- Email: [email protected]
Producing selenomethionine-labeled proteins with a baculovirus expression vector system
Ways & Means
R263
Producing selenomethionine-labeled proteins with a baculovirus expression vector system John J Bellizzi III1, Joanne Widom1, Christopher W Kemp2 and Jon Clardy1* Addresses: 1Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853-1301, USA and 2Kemp Biotechnologies, Inc., Frederick, MD 21704, USA.
Figure 1
25
*Corresponding author. E-mail: [email protected]
SeMet MAD Baculovirus
Structure November 1999, 7:R263–R267
Introduction
Both selenomethionine (SeMet) based multiwavelength anomalous diffraction (MAD) phasing and protein production in baculovirus expression vector systems have greatly enhanced the scope and speed of macromolecular crystallography, but the apparent incompatibility of these two powerful techniques presents an obstacle to many structure determinations. The use of either technique alone is essentially routine, and although their individual use is growing dramatically (Figure 1), their combined use is virtually unknown. As of 14th July 1999, the Protein Data Bank contained no deposited coordinates for structures solved using MAD phasing on baculovirus produced SeMet-labeled proteins. SeMet-based MAD phasing
The MAD approach to obtaining phase information has proved to be one of the most powerful in protein crystallography, and its use for a variety of problems has been expanding [1,2]. The increasing availability of synchrotron sources [3], improved algorithms for extracting phase information from MAD data [2,4,5], the development of cryocrystallography [6], and advances in area detector technology [7,8] have all contributed to the rapid and widespread popularity of MAD techniques. While MAD can be employed with various elements, selenium has become the atom of choice. Selenium has an experimentally convenient X-ray absorption edge (0.98 Å) and can be introduced into proteins biosynthetically through the substitution of selenomethionine for methionine. SeMetlabeled proteins usually differ only slightly, or not at all, from their native counterparts [9]. Baculovirus expression
The overexpression of SeMet-labeled proteins in Escherichia coli has become routine [10,11]. However, many interesting proteins cannot be satisfactorily produced in bacterial expression systems. These proteins typically require the post-translational processing machinery of eukaryotic cells, and their expression in bacterial systems leads to proteins that are insoluble, improperly
Number of structures
0969-2126/99/$ – see front matter © 1999 Elsevier Science Ltd. All rights reserved.
20
15
10
5
0
1993
1994
1995
1996 Year
1997
1998 Structure
The annual number of structures solved using proteins expressed from baculovirus and the annual number of structures solved by SeMet-based MAD phasing. Data were obtained from http://www.biomednet.com/db/msd/ (Macromolecular Structures Database).
folded and/or degraded. The baculovirus expression vector system has emerged as a powerful method for protein production in eukaryotic cells [12]. In this system, insect cells are infected with a baculovirus containing the DNA encoding the protein of interest. Expression of the desired gene product is often controlled by the viral polyhedrin promoter, which leads to very high protein expression (up to 50% of total cellular protein late in the infectious cycle). Baculovirus overexpression has been a successful means to produce proteins that require post-translational modifications, such as glycosylation, or that are too large to express in bacteria. This system has not proved generally useful for producing SeMet-labeled proteins. One obstacle to the production of SeMet-labeled proteins using the baculovirus system is the greater fragility of insect cells compared to bacteria. The insect cells often die or produce only small amounts of protein in response to mechanical stress. A second problem associated with the insect cell expression system is that the successful labeling of the protein requires the absence of the unlabeled form of the molecule. It is difficult, or impossible, to
R264
Structure 1999, Vol 7 No 11
purify SeMet-labeled protein from a mixture of labeled and unlabeled protein, and methionine will be preferentially used in protein biosynthesis over SeMet. As a consequence, SeMet-labeled protein must be expressed in a system without methionine. This is easily achieved with bacteria, which have the ability to grow in minimal media, but insect cells require a complex mixture of amino acids, peptides and lipids to support growth. It has proven difficult to devise a medium that will allow the cells to proliferate at a rate that supports viral infection and protein expression, while simultaneously maintaining a depleted level of an essential amino acid.
sites on PPT1. Upon completion of the structure determination, it became apparent that one intermolecular contact involves the glycosylated regions of two adjacent symmetry-related molecules.
Although there have been at least two reports of SeMetlabeled proteins produced in insect cells [13,14], the ability to produce MAD-quality protein crystals of these proteins has not been reported, as the structures of both were solved without using MAD techniques on the insect cell expressed protein [14,15]. We report here a protocol for the overexpression of SeMet-labeled proteins in the baculovirus/insect cell system; the labeled protein was used to solve the structure of palmitoyl protein thioesterase 1 (PPT1) using MAD phasing. A systematic study of conditions for cell growth was necessary to determine a procedure that produced protein in sufficient quantities with a high level of SeMet incorporation. We believe that this is the first such systematic analysis, and we expect that the protocol should be generally applicable to a variety of proteins.
SeMet-labeling protocol
Case study — PPT1
Project background
PPT1 is a 279-residue glycosylated lysosomal hydrolase that can be overexpressed in soluble form in high yield in insect cells, but not in E. coli [16,17]. Crystals of native PPT1 were obtained that diffracted to 2.25 Å resolution, but phase determination proved to be difficult. Severe nonisomorphism between native crystals manifested itself in significant variations of the unit-cell dimensions and unacceptably high merging R factors (greater than 25%) between datasets from different crystals. This nonisomorphism and the lack of any known structure with significant homology precluded the use of both multiple isomorphous replacement and molecular replacement techniques. Variations in the post-translational processing of glycoproteins often leads to an ensemble of molecules with different sugar chains that cannot easily be purified from one another. Although N-linked glycosylation in insect cells is primarily made up of high-mannose sugar chains, heterogeneity is still a problem due to variation in the number of sugar residues on different molecules. Such heterogeneous glycosylation can prevent crystallization in severe cases, and may lead to nonisomorphism if the variable sugar regions are involved in crystal packing. We suspect that the nonisomorphism of PPT1 crystals was due at least in part to the three N-linked glycosylation
MAD phasing seemed to be the most practical approach to structure determination, and attempts were made to obtain heavy-atom derivatives for data collection at the gold and mercury LIII absorption edges without success. The sequence of PPT1 suggested that it would be a good candidate for SeMet MAD phasing, having eight methionines among its 279 residues.
PPT1 with a high level of SeMet can be produced using a five-step protocol developed for insect cells in bioreactor culture: cell growth, infection, methionine depletion, SeMet labeling and harvest (Figure 2). With the exception of the depletion and labeling steps, this is a standard baculovirus expression protocol. A stirred tank bioreactor with a 10 L working volume was used for the large-scale production of labeled PPT1 (Bellco Glass Inc., Vineland NJ). The bioreactor consists of a glass vessel fitted with probes and hardware used to control the dissolved oxygen level and temperature of the culture. The strict control of these environmental conditions along with the use of gentle agitation greatly reduces cellular stress in bioreactor experiments as compared to shaker flasks. Sf21 cells were inoculated into the bioreactor at an initial cell density of 1 × 105 cells ml–1 in 10 L of serum-free medium (Cyto-SF9, Kemp Biotechnologies Inc., Frederick MD). The temperature of the culture was maintained at 27°C and the dissolved oxygen level at 60% air saturation. The cells were expanded to a density of 1 × 106 cells ml–1 and infected with the recombinant PPT1 baculovirus using a multiplicity of infection of five (five virus particles per cell). Thirty-six hours after infection the cells were aseptically removed from the bioreactor and dispensed into sterile 1 L centrifuge bottles. The cells were sedimented at 300 × g and 20°C using a swinging-bucket rotor. The cell pellets were resuspended in 2 L of Grace’s Medium without methionine (LTI, Gaithersburg MD) supplemented with 10% v/v dialyzed fetal bovine serum (Kemp Biotechnologies, Inc.) and returned to the bioreactor. The volume in the vessel was adjusted to 10 L using the methionine-deficient medium with dialyzed serum, and the culture was maintained for 4 h to allow the cells to deplete intracellular pools of methionine. Following the incubation period, the cells were aseptically removed from the bioreactor and sedimented at 300 × g and 20°C as described above. The cell pellets were resuspended in 2 L of Grace’s Medium without methionine, supplemented with 10% v/v dialyzed fetal bovine serum and 50 mg l–1 SeMet (Calbiochem, San Diego CA). The cell suspension was returned to the bioreactor and the volume in the vessel was adjusted to 10 L using the
Ways & Means Producing SeMet-labeled proteins from baculovirus Bellizzi et al.
R265
Figure 2 Protocol for the expression of SeMet-labeled PPT1 in baculovirus. See text for details. The media used in the different stages are color coded: yellow, Cyto-SF9 serum-free medium; red, Grace’s medium without methionine plus 10% (v/v) dialyzed fetal bovine serum; blue, Grace’s medium without methionine plus 10% (v/v) dialyzed fetal bovine serum and 50 mg l–1 SeMet.
Inoculate bioreactor with Sf21 cells
Infect with baculovirus
Cell expansion
Post-infection cell growth
Harvest
Methionine depletion
Change media
SeMet labeling
Change media
Structure
methionine-deficient medium with dialyzed serum and SeMet. The culture was maintained for an additional 36 h and the cells were sedimented at 800 × g for 10 min at 4°C. PPT1 is secreted into the culture media, so the pellets were discarded and the supernatant pooled and held at 4°C prior to purification.
using the same method as for PPT1, and ~95% for the peptide exchange factor H2-M from mouse [14], which was estimated by loss of methionine. These estimates are subject to considerable error, as loss of methionine due to hydrolysis and oxidation can lead to underestimates of the methionine content by as much as 30% [18,19].
Results of SeMet-labeled PPT1 production
We crystallized SeMet-labeled PPT1 under conditions nearly identical to those of native PPT1 (data not shown). X-ray fluorescence measurements on SeMet PPT1 crystals at the Cornell High Energy Synchrotron Source (CHESS) F2 beamline indicated a strong absorption edge at 12.666 KeV, confirming the successful incorporation of selenium into the crystal. The absorption peak was shifted ~5 eV higher in energy than the peak of the reference selenium foil. We collected diffraction data on a crystal of ~0.2 mm × 0.2 mm × 0.2 mm using an area detector systems corporation quantum-4 CCD detector [8] at CHESS F2 using three wavelengths (λ1 = 0.9791 Å, edge; λ2 = 0.9788 Å, peak; λ3 = 0.9633 Å, remote) and the inverse beam method. The data were reduced using MOSFLM [20] and the CCP4 software package [21]. The Rsym for data to 3.0 Å was 7.2% for λ1 and 7.8% for λ2 and λ3. Completeness was greater than 99% at all three wavelengths.
The purification protocol for SeMet-labeled PPT1 was identical to that for native PPT1 (data not shown). Because we had already expressed the protein in an oxidative environment, we did not take any special precautions to prevent oxidation of SeMet in PPT1 during purification and crystallization. Levels of SeMet incorporation in bacterial expression systems are usually measured by mass spectrometry (MS) techniques, but MS is not useful with variably glycosylated proteins like PPT1. The mass differences resulting from the different numbers of sugars on individual protein molecules can exceed the mass differences due to substitution of SeMet for methionine. Amino acid analysis can also be used to establish the incorporation of SeMet. Methionine and SeMet are sensitive to the conditions of the analysis, however, and can be lost due to oxidation and hydrolysis. To partially account for this sensitivity, we estimated the degree of substitution by comparing the methionine recovered from analysis of SeMet-labeled PPT1 to the methionine recovered from analysis of native PPT1, which gave an incorporation ratio of ~76%. The two prior reports of SeMet-labeled protein expression in insect cells reported substitution levels of ~84% for human choriogonadotropin [13], which was estimated
The difference Patterson methods implemented in SOLVE [22] (http://www.solve.lanl.gov) were used to locate the selenium atom positions. The program located eight peaks with an overall figure of merit of 0.65. Phasing with SHARP [4] followed by density modification with SOLOMON [23] improved the figure of merit to 0.91. The resulting solvent-flattened electron-density map was readily interpretable, and the initial model was built using
R266
Structure 1999, Vol 7 No 11
the program O [24]. Full details on the structure determination of PPT1 will be published elsewhere. Discussion
The production strategy presented above was developed over time. We originally attempted to adapt the Sf21 cell line to grow in a medium containing dialyzed serum and SeMet. The cells exhibited a decreased growth rate in this medium, which caused a decrease in product expression. We abandoned this protocol in favor of one where we replaced the normal growth medium with the labeling medium immediately prior to the addition of the virus. This protocol also resulted in a low level of product expression that may have been due to stresses placed on the cell population by the medium exchange. We reasoned that the yield might be improved by waiting to introduce the stress of a medium exchange until after the infection cycle was initiated. We allowed the infection of the cells and the initial 36 h of the infection cycle to proceed in the growth medium, and then changed the medium to the methionine-deficient Grace’s medium. This ‘starvation’ medium, introduced in an attempt to lower the intracellular pool of methionine, was not a true minimal medium, as supplementation with 10% dialyzed fetal bovine serum was essential for cell viability. We tested starvation times of 4, 8 and 16 h and found that cell viability decreased in the samples incubated for 8 or 16 h. The 4 h incubation was selected on the basis of this observation. Omission of the starvation step led to a low level of SeMet incorporation. After the 4 h incubation, the medium was exchanged for the labeling medium and the expression was allowed to proceed to the post-infection time that was determined to be optimal for this protein (72 h post-infection). The general success of this protocol depends on defining conditions that maintain the viability of the culture under the stressful conditions imposed by the labeling medium. We would suggest that investigators applying this protocol to other proteins maintain 24 h as the maximum initial post-infection incubation phase and 4 h as the starvation period. The 24 h time period is of sufficient length to allow the infection to initialize and protein expression to begin. Waiting longer may dilute the labeled protein population with unlabeled material. On the basis of the PPT1 results, a 4 h starvation period is able to deplete methionine reserves enough to produce product that is sufficiently enriched in SeMet for MAD phasing. The optimal length of the incubation in the labeling medium should be determined for each protein of interest, and will depend on the dynamics of expression of that particular protein. The mechanical and nutritional sensitivity of the insect cells must not be overlooked. In addition to the requirement for fetal bovine serum in the starvation and labeling
media, we found that higher concentrations of SeMet (100 mg l–1) led to premature cell death, but the cells could grow in 50 mg l–1 SeMet. Extreme care has to be taken when handling the cells during media exchange, because in their stressed state they are extremely susceptible to mechanical rupture. We found that the controlled environmental conditions and gentle agitation rates achievable using a bioreactor were ideal for this application and that our best results were obtained using this instrument. Trials using static culture, roller bottles, shaker flasks and spinner flasks produced unsatisfactory results. Baculovirus expression and MAD phasing are both critical techniques in this era of high-throughput structural biology, expanding the variety of proteins that can be overexpressed and streamlining the structure determination process. The marriage of these two techniques, SeMet labeling in insect cells and MAD phasing, is likely to be widely adopted. Note added in proof
After submission of this article, the production of a SeMetlabeled fragment of the human tumor necrosis factor TRAF2 in baculovirus with ~40% incorporation, as measured by mass spectrometry, was reported (McWhirter, S.M., Pullen, S.S., Holton, J.M., Crute, J.J., Kehry, M.R. & Alber, T. (1999). Crystallographic analysis of CD40 recognition and signaling by human TRAF2. Proc. Natl Acad. Sci. 96, 8408-8413). The authors used a strategy for SeMet labeling similar to the one described in this article, except that their methionine depletion time was only 1 h compared to the 4 h in our protocol, and they used a concentration of 100 mg l–1 SeMet rather than the 50 mg l–1 SeMet used in our protocol. They determined the structure of TRAF2 complexed with a CD40 peptide using MAD phasing on both an Hg derivative and a SeMet derivative. Acknowledgements We would like to thank Wayne Hendrickson for directing our attention to [14]. This work was supported by NIH grant CA59021 (JC) and an NIH training grant in molecular physics of biological systems (JJB). This work is based in part on research conducted at CHESS, which is supported by the National Science Foundation under award DMR-9311772, using the macromolecular diffraction at CHESS (MacCHESS) facility, which is supported by award RR-01646 from the NIH.
References 1. Hendrickson, W.A. (1991). Determination of macromolecular structures from anomalous diffraction of synchrotron radiation. Science 254, 51-58. 2. Hendrickson, W.A. & Ogata, C.M. (1997). Phase determination from multiwavelength anomalous diffraction measurements. Methods Enzymol. 276, 494-523. 3. Mitchell, E., Kuhn, P. & Garman, E. (1999). Demystifying the synchrotron trip: a first time user’s guide. Structure 7, R111-R122. 4. de La Fortelle, E. & Bricogne, G. (1997). Maximum-likelihood heavyatom parameter refinement in the MIR and MAD methods. Methods Enzymol. 276, 472-494. 5. Terwilliger, T.C. & Berendzen, J. (1997). Bayesian MAD phasing. Acta Crystallogr. D 53, 571-579. 6. Rogers, D.W. (1994). Cryocrystallography. Structure 2, 1135-1140.
Ways & Means Producing SeMet-labeled proteins from baculovirus Bellizzi et al.
7. Walter, R.L., et al., & Ealick, S.E. (1995). High-resolution macromolecular structure determination using CCD detectors and synchrotron radiation. Structure 3, 835-844. 8. Szebenyi, D.M.E., Arvai, A., Ealick, S., Laluppa, J.M. & Nielsen, C. (1997). A system for integrated collection and analysis of crystallographic diffraction data. J. Sync. Rad. 4, 128-135. 9. Smith, J.L. & Thompson, A. (1998). Reactivity of selenomethionine — dents in the magic bullet? Structure 6, 815-819. 10. Hendrickson, W.A., Horton, J.R. & Lemaster, D.M. (1990). Selenomethionyl proteins produced for analysis by multiwavelength anomalous diffraction (MAD): a vehicle for direct determination of three-dimensional structure. EMBO J. 9, 1665-1672. 11. Doublie, S. (1997). Preparation of selenomethionyl proteins for phase determination. Methods Enzymol. 276, 523-530. 12. Possee, R.D. (1997). Baculoviruses as expression vectors. Curr. Opin. Biotech. 8, 569-572. 13. Chen, W. & Bahl, O.P. (1990). Recombinant carbohydrate and selenomethionyl variants of human choriogonadotropin. J. Biol. Chem. 266, 8192-8197. 14. Fremont, D.H., Crawford, F., Marrack, P., Hendrickson, W.A. & Kappler, J. (1998). Crystal structure of mouse H2-M. Immunity 9, 385-393. 15. Wu, H., Lustbader, J.W., Liu, Y., Canfield, R.E. & Hendrickson, W.A. (1994). Structure of human chorionic gonadotropin at 2.6 Å resolution from MAD analysis of the selenomethionyl protein. Structure 2, 545-558. 16. Camp, L.A., Verkruyse, L.A., Afendis, S.J., Slaughter, C.A. & Hofmann, S.L. (1994). Molecular cloning and expression of palmitoyl-protein thioesterase. J. Biol. Chem. 269, 23212-23219. 17. Camp, L.A. & Hofmann, S.L. (1993). Purification and properties of a palmitoyl-protein thioesterase that cleaves palmitate from H-Ras. J. Biol. Chem. 268, 22566-22574. 18. Schegg, K., et al., & Mann, K. (1997). Quantitation and identification of proteins by amino acid analysis: ABRF-96AAA collaborative trial. In Techniques in Protein Chemistry VIII. (Marshak, D.R., ed.), pp. 207-216, Academic Press, San Diego, CA. 19. Dunn, B.M. (1995). Quantitative amino acid analysis. In Current Protocols in Protein Science. (Coligan, J.E., Dunn, B.M., Ploegh, H.L., Speicher, D.W. & Wingfield, P.T., eds), pp. 3.2.1-3.2.3, John Wiley & Sons, New York, NY. 20. Leslie, A.G.W. (1994). Mosflm User Guide, Mosflm Version 5.30. MRC Laboratory of Molecular Biology, Cambridge, UK. 21. Collaborative Computational Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D 50, 760-763. 22. Terwilliger, T.C., Kim, S.-H. & Eisenberg, D. (1987). Generalized method of determining heavy-atom positions using the difference Patterson function. Acta Crystallogr. A 43, 1-5. 23. Abrahams, J. & Leslie, A. (1996). Methods used in the structure determination of bovine mitochondrial F1 ATPase. Acta Crystallogr. D 52, 30-42. 24. Jones, T.A., Zou, J.-Y., Cowan, S. & Kjeldgaard, M. (1991). Improved methods for binding protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110-119.
R267
R263
Producing selenomethionine-labeled proteins with a baculovirus expression vector system John J Bellizzi III1, Joanne Widom1, Christopher W Kemp2 and Jon Clardy1* Addresses: 1Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853-1301, USA and 2Kemp Biotechnologies, Inc., Frederick, MD 21704, USA.
Figure 1
25
*Corresponding author. E-mail: [email protected]
SeMet MAD Baculovirus
Structure November 1999, 7:R263–R267
Introduction
Both selenomethionine (SeMet) based multiwavelength anomalous diffraction (MAD) phasing and protein production in baculovirus expression vector systems have greatly enhanced the scope and speed of macromolecular crystallography, but the apparent incompatibility of these two powerful techniques presents an obstacle to many structure determinations. The use of either technique alone is essentially routine, and although their individual use is growing dramatically (Figure 1), their combined use is virtually unknown. As of 14th July 1999, the Protein Data Bank contained no deposited coordinates for structures solved using MAD phasing on baculovirus produced SeMet-labeled proteins. SeMet-based MAD phasing
The MAD approach to obtaining phase information has proved to be one of the most powerful in protein crystallography, and its use for a variety of problems has been expanding [1,2]. The increasing availability of synchrotron sources [3], improved algorithms for extracting phase information from MAD data [2,4,5], the development of cryocrystallography [6], and advances in area detector technology [7,8] have all contributed to the rapid and widespread popularity of MAD techniques. While MAD can be employed with various elements, selenium has become the atom of choice. Selenium has an experimentally convenient X-ray absorption edge (0.98 Å) and can be introduced into proteins biosynthetically through the substitution of selenomethionine for methionine. SeMetlabeled proteins usually differ only slightly, or not at all, from their native counterparts [9]. Baculovirus expression
The overexpression of SeMet-labeled proteins in Escherichia coli has become routine [10,11]. However, many interesting proteins cannot be satisfactorily produced in bacterial expression systems. These proteins typically require the post-translational processing machinery of eukaryotic cells, and their expression in bacterial systems leads to proteins that are insoluble, improperly
Number of structures
0969-2126/99/$ – see front matter © 1999 Elsevier Science Ltd. All rights reserved.
20
15
10
5
0
1993
1994
1995
1996 Year
1997
1998 Structure
The annual number of structures solved using proteins expressed from baculovirus and the annual number of structures solved by SeMet-based MAD phasing. Data were obtained from http://www.biomednet.com/db/msd/ (Macromolecular Structures Database).
folded and/or degraded. The baculovirus expression vector system has emerged as a powerful method for protein production in eukaryotic cells [12]. In this system, insect cells are infected with a baculovirus containing the DNA encoding the protein of interest. Expression of the desired gene product is often controlled by the viral polyhedrin promoter, which leads to very high protein expression (up to 50% of total cellular protein late in the infectious cycle). Baculovirus overexpression has been a successful means to produce proteins that require post-translational modifications, such as glycosylation, or that are too large to express in bacteria. This system has not proved generally useful for producing SeMet-labeled proteins. One obstacle to the production of SeMet-labeled proteins using the baculovirus system is the greater fragility of insect cells compared to bacteria. The insect cells often die or produce only small amounts of protein in response to mechanical stress. A second problem associated with the insect cell expression system is that the successful labeling of the protein requires the absence of the unlabeled form of the molecule. It is difficult, or impossible, to
R264
Structure 1999, Vol 7 No 11
purify SeMet-labeled protein from a mixture of labeled and unlabeled protein, and methionine will be preferentially used in protein biosynthesis over SeMet. As a consequence, SeMet-labeled protein must be expressed in a system without methionine. This is easily achieved with bacteria, which have the ability to grow in minimal media, but insect cells require a complex mixture of amino acids, peptides and lipids to support growth. It has proven difficult to devise a medium that will allow the cells to proliferate at a rate that supports viral infection and protein expression, while simultaneously maintaining a depleted level of an essential amino acid.
sites on PPT1. Upon completion of the structure determination, it became apparent that one intermolecular contact involves the glycosylated regions of two adjacent symmetry-related molecules.
Although there have been at least two reports of SeMetlabeled proteins produced in insect cells [13,14], the ability to produce MAD-quality protein crystals of these proteins has not been reported, as the structures of both were solved without using MAD techniques on the insect cell expressed protein [14,15]. We report here a protocol for the overexpression of SeMet-labeled proteins in the baculovirus/insect cell system; the labeled protein was used to solve the structure of palmitoyl protein thioesterase 1 (PPT1) using MAD phasing. A systematic study of conditions for cell growth was necessary to determine a procedure that produced protein in sufficient quantities with a high level of SeMet incorporation. We believe that this is the first such systematic analysis, and we expect that the protocol should be generally applicable to a variety of proteins.
SeMet-labeling protocol
Case study — PPT1
Project background
PPT1 is a 279-residue glycosylated lysosomal hydrolase that can be overexpressed in soluble form in high yield in insect cells, but not in E. coli [16,17]. Crystals of native PPT1 were obtained that diffracted to 2.25 Å resolution, but phase determination proved to be difficult. Severe nonisomorphism between native crystals manifested itself in significant variations of the unit-cell dimensions and unacceptably high merging R factors (greater than 25%) between datasets from different crystals. This nonisomorphism and the lack of any known structure with significant homology precluded the use of both multiple isomorphous replacement and molecular replacement techniques. Variations in the post-translational processing of glycoproteins often leads to an ensemble of molecules with different sugar chains that cannot easily be purified from one another. Although N-linked glycosylation in insect cells is primarily made up of high-mannose sugar chains, heterogeneity is still a problem due to variation in the number of sugar residues on different molecules. Such heterogeneous glycosylation can prevent crystallization in severe cases, and may lead to nonisomorphism if the variable sugar regions are involved in crystal packing. We suspect that the nonisomorphism of PPT1 crystals was due at least in part to the three N-linked glycosylation
MAD phasing seemed to be the most practical approach to structure determination, and attempts were made to obtain heavy-atom derivatives for data collection at the gold and mercury LIII absorption edges without success. The sequence of PPT1 suggested that it would be a good candidate for SeMet MAD phasing, having eight methionines among its 279 residues.
PPT1 with a high level of SeMet can be produced using a five-step protocol developed for insect cells in bioreactor culture: cell growth, infection, methionine depletion, SeMet labeling and harvest (Figure 2). With the exception of the depletion and labeling steps, this is a standard baculovirus expression protocol. A stirred tank bioreactor with a 10 L working volume was used for the large-scale production of labeled PPT1 (Bellco Glass Inc., Vineland NJ). The bioreactor consists of a glass vessel fitted with probes and hardware used to control the dissolved oxygen level and temperature of the culture. The strict control of these environmental conditions along with the use of gentle agitation greatly reduces cellular stress in bioreactor experiments as compared to shaker flasks. Sf21 cells were inoculated into the bioreactor at an initial cell density of 1 × 105 cells ml–1 in 10 L of serum-free medium (Cyto-SF9, Kemp Biotechnologies Inc., Frederick MD). The temperature of the culture was maintained at 27°C and the dissolved oxygen level at 60% air saturation. The cells were expanded to a density of 1 × 106 cells ml–1 and infected with the recombinant PPT1 baculovirus using a multiplicity of infection of five (five virus particles per cell). Thirty-six hours after infection the cells were aseptically removed from the bioreactor and dispensed into sterile 1 L centrifuge bottles. The cells were sedimented at 300 × g and 20°C using a swinging-bucket rotor. The cell pellets were resuspended in 2 L of Grace’s Medium without methionine (LTI, Gaithersburg MD) supplemented with 10% v/v dialyzed fetal bovine serum (Kemp Biotechnologies, Inc.) and returned to the bioreactor. The volume in the vessel was adjusted to 10 L using the methionine-deficient medium with dialyzed serum, and the culture was maintained for 4 h to allow the cells to deplete intracellular pools of methionine. Following the incubation period, the cells were aseptically removed from the bioreactor and sedimented at 300 × g and 20°C as described above. The cell pellets were resuspended in 2 L of Grace’s Medium without methionine, supplemented with 10% v/v dialyzed fetal bovine serum and 50 mg l–1 SeMet (Calbiochem, San Diego CA). The cell suspension was returned to the bioreactor and the volume in the vessel was adjusted to 10 L using the
Ways & Means Producing SeMet-labeled proteins from baculovirus Bellizzi et al.
R265
Figure 2 Protocol for the expression of SeMet-labeled PPT1 in baculovirus. See text for details. The media used in the different stages are color coded: yellow, Cyto-SF9 serum-free medium; red, Grace’s medium without methionine plus 10% (v/v) dialyzed fetal bovine serum; blue, Grace’s medium without methionine plus 10% (v/v) dialyzed fetal bovine serum and 50 mg l–1 SeMet.
Inoculate bioreactor with Sf21 cells
Infect with baculovirus
Cell expansion
Post-infection cell growth
Harvest
Methionine depletion
Change media
SeMet labeling
Change media
Structure
methionine-deficient medium with dialyzed serum and SeMet. The culture was maintained for an additional 36 h and the cells were sedimented at 800 × g for 10 min at 4°C. PPT1 is secreted into the culture media, so the pellets were discarded and the supernatant pooled and held at 4°C prior to purification.
using the same method as for PPT1, and ~95% for the peptide exchange factor H2-M from mouse [14], which was estimated by loss of methionine. These estimates are subject to considerable error, as loss of methionine due to hydrolysis and oxidation can lead to underestimates of the methionine content by as much as 30% [18,19].
Results of SeMet-labeled PPT1 production
We crystallized SeMet-labeled PPT1 under conditions nearly identical to those of native PPT1 (data not shown). X-ray fluorescence measurements on SeMet PPT1 crystals at the Cornell High Energy Synchrotron Source (CHESS) F2 beamline indicated a strong absorption edge at 12.666 KeV, confirming the successful incorporation of selenium into the crystal. The absorption peak was shifted ~5 eV higher in energy than the peak of the reference selenium foil. We collected diffraction data on a crystal of ~0.2 mm × 0.2 mm × 0.2 mm using an area detector systems corporation quantum-4 CCD detector [8] at CHESS F2 using three wavelengths (λ1 = 0.9791 Å, edge; λ2 = 0.9788 Å, peak; λ3 = 0.9633 Å, remote) and the inverse beam method. The data were reduced using MOSFLM [20] and the CCP4 software package [21]. The Rsym for data to 3.0 Å was 7.2% for λ1 and 7.8% for λ2 and λ3. Completeness was greater than 99% at all three wavelengths.
The purification protocol for SeMet-labeled PPT1 was identical to that for native PPT1 (data not shown). Because we had already expressed the protein in an oxidative environment, we did not take any special precautions to prevent oxidation of SeMet in PPT1 during purification and crystallization. Levels of SeMet incorporation in bacterial expression systems are usually measured by mass spectrometry (MS) techniques, but MS is not useful with variably glycosylated proteins like PPT1. The mass differences resulting from the different numbers of sugars on individual protein molecules can exceed the mass differences due to substitution of SeMet for methionine. Amino acid analysis can also be used to establish the incorporation of SeMet. Methionine and SeMet are sensitive to the conditions of the analysis, however, and can be lost due to oxidation and hydrolysis. To partially account for this sensitivity, we estimated the degree of substitution by comparing the methionine recovered from analysis of SeMet-labeled PPT1 to the methionine recovered from analysis of native PPT1, which gave an incorporation ratio of ~76%. The two prior reports of SeMet-labeled protein expression in insect cells reported substitution levels of ~84% for human choriogonadotropin [13], which was estimated
The difference Patterson methods implemented in SOLVE [22] (http://www.solve.lanl.gov) were used to locate the selenium atom positions. The program located eight peaks with an overall figure of merit of 0.65. Phasing with SHARP [4] followed by density modification with SOLOMON [23] improved the figure of merit to 0.91. The resulting solvent-flattened electron-density map was readily interpretable, and the initial model was built using
R266
Structure 1999, Vol 7 No 11
the program O [24]. Full details on the structure determination of PPT1 will be published elsewhere. Discussion
The production strategy presented above was developed over time. We originally attempted to adapt the Sf21 cell line to grow in a medium containing dialyzed serum and SeMet. The cells exhibited a decreased growth rate in this medium, which caused a decrease in product expression. We abandoned this protocol in favor of one where we replaced the normal growth medium with the labeling medium immediately prior to the addition of the virus. This protocol also resulted in a low level of product expression that may have been due to stresses placed on the cell population by the medium exchange. We reasoned that the yield might be improved by waiting to introduce the stress of a medium exchange until after the infection cycle was initiated. We allowed the infection of the cells and the initial 36 h of the infection cycle to proceed in the growth medium, and then changed the medium to the methionine-deficient Grace’s medium. This ‘starvation’ medium, introduced in an attempt to lower the intracellular pool of methionine, was not a true minimal medium, as supplementation with 10% dialyzed fetal bovine serum was essential for cell viability. We tested starvation times of 4, 8 and 16 h and found that cell viability decreased in the samples incubated for 8 or 16 h. The 4 h incubation was selected on the basis of this observation. Omission of the starvation step led to a low level of SeMet incorporation. After the 4 h incubation, the medium was exchanged for the labeling medium and the expression was allowed to proceed to the post-infection time that was determined to be optimal for this protein (72 h post-infection). The general success of this protocol depends on defining conditions that maintain the viability of the culture under the stressful conditions imposed by the labeling medium. We would suggest that investigators applying this protocol to other proteins maintain 24 h as the maximum initial post-infection incubation phase and 4 h as the starvation period. The 24 h time period is of sufficient length to allow the infection to initialize and protein expression to begin. Waiting longer may dilute the labeled protein population with unlabeled material. On the basis of the PPT1 results, a 4 h starvation period is able to deplete methionine reserves enough to produce product that is sufficiently enriched in SeMet for MAD phasing. The optimal length of the incubation in the labeling medium should be determined for each protein of interest, and will depend on the dynamics of expression of that particular protein. The mechanical and nutritional sensitivity of the insect cells must not be overlooked. In addition to the requirement for fetal bovine serum in the starvation and labeling
media, we found that higher concentrations of SeMet (100 mg l–1) led to premature cell death, but the cells could grow in 50 mg l–1 SeMet. Extreme care has to be taken when handling the cells during media exchange, because in their stressed state they are extremely susceptible to mechanical rupture. We found that the controlled environmental conditions and gentle agitation rates achievable using a bioreactor were ideal for this application and that our best results were obtained using this instrument. Trials using static culture, roller bottles, shaker flasks and spinner flasks produced unsatisfactory results. Baculovirus expression and MAD phasing are both critical techniques in this era of high-throughput structural biology, expanding the variety of proteins that can be overexpressed and streamlining the structure determination process. The marriage of these two techniques, SeMet labeling in insect cells and MAD phasing, is likely to be widely adopted. Note added in proof
After submission of this article, the production of a SeMetlabeled fragment of the human tumor necrosis factor TRAF2 in baculovirus with ~40% incorporation, as measured by mass spectrometry, was reported (McWhirter, S.M., Pullen, S.S., Holton, J.M., Crute, J.J., Kehry, M.R. & Alber, T. (1999). Crystallographic analysis of CD40 recognition and signaling by human TRAF2. Proc. Natl Acad. Sci. 96, 8408-8413). The authors used a strategy for SeMet labeling similar to the one described in this article, except that their methionine depletion time was only 1 h compared to the 4 h in our protocol, and they used a concentration of 100 mg l–1 SeMet rather than the 50 mg l–1 SeMet used in our protocol. They determined the structure of TRAF2 complexed with a CD40 peptide using MAD phasing on both an Hg derivative and a SeMet derivative. Acknowledgements We would like to thank Wayne Hendrickson for directing our attention to [14]. This work was supported by NIH grant CA59021 (JC) and an NIH training grant in molecular physics of biological systems (JJB). This work is based in part on research conducted at CHESS, which is supported by the National Science Foundation under award DMR-9311772, using the macromolecular diffraction at CHESS (MacCHESS) facility, which is supported by award RR-01646 from the NIH.
References 1. Hendrickson, W.A. (1991). Determination of macromolecular structures from anomalous diffraction of synchrotron radiation. Science 254, 51-58. 2. Hendrickson, W.A. & Ogata, C.M. (1997). Phase determination from multiwavelength anomalous diffraction measurements. Methods Enzymol. 276, 494-523. 3. Mitchell, E., Kuhn, P. & Garman, E. (1999). Demystifying the synchrotron trip: a first time user’s guide. Structure 7, R111-R122. 4. de La Fortelle, E. & Bricogne, G. (1997). Maximum-likelihood heavyatom parameter refinement in the MIR and MAD methods. Methods Enzymol. 276, 472-494. 5. Terwilliger, T.C. & Berendzen, J. (1997). Bayesian MAD phasing. Acta Crystallogr. D 53, 571-579. 6. Rogers, D.W. (1994). Cryocrystallography. Structure 2, 1135-1140.
Ways & Means Producing SeMet-labeled proteins from baculovirus Bellizzi et al.
7. Walter, R.L., et al., & Ealick, S.E. (1995). High-resolution macromolecular structure determination using CCD detectors and synchrotron radiation. Structure 3, 835-844. 8. Szebenyi, D.M.E., Arvai, A., Ealick, S., Laluppa, J.M. & Nielsen, C. (1997). A system for integrated collection and analysis of crystallographic diffraction data. J. Sync. Rad. 4, 128-135. 9. Smith, J.L. & Thompson, A. (1998). Reactivity of selenomethionine — dents in the magic bullet? Structure 6, 815-819. 10. Hendrickson, W.A., Horton, J.R. & Lemaster, D.M. (1990). Selenomethionyl proteins produced for analysis by multiwavelength anomalous diffraction (MAD): a vehicle for direct determination of three-dimensional structure. EMBO J. 9, 1665-1672. 11. Doublie, S. (1997). Preparation of selenomethionyl proteins for phase determination. Methods Enzymol. 276, 523-530. 12. Possee, R.D. (1997). Baculoviruses as expression vectors. Curr. Opin. Biotech. 8, 569-572. 13. Chen, W. & Bahl, O.P. (1990). Recombinant carbohydrate and selenomethionyl variants of human choriogonadotropin. J. Biol. Chem. 266, 8192-8197. 14. Fremont, D.H., Crawford, F., Marrack, P., Hendrickson, W.A. & Kappler, J. (1998). Crystal structure of mouse H2-M. Immunity 9, 385-393. 15. Wu, H., Lustbader, J.W., Liu, Y., Canfield, R.E. & Hendrickson, W.A. (1994). Structure of human chorionic gonadotropin at 2.6 Å resolution from MAD analysis of the selenomethionyl protein. Structure 2, 545-558. 16. Camp, L.A., Verkruyse, L.A., Afendis, S.J., Slaughter, C.A. & Hofmann, S.L. (1994). Molecular cloning and expression of palmitoyl-protein thioesterase. J. Biol. Chem. 269, 23212-23219. 17. Camp, L.A. & Hofmann, S.L. (1993). Purification and properties of a palmitoyl-protein thioesterase that cleaves palmitate from H-Ras. J. Biol. Chem. 268, 22566-22574. 18. Schegg, K., et al., & Mann, K. (1997). Quantitation and identification of proteins by amino acid analysis: ABRF-96AAA collaborative trial. In Techniques in Protein Chemistry VIII. (Marshak, D.R., ed.), pp. 207-216, Academic Press, San Diego, CA. 19. Dunn, B.M. (1995). Quantitative amino acid analysis. In Current Protocols in Protein Science. (Coligan, J.E., Dunn, B.M., Ploegh, H.L., Speicher, D.W. & Wingfield, P.T., eds), pp. 3.2.1-3.2.3, John Wiley & Sons, New York, NY. 20. Leslie, A.G.W. (1994). Mosflm User Guide, Mosflm Version 5.30. MRC Laboratory of Molecular Biology, Cambridge, UK. 21. Collaborative Computational Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D 50, 760-763. 22. Terwilliger, T.C., Kim, S.-H. & Eisenberg, D. (1987). Generalized method of determining heavy-atom positions using the difference Patterson function. Acta Crystallogr. A 43, 1-5. 23. Abrahams, J. & Leslie, A. (1996). Methods used in the structure determination of bovine mitochondrial F1 ATPase. Acta Crystallogr. D 52, 30-42. 24. Jones, T.A., Zou, J.-Y., Cowan, S. & Kjeldgaard, M. (1991). Improved methods for binding protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110-119.
R267