The Occurrence of 0-Acylation during Biotinylation of Gonadotropin-releasing Hormone and Analogs EVIDENCE FOR A REACTIVE SERINE*

Gonadotropin-releasing hormone (GnRH) and two of its analogs ([D-Lys’IGnRH and de~-Gly’~-[o-Trp~]- GnRH) were reacted with sulfonated N-hydroxysuc-cinimide esters of biotin that have been reported to react specifically with primary amino groups. Frac- tionation by reversed-phase high performance liquid chromatography demonstrated the Occurrence of mul- tiple biotinylated derivatives for each reacted peptide. hydroxylamine significant 0-biotinylation had the observed 0-biotinylation was dependent on peptide conformation. All the 0-biotinylated derivatives displayed significantly reduced bioactivity. Taken together, these results give strong evidence that the Ser4 hydroxyl of GnRH has a significantly elevated intrinsic reactivity, which raises new ques- tions concerning its putative role in the conformation and mode of action of the hormone. These results also demonstrate for the first time that the N-hydroxysuc-cinimide-biotin esters are capable of significant 0- acylation and may be generally useful reagents for detecting highly reactive hydroxyamino acid residues.

Gonadotropin-releasing hormone (GnRH) and two of its analogs ([D-Lys'IGnRH and de~-Gly'~-[o-Trp~]-GnRH) were reacted with sulfonated N-hydroxysuccinimide esters of biotin that have been reported to react specifically with primary amino groups. Fractionation by reversed-phase high performance liquid chromatography demonstrated the Occurrence of multiple biotinylated derivatives for each reacted peptide. These results were unexpected since GnRH and de~-Gly'~-[~-Trp']GnRH contained no reactive amino groups and [D-LYS'IG~RH had only one. Reaction of the biotinylated derivatives with hydroxylamine indicated that significant 0-biotinylation had occurred. Mass spectrometric analyses established the stoichiometry of biotinylation and confirmed that substantial 0-biotinylation of residue Ser4, and to a minor extent Tyr', of GnRH and the two analogs had occurred. In contrast, the biotinylation of selected peptides unrelated to GnRH under identical reaction conditions indicated no significant evidence of 0-acylation of seryl residues. Strikingly, biotinylation of GnRH under denaturing conditions largely abolished 0-acylation, indicating that the observed 0-biotinylation was dependent on peptide conformation. All the 0-biotinylated derivatives displayed significantly reduced bioactivity. Taken together, these results give strong evidence that the Ser4 hydroxyl of GnRH has a significantly elevated intrinsic reactivity, which raises new questions concerning its putative role in the conformation and mode of action of the hormone. These results also demonstrate for the first time that the N-hydroxysuccinimide-biotin esters are capable of significant 0acylation and may be generally useful reagents for detecting highly reactive hydroxyamino acid residues.
GnRH' is a linear decapeptide produced by neurosecretory cells in the anterior hypothalamus and in extrahypothalamic neuronal systems of several species (Sherwood, 1987). Released in discrete pulses from hypothalamic nerve terminals, GnRH is transported to the anterior pituitary, where it effects the release of LH and FSH (Fink, 1988). In addition to its critical endocrine actions on the pituitary, GnRH has important effects in the brain, where it modulates certain aspects of reproductive behavior (Moss and Dudley, 1989). The essential role of GnRH in mammalian reproductive physiology has led to extensive investigations of the structure/function of this peptide hormone and the synthesis of more than 2000 analogs, some of which have found widespread clinical applications (see review by Karten and Rivier (1986)). These studies have exhaustively documented the relative importance of each of the 10 GnRH residues in various aspects of receptor binding and bioactivity. Similar investigations have determined which residues are required for metabolic stability and have shed light on the in vivo enzymatic degradation of the hormone. The majority of the reported residue substitution and peptide modification experiments have not extensively addressed questions concerning the intrinsic chemical reactivity of functional groups on GnRH and its analogs. However, knowledge of intrinsic residue reactivity can provide significant information relating to hormonal mode of action and can be valuable in interpreting results obtained from threedimensional structural analysis.
The use of the avidin-biotin system to study molecular and biological interactions has expanded significantly over the last decade (see reviews by Bayer and Wilchek (1980) and Wilchek and Bayer (1990)). Biotinylation reagents have been widely employed to acylate amino groups on lysyl and aamino-terminal residues of proteins and peptides (Hofmann and Finn, 1985;Hochhaus et al., 1988;Wilchek and Bayer, 1988). These published reports have emphasized that the specificity of NHS-biotin esters is directed to amino groups (Yem et al., 1989), and there have been no previous reports of 0-acylation by these reagents. In this report, we describe that, in the case of GnRH and related analogs, considerable 0acylation can occur under reaction conditions normally employed for peptide biotinylation. In addition to describing the The abbreviations used are: GnRH, gonadotropin-releasing hormone; NHS-biotin, N-hydroxysuccinimide esters of biotin; sulfo-NHS-cAhx-biotin, sulfosuccinimidyl-6-(biotinamido) hexanoate; HPLC, high performance liquid chromatography; HEPES, N-2-hydroxyethylpiperazine N-2-ethanesulfonic acid; LH, luteinizing hormone; FSH, follicle-stimulating hormone; EMIP, epidermal mitosis inhibitory pentapeptide; MS, mass spectrometry; MS/MS, tandem mass spectrometry; FAB, fast atom bombardment; CID, collisioninduced dissociation; MS-1 and MS-2, the first and second spectrometers of a tandem high resolution mass spectrometer; [M + HI+, protonated molecular ion; [m+Na]+, sodium adduction; RIA, radioimmunoassay. 5060 0-Acylation of Serine in GnRH and Related Peptides 5061 modification, purification, and detailed structural analysis of these unexpected derivatives, we have also evaluated the relative bioactivity of the various monobiotinylated GnRH species isolated by HPLC, Furthermore, we have compared the results of acylation of GnRH and its analogs with unrelated but structurally similar peptides. Finally, we discuss the implications of these findings in the general design and interpretation of peptide biotinylation investigations that require the preparation of biotinylated peptide ligands.
Biotinylation-Biotinylation reactions were conducted in 50 mM sodium bicarbonate (250 pl) at peptide concentrations of 1 mg/ml, pH 8.2, containing sulfo-NHS-rAhx-biotin at indicated reagent: peptide molar ratios. The sulfo-NHS-cAhx-biotin was added dry to minimize excessive reagent hydrolysis by water. The reaction mixtures were gently shaken every 5-10 min during the course of the biotinylation. In time course biotinylation experiments, aliquots of the reaction mixture were removed at various intervals as indicated, acidified to pH 3-4 with 0.1% trifluoroacetic acid, and rapidly frozen. All aliquots and reaction mixtures were stored at -20 "C until HPLC analysis. For biotinylation reactions in the presence of guanidine HC1, identical aliquots of GnRH were dissolved in 0.1 M sodium bicarbonate containing 6 M guanidine HC1, pH 8.2, and incubated at 25 or 50 "C for 60 min or 100 "C for 5 min. These solutions were then biotinylated for 40 min at 25 "C with sulfo-NHS-cAhx-biotin. The entire reaction mixtures were then acidified and subjected to HPLC analysis. A separate set of experiments was conducted to examine the stability of the derivatives. Biotinylated peptide fractions were heated at 37 "C for 2 h in 0.1% trifluoroacetic acid and re-analyzed by C18 reversed-phase HPLC. As a point of information, we have observed batch-to-batch variation in the reactivity of commercial biotinylation reagents. Therefore, all comparative experiments were conducted with the same batch of reagent.
Hydroxylamine Reaction-Hydroxylamine hydrochloride solutions (0.8 M in 0.1 M boric acid) were adjusted to pH 9.2 with NaOH. Aliquots of HPLC-purified, biotinylated peptide were dissolved in a small volume of 0.1% trifluoroacetic acid and added to the buffered hydroxylamine solution. The reaction was allowed to proceed for 4 h at 25 "C. Control incubations were carried out in 0.1 M boric acid, pH 9.2. The reaction mixtures were then acidified with trifluoroacetic acid and analyzed by C18 reversed-phase HPLC.
HPLC-Lyophilized peptides and reaction mixtures were dissolved in 0.1% trifluoroacetic acid and applied to a Vydac C18 reversedphase HPLC semipreparative column. Peptides were eluted at 25 "C with a linear gradient of solvent A (0.1% trifluoroacetic acid) and solvent B (100% acetonitrile containing 0.1% trifluoroacetic acid) at a flow rate of 1.75 ml/min. Our standard linear gradient for the GnRH analogs and their biotinylated derivatives was 0-45% solvent B over 120 min. An identical gradient was used during chromatography of EMIP. For HPLC of eledoisin and physalaemin, a linear gradient of 12-57% solvent B over 120 min was employed. The column eluate was monitored at 215 nm, and 1-min fractions were collected. Fractions were pooled based on absorbance and were subjected to amino acid compositional analysis, mass spectrometry, and bioassays.
All fractions and aliquots were lyophilized and stored at -20 "C.
Amino Acid Analysis-Samples were hydrolyzed with 6 N HCl in vacuo at 107 "C for 22-24 h. Amino acid compositional analysis was carried out on a Beckman 121MB analyzer employing single column methodology on Beckman W-2 resin or on a Beckman 6300 analyzer. Alternatively, samples were applied directly to an Applied Biosystems 420H derivatizer-hydrolyzer that provided on-line hydrolysis and phenylthiocarbamyl derivatization. The Applied Biosystems 420H analyzer was employed for the quantification of the 6-amino hexanoic acid spacer arm, which provided information about the extent of biotinylation of the various peptide derivatives .
Mass Spectrometry-Mass spectrometric analyses were performed at the Mass Spectrometry Facility, Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts. FAB/ MS analyses of the GnRH-related biotinylated derivatives were carried out in the first of two mass spectrometers of a tandem high resolution mass spectrometer (JEOL HXllO/HX110) as previously described (Barber et al., 1981). Tandem MS was carried out using all four sectors of the JEOL HXllO/HX110, essentially according to Sat0 et al. (1987). Briefly, the CID of protonated peptide molecules, selected with MS-1, took place in the field-free region after B1, thus operating both MS-1 and MS-2 as double-focusing instruments. The CID mass spectra were recorded with 100 Hz filtering at a rate that corresponds to a scan from m/z 0 to 6000 in 1.5 min. MS-1 was operated at a resolution set to transmit only the species of the protonated peptide molecule to be analyzed. MS-2 was operated at a resolution of 1:lOOO and was calibrated with a mixture of CsI, NaI, KI, RbI, and LiC1.
Bioassays-Pituitary glands of intact, adult male Sprague-Dawley rats were removed within 1 min of sacrifice and placed in ice-cold Ca2+-and Mg+-free Hanks' balanced salt solution supplemented with 0.4% bovine serum albumin, 25 mM HEPES, and gentamicin sulfate (5 mg/100 ml). The neurointermediate lobes were removed, and the anterior pituitaries were washed, quartered, diced with a sterile razor blade, and washed again with Hanks' salt solution. The tissues were gently agitated for 25 min at 25 "C in a solution of 0.3% trypsin containing 2.5 pg DNAase/anterior pituitary, and then mechanically disrupted with a Pasteur pipette (40X) in 1 ml of Medium-199 containing trypsin inhibitor and DNAase (each 25 pglml). Dispersed cells were recovered by low speed centrifugation, resuspended, and examined for cell number and viability using a hemacytometer and the trypan blue exclusion test (Freshney, 1983). Viabilities greater than 90% and cell counts of 1-2 X lO'/anterior pituitary were routinely observed. Cells were cultured in a moisturized atmosphere of air:COz (95:5) at 37 "C in Falcon multiwell tissue culture plates (24 wells/plate) at a density of 4 X lo4 cells/well and incubated in Medium-199 containing Earle's balanced salts, 2 mM CaC12, 13 mM HEPES, 17 mM NaH2C03, 1000 units/ml penicillin, 100 mg/liter streptomycin, and 10% fetal calf serum.
HPLC-purified GnRH, GnRH analogs, and their biotinylated derivatives were dissolved in a small volume of 0.1% trifluoroacetic acid and diluted to their indicated final concentrations in culture medium in which the fetal calf serum was replaced by bovine serum albumin (100 mg/liter). All peptide concentrations were determined by amino acid analysis. Preincubation medium was removed from 48-h-dispersed cells and replaced with 1 ml of incubation media that contained the peptide derivative to be tested (three wells/peptide). After incubation for 4 h at 37 "C, media were removed, and centrifuged at 10,000 X g at 4 "C, and supernatant fluids were collected and stored at -20 "C prior to radioimmunoassay.
Measurements of LH and FSH in the cell media were performed using specific radioimmunoassays as previously described (Parkening et al., 1982;Westlund et al., 1984), employing kits and standard hormones supplied by the National Institutes of Health (see "Materials"). The sensitivities of these assays were approximately 0.3 ng for LH and 15 ng for FSH. Intra-assay variation was less than 10% for LH and less than 7% for FSH. Neither the synthetic peptides nor their biotinylated derivatives, at the concentrations used in these studies, interfered with the RIA. Statistically significant ( p < 0.05) differences in hormone release were determined by a two-way analysis of variance for repeated measurements, followed by a Newman-Keul's test (Bruning and Kintz, 1977).

The primary structures of the peptides used in these studies
are shown in Table I.
Ethylamide-containing carboxyl terminus. Sequence taken from Bernardi et al. (1966). e Sequence taken from Reichelt et al. (1987). GnRH with sulfo-NHS-tAhx-biotin at a 4:l molar ratio of reagent:peptide for 10 min at 25 "C resulted in six major fractions, as evidenced by HPLC (Fig. lA). Fraction A1 appeared at the same position as unmodified [D-L~s'IG~RH, whereas fractions A2-A6 appeared to be chemically distinguishable biotinylated derivatives. The multiple peptide derivatives illustrated in Fig. L4 were consistently observed during the chromatography of many different biotinylation reaction mixtures carried out under various reaction conditions. Amino acid compositional analysis indicated that each fraction contained a peptide identical in composition with the original [D-Lys'IGnRH. When [D-Lys']GnRH was reacted with sulfo-NHS-tAhx-biotin at a 20:1 molar ratio of reagent:peptide for 60 min at 25 "C, no unbiotinylated peptide (fraction A l ) remained, and the biotinylated derivatives represented by fractions A2, A3, and A5 were absent (Fig. 1B). Fig. 1C shows the rechromatography of pooled aliquots of peptide fraction A6 that was obtained by HPLC after treatment of the modified peptide with 0.1% trifluoroacetic acid for 2 h at 37 "C. The acid treatment had no effect on the HPLC elution time of fraction A6. Identical acid treatment of fractions A 2 4 5 also had no effect on their HPLC elution times (results not shown).
To establish whether or not the biotin moieties were attached via ester linkages to hydroxyamino acids, the HPLCpurified material in the major fractions was reacted with hydroxylamine at alkaline pH and subsequently rechromatographed. These experiments demonstrated that the biotinyl- ated derivatives in all the fractions except for A1 and A3 ( Fig.  1) were hydroxylamine-labile. A representative example of the experiments with hydroxylamine is illustrated in Fig. 2. Fig. 2A shows the repeat HPLC elution of pooled material from fraction A6 (Fig. L4) incubated in buffer without hydroxylamine. Reaction with hydroxylamine ( Fig. 2B) resulted in a shift in the elution time of the fraction A6 peptide to a position corresponding to fraction A3. HPLC analyses of the derivatives in the other fractions both before and after hydroxylamine treatment were obtained in a similar fashion. The results of these analyses are summarized in Table 11.
Mass Spectrometric Analysis-HPLC-purified peptide  Table I. The spectra of derivatives A2 and A3 (Fig. 3, B and C) indicated major ions at m/z 1592.9 and were in agreement with the calculated value for a monobiotinylated [D-Lys'] GnRH peptide, whereas the spectrum of fraction A4 (Fig. 3 0    ence of a major ion at m/z 2272.1. Representative CID/MS/MS analyses for the identification of the modified amino acid residues in specific biotinylated species are shown in Fig. 4. The CID spectrum shown in Fig.   4 A , obtained from the protonated peptide molecule at m/z 1253.8 (Fig 3A) is that of unmodified [D-L~s'IG~RH from HPLC fraction Al. As expected, this spectrum exhibited the characteristics of a peptide containing one or more basic amino acids at or near the carboxyl terminus. Due to the presence of arginine as the third amino acid from the carboxyl terminus, the spectrum was dominated by the wn and vn ions. The fragment nomenclature used here has been described previously (Biemann, 1988).
The biotinylated derivative from HPLC fraction A2 (FAB spectrum shown in Fig. 3B) yielded a CID spectrum that was only compatible with reaction at serine (Fig. 4B). The region below m/z 400 was similar to that shown in Fig. 4.4. While the w5 ion was the same because it had lost the side chain of lysine, the occurrence of other lysine-containing carboxylterminal ions, 26, &, and xg, demonstrated that this residue was not modified. This conclusion was also supported by the abundant w7 ion. However, the most compelling reason for assigning this isomer the structure of a peptide biotinylated at serine was the abundant ion at mlz 1235.5, which indicated the conversion of serine to dehydroalanine as a result of the elimination of the biotin moiety and the loss of the serine hydroxyl. The CID spectrum illustrated in Fig. 4C is that of the isomeric protonated peptide at mlz 1592.9 shown in Fig.  3C (from HPLC fraction A3). The relatively wide region devoid of significant peaks indicated that the long chain biotin moiety was attached somewhere in the center of the peptide, consistent with modification at the D -L~' residue. Since the formation of w, ions involved elimination of part of the lysine side chain, w5 had lost the biotin moiety and, therefore, was of the same mass as the underivatized peptide. The mlz values of the peaks designated y5, G , x69 and w7, required that all of these still contained the biotin substituent, indicating biotinylation at D-LYs'. The two CID spectra shown in Fig. 4, B and C, therefore, demonstrate that the former represents that of a peptide in which acylation of the serine has taken place, whereas the latter is that of a peptide containing biotinylated lysine.
Biotinylation of Des-Gly'o-[D-Trp'lGnRH and GnRH- Fig.  5A illustrates the results obtained from the reversed-phase HPLC of the peptide derivatives from a biotinylation reaction mixture containing des-Gly"-[~-Trp']GnRH and sulfo-NHS-tAhx-biotin (1O:l reagenkpeptide molar ratio). Fraction B1 appeared at the same retention time as unmodified synthetic des-Gly'O-[o-Trp']GnRH. Amino acid analysis indicated that each fraction contained a peptide identical in composition with synthetic de~-Gly'~-[~-Trp']GnRH. The HPLC elution times of fractions B2, B3, and B4 were shifted to that of fraction B1 after reaction with hydroxylamine (Table 11).
The biotinylated products generated after reaction of GnRH with sulfo-NHS-tAhx-biotin, as evidenced by HPLC, are shown in Fig. 5B. Amino acid analysis confirmed that the compositions of fractions C1, C2, C3, and C4 were identical to that of GnRH. After reaction with hydroxylamine, the HPLC elution times of fractions C2, C3, and C4 were shifted to that of fraction C1, corresponding to unreacted peptide (Table 11).
HPLC-purified peptide fractions from the biotinylations of both GnRH and de~-Gly'~-[~-Trp']GnRH were analyzed by FAB/MS and CID tandem MS, as described above. As with [D-L~s'IG~RH, these analyses, coupled with the results of hydroxylamine treatment, established the number of biotin moieties present in each derivative and identified the specific residues modified. Table I11 summarizes the results of these mass spectrometric analyses for GnRH, both GnRH analogs, and each of their major HPLC-purified, biotinylated derivatives.
Biotinylation of Peptides Unrelated to GnRH-To compare the reactivity of the hydroxyamino acids in GnRH and its analogs with the reactivity of similar residues in unrelated peptides, we performed a series of biotinylation reactions with the peptides eledoisin and physalaemin. These peptides, in addition to being structurally similar to GnRH, allowed us to compare the relative occurrence of 0-acylation and N-acylation within each peptide. Eledoisin and physalaemin are both 11 residues in length, have pyroglutamyl residues as amino termini, and have amidated C-terminal residues. Both peptides contain a single lysine, with eledoisin having a single serine at position 3 and no tyrosine, whereas physalaemin has one tyrosine at position 8 and no serine. The reaction of synthetic eledoisin with sulfo-NHS-tAhx-biotin at a 4:l reagenkpeptide molar ratio for 10 min at 25 "C resulted in one major biotinylated species, as illustrated in the HPLC profile in Fig. 6A. Amino acid compositional analysis confirmed that fraction D l was unmodified eledoisin, whereas fraction D2 was a monobiotinylated derivative. Reaction of fraction D2 with hydroxylamine had no effect on HPLC retention time, indicating that the biotin moiety was attached to the Lys residue at position 4. There was no indication of significant 0-acylation of the seryl residue. Fig. 6B shows the HPLC elution of synthetic physalaemin after reaction with sulfo-NHS-cAhx-biotin at a 4:l reagent:peptide molar ratio for 10 min at 25 "C. Amino acid compositional analysis indicated that fraction E l was unmodified physalaemin, whereas fractions E2 and E3 were monobiotinylated derivatives. Fraction E4 was a dibiotinylated species. Reaction of fraction E2 with hydroxylamine had no effect on its HPLC elution time, indicating that it was monobiotinylated on Lys'. By similar reasoning, we deduced that fraction E3 was monobiotinylated on Tyr' and fraction E4 was dibiotinylated on Lys' and Tyr'.
The time course biotinylation of specific residues in [D-L~s'IG~RH, eledoisin, and physalaemin are shown in Fig.  7. The results illustrated in Fig. 7 were generated by reacting equivalent amounts of peptide with sulfo-NHS-eAhx-biotin under identical reaction conditions. In each peptide reaction, approximately 60% of the lysyl residues had reacted by 10 min. Strikingly, the reactivity of Set4 in [D-L~s'IG~RH was significantly high and somewhat paralleled that of the lysyl residue. Over 45% of the serine was biotinylated within 10 min (75% relative to lysine). The reactivity of the [D-Lys'] GnRH serine was in marked contrast to the results obtained from biotinylation of eledoisin, in which case less than 2% of the seryl residue was acylated in the first 10 min of the reaction. By comparison, the reactivity of the tyrosyl residues in [D-L~s'IG~RH and physalaemin was relatively low and roughly equivalent. A molar comparison of serine acylation after reaction of [D-L~s'IG~RH, eledoisin, and EMIP with sulfo-NHS-cAhxbiotin at a relatively high 20:l reagenkpeptide molar ratio is shown in Fig. 8. Under these conditions, all of Ser4 in the GnRH analog was biotinylated, whereas only about 6% of Ser' in eledoisin was modified. Virtually no seryl acylation was detected in the pentapeptide EMIP, which contained 1 seryl residue and no available amino groups.

O-Acylation of Serine in GnRH and Related Peptides
Biotinylation under Denaturing Conditions-Biotinylation of GnRH with sulfo-NHS-tAhx-biotin at 25 "C in the presence of 6 M guanidine HCl, after heating the peptide in the denaturing solution at 100 "C for 5 min, reduced O-biotinylation substantially (Fig. 9 D ) when compared with the control (Fig. 9A). Incubation of GnRH in 6 M guanidine HCI for 1 h at 25 or 50 "C significantly reduced O-biotinylation, but appreciable amounts of monobiotinyl serine and monobiotinyl tyrosine were still evident (Fig. 9, B and C) when compared with the biotinylation reaction with prior heating at 100 "C ( Fig. 9D). Incubation of [biotinyl-Ser4]GnRH in biotinylating reaction buffer containing 6 M guanidine HCl for 40 min prior to HPLC analysis showed no evidence of deacylation (results not shown).
Bioactivity Studies- Fig. 10 illustrates the effects of native GnRH, [D-L~s'IG~RH, and their monobiotinylated derivatives on the release of LH from dispersed rat anterior pituitary cells. GnRH caused a significant release of LH at concentrations greater than 1 nM, whereas [biotinyl-Ser4]GnRH and [biotinyl-Tyr6]GnRH (Fig. 523, derivatives C2 and C3, respectively) were effective only at 100 nM, the highest concentration tested (Fig. 1OA). However, even at this concentration, the monobiotinylated derivatives were only 40% as active as unmodified GnRH. By comparison, a notable difference in bioactivity of the monobiotinylated derivatives of [D-Lys'] GnRH was observed (Fig. 1023) in repeated bioassays. In general, the release of FSH from the pituitary cultures after stimulation with both peptides and their biotinylated derivatives closely paralleled the results from the LH radioimmunoassays (results not shown).

DISCUSSION
Reaction of [D-L~s'IG~RH with a sulfonated N-hydroxysuccinimide ester of biotin generated multiple forms of biotinylated peptide (Fig. 1). Since [D-L~s'IG~RH contained only one available amino group, it was evident that other functional groups on the peptide had reacted with the sulfo-NHS-tAhx-biotin reagent. Because we observed that relatively large amounts of these unexpected derivatives appeared throughout the course of the reaction, we undertook an extensive structural characterization of all major reaction products. This characterization involved HPLC purification, amino acid compositional analysis, hydroxylamine reaction, and analysis by both FAB/MS and CID tandem MS. The results of these analyses clearly demonstrated that, in addition to the expected modification at D-LYS', biotin moieties were also attached to [D-Lys'IGnRH via ester linkages on Ser4 and Tyr'. Evidence of similar 0-acylation was obtained from separate experiments with a second GnRH analog, de~-Gly'~-[~-Trp'] GnRH (Fig. 5A), as well as with synthetic GnRH (Fig. 5 B ) , neither of which contain reactive amino groups. In addition,

2,500
m/z 0-acylated derivatives of GnRH and analogs were obtained after reaction with the nonsulfonated NHS-tAhx-biotin reagent (results not shown). The most striking finding from the analyses of the biotinylation reaction products was the evidence of a highly reactive seryl residue (Ser4) in GnRH and both analogs. In the case of [D-L~s'IG~RH, a significant amount of [biotinyl-Ser4]-[~-Lys']GnRH was generated in the first 10 min of reactions carried out at a 4:l reagent:peptide molar ratio (Fig.  1). In this GnRH analog, the serine reactivity closely paralleled the reactivity of the lysyl residue (Fig. 7). In addition, after 1 h at a reagent:peptide molar ratio of 20:1, all of the serine in [D-L~s'IG~RH could be modified (Figs. 1B and 8).
By comparison, the reactivity of Tyr' to biotinylation was considerably less than that of Ser4 in both [D-L~s'IG~RH and de~-Gly'~-[~-Trp']GnRH (Figs. 1 and 5A), although tyrosine modification increased at higher molar ratios of reagenkpeptide. For example, at a reagent:peptide molar ratio of 101, the yield of [biotinyl-Tyr'IGnRH approached that of [biotinyl-Ser4]GnRH (Fig. 5B). We found no evidence of monobiotinylated [D-L~s'IG~RH modified on Tyr'. This derivative may have been one of the minor HPLC components not evaluated. It is important to note that none of the major 0-acylated peptide derivatives were short lived reaction intermediates. In fact, these biotinylated peptides were stable 5066 0-Acylation of Serine in GnRH and Related Peptides GnRH; B, synthetic GnRH. Peptides were reacted at a 101 reagenkpeptide molar ratio for 20 min at 25 "C. Fractions B1 and C1 represent unmodified peptides. Fractions B2 and C2 were biotinylated at Ser', B3 and C3 at TyP, and B4 and C4 at both Ser' and TyP.
during dry storage at -20 "C for several months, as evidenced by rechromatography by HPLC. Moreover, as a further test of stability, biotinylated [D-L~s'IG~RH derivatives were treated with 0.1% trifluoroacetic acid for 2 h at 37 "C and subsequently rechromatographed. As typified in Fig. 1C for m/z the tribiotinylated species (fraction A6), no significant deacylation of any of the [D-L~s'IG~RH derivatives was observed. This acid stability contrasts with the relative instability of acylated tyrosyl residues, as previously reported in the case of pepsinogen succinylation (Gounaris and Perlmann, 1967).
Although the occurrence of 0-acylation has been reported under reaction conditions typically employed for N-acylation for a number of reagents, this is the first report of such reactivity for the N-hydroxysuccinimide esters of biotin. In fact, many recent in-depth reports on the structural characterization of biotinylated derivatives of peptide hormones, such as that of @endorphin (Hochhaus et al., 1988), parathyroid hormone (Newman et al., 1989), interleukin-la (Yem et al., 1989), porcine relaxin (Bullesbach and Schwabe, 1990), and egglaying hormone , indicated no evidence of 0-acylation even when a large molar excess of reagent was employed (e.g. 101 reagenkprotein for interleukin-1B). In the case of the 36-residue egg-laying hormone of Aplysiu, which has 1 seryl, 2 threonyl, and 1 tyrosyl residue, 0-acylation using NHS-tAhx-biotin was specifically investigated . Moreover, some reports categorically state that the NHS-biotin reagents are highly specific for amino groups (Yem et al., 1989). The fact that the biotinylated residues in many of these modified peptides have been identified and no evidence of 0-biotinylation was reported emphasizes the likelihood that the serine residues in GnRH FIG. 6. Reversed-phase HPLC purification of eledoisin, physalaemin, and reaction products after reaction with sulfo-NHS-cAhx-biotin, A, eledoisin; B, physalaemin. Peptides were reacted at a 4:l reagentpeptide molar ratio for 10 min at 25 "C. Fraction D l is unmodified eledoisin; fraction D2 is its monobiotinylated derivative. Fraction E l is unmodified physalaemin; fractions E2 and E3 are monobiotinylated derivatives. Fraction E4 was dibiotinylated. and its analogs are uniquely reactive toward activated esters of biotin. Further evidence in this regard was obtained when we compared the time course of biotinylation of serine and lysine in eledoisin and [D-L~s'IG~RH (Fig. 7). Under identical reaction conditions that demonstrated comparable lysyl reactivity in both peptides, over 45% of Ser' in [D-L~s'IG~RH was acylated in the first 10 min, whereas less than 2% of the eledoisin serine was biotinylated. Moreover, when equivalent reaction mixtures of [D-L~s'IG~RH, eledoisin, and EMIP were biotinylated at a high reagenkpeptide molar ratio of 20:1, all of Ser' in [D-L~s'IG~RH was acylated, whereas only -6% of the eledoisin serine and 4 % of the EMIP serine was modified (Fig. 8). The result from the biotinylation of EMIP was especially notable, since this pentapeptide contained the active site sequence (Asp-Ser-Gly) typically found in serine proteases (Blow et al., 1969).

0-Acylation of Serine in GnRH and Related Peptides
The unusually high reactivity of the Ser' residue in GnRH  Peptides were reacted at a 201 reagent:peptide molar ratio for 60 min at 25 "C. Bars indicate total serine acylation for each peptide as determined from HPLC-purified derivatives. and its analogs prompted us to investigate 0-acylation under denaturing conditions to inquire whether or not the observed reactivity was associated with conformation. Carrying out the biotinylation reaction in the presence of 6 M guanidine HCl after heating the peptide at 100 "C adversely affected 0biotinylation (Fig. 9D), indicating that the acylation of serine and tyrosine was dependent on peptide conformation. It was noteworthy that denaturation of GnRH in 6 M guanidine HC1 at lower temperatures (25 and 50 "C) was incomplete after 1 h (Fig. 9, B and C), suggesting that the native conformation of GnRH surprisingly was relatively stable under these conditions. We found no indication that the guanidine treatment itself hydrolyzed ester linkages between biotin and GnRH, and the observed decrease in biotinylation with increasing temperature gave added evidence that the reduction in 0biotinylation was not due to deacylation.
A likely explanation for the observed superreactivity of the seryl residue is increased nucleophilicity of the hydroxyl oxy- gen due to the hydrogen bonding of the hydroxyl hydrogen to some as yet undetermined acceptor. The occurrence of hydrogen bonding may derive from secondary structural elements. Both theoretical and experimental evidence has been presented supporting a folded conformation for GnRH due to a modified Type I1 @-turn centered at Tyr6-Gly6-Leu7-Ar$ (Momany, 1976a(Momany, , 1976bDonzel et al., 1977;Freidinger et dl., 1980). In the folded conformation, the Ser4 side chain, and perhaps Tyr', may be participating in hydrogen bonding. This conclusion is supported by the reduction of 0-acylation observed when biotinylation is carried out in the presence of guanidine HC1, as discussed above. The increased nucleophilicity of serine as a result of hydrogen bonding has been well documented in the case of Ser19' of the chymotrypsinogen family of serine proteases (Blow et al., 1969). The significance of the 0-acylation of Tyr6 of GnRH and its analogs remains somewhat of an open question. Although reactive (Fig. l), especially at higher reagent:peptide ratios, Tyr6 was clearly less reactive to biotinylation than Ser4 (Fig.  7). Since the side chain phenolic oxygen of tyrosine is typically more nucleophilic than the hydroxyl oxygen of serine, one would expect Tyr' to be more readily acylated than Ser4 in GnRH under normal conditions. Thus, the observed superreactivity of Ser' is further underscored. The time course of 0-acylation of Tyr' in GnRH and that of the single tyrosyl residue in physalaemin were comparable and may represent the general baseline reactivity to be expected with NHSbiotin reagents. Further studies of other tyrosyl containing peptides will be required to establish this point more definitively. Preliminary pseudo first-order plots of log percent residual unmodified residue versus time indicated that the 0biotinylation of Tyr' was about 20-fold slower than the Nbiotinylation of Lys' in [D-Lys'IGnRH. An in-depth study of the chemical kinetics of the reaction of GnRH and analogs to NHS-biotin esters, including comparison to model compounds, is currently being carried out and will be reported elsewhere. Finally, it is worthwhile mentioning that in a separate experiment, using similar reaction conditions employed with [D-L~s'IG~RH, we were unable to biotinylate a synthetic fibronectin fragment (Gly-Arg-Gly-Asp-Thr-Pro) on its threonyl residue, as evidenced by nonreactivity with hydroxylamine.
All of the monobiotinylated 0-acylated derivatives of [D-L~s'IG~RH and GnRH were assayed for their ability to release LH from cultured pituitary cells. In these assays, 0acylation of both analogs shifted the dose-response curve to the right by approximately 2 orders of magnitude (Fig. 10). Since the Ser' residue of GnRH is one of 5 highly conserved residues in the GnRH family of peptides (Sherwood, 19871, modifications at this position appear to be critical to bioactivity. An [Ala'IGnRH analog was reported to have only 5% of the activity of native GnRH (Geiger et aL, 1972;Yanaihara et al., 1973). Similarly, modification of the Tyr' residue also results in marked decreases in bioactivity (Yanaihara et al., 1973). These reported potency estimates for Ser4 and Tyr' replacement analogs are consistent with our bioassay results. In addition, similar results of reduced potency were obtained from in vivo studies using 0-acetylated derivatives of GnRH .
Biotinylation of the Lys' residue in [D-L~s'IG~RH had no detrimental effect on the ability of the peptide to stimulate LH secretion, and in fact, increased LH secretion when compared with the unmodified peptide (Fig. 1OE). A similar result was reported by Tibolt and Childs (19851, using a different biotinylating reagent that did not contain an extended spacer arm. The unusually high reactivity of Ser' in GnRH, coupled with the significant loss of bioactivity associated with Ser4 biotinylation, or with residue substitution, invites further inquiry concerning the potential role of this residue in the mechanism of action of GnRH. For example, is the observed reactivity of Ser4 primarily the result of its participation in 0-Acylation of Serine in GnRH and Related Peptides 5069 maintenance of conformation or could this residue have a more direct role in GnRH action? Clearly, the results obtained herein concerning the intrinsic reactivity of Ser4 in GnRH raise important new questions concerning the potential role of this residue in structure/function considerations of GnRH. Knowledge of the unique reactivity of Ser4 will be of value in interpreting the three-dimensional structure of GnRH when it becomes available. For example, the determination of the unique reactivity of Ser'% of the chymotrypsinogen family of serine proteases to diisopropyl fluorophosphate was highly important in developing the hypothesis that the observed charge relay in chymotrypsin that centered at Serlg5 was part of the catalytic mechanism of the active site (Balls and Jansen, 1952;Schaffer et al., 1953;Blow et al., 1969). Although synthetic residue substitution experiments and similarly sitedirected mutagenesis give us valuable information about structure and/or function, they do not address intrinsic reactivity of residue side chains.
In summary, chemical modification of GnRH and related peptides using sulfo-NHS-cAhx-biotin yielded a number of biotinylated peptides with various combinations of modification on lysyl, seryl, and tyrosyl residues. A time course of biotinylation was notable in demonstrating that the seryl residue at position 4 was especially reactive. Under identical reaction conditions, seryl residues in selected, structurally similar peptides unrelated to GnRH demonstrated little, if any, reactivity. Peptide denaturation experiments gave strong evidence that the unique reactivity of Ser4 is in all likelihood related to conformational parameters. Biotinylation of [D-Lys'IGnRH at D-LYS' resulted in a derivative with increased in vitro activity, whereas acylation of the single seryl residue in GnRH and [D-Lys'IGnRH, and the single tyrosyl residue in GnRH, greatly reduced bioactivity. The procedures we have described for isolating and chemically characterizing peptide biotinylation reaction products, including 0-acylated derivatives, have general applicability to other peptides and can be carried out on relatively small amounts of material. In addition, the occurrence of 0-acylation during biotinylation using NHS-biotin reagents could explain the unexpected loss of biological activity in those cases where N-acylated derivatives were expected to remain bioactive. These results emphasize the potential value of employing NHS-biotin esters as general chemical modification reagents for seryl and tyrosyl hydroxyl groups. There is a reasonable likelihood, as evidenced by results described herein, that these reagents have a fortuitous reactivity that is capable of discriminating between especially reactive side chain hydroxyl groups.