with the percentage of intact hCG (94% correlation). Similar results were observed in an assay using B107, another hCG dimer-specific antibody. This indicated that the differential recognition of intact and nicked hCG may be a property of most antibodies specific for hCG dimer. Varying results have been reported with the same sample in different hCG immunoassays and by different clinical laboratories. The problems and confusion created by these differences is exemplified by a recent article by Painter, "Discordant hCG results in pregnancy: a method in crisis" (33), and in several other recent articles (3436). In general, the competitive nature of RIAs, such as the hCG/? RIA, renders them more sensitive to small peptide or carbohydrate structural differences. As shown here, results from different types of immunometric assays, such as the anti-a:anti-/? (measures total hCG) and the anti-hCG dimer type (distinguished nicked and intact molecules), can also conflict and be confusing. Furthermore, the international hCG standards (First International Reference Preparation, Third International Standard, and CR series) are 10-20% nicked (9) and can themselves confuse results in different competitive and
Endo • 1991 Voll29«No3
immunometric assays. We concluded that variable nicking in individual serum and urine samples and in standard hCG preparations is a major contributor to the microheterogeneity of hCG and may be a cause of discordant results in different hCG immunoassays. Clearly, some assays, such as our in-house anti-a:anti-/? test, the Hybritech Tandem kit, and possibly numerous other similar assays, are unaffected by nicking and thus measure total hCG. Other assays, such as the Columbia antihCG test, the B109:anti-/3 peroxidase assay, and test kits that use this common type of antibody, are specific for intact or nonnicked hCG. While the anti-a:anti-/3 test, which measures total hCG, may be optimal for following hCG production, the latter, the intact hCG assay, may be preferable for measuring functional or biologically active hormone. We considered methods other than purification and sequence analysis to investigate the occurrence of nicking in hCG samples. Nishimura et al. (4), Bidart and colleagues (5), Sakakibara et al. (8), and Puisieux and colleagues (7) all used electrophoresis of reduced hCG and Western blotting with C-terminal peptide antisera to detect nicked molecules. Using this method we were able to demonstrate the broad occurrence of nicked molecules in pregnancy and trophoblast disease fluids, detecting nicks in 19 of 24 urine and 10 of 12 serum hCG samples. Quantitative methods were needed for investigating the amount of nicking in serum and urine samples. Using the Hybritech Tandem assay to measure total hCG and the B109:anti-/?-peroxidase assay to detect intact molecules, we were able to deduce the amount of nicked hCG (total minus intact) in serum and urine samples. In 38 parallel sets of serum and urine samples, 20 from pregnancy and 18 from patients with trophoblast disease, a similar percentage of intact hCG was detected in serum and urine (mean ± SD, 72 ± 22% and 76 ± 35%, respectively; P > 0.5). The percentage of intact hCG was not significantly different in pregnancy (71 ± 26%) and trophoblast disease (77 ± 33%) samples (P > 0.3). We deduced from these findings that approximately onequarter of hCG molecules, whether from urine or serum, pregnancy or trophoblast disease, are nicked. The finding of a major hCG component in pregnancy and trophoblast disease serum and urine with only minimal biological activity is novel. It raises questions concerning the physiology of hCG and the possibility that nicking is a specific regulatory process. Questions are also raised about the clinical significance of nicked hCG and whether the extent of nicking could mark gestational disorders.
Acknowledgments The authors wish to thank Dr. H. C. Chen at NIH for the gift of hCG/3 C-terminal peptide antisera (CC 11), and Dr. Fu
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
OCCURRENCE AND ACTIVITIES OF NICKED hCG Chai-ying and the medical staff in the Department of Obstetrics and Gynecology at Yale University for providing the parallel urine and serum samples used in this paper. Thanks also go to Kay Shirey at Columbia University for proofreading this manuscript before submission.
17. 18.
References 1. Shome B, Parlow AF 1973 The primary structure of the hormonespecific, beta subunit of human pituitary luteinizing hormone (hLH). J Clin Endocrinol Metab 36:618-624 2. Ward DN, Glenn SD, Nahm HS, Wen T 1986 Characterization of cleavage products in selected human lutropin preparations. Int J Peptide Protein Res 27:70-78 3. Hartree AS, Lester JB, Shownkeen RC 1985 Studies of the heterogeneity of human pituitary LH by fast protein liquid chromatography. J Endocrinol 105:405-414 4. Nishimura R, Ide K, Utsunomiya T, Kitajima T, Yuki Y, Mochizuki M 1988 Fragmentation of the /9-subunit of human chorionic gonadotropin produced by choriocarcinoma. Endocrinology 123:420425 5. Bidart J-M, Puisieux A, Troalen F, Foglietti MJ, Bohuon C, Bellet D 1988 Characterization of the cleavage product in the human choriogonadotropin /8-subunit. Biochem Biophys Res Commun 154:626-632 6. Cole LA, Kardana A, Birken S 1989 The isomers, subunits and fragments of hCG. Serono Symp 65:59-78 7. Puisieux A, Bellet D, Troalen F, Razafindratsita A, Lhomme C, Bohoun C, Bidart J-M 1990 Occurrence of fragmentation of free and combined forms of the 0 subunit of human chorionic gonadotropin. Endocrinology 126:687-694 8. Sakakibara R, Miyazaki S, Ishiguro M 1990 A nicked /3-subunit of human chorionic gonadotropin purified from pregnancy urine. J Biochem 107:858-862 9. Lustbader JW, Birken S, Pileggi NF, Kolks MAG, Pollak S, Cuff ME, Yang W, Hendrickson WA, Canfield RE 1989 Crystallization and characterization of human chorionic gonadotropin in chemically deglycosylated and enzymatically desialylated states. Biochemistry 28:9239-9243 10. Birken S, Gawinowicz MA, Kardana A, Cole LA 1991 The heterogeneity of hCG. II. Characteristics and origins of nicks in hCG reference standards. Endocrinology 129:1551-1558 11. Kardana A, Elliott ME, Gawinowicz MA, Birken S, Cole LA 1991 The heterogeneity of hCG. I. Characterization of peptide variations in 13 individual preparations of hCG. Endocrinology 129:15411550 12. Laemmli UK 1970 Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680-685 13. Burnette WN 1981 "Western blotting:" electrophoretic transfer of proteins from sodium dodecyl sulfate-polyacrylamide gels to unmodified nitrocellulose and radiometric detection with antibody and radioiodinate protein A. Anal Biochem 112:195-201 14. Krichevsky A, Armstrong EG, Schlatterer J, Birken S, O'Connor J, Bikel K, Silverberg S, Lustbader JW, Carfield R 1988 Preparation and characterization of antibodies to the urinary fragment of the human chorionic gonadotropin /3-subunit. Endocrinology 123:584-593 15. Cole LA, Birken S 1988 Origin and occurrence of human chorionic gonadotropin /3-subunit core fragment. Mol Endocrinol 2:825-830 16. O'Connor J, Schlatterer JP, Birken S, Krichevsky A, Armstrong
19. 20. 21. 22. 23.
24.
25.
26. 27. 28. 29. 30.
31. 32. 33. 34. 35. 36.
1567
EG, McMahon D, Canfield R 1988 Development of highly sensitive immunoassays to measure human chorionic gonadotropin, its fisubunit, and /3 core fragment in the urine: application of malignancies. Cancer Res 48:1361-1366 Wilcox AJ, Weinberg CR, O'Connor JF 1988 Incidence of early pregnancy loss. N Engl J Med 319:189-194 Behrman HR, Preston SL, Hall AK 1980 Cellular mechanism of the antigonadotropic action of luteinizing hormone-releasing hormone in the corpus luteum. Endocrinology 107:656-664 Greenwood FC, Hunter WM, Glover JS 1963 The preparation of 12S I-labeled human growth hormone of high specific radioactivity. Biochemistry 89:114-123 Lee CY, Ryan RJ 1973 Interactions of ovarian receptors with luteinizing hormone and human chorionic gonadotropin. Biochemistry 12:4609-4615 Lee CY, Ryan RJ 1972 Luteinizing hormone receptors: specific binding of human luteinizing hormone to homogenates of luteinized rat ovaries. Proc Natl Acad Sci USA 69:3520 Bousfield GR, Ward DN 1988 Selective proteolysis of ovine lutropin or its j8-subunit by endoproteinase arg-C. J Biol Chem 263:12602-12607 Keutmann HT, Charlesworth MC, Mason KA, Ostrea T, Johnson L, Ryan RJ 1987 A receptor-binding region in human choriogonadotropin/lutropin beta subunit. Proc Natl Acad Sci USA 84:2038-2042 Keutmann HT, Charlesworth MC, Mason KA, Johnson L, Ryan RJ 1988 Primary and secondary structural determinants in the receptor binding sequence beta (38-57) from human luteinizing hormone. Biochemistry 27:8939-8944 Vaitukaitis JL 1978 Glycoprotein hormones and their subunitsimmunological and biological characterization. In: McKerns KW (ed) Structure and Function of the Gonadotropins. Plenum Press, New York, pp 339-360 Fauser BCJ M, Hsueh AJW 1989 Bioassays of gonadotropins: background and clinical applications. Serono Symp 65:123-136 Grotjan HE, Cole LA 1988 Human chorionic gonadotropin microheterogeneity. In: Keel B, Grotjan H (eds) Microheterogeneity of Glycoprotein Hormones. CRC Press, Boca Raton, pp 219-237 Kalyan NK, Lippes HA, Bahl OP 1982 Role of carbohydrate in human chorionic gonadotropin. J Biol Chem 257:12624-12631 Moyle WR, Bahl OP, Marz L 1975 Role of carbohydrate of human chorionic gonadotropin in the mechanism of hormone action. J Biol Chem 23:9163-9169 Birken S, Agosto G, Amr S, Nisula B, Cole L, Lewis J, Canfield R 1988 Characterization of antisera distinguishing carbohydrate structures in the /3-carboxyl-terminal region of human chorionic gonadotropin. Endocrinology 123:572-583 Amir SM, Kasagi K, Ingbar SH 1987 The role of subunit sialic acid in the thyrotropic and gonadotropic activities of human chorionic gonadotropin. Endocrinology 121:160-166 Catt KJ, Dufau ML 1973 Spare gonadotropin receptors in rat testis. Nature [New Biol] 244:219-221 Painter PC 1989 Discordant hCG results in pregnancy: a method in crisis. Clin Casebook 27:20-24 Hussa RO 1987 The Clinical Marker hCG. Praeger Press, New York Chen HC, Matsuura S, Ohashi M 1980 Limitations and problems of hCG-specific antisera. In: Segal SJ (ed) Chorionic Gonadotropin. Plenum Press, New York, pp 231-252 Hussa RO, Rinke ML, Schweitzer PG 1985 Discordant human chorionic gonadotropin results: causes and solutions. Obstet Gynecol 65:211-219
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 09 June 2015. at 00:07 For personal use only. No other uses without permission. . All rights reserved.
