ANALYTICAL BIOCHEMISTRY ARTICLE NO.
264, 41– 46 (1998)
AB982826
Muramyl Peptide Probes Derived from Tracheal Cytotoxin of Bordetella pertussis Tod A. Flak1 and William E. Goldman2 Department of Molecular Microbiology, Washington University School of Medicine, 660 South Euclid Avenue, St. Louis, Missouri 63110
Received April 13, 1998
A novel semisynthetic scheme was developed to couple amine-reactive labeling reagents to the muramyl peptide tracheal cytotoxin (TCT) without affecting a critical amine group. Tracheal cytotoxin, N-acetylglucosaminyl-1,6-anhydro-N-acetylmuramylAla-g-Glu-A2pmAla (A2pm, diaminopimelic acid), is released by Bordetella pertussis, the etiologic agent of whooping cough. This glycopeptide reproduces the specific ciliated cell damage observed in the respiratory tract during B. pertussis infection. To examine binding of TCT to target respiratory cells, we have produced labeled TCT analogs. Structure–function studies have shown that the primary amine of the A2pm side chain is essential for TCT toxicity in respiratory tissue. The methodology described here allows coupling of amine-reactive reagents to TCT without affecting this essential amine. The terminal N-acetylglucosamine ring is opened by oxidation with periodic acid, a dihydrazide linker is coupled to the oxidized ring, and pH control is used to selectively derivatize the free hydrazide with an N-hydroxysuccinimide ester, while the A2pm side-chain amine remains free. Using this method, we have coupled the Bolton– Hunter reagent to TCT, producing a biologically active 125I-labeled TCT analog. © 1998 Academic Press
Tracheal cytotoxin (TCT)3 is a 921-Da disaccharide tetrapeptide produced by Bordetella pertussis. This 1 Present address: Acacia Biosciences, 4136 Lakeside Dr., Richmond, CA 94806. E-mail:
[email protected]. 2 To whom correspondence should be addressed. goldman@borcim. wustl.edu. 3 Abbreviations used: TCT, tracheal cytotoxin; MDP, muramyl dipeptide; A2pm, diaminopimelate; TFA, trifluoroacetic acid; FAB-MS, fast atom bombardment mass spectrometry; rt, retention time; SHPP, N-succinimidyl-3-(4-hydroxyphenyl)propionate; HTE, hamster trachea epithelial; GlcNAc, N-acetylglucosamine; oxTCT, oxidized TCT; dgTCT, deglycosylated TCT; AADH-TCT, adipic acid dihydrazide-TCT.
0003-2697/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
glycopeptide causes respiratory epithelial cytopathology equivalent to that observed in the pertussis syndrome (1). The structure of TCT is N-acetylglucosaminyl-1,6-anhydro-N-acetylmuramyl-(L)-alanyl-g(D)-glutamyl-meso-diaminopimelyl-(D)-alanine (2) and is shown in Fig. 1A. TCT is therefore a member of the muramyl peptide family, small fragments of bacterial peptidoglycan implicated in a wide range of biological activities, including somnogenicity, arthritogenicity, pyrogenicity, and adjuvanticity (for a review see Ref. 3). Most work involving muramyl peptides is centered upon muramyl dipeptide (MDP), originally defined as a minimally active analog for adjuvanticity (4). There are many derivatives of MDP (reviewed in Refs. 3 and 5), including some used for receptor studies (6, 7). These MDP analogs generally take advantage of the fact that MDP can tolerate major additions to the Cterminal end of the peptide without affecting its adjuvant activity (8). Unfortunately, structure–activity studies of TCT have demonstrated that the functional groups available for derivatization using typical labeling methods (e.g., the amine or carboxylates of the peptide portion) are necessary for biological activity of the toxin in respiratory epithelial cells (9). However, in contrast to other muramyl peptide activities, our analysis of structural requirements for TCT activity demonstrated that the disaccharide portion of the molecule is dispensable for cytotoxic activity in respiratory tissue (10). Since there are no functional groups available for derivatization in this region, we used a semisynthetic approach to augment the native structure with a versatile hydrazide linker moiety. This particular functional group has the advantage of permitting derivatization by any of a large variety of amine-reactive coupling agents (i.e., succinimidyl esters), while allowing the protection of the primary amine on the diaminopimelate (A2pm) side chain by selective pH control of the site of reaction. Using this hydrazide intermediate, we have produced a 41
42
FLAK AND GOLDMAN
FIG. 1. Progression of TCT derivatives. (A) The full structure of TCT is shown. For the subsequent molecules, the region not involved in the reaction is indicated by abbreviated text. (B) Oxidized TCT (oxTCT). (C) TCT with adipic acid dihydrazide linker arm (AADH-TCT). (D) Bolton–Hunter derivative of TCT (HPP-TCT).
broad range of labeled TCT analogs that retain their biological activity. MATERIALS AND METHODS
Buffer A. A gradient was developed beginning 5 min after sample injection by linearly increasing the fraction of Buffer B by 0.5%/min. In-line UV detection was employed at 214 nm.
Materials Trifluoracetic acid (TFA), Iodobeads, and all N-hydroxysuccinimide esters were from Pierce Chemical (Rockford, IL). HPLC-grade acetonitrile and methanol were from EM Science (Gibbstown, NJ). Na125I was from Amersham (Arlington Heights, IL), and [methyl3 H]thymidine was from ICN (Costa Mesa, CA). All other chemicals were from Sigma Chemical Co. (St. Louis, MO).
Mass Spectrometry Fast atom bombardment mass spectrometry (FABMS) was performed (by E. Kolodziej, Monsanto Corp.) essentially as previously described (9). For most samples the deposition matrix included glycerol, thioglycerol, HCl, m-nitrobenzyl alcohol, and heptafluorobutyric acid. The mass of the observed molecular ion species is reported as “M 1 H.”
HPLC HPLC was performed on an RP-300 C8 column (Applied Biosystems), 4.6 mm 3 10 cm, with a coupled 3-cm guard column cartridge. Buffer A was 0.1% TFA; Buffer B was 60 parts acetonitrile to 40 parts 0.25% TFA. The flow rate was 1 ml/min, beginning with 100%
Amino Acid Composition For the initial quantitation of products, amino acid composition analysis was routinely performed as described (10), from which molar extinction coefficients were determined (based upon HPLC peak areas). All
MURAMYL PEPTIDE PROBES FROM TRACHEAL CYTOTOXIN
extinction coefficients given are at 214 nm in HPLC buffer. Oxidation of TCT Native TCT was prepared as previously described (1). For oxidation trials, 20 –50 nmol of TCT was treated with periodic acid for various times. At the end of the reaction, TFA was added to a concentration of 1%. A C18 Sep-pak (Waters) solid-phase extraction cartridge was prepared with 10 ml MeOH and 30 ml 0.1% TFA. The sample was applied, rinsed with 20 ml 0.1% TFA, eluted with 2 ml MeOH, and dried in a SpeedVac (Savant). The eluate was analyzed by HPLC, and the collected peaks were subjected to FAB-MS for identification. The identified peaks (with HPLC retention times (rt), observed FAB-MS molecular ion masses, and extinction coefficients) were deglycosylated TCT (rt 5 22.3 min, M 1 H 5 719.5); unreacted native TCT (rt 5 25.4 min, M 1 H 5 922.5, e214nm 5 4.0 3 103M21cm21); oxidized TCT (rt 5 25.7 min, M 1 H 5 920.3). In large-scale production for use in subsequent steps, 500 nmol TCT was oxidized in an unbuffered 400 mM aqueous solution of periodic acid for 2.5 h at room temperature in the dark. The reaction products were collected by solid-phase extraction on a Sep-pak cartridge as described above, and the eluate was dried by evaporation in a SpeedVac. Typical yield from this procedure was 30 –50% of the starting material. Coupling of Dihydrazide Linker To form the Schiff’s base intermediate, a 50-fold excess of adipic acid dihydrazide was added to the dry oxidized TCT in 100 ml of an aqueous solution buffered with 200 mM potassium oxalate, pH 4.0. The mixture was vortexed and allowed to stand at room temperature for 12 h. The reduction of the hydrazone bond was accomplished with 5 ml of sodium cyanoborohydride, freshly prepared in water at 20 mg/ml. After 8 h at room temperature, the reaction was brought up to 500 ml and to a final concentration of 1% TFA for HPLC purification. The product, adipic acid dihydrazide TCT (AADH-TCT), eluted as a broad trailing peak (rt 5 29.7 min, e214nm 5 8.6 3 103 M21 cm21), and its identity was confirmed by FAB-MS (M 1 H 5 1063). Typical yield from this procedure was 50% of the starting material. Reaction with N-Hydroxysuccinimide Esters Dried HPLC-purified AADH-TCT was resuspended in 200 ml 100 mM sodium acetate, pH 5.5. Approximately 4 mg N-succinimidyl-3-(4-hydroxyphenyl)propionate (SHPP, Bolton–Hunter reagent) was added. Solubilization of the SHPP reagent was assisted by immersion in an ultrasonic bath for 30 s, although it never completely dissolved. The mixture was held at 37°C, with remixing and ultrasound treatment after
43
about 30 min. After 1 h the undissolved SHPP reagent was pelleted by brief centrifugation, and the supernatant was brought to 1% TFA and purified by HPLC to arrive at HPP-TCT (rt 5 48.7 min, e214nm 5 12.7 3 103 M21 cm21). FAB-MS of the collected peak confirmed the proper derivatization (M 1 H 5 1210.5). Typical yield from this procedure was 50 – 80% of the starting material. Other derivatives were produced in a similar manner using the N-hydroxysuccinimide esters of the desired functional moiety (all from Pierce). Iodination of Bolton–Hunter Derivative HPP-TCT (2-20 nmol) was resuspended in 0.5 ml of 0.5 M sodium phosphate buffer at pH 7.5. Carrier-free Na125I (74 MBq, 2 mCi, ; 1 nmol; Amersham) was added, along with two Iodobeads (Pierce), and the reaction mixed by inversion for 10 min. The liquid was transferred away from the Iodobeads to a tube containing 120 ml of 10% TFA, and this was applied to a C18 Sep-pak which was processed as above except that the 0.1% TFA wash solution also contained 1 M NaI to help displace any unreacted 125I2 being retained. The product was eluted in 2 ml MeOH and dried by evaporation under a nitrogen stream. All manipulations involving free 125I2 were performed in a fume hood. The iodination products were purified by HPLC, and the identities of nonradioactive versions of the products were confirmed by FAB-MS (monoiodo-TCT: rt 5 61.6 min, M 1 H 5 1336.0, e214nm 5 19.3 3 103 M21 cm21; diiodo-TCT: rt 5 70.2 min, M 1 H 5 1461.8, e214nm 5 31.6 3 103 M21 cm21). Biological Assays Biological response to TCT and analogs was assessed by measuring inhibition of DNA synthesis by hamster trachea epithelial (HTE) cells, as previously described (9). Briefly, HTE cells, which are a homogeneous proliferating epithelial cell population derived from hamster trachea (11), were synchronized in their growth cycle by serum deprivation. TCT and bacterial LPS (Escherichia coli strain 026:B6), a required cofactor (12), were added to the cells in a serum-free medium, and after 4 h [methyl-3H]thymidine was supplied along with serum to promote resumption of cell growth. After 26 h DNA synthesis was assessed by scintillation counting of trichloroacetic acid-precipitable material. RESULTS
Oxidation of TCT To create a site within the disaccharide portion of TCT for attachment of a linker arm, we chose to open the N-acetylglucosamine (GlcNAc) ring by oxidation with periodic acid, which selectively acts upon vicinal diols. Purified, native TCT was treated with periodic acid for various amounts of time and the reaction prod-
44
FLAK AND GOLDMAN
zides to periodate-oxidized saccharides has been described (14–16). The reaction of the aldehydes with the hydrazide resulted in two hydrazone bonds (RNH-RNCHOH-R) which completed a seven-member ring. This intermediate could be observed by HPLC, but was unstable to drying. Therefore the hydrazone bonds were reduced with sodium cyanoborohydride (17), to form AADH-TCT (Fig. 1C). The product was purified by HPLC, and its identity was confirmed by FAB-MS. Reaction with N-Hydroxysuccinimide Esters
FIG. 2. HPLC chromatograms of the time course of TCT oxidation. Samples of a TCT oxidation reaction were analyzed at 1-h intervals by HPLC. The chromatograms for each sample are shown in the region of interest (20 –30 min), aligned to a common time scale at the bottom. Native TCT elutes at 25.4 min, oxTCT at 25.7 min, and dgTCT at 22.3 min. The identity of the byproduct peak at 27.2 is not known. Note the shoulder that forms on the trailing side of TCT beginning with the 1-h time point.
ucts were resolved by HPLC. As shown in Fig. 2, over time several new peaks were generated in the region of the chromatogram near TCT. By subjecting the collected peaks to FAB-MS, we found that the oxidized TCT (oxTCT, Fig. 1B) eluted as a trailing shoulder (at ;25.7 min) behind the TCT peak (at ;25.4 min). A minor undesired product at 22.3 min was found by FAB-MS to be TCT that had lost the entire terminal GlcNAc unit (deglycosylated TCT, dgTCT). The peak at 27.2 min was not conclusively identified. Based upon this time course experiment, it appeared that the best yield of oxidized product was obtained after 2.5 h of treatment. Oxidation by periodic acid is also dependent upon temperature and pH (13), and optimization of these variables is reflected in the procedure presented under Materials and Methods. Notably, oxidation by sodium periodate in a pH 7.5 bicarbonate buffer resulted in the production of a large proportion of the undesired deglycosylated TCT (data not shown).
The large array of labeling compounds available as N-hydroxysuccinimide esters are designed for coupling the desired label to a primary or secondary amine. Due to the similar electronic configuration of the terminal nitrogen of the hydrazide, it is also reactive with these esters. To act as a nucleophile the nitrogen must be in its unprotonated state. While the pKa for a primary amine is ;7– 8, the pKa of a hydrazide is ;3 (data not shown and Ref. 18). This difference allows for selective derivatization of the hydrazide group while the primary amine on the side chain of the diaminopimelic acid remains underivatized. As an example of this reaction, SHPP (Bolton– Hunter reagent) was employed to attach a tyrosyl-like residue to the free hydrazide end of AADH-TCT. The product, HPP-TCT (Fig. 1D), was purified by HPLC. FAB-MS demonstrated a molecular ion of the expected mass (M 1 H 5 1210.5). Examination of the ions generated by intramolecular fragmentation revealed a peak at M 1 H 5 719.6, which is indicative of the TCT fragment broken at the GlcNAc-MurNAc glycosidic bond (2). Had the Bolton–Hunter moiety been added to the A2pm side-chain amine, this fragment would not have been present. This demonstrated that the pH of the reaction allowed selective control over the site of derivatization. Amino acid composition analysis also confirmed that the A2pm residue was unmodified (data not shown). Using the AADH-TCT intermediate, we have also produced TCT derivatives coupled to biotin, fluorescein, and azidosalicylic acid (data not shown). These compounds were all derived by reaction of the N-hydroxysuccinimide ester of the desired functionality with AADH-TCT as described, followed by HPLC purification. The azidosalicylic acid derivative was processed further to introduce 125I, as done for HPP-TCT. This compound constitutes a reagent to allow labeling of a putative TCT receptor by the photoactivated formation of a nonspecific covalent bond. Iodination
Coupling of Dihydrazide Linker To introduce a reactive site, we coupled a dihydrazide linker arm to oxTCT. The linkage of amines and hydra-
We performed iodination of HPP-TCT using Iodobeads (Pierce) as described by the manufacturer. Depending upon reaction conditions, it is possible to add one or two
MURAMYL PEPTIDE PROBES FROM TRACHEAL CYTOTOXIN
FIG. 3. Toxicity of iodinated analogs for HTE cells. HTE cells were treated with TCT (h), monoiodinated-TCT (F), or di-iodinated-TCT (■). DNA synthesis was assessed by incorporation of [3H]thymidine, and the inhibition of DNA synthesis was calculated relative to untreated controls. Both of the iodinated analogs retain toxic effect nearly equivalent to the unmodified toxin.
iodine atoms to a tyrosyl ring. Initially, we used a molar excess of nonradioactive sodium iodide to produce a substantial amount of nonradioactive products for use in characterization and biological assays. We found that both the di- and monoiodinated products could be easily resolved from the starting material by HPLC, and the identity of these resolved peaks was confirmed by FABMS. Therefore, using carrier-free Na125I, HPLC purification of the iodinated product resulted in radiolabeled TCT containing no unlabeled starting material and possessing specific activity limited only by the purity of the commercially available 125I (typically 80 to 98% of the potential pure isotopic activity of 80.5 3 1015 Bq/mol). Under the reactions conditions presented under Materials and Methods, we observed that the monoiodinated product was greatly favored over the diiodinated form. Biological Activity of the Iodinated Analog To determine if this labeling procedure preserved the biological activity of TCT, we tested the nonradioactive iodinated TCT products in an assay of DNA synthesis inhibition in hamster trachea epithelial cells. In this assay, TCT, when combined with bacterial LPS, typically results in inhibition of DNA synthesis of greater than 80% relative to untreated controls. As shown in Fig. 3, both the di- and monoiodinated TCT analogs retained biological activity at concentrations very similar to the native molecule. DISCUSSION
For studies of TCT binding to respiratory epithelial cells, we considered numerous methods of generating muramyl peptide probes. The primary amine on the
45
A2pm side chain of TCT provides a natural location for labeling of the molecule by typical amine-reactive reagents. For example, we previously have used the Bolton–Hunter reagent to introduce an iodinated tyrosyl moiety at this location. While producing a ligand with high specific activity, this derivatization resulted in a biologically inactive molecule (data not shown). Subsequent studies revealed that this amine is critical for the biological activity of TCT in respiratory epithelium (9). We also found that modification of the C-terminal carboxylate of TCT affected the toxicity of the molecule, making that site unsuitable for derivatization. An alternative approach to synthetically producing a labeled molecule is metabolic labeling. TCT is a monomeric unit of bacterial peptidoglycan and is most likely produced by the action of a transglycosylase in the continual process of peptidoglycan remodeling (19). Therefore, the metabolic production of labeled TCT necessitates introduction of a radioactive precursor into the large pool of peptidoglycan. A unique substrate utilized by the synthetic pathway for peptidoglycan precursor peptides is A2pm. While A2pm is also the precursor of lysine, lysA mutants defective for this conversion preclude radiolabeled A2pm from being incorporated into protein. We chose an existing lysA mutant of Salmonella typhimurium, which is particularly efficient at utilizing exogenous A2pm (20), and grew it in medium containing 3H-A2pm for several generations. Unlike Bordetella, Salmonella does not release measurable amounts of TCT during growth; but all gram-negative bacteria will degrade their peptidoglycan to small fragments during autolysis. We therefore triggered Salmonella autolysis by brief treatment with trichloroacetic acid and subsequently purified 3H-TCT by solid-phase extraction and HPLC. However, since 3 H-A2pm could not be introduced into all of the peptidoglycan, this approach yielded only a low fraction of labeled molecules (approximately 4%), making this preparation unsuitable for receptor binding studies. Thus, it was necessary to devise the described semisynthetic route for the labeling of the molecule. The introduction of a hydrazide linker arm fulfilled two requirements. First, it provided a site of derivatization in a portion of the molecule that we knew was not required for biological activity of TCT in respiratory epithelium. Modification of the saccharide portion of the molecule kept the added moieties well removed from the critical peptide portion to minimize the chance of steric hindrance. Second, the choice of a hydrazide functionality resolved the issue of how to protect the peptide functional groups critical for biological activity. This method is distinguished from other hydrazide-based coupling techniques (14–16) in the recognition that pH control can play an important role in the site-selective derivatization of the molecule with an amine-reactive reagent. By performing the reaction of the hydrazide with an N-hydroxysuccinimide ester at a low pH, the primary amine is
46
FLAK AND GOLDMAN
protonated and is unreactive. Using this partial synthesis strategy, we were able to produce a biologically active radioiodinated probe, completely free of unlabeled material. Other TCT analogs produced in this fashion contained fluorescein, biotin, or a photoactivatable crosslinker (data not shown). The derivatization method presented here is widely applicable to any periodate-sensitive glycopeptide or glycoprotein that contains critical amine groups which must be protected. This method permits the production of a biologically active labeled compound when total chemical synthesis is not feasible. Given the wide variety of labels available as N-hydroxysuccinimide esters, many possible derivatives can be formed easily from the dihydrazide intermediate. It would be interesting to know whether the described modification would affect other activities associated with muramyl peptides, such as adjuvant activity or macrophage activation. Most of the work on muramyl peptide structure–function relationships has focused on derivatives of simpler monosaccharide-peptides, such as MDP. As reviewed by Adam et al. (5), major additions to the sugar residue of such muramyl peptides do not interfere with adjuvanticity. Therefore, it is likely that our method of TCT derivatization generates a probe that will be useful not just for pertussis research, but for many different types of studies with muramyl peptides. ACKNOWLEDGMENTS Production of 3H-TCT in Salmonella was performed by Brad T. Cookson. We thank Eric Kolodziej, Monsanto Corp., for performing mass spectrometry analysis. This work was supported by Public Health Service Grants AI22243 (to W.E.G.) and AI07172 (to Washington University).
REFERENCES 1. Cookson, B. T., Cho, H.-L., Herwaldt, L. A., and Goldman, W. E. (1989) Infect. Immun. 57, 2223–2229.
2. Cookson, B. T., Tyler, A. N., and Goldman, W. E. (1989) Biochemistry 28, 1744 –1749. 3. Adam, A., and Lederer, E. (1984) Med. Res. Rev. 4, 111–152. 4. Ellouz, F., Adam, A., Ciorbaru, R., and Lederer, E. (1974) Biochem. Biophys. Res. Commun. 59, 1317–1325. 5. Adam, A., Petit, J. F., Lefrancier, P., and Lederer, E. (1981) Mol. Cell Biochem. 41, 27– 47. 6. Sumaroka, M. V., Litvinov, I. S., Khaidukov, S. V., Golovina, T. N., Kamraz, M. V., Komal’eva, R. L., T. M., A., Makarov, E. A., Nesmeyanov, V. A., and Ivanov, V. T. (1991) FEBS Lett. 295, 48 –50. 7. Golovina, T. N., Sumaroka, M. V., Samokhvalova, L. V., Shebzukhov, Y. V., Andronova, T. M., and Nesmeyanov, V. A. (1994) FEBS Lett. 356, 9 –12. 8. Lederer, E. (1980) J. Med. Chem. 23, 819 – 824. 9. Luker, K. E., Tyler, A. N., Marshall, G. R., and Goldman, W. E. (1995) Mol. Microbiol. 16, 733–743. 10. Luker, K. E., Collier, J. L., Kolodziej, E. W., Marshall, G. R., and Goldman, W. E. (1993) Proc. Natl. Acad. Sci. USA 90, 2365– 2369. 11. Goldman, W. E., and Baseman, J. B. (1980) In Vitro 16, 313–319. 12. Heiss, L. N., Moser, S. A., Unanue, E. R., and Goldman, W. E. (1993) Infect. Immun. 61, 3123–3128. 13. O’Shannessy, D. J., and Quarles, R. H. (1987) J. Immunol. Methods 99, 153–161. 14. Robberson, D. L., and Davidson, N. (1972) Biochemistry 11, 533–537. 15. Wilchek, M., and Bayer, E. A. (1987) Methods Enzymol. 138, 429 – 442. 16. Wilchek, M., and Lamed, R. (1974) Methods Enzymol. 34, 475– 479. 17. Borch, R. F., Bernstein, M. D., and Durst, H. D. (1971) J. Am. Chem. Soc. 93, 2897–2904. 18. Inman, J. K., and Dintzis, H. M. (1969) Biochemistry 8, 4074 – 4082. 19. Ho¨ltje, J. V., Mirelman, D., Sharon, N., and Schwarz, U. (1975) J. Bacteriol. 124, 1067–1076. 20. Cooper, S., and Metzger, N. (1986) FEMS Microbiol. Lett. 36, 191–194.
