504
Bioconjugafe Chem. 1994, 5, 504-507
Site-Specific Conjugation of a Temperature-Sensitive Polymer to a Genetically-EngineeredProtein? Ashutosh Chilkoti,t Guohua Chen,g Patrick S. Stayton,*,$and Allan S. Hoffman*J Center for Bioengineering, University of Washington, WD-12/FL-20, Seattle, Washington 98195. Received June 8, 1993@
A genetically-engineered mutant of cytochrome b5, incorporating a unique cysteine residue, was conjugated to maleimide-terminated oligo(N-isopropylacrylamide). The conjugation of the protein by reaction of the cysteine residue, precisely positioned by site-directed mutagenesis techniques, with an activated oligomer containing only one reactive end group in the oligomer chain permits the sitespecific and stoichiometric conjugation of the oligomer with the protein. The protein-oligomer conjugate was shown to exhibit lower critical solution temperature (LCST) behavior, similar to the free oligomer. Furthermore, the LCST behavior of the protein-oligomer conjugate is reversible and allows selective precipitation of the conjugate above its LCST.
Poly(N-isopropylacrylamide)(poly(N1PAA.M))is a temperature-sensitive polymer that exhibits lower critical solution temperature (LCST) behavior in water. Below its LCST of 32 "C, poly(N1PAA.M) is readily soluble in water, while above its LCST the polymer sheds much of its bound water and becomes hydrophobic, which leads to collapse and aggregation of the polymer chains and subsequent precipitation of the polymer (2). This phenomenon is reversible and occurs within a sharp transition range (typically 1-2 "C) (1). The LCST behavior of poly(N1PAA.M) is fully maintained upon conjugation to proteins (2), making such protein-polymer conjugates attractive for affinity separations and immunoassays (38).
Conventional protein-polymer conjugation schemes utilize the reactive amino group of lysine residues to attach proteins to activated soluble polymers. Limitations in controlling the conjugation chemistry frequently arise because the number and location of lysine residues vary greatly in natural proteins and because the location of reactive groups along the polymer chain is random. Thus, the stoichiometry of the protein-polymer conjugate and the attachment site(s) of the polymer to the protein cannot be precisely controlled, factors that may have important implications for protein stability and function. These limitations may be circumvented by appropriate design of both the protein and the activated polymer. Protein engineering techniques permit the design of unique attachment sites on the protein surface for polymer conjugation. The synthesis of a polymer with one activated group per polymer chain, and at one end of the molecule, then allows the stoichiometrically-precise attachment of the polymer to the protein of interest. Furthermore, genetic engineering techniques allow the attachment site to be precisely defined on the protein molecule using site-directed mutagenesis techniques.
* Authors to whom correspondence should be addressed. Tel (P.S.S.): (206) 685-8148. Tel (A.S.H.): (206) 543-9423. Fax: (206) 685-3300. t Supported by Whitaker Foundation (P.S.S) and NSF Grant NO.BCS-9101716 (A.S.H). WD-12. 9 FL-20. Abstract published in Advance ACS Abstracts, August 15, 1994.
*
@
1043-1802/94/2905-0504$04.50/0
We report here the stoichiometrically-precise, sitespecific attachment of oligo(N1PAA.M)by one end group only to a genetically-engineered protein containing a unique thiol functionality at a defined surface site. The protein is conjugated by reaction of the protein sulfhydryl group with a polymer containing a single maleimide end group. The protein selected for these experiments is cytochrome bg, a small (molecular mass: -11 000 Da), bis-imidazole-ligated heme protein (9-1 1). The absence of cysteine residues in the native protein, and the availability of detailed structural information for this protein (12, 13), make it an attractive candidate for the design of site-directed cysteine mutants. Furthermore, the unique electronic and optical properties associated with the heme prosthetic group provide a convenient spectroscopic probe to monitor the thermally induced reversible precipitation of the protein-polymer conjugate (14). A unique thiol group was introduced by the replacement of a threonine residue at amino acid position 8 with a cysteine utilizing site-directed mutagenesis (T8C) (15, 16). The maleimide (MI)-terminated oligo(N1PAA.M) (MI-oligo(N1PM.M))' was conjugated to T8C in solution; the reaction scheme is shown in Figure 1.2 Native cytochrome b5 did not react with MI-oligo(N1PAA.M) The maleimide-terminated oligo(N1PAAM) [MI-oligo(NIPAAM)] was synthesized as follows: first, the aminoterminated oligo(N1PAAM) [A-oligo(NIPAAM)] with a molecular weight of 1900 was synthesized by free radical polymerization of NIPAAM using 2,2'-azobisisobutyronitrile and 2-aminoethanethiol hydrochloride as initiator and chain transfer reagent, respectively, at 60 "C for 4 h. Then, the amino end group was reacted with succinimidyl4-(N-maleimidomethyl)cyclohexane-1-carboxylate, resulting in MI-oligo(N1PAAM). The following procedure was performed to conjugate T8C to MI-oligo(N1PAAM). Typically, 160 pL of a 1 mM protein solution in 50 mM phosphate, 1 mM EDTA buffer, pH 8.0 was reduced with 1mM dithiothreitol (DTT) for 10 min at 4 "C. The mixture was passed over a Sephadex G-25 gel filtration column equilibrated with the same buffer to recover the protein free of DTT. The protein band was collected in a 15 mL centrifuge tube containing a 10-fold molar excess of MI-oligo(N1PAAM). The reaction was allowed to proceed for 4 h at room temperature with gentle shaking to ensure complete mixing of the reactants. The T8C/MI-oligo(NIPAAM) conjugate was separated from unreacted T8C by the following procedure: 10%( v h ) saturated (NH&S04 was added to the mixture to depress the LCST from
0 1994 American Chemical Society
Bioconjugate Chem., Vol. 5, No. 6, 1994 505
Letters
SH
+
($-CH2O!-NHCH2CH2S+&--&kH 0
0
H
H
I H
I C=O
I
NH I
CH H3C/ \CH3
H
H
I
P;JH
CH H 3 d 'CH3
Figure 1. Reaction scheme of stoichiometrically-precise conjugation of a protein to activated poly(N1PAAM) by reaction of the protein sulfhydryl group with a maleimide group, located at one end of the polymer chain.
under the reaction conditions ~ t i l i z e d ,indicating ~ that the maleimide end groups react selectively with thiol groups in the protein. Given the presence of only one thiol group in T8C, these results strongly suggest a 1:l stoichiometry of protein and oligomer in the T8C-oligo(NIPAAM) conjugate. These conclusions are supported by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (MS).4 MALDI-TOF MS of T8C/MIoligo(N1PAAM)conjugate showed a peak a t -11 000 Da [(M H)+],and a minor peak at -13 000 Da. The latter peak was not observed in the spectrum of protein alone (native b5, T8C) or the spectrum of a physical mixture of native b5 and MI-oligo(N1PAAM). The MALDI-TOF MS spectrum of MI-oligo(N1PAAM) displays a range of peaks differing by 113 Da, suggesting that they can be assigned to [nM HI+ ions where M is the monomer
+
+
32 "C to -20 "C, and the reaction mixture was warmed to 37 "C (> LCST) to selectively precipitate the TSC/MI-oligo(NIPAAM) conjugate. The precipitate was then separated by centrifugation (lOOOOg, ambient temperature) to produce a pinkcolored pellet. The pellet was redissolved in buffer, and the precipitation step was repeated twice to ensure complete removal of the unreacted T8C. There is probably some unreacted oligo(N1PAAM) that is precipitated along with the T8C/MIoligo(N1PAA.M)conjugate. One hundred pL of 0.84 mM native cytochrome bg was mixed with a 10-fold molar excess of MI-oligo(N1PAA.M) following the same procedure as for T8C conjugation. No conjugation of MIoligo(N1PAA.M) with native b5 was achieved, based on the absence of a visible peak at 412 nm in the redissolved pellet, indicating that no protein was entrapped in the precipitated polymer. 4MALDI-TOF mass spectra were acquired on a Finnigan MAT LaserMAT LD-TOF instrument. Samples were prepared by depositing -1 pL of sample on the center of a gold-plated metal target. All samples were at 5-10 pm concentration in distilled, deionized water except the T8C/MI-oligo(NIPAAM) conjugate, which was at a concentration of 1pM. The matrix used was 2,5-dihydroxybenzoic acid, dissolved in 60% acetonitrile, 40% 0.1% TFA at a concentration of -5-10 mg/mL. The molar ratio of analyte to matrix was typically 1:lO 000. The samples were allowed to air dry and placed in the spectrometer. Multiple laser shot spectra were acquired after spectral optimization, which involved determining the minimum laser power density required to observe analyte ions from one of four predefined target positions. The instrument was calibrated using sperm whale apomyoglobin as a calibrant.
(minus the maleimide endgroup). The observation of these peaks is consistent with the distribution of oligomer chain lengths expected from free radical polymerization. More importantly, however, the molecular weight distribution of the oligomer is centered a t -1900 Da, which is consistent with the difference in mass of the protonated protein and the unique peak at -13 000 Da (11000 1900 Da) observed only in the spectrum of the conjugate. These results strongly support an o1igomer:protein stoichiometry of 1:l. We note that, given the distribution of oligomer chain lengths, we would expect a series of peaks corresponding to oligomers with different chain lengths conjugated to T8C but the low signal to noise in the mass spectrum of the conjugate precludes their assignment. The LCST of the T8C/MI-oligo(NIPAAM) conjugate in water, determined spectrophotometrically by cloud point mea~urement,~ was 32 "C, which is in agreement with the free oligomer LCST. The reversible precipitation of the TSC/MI-oligo(NIPAAM) conjugate, monitored by the solution absorbance of the heme group at 412 nm, is shown in Figure 2.6 The T8C/MI-oligo(NIPAAM) conjugate was cooled at 4 "C, and the absorbance at 412 nm was measured to determine the concentration of the conjugate (cooling cycle 1,filled bar). TheTSC/MI-oligo(NIPAAM)conjugate was then heated at 37 "C for 5 min, followed by centrifugation at room temperature at lOOOOg for 10 min to precipitate the T8C/MI-oligo(NIPAAM) conjugate. The fraction of unprecipitated T8C/MI-oligo(NIPAAM)conjugate was then determined by measuring the absorbance at 412 nm of the supernatant (heating cycle 1, filled bar). A mixture of T8C and aminoterminated oligo(NIPAAM)[A-oligo(NIPAAM)] was simi-
+
The cloud point of the T8C/MI-oligo(NIPAAm) conjugate was measured spectrophotometrically in deionized-distilled water with a temperature rate increase of 0.4 "C min-l. The temperature a t 90%light transmittance (at 500 nm) was defined as the cloud point. Because (NH4)2S04 depresses the LCST of the conjugate, and the presence of free oligomer (precipitated and collected along with the conjugate by thermally-induced precipitation and centrifugation at 37 "C) can interfere with the measurement of the LCST of the protein-oligomer conjugate, (NH4)2S04and unreacted MI-oligo(N1PAAM) were removed from the TSC/MI-oligo(NIPAAM) conjugate by ultrafiltration through a 10 000 MW cutoff filter (Amicon).
506
Bioconjugafe Chem., Vol. 5, No. 6,1994
-5
TBUMl-Oligo(NIPA4M) Conjugate
2.5 Q
TBC+AOIigo(NIPAAM) MiXture
2.0
v
.- 1.5 e c 5 1.0 e
0
5
0.5
0.0
Cool 1
Heat1
Cool2 Heat2 Cool3 Cycle #
Figure 2. Concentration of the T8C/MI-oligo(NIPAAM) conjugate (filled bars) and the control, a mixture of T8C and A-oligo(N1PAA.M) (hatched bars) plotted versus cycle number. For the cooling cycles 1-3, the sample was assayed below the LCST, and for heating cycles 1 and 2, the sample was assayed above its LCST (see footnotes 6 and 7 for details).
larly assayed7 as a control for the effect of physical coprecipitation of the protein due to the presence of A-oligo(N1PAAM) and (NH4)2S04(17) (hatched bar, cooling cycle 1 and heating cycle 1, respectively). This process was then repeated twice to examine the reversibility of this phenomenon and the short term stability of the protein-oligomer conjugate. Three important observations can be derived from these data. First, the TSC/MI-poly(NIPAAM) conjugate can be precipitated above the LCST of the proteinoligomer conjugate. Below its LCST, the protein-oligomer conjugate is soluble (cooling cycle 1,filled bar). Upon heating at 37 "C the conjugate precipitates, which is shown by the significantly lower concentration of the conjugate in the supernatant (heating cycle 1,filled bar). This process is -90% efficient in selectively fractionating the protein-oligomer conjugate from the aqueous phase, as shown by the difference in the concentration of soluble protein-oligomer conjugate in cycles 2 and 3 (filled bars). Second, this behavior is fully reversible, shown by the repeated cycling of the same sample between the hydrated, water-soluble state and the precipitated, insoluble state. Third, this phenomenon requires covalent conjugation of the oligomer and protein, as shown by the much lower ( < 10%)degree of precipitation of a physical mixture of protein (T8C) and A-oligo(NIPAAM), even with added 6 A 0.95 mL portion of 0.8 ,uM T8C/MI-oligo(NIPAAM) conjugate in 50 mM phosphate, 1 mM EDTA, pH 8.0 buffer with 10%( v h ) (NH&S04 was cooled at 4 "C for 5 min. The saturated (NH&S04 (10% v/v) was added to prevent resolubilization of the conjugate during centrifugation at room temperature. A 0.5 mL portion of this solution was pipetted into a cuvette, and the absorbance at 412 nm was determined spectrophotometrically (cooling cycle 1, filled bar). The T8C/MI-oligo(NIPAAM) conjugate solution was then carefully removed from the cuvette and added back to the original sample. The sample was then warmed to 37 "C for 10 min to precipitate the conjugate, followed by centrifugation at room temperature. A 0.5 mL portion of the supernatant was assayed spectrophotometrically a t room temperature to determine the concentration of the conjugate that remained in solution (heating cycle 1, filled bar). The supernatant was then added back to the original sample and the conjugate was redissolved by cooling to 4 "C, and the process was repeated twice to monitor the reversible precipitation of the protein-oligomer conjugate (cycles 2 and 3). A physical mixture of 1mg of A-oligo(N1PAAM) and 50 pL of 1 mM T8C in 1mL of buffer (50 mM phosphate, 1 mM EDTA, pH 8.0) with 10% (v/v) saturated (NH&S04 was similarly assayed (hatched bars, Figure 1) as in footnote 6.
Chilkoti et al.
(NH4)2S04. Finally, we note that the degree of conjugation required for precipitation of protein-poly(N1PAAM) conjugates has, heretofore, been difficult to control, largely due to the inherent lack of control in the number and location of the active sites on the oligomer chains and their subsequent reaction with native proteins. The results in this study provide clear evidence that a 1:l conjugation of poly(N1PAAM)with an average molecular weight approximately 6-fold smaller than the protein is sufficient to allow efficient ( > 90%) and reversible precipitation of the protein-oligomer conjugate. ACKNOWLEDGMENT
We thank Stephen G. Sligar and Mark McLean (Department of Biochemistry, University of Illinois at Urbana-Champaign, Urbana, IL) for providing native cytochrome bg and the T8C mutant. The oligomer synthesis and end-conjugation to native proteins has beeen developed under support of the National Science Foundation, Grant No. BCS-9101716 (A.S.H). We also thank John Yates (National Science Foundation Center for Molecular Biotechnology, Department of Molecular Biotechnology, University of Washington) for access to the MALDI-TOF MS instrumentation. LITERATURE CITED (1) Heskins, M., and Guillet, J. E. (1968) Solution properties of poly(N-isopropylacrylamide). J . Macromol. Sci. Chem. A2(S), 1441. (2) Chen, J. P., Yang, H. J., Hoffman, A. S. (1990) Polymerprotein conjugates I. Effect of protein conjugation on the cloud point of poly(N4sopropylacrylamide). Biomaterials 11, 625. (3) Hoffman, A. S. (1987) Applications of thermally reversible polymers and hydrogels in therapeutics and diagnostics. J . Controlled Release 6, 297. (4) Monji, N., and Hoffman, A. S. (1987) A novel immunoassay system and bioseparation process based on thermal phase separation polymers Appl. Biochem. Biotechnol. 14, 107. (5) Chen, J. P., and Hoffman, A. S. (1990) Protein-polymer conjugates 11. Affinity precipitation separation of immunogammaglobulin by a poly(N-isopropylacry1amide)-proteinA conjugate Biomaterials 11, 631. (6) Chen, G. H., and Hoffman, A. S. (1993) Preparation and properties of thermoreversible, phase-separating enzymeoligo(isopropylacry1amide)conjugates. Bioconjugate Chem. 4 , 509. (7) Chen, G. H., and Hoffman, A. S. (1994) Synthesis of carboxylated poly(N1PAAM) oligomers and their application to form thermo-reversible polymer-enzyme conjugates. J. Biomat. Sci. Polym. Ed. 5, 371. (8) Takei, Y. G., Aoki, T., Sanui, K., Ogata, N., Okano, T., and Sakurai, Y. (1993) Temperature-responsive bioconjugates. 1. Synthesis of temperature-responsive oligomers with reactive end groups and their coupling to biomolecules. Bioconjugate Chem. 4 , 42. (9) Salemme, F. R. (1976) A hypothetical structure for an intermolecular complex of cytochrome c and b5. J . Mol. Biol. 102, 563. (10) Poulos, T. L., and Mauk, A. G. (1983) Models for the complexes formed between cytochrome bg and the subunits of methemoglobin. J. Biol. Chem. 258, 7369. (11) Wendolski, J. J., Mathew, J. B., Weber, P. C., and Salemme, F. R. (1987) Molecular dynamics of a cytochrome c-cytochrome b5 electron transfer complex. Science 238, 794. (12) Mathew, F. S., Levine, M., and Argos, P. (1972) Three dimensional Fourier synthesis of calf liver cytochrome b5 at 2.8 A resolution. J. Mol. Biol. 64, 229. (13) Pochapsky, T. C., Sligar, S. G., McLachlan, S. J.,and La Mar, G. N. (1990) Relationship between heme binding site
Letters structure and heme orientation of two ferrocytochrome b5s. A study in prosthetic group recognition. J . Am. Chem. SOC. 112, 5258. (14) Gouterman, M. (1978) In The Porphyrins (D. Dolphin, Ed.) pp 1-165, Academic Press Inc., New York. (15) Beck von Bodman, S., Schuler, M. A., Jollie, D. R., Sligar, S. G. (1986) Synthesis, bacterial expression and mutagenesis of the gene coding for mammalian cytochrome b5. Proc. Natl. Acad. Sci. U.S.A. 83, 9443-9447.
Bioconjugate Chem., Vol. 5, No. 6, 1994 507
(16) Stayton, P. S., Fisher, M. T., and Sligar, S. G. (1988) Determination of cytochrome b5 association reactions. Characterization of methemoglobin and cytochrome P-450,, binding to genetically engineered cytochrome b5. J.Biol. Chem. 263, 13544. (17) Yang, H. J . (1989) Investigation into an affinity precipitation system based on the thermally reversible solution behavior of poly(N4sopropylacrylamide). Ph.D. dissertation, University of Washington.
