PEDS Advance Access originally published online on February 14, 2006
Protein Engineering Design and Selection 2006 19(4):147-153; doi:10.1093/protein/gzj013
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Enhanced stability of recombinant keratinocyte growth factor by mutagenesis
1Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, 2Alliance Protein Laboratories, Camarillo, CA and 3Chiron Corporation 4560 Horton Street Emeryville, CA 94583, USA
4 To whom correspondence should be addressed. E-mail: lnarhi{at}amgen.com
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Native sequence keratinocyte growth factor (KGF) is fairly unstable, as manifested by the loss of the monomeric native protein accompanied by the accumulation of aggregated species during storage at moderate temperatures. Several different types of analogs were generated and the storage stability of the protein assessed. In the first type of analog one or more of the five cysteinyl residues in KGF were replaced; in the second class the N-terminal residues that included the first disulfide bond were deleted. Both of these types of analogs involved removal of the disulfide bond between cysteines 1 and 15. The third group involved mutating one of the basic amino acids located in a cluster of positive charges (involved in heparin binding) around Arg144 to a neutral or acidic amino acyl residue. Among the cysteine replacement analogs, the double mutation of Cys1 and 15 to Ser resulted in significantly increased stability without compromising the mitogenic activity, while Cys to Ser mutations at other positions were either destabilizing or had no effect. Deletion of the 15, 23 or 27 N-terminal amino acyl residues also increased the stability of the protein. The activity of the analogs was not affected by the deletion of 15 or 23 amino acids, but it was significantly decreased upon removal of the 27 N-terminal amino acyl residues. Much greater stability was achieved by mutation of the basic amino acids, especially Arg144, to Glu or Gln, but this increase in stability was accompanied by large decrease in activity. The analog with the 23 N-terminal amino acyl residues deleted represents one of the best compromises between increased stability and retention of activity.
Keywords: KGF/protein stability/heparin binding/cysteine replacement
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The fibroblast growth factor (FGF) family consists of at least 22 members and 4 FGF-related growth factor members (Itoh and Ornitz, 2004
). Among them, keratinocyte growth factor (KGF) (FGF-7 and FGF-10) acts as a growth factor exclusively through the FGF receptor-2IIIb variant expressed by epithelial cells (Ornitz et al., 1996) and hence can protect these cells from various insults, such as anti-cancer drugs (Farrell et al., 2002; auf dem Keller et al., 2004; Finch and Rubin, 2004) and skin or organ injuries (Finch and Rubin, 2004). As a family these are relatively unstable proteins; this has been demonstrated for acidic FGF and KGF (Tsai et al., 1993; Volkin et al., 1993; Chen et al., 1994a,b; Chen and Arakawa, 1996). In thermal unfolding experiments, KGF denaturation commences at 
40°C, on the low end of the temperature range observed for most globular proteins. Shortly after it begins unfolding the protein starts to aggregate, leading to the formation of particulates and precipitates (irreversible products, Chen et al., 1994a; Chen and Arakawa, 1996). In order to improve the storage stability of recombinant KGF, we have previously investigated the effects of various additives, including heparin, polymers and osmolytes (Chen et al., 1994a; Chen and Arakawa, 1996). In this study, we have taken a different approach, and have attempted to generate more stable mutants of KGF.
KGF contains five cysteinyl residues at positions 1, 15, 40, 102 and 106. Although the complete disulfide structure has not been determined on the natural protein, analysis of the recombinant protein suggested disulfide bonds between Cys 1 and 15 and between Cys102 and 106, with Cys 40 present in the free sulfhydryl form (Culajay et al., 2000). Mutation of the cysteine to alanine residues has been reported previously (Culajay et al., 2000); however, serine is a more conservative substitution and could give different results. A set of mutations where the N-terminal region containing the first disulfide bond was deleted was also generated.
FGF family members, including KGF, are characterized by avid binding to heparin (Bottaro et al., 1990; Wen et al., 1996; Hsu et al., 2000). The X-ray crystallographic structure of acidic and basic FGF revealed a cluster of four positive charges (Ago et al., 1991; Eriksson et al., 1991; Zhang et al., 1991; Zhu et al., 1991, 1993; Faham et al., 1996; Osslund et al., 1998) that are also seen in the structure of KGF (Osslund et al., 1998) (Figure 1). Chen and co-workers showed that addition of heparin increases the melting temperature of KGF and its shelf life in accelerated stability studies (Chen et al., 1994b; Chen and Arakawa, 1996). The mechanism of this stabilization may be due to the neutralization of these positive charges. If this is true, then one may speculate that these basic residues electrostatically destabilize the protein and hence mutation of at least one of these four basic residues to an acidic or a neutral amino acid would decrease the charge repulsion. We have therefore mutated Arg144 and several other basic residues in this region to various amino acids, either on the framework of the native sequence protein, on the cysteine mutants, or on the deletion mutants described above.
|
We have previously shown that the substitution of glycine for His93 in aFGF increased the stability of the protein (Arakawa et al., 1993
). Therefore the corresponding histidine at position 116 in KGF was mutated to asparagine or glycine. Thus three different types of mutations were introduced in an attempt to generate more stable but active analogs; in the first family of mutations cysteines were mutated to serines, in the second family different lengths of the N-terminal domain containing Cys1 and 15 were deleted and in the third family the charge in the basic heparin binding region was decreased by the introduction of an acidic residue. Table I lists the analogs studied, with their specific mutations, and the family that they belonged to. In this paper, we report the characterization of these three different types of mutations with regard to their mitogenic activity, conformational and storage stabilities, and other biochemical properties.
|
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Materials
The cloning, expression and purification of KGF have been described previously (Bottaro et al., 1990; Spahr et al., 1997; Ossulund et al., 1998).
All chemicals were of the analytical grade and obtained from Sigma Chemical Co. Dulbecco's phosphate-buffered saline (PBS) was obtained from Gibco BRL Life Technologies Inc., Gaithersburg, MD.
Chromatographic analysis
Ion-exchange chromatography was performed on a Hewlett-Packard 1050 liquid chromatography system equipped with an HP 3D Chemstation for data acquisition. The ion-exchange chromatography was carried out using a Pharmacia Mono-S column as described previously (Chen et al., 1994b; Chen and Arakawa, 1996).
Heparin chromatography of KGF was performed on a Toso Haas heparin5PW TSK column (8 mm x 7.5 cm, 10 µm). Injected samples were first equilibrated for 5 min in 95% buffer A (20 mM Na phosphate, pH 7.5) and 5% buffer B (20 mM Na phosphate, pH 7.5 and 2 M NaCI). The protein was subsequently eluted at a 1 ml/min flow rate using a linear gradient of 5–95% buffer B in 40 min and detected by absorbance at 230 nm monitored on an HP diode-array detector.
In vitro mitogenic assay
The in vitro mitogenic assay was performed using a mouse epidermal keratinocyte cell line, Balb/MK, essentially according to Falco et al. (1988).
Preliminary accelerated stability testing
Several of the analogs were subjected to accelerated stability testing. They were incubated in PBS at 37°C for 20 min and centrifuged to remove the precipitate. The amount of soluble protein remaining was quantified by the Bradford protein assay and by the area of the RP-HPLC elution peak. Owing to the limited amounts of protein available for many of the analogs, replicate samples could not be assayed, and each point is the result of a single sample.
Formal accelerated stability test
Samples for storage stability tests contained 0.5 mg/ml KGF prepared in PBS. These samples (0.5 ml each) were aliquoted into 3 ml type 1 glass vials in a sterile laminar flow hood. The vials were sealed with rubber stoppers (1888 Teflon, West Co.) and 13 mm flip-off aluminum seals were crimped in place. Finally, the vials were placed in incubators set at 37°C. After storage at 37°C, samples were withdrawn and filtered through 0.22 µm Costar Spin-X filter units to separate soluble proteins from precipitates. The soluble proteins were subjected to ion-exchange chromatography or RP-HPLC. Then the amount of soluble KGF was estimated from the integrated area of the monomeric KGF peak in the chromatogram (Chen et al., 1994a,b). Kinetic parameters such as rate constants and half-lives for loss of soluble monomeric KGF were determined by fitting the plot of the integrated peak area against the storage period to a single exponential decay. Due to the limited amounts of protein available for many of the analogs, replicates could not be assayed. However, the normal variability of this assay was 5–10%.
CD
Far and near UV spectra were obtained on a Jasco J-710 spectropolarimeter, using a cuvette with a pathlength of 0.02 cm for the far UV region and a 1.0 cm pathlength for the near UV region. Molar ellipticity was calculated assuming a mean residue weight of 115.8. Protein concentrations were determined from the absorbance at 280 nm of 0.1 assuming an E 280 of 1.16.
Thermal stability was assessed by following changes in the ellipticity at 230 nm on a Jasco J-720 spectropolarimeter equipped with a Peltier thermal cuvette holder and a JTC 345 control unit, using a protein concentration at 0.5 mg/ml, a heating rate of 100°C/h and a 0.1 cm pathlength thermal cuvette. The thermal unfolding of KGF is accompanied by aggregation and precipitation, resulting in light scattering that interferes with the CD signal. Therefore the temperature at which melting commences was measured, and was reported as an apparent transition temperature.
DSC
KGF samples at 0.5–3.5 mg/ml in PBS were heated at 100°C per h in a Hart differential scanning calorimeter. Thermally induced denaturation of KGF is accompanied by precipitation, resulting in a heating profile that consists of a single large exotherm. The temperature at which this exotherm occurred was measured using the software provided by the manufacturer and reported as the apparent transition temperature by this technique. The melting of KGF is an irreversible reaction and therefore no thermodynamic calculations were performed using this apparent transition temperature (Tm).
Stability to denaturants
CD was used to follow urea-induced denaturation in PBS. The denaturation was initiated by manually mixing a constant amount of protein with varying aliquots of 5 M urea in PBS and PBS, to obtain the desired denaturant concentration, and changes in the CD signal at 231 nm were then followed with time. The time course of unfolding was fit to a single exponential curve, and the difference in the half-life relative to that of the native sequence was reported.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Conformation
The native sequence KGF and analogs (shown in Table I) were purified as previously described from the supernatant of mechanically lysed Escherichia coli cells and hence were never subjected to solubilization and refolding processes. It is therefore likely that the purified proteins were folded into a distinct tertiary structure. However, recovery of the protein varied depending on the mutations, probably due to variations in the expression level and stability. To confirm that all of the proteins studied had similar secondary and tertiary structures, far and near UV CD spectra were determined for the native sequence KGF and the mutants. The far and near UV CD spectra of KGF and D23-KGF are shown in Figure 2; the spectra of the other analogs were similar. They all show a maximum at 230 nm and a minimum at 200 nm with a shoulder in the 210 nm region. These spectra are similar to that seen for other members of the FGF family, with the 230 nm feature being especially sensitive to changes in structure (Chen et al., 1994a; Danilenko et al., 1999). The far UV CD spectra indicate that the analogs all have similar secondary structures, and consist of primarily beta-sheet structures, with no evidence for any alpha-helices. The native sequence and D23 proteins have distinct signals in the near UV region, suggesting that they are folded. Although KGF has very weak signals in this region, deletion of the 23 N-terminal amino acids resulted in reproducible differences in the near UV spectra. All the deletion mutants showed spectra similar to D23-KGF. All the analogs had very similar far and near UV CD spectra, varying within ±10% from the spectrum of the native sequence KGF. This indicates that these mutants are folded into a structure resembling that of the native protein.
|
Conformational stability
In order to test the effect of the different types of mutations on the stability of the KGF, the thermal stability of the analogs was probed. As described previously, KGF aggregates and precipitates upon melting (Chen et al., 1994a,b). Therefore, the thermal stability was determined in several different ways in order to ensure that we were comparing differences in stability rather than simply differences in solubility. The typical DSC profile for KGF consists of an exotherm resulting from the precipitation. The Tm determined from this is fairly independent of protein concentration, suggesting that it is a reliable indication of stability and that unfolding is not driven by aggregation.
Representative DSC profiles, of the native sequence, D23 and R144Q KGF, are shown in Figure 3, while the apparent Tms for all of the analogs are summarized in Table II. The Tm for the native sequence KGF determined by DSC was 57°C, while the apparent Tm of the R11Q analog was increased by 2°. The C(1,15)S showed a 5°C higher melting temperature, while C(1,15,40)S had a 6°C lower melting temperature. All of the N-terminal deletion mutants, D15-KGF, D23-KGF and D27-KGF, exhibited slightly increased melting temperatures relative to the native sequence KGF.
|
|
Thermal stability was also assessed by following changes in the ellipticity of the far UV CD spectra at 230 nm with temperature. The aggregation that results following the thermally induced unfolding of KGF interferes with the determination of the end point of the thermal transition, making the mid-transition temperature difficult to be determined accurately. Therefore the temperature at which the ellipticity begins to change (the onset of melting) was compared; this is 48°C for native sequence KGF. Representative CD melting curves for the native sequence, D23 and R144Q proteins are shown in Figure 4, while the temperature at which unfolding begins obtained from the CD analysis is given in Table II.
|
The absolute increase in the thermal stability is greater by DSC than by CD, most likely due to the differences between onset of melting and apparent Tm, and due to the differences in what changes are being monitored by the two different techniques. The far UV CD can detect small changes in the secondary structure that occur before the protein aggregates. These changes do not appear to cause differences in the energy absorbed by the solution that are detectable by DSC. However, the relative stabilities of the analogs were similar regardless of the technique used. Both techniques indicate that the C(1,15)S and D23-KGF substitutions result in an increase in thermal stability. However, substitution at the C40 or C102 positions resulted in decreased stability, suggesting that both the disulfide bond involving the C102 and the free cysteine at 40 stabilize the protein. Replacement of R144 with a neutral or acidic amino acid, as shown in Table II, dramatically increased the thermal stability of KGF. On the other hand, mutations of other basic amino acids in the potential heparin binding site, such as Lys139 and Lys153, to a neutral or an acidic residue showed little effect on the thermal stability. This suggests that the destabilizing effect of Arg144 due to charge/charge repulsion of basic residues is the most significant among the three basic residues tested. Alternatively, mutation at this particular position could induce a local conformational change, undetectable by CD spectroscopy, which results in enhanced thermal stability. A combination of the removal of the cysteines at positions 1 and 15 and the neutralization of the acidic residue at 144 further increases the stability. Thus these two substitutions appear to be additive, suggesting that they are acting by different mechanisms.
The urea-induced unfolding of KGF is also irreversible; therefore, stability to 5 M urea was assessed by following the decrease in ellipticity at 231 nm with time. Figure 5 shows the time course of the denaturation of the native KGF, the C(1,15)S, and the C(1,15)S/R144E, while the half-life of the mutants relative to that of the native sequence KGF is also shown in Table II. The stability to urea-induced denaturation of all of the Cys analogs examined was equivalent to that of the native sequence KGF, while the R144Q or R144E mutant exhibited greatly reduced rates of urea-induced unfolding, corresponding to an increase in stability.
|
Heparin binding
KGF is known to bind heparin, and its stability against thermal denaturation and storage stability is greatly increased upon heparin binding (Chen et al., 1994a; Danilenko et al., 1999). Heparin binding also mediates receptor binding (Hsu et al., 2000). Heparin binding of these mutants was assessed using heparin chromatography, performed at pH 7.0. The stronger a protein binds to heparin the higher the salt concentration necessary to elute the protein from the heparin column.
The salt concentrations required to elute KGF and mutants from the heparin column, along with the charged state of that analog at pH 7, are listed in Table III. The C(1,15)S and D23-KGF eluted at a salt concentration equivalent to that at which the native sequence KGF elutes, consistent with their approximate charged state of +8 or +9 at pH 7.0. The D15- and D27-mutants exhibited greater heparin binding, most likely reflecting a non-specific electrostatic effect, since their charged states were increased to +10 at pH 7.0. In contrast, all of the R144Q or R144E mutants, independent of the background sequence, showed much lower heparin binding affinity than the native sequence KGF. The charged states of these analogs were only slightly different than the native KGF. These results suggest that residue 144 is directly involved in heparin binding, or alternatively mutation of R144 induces undetectable conformational changes that result in a decreased affinity for heparin.
|
The effect of a 5-fold molar excess of heparin on thermal unfolding was determined by DSC, and the DSC profiles of several analogs in the presence of heparin are shown in Figure 6. The temperature at which thermal denaturation occurs was increased by
5°C by the addition of heparin to the native protein. The analogs involving deletion mutations and the D23 analog were also stabilized similarly by the addition of heparin. Thus the effect of heparin and the removal of the Cys appear to be acting by different mechanisms. The onset of melting of the R144Q and R144E analogs in the presence of heparin was not significantly different from that in the absence of heparin. This indicates no stabilization effect of heparin on these mutants. This is consistent with the decreased binding affinity for heparin demonstrated earlier by the lower concentration of NaCl required to elute this analog from the heparin column.
|
Bioactivity
The dose dependence of the in vitro mitogenic activity of KGF and the mutants was determined using a Balb/MK cell line. The activity varied significantly from one assay to another and hence always was performed in comparison with native KGF and reported as the percentage activity of native protein. The bioactivity is shown in Table IV. The cysteine mutants, C40S, C40A, C102S, C(1,15)S, C(1,15,40)S, C(1,15,102)S, C(1,15,102,106)S and C(1,15,40,102,106)S, and the D15 and D23 analogs, all had activity that was equivalent, within experimental error, to the native sequence KGF. The D27 and H116G mutants showed a decrease in activity, as they exhibited lower thymidine uptake rates over the entire range of protein concentrations tested. The R144E mutation on the C(1,15)S backbone also resulted in decreased activity.
|
Long-term (storage) stability
Next, the kinetics of loss of soluble protein was determined at 37°C. Figure 7 shows the percentage of soluble monomeric KGF remaining as a function of incubation time at 37°C. The half-life for loss of soluble protein was obtained by performing a single exponential decay analysis and is listed in Table II, with representative curves shown in Figure 8. The native sequence KGF showed the fastest loss of protein, with a half-life of about half a day. The half-life of the C(1,15)S and the D23 analogs were increased, by 2-fold, to 1.2 days. C(1,15)S/R144Q and C(1,15)S/R144E lose their monomeric protein with much slower rate constants. The half-lives for loss of soluble protein have been extended to 13 and 38 days, respectively. Compared with the KGF, this represents an increase of 20-fold for C(1,15)S/R144Q and 60-fold for C(1,15)SR/144E. The D15-KGF showed an increase in half-life relative to native KGF of 10-fold when incubated in 100 mM Na phosphate, 140 mM NaCl, pH 7.0 at 37°C (data not shown).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
We have shown here and in the previous papers (Chen et al., 1994a
,b; Chen and Arakawa, 1996) that recombinant native sequence KGF is unstable to long-term storage. Unfolding and aggregation occur during storage due at least in part to disulfide scrambling and formation of dimers. Among the five cysteines in KGF, only Cys40 exists in the free form (Culajay et al., 2000). The presence of this free disulfide could result in disulfide scrambling. However, thermal and storage stability showed that the mutation of this cysteine to alanine greatly reduced the stability of the protein. This suggests that Cys40 is not solvent-exposed and plays an important role in maintaining the structural integrity of the protein.
Conversely, simultaneous mutations of Cys1 and 15 to Ser or the deletion of this N-terminal region of KGF resulted in significantly enhanced stability, in particular against long-term storage. This is surprising, considering the general stabilizing action of disulfide bonds. These mutations eliminate a disulfide bond occurring between these two residues, suggesting that this bond contributes little to the stability of the overall protein fold. It appears to be susceptible to disulfide rearrangement, perhaps interacting with the free Cys40. We have observed that in E.coli in addition to the cleavage of the N-terminal methionine Cys1 is also sometimes cleaved, resulting in a mixture of KGF with and without Cys1 (E. Hsu and W. C. Kenney, unpublished data). Missing Cys1 will leave Cys15 free, and this free SH could then cause disulfide scrambling. The importance of Cys40 was further demonstrated by introducing an additional Cys40 to an Ala mutation into the above C(1,15)S mutant; this resulted in greatly decreased stability against heat, consistent with our observation for the C40A mutation alone. This is similar to the results obtained with aFGF, where mutations of the three cysteines to serine resulted in decreased thermodynamic stability but increased physiological half-life due to the elimination of sulfhydryl reactions (Culajay et al., 2000).
Mutations of Cys102 and Cys106 showed a detrimental effect on the heat stability of the protein. We also observed that the mutation of these two cysteines alone or together caused decreased stability. Removal of both Cys1 and Cys15 by generating deletion mutations lacking the N-terminal region containing these two residues (KGF D15, D23, and D27) resulted in enhanced stability. This is consistent with the view that processing of only Cys1 causes disulfide scrambling and hence aggregation. These three deletion mutants lack the destabilizing Cys1 and Cys15 but retain the structurally important Cys residues at positions 40, 102 and 106. Consistent with these observations, crystallization of the full-length KGF was unsuccessful due to precipitation, even with an additional stabilizing mutation on KGF (Arakawa et al., 1993). On the other hand, the d23 mutant resulted in a high quality crystal, from which the high-resolution crystal structure could be determined (Osslund et al., 1998).
The in vitro bioassay performed here demonstrated that none of these cysteines, or the N-terminal 23 residues, is involved in receptor binding and biological function. This suggests that the structure of the D23 mutant determined by X-ray diffraction should contain all the biologically relevant information. These analogs also bind to a heparin affinity column and show the stabilization to heat-denaturation in the presence of heparin that is seen with native KGF. This suggests that the mutations have not affected heparin binding either.
The largest increase in stability was observed by the neutralizing mutation of Arg144, while no stabilization was observed by mutations of Lys139 or Lys153. These residues are located in the putative heparin binding region, which contains four basic residues. Since we have not solved the structure of D23-KGF without the R144 mutation it is impossible to know how the R144Q mutation impacted the electrostatic interactions in this region. However, the D23/R144Q KGF crystal structure shows that the mutation allows the formation of several hydrogen bonds, which lock the two loops of the protein into a relatively rigid conformation. Hydrogen bonds are formed with Glu93 and Lys146, while Glu93 itself forms hydrogen bonds with Lys92 and 146 (Osslund et al., 1998).
We have observed greatly decreased heparin binding affinity by the R144 mutants, consistent with the prediction from the bFGF structure that these basic residues are important in heparin binding. It has been shown that heparin binding is essential for appropriate interactions between KGF and its receptor (Hsu et al., 2000). The observed decreased activity for the R144E mutant is consistent with this observation. Thus, though this analog is the most stable, it would also be the least effective as a therapeutic. These results suggest that the D23 mutant represents an optimal balance between stability and activity.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The authors thank Ann (Youh-Rong) Hsu for the helpful discussions and Mary Beth Abate and Mariana Dimitrova for the help in preparation of the manuscript and figures.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Ago,H., Kitagawa,Y., Fujishima,A., Matsuura,Y. and Katsube,Y. (1991) J. Biochem., 110, 360–363.
Arakawa,T., Horan,T.P., Narhi,L.O., Rees,D.C., Schiffer,S.G., Hoist,P.L., Prestrelski,S.J., Tsai,L.B. and Fox,G.M. (1993) Protein Eng., 6, 541–546.
auf dem Keller,U., Krampert,M., Kumin,A., Braun,S. and Werner,S. (2004) Eur. J. Cell Biol., 83, 607–612.[CrossRef][Web of Science][Medline]
Bottaro,D.P., Rubin,J.S., Ron,D., Finch,P.W., Florio,C. and Aaronson,S.A. (1990) J. Biol. Chem., 256, 12767–12770.
Chen,B.L. and Arakawa,T. (1996) J. Pharm. Sci., 85, 419–422.[CrossRef][Web of Science][Medline]
Chen,B.L., Arakawa,T., Hsu,E., Narhi,L.O., Tressel,T.J. and Chien,S.L. (1994a) J. Pharm. Sci., 83, 1657–1661.[CrossRef][Web of Science][Medline]
Chen,B.L., Arakawa,T., Morris,F.C., Kenney,W.C., Wells,C.M. and Pitt,C.G. (1994b) Pharm. Res., 11, 1581–1587.[CrossRef][Web of Science][Medline]
Culajay,J.F., Blaber,S.I., Khurana,A. and Blaber,M. (2000) Biochemistry, 38, 7153–7158.
Danilenko,D.M. et al. (1999) Arch. Biochem. Biophysics, 361, 34–46.[CrossRef][Web of Science][Medline]
Eriksson,E.A., Cousens,L.S., Weaver,L.H. and Matthews,B.W. (1991) Proc. Natl Acad. Sci. USA, 88, 3441–3445.
Faham,S., Hileman,R.E., Fromm,J.R., Linhardt,R.J. and Rees,D.C. (1996) Science, 5252, 1116–1120.
Falco,J.P., Taylor,W.G., DiFiore,P.P., Weissman,B.E. and Aaronson,S.A. (1988) Oncogene, 2, 573–578.[Web of Science][Medline]
Farrell,C.L., Rex,K.L., Chen,J.N., Bready,J.V., DiPalma,C.R., Kaufman,S.A., Rattan,A., Scully,S. and Lacey,D.L. (2002) Cell Prolif., 35, 78–85.
Finch,P.W. and Rubin,J.S. (2004) Adv. Cancer Res., 91, 69–136.[Web of Science][Medline]
Hsu,Y.-R., Hsu,E.W.-J., Katta,V., Brankow,D., Tseng,J., Hu,S., Morris,C.F., Kenney,W.C. and Lu,H.S. (2000) Biochemistry, 38, 2523–2534.
Itoh,N. and Ornitz,D.M. (2004) Trends Genet., 20, 563–569.[CrossRef][Web of Science][Medline]
Ornitz,D.M., Xu,J., Colvin,J.S., McEwen,D.M., MacArthur,C.A., Coulier,F., Gao,G. and Goldfarb,M. (1996) J. Biol. Chem., 271, 15292–15297.
Osslund,T.D. et al. (1998) Protein Sci., 7, 1681–1690.[Web of Science][Medline]
Spahr,C.S., Narhi,L.O., Speakman,J., Lu,H.S. and Hsu,Y.-R. (1997) In Marshuk,D.R. (ed.), Techniques in Protein Chemistry VIII, Academic Press, San Diego, CA, pp. 299–308.
Tsai,P.K., Volkin,D.B., Debova,J.B., Thompson,K.C., Bruner,M.W., Gress,J.O, Matuszernska,B., Koegan,M., Bondi,J.V. and Middaugh,C.R. (1993) Pharm. Res., 10, 649–659.[CrossRef][Web of Science][Medline]
Volkin,D.B., Tsai,P.K., Debora,J.M., Bress,J.O., Burke,C.J., Linhardt,R.J. and Middaugh,C.R. (1993) Arch. Biochem. Biophys., 300, 30–41.[CrossRef][Web of Science][Medline]
Wen,J., Hsu,E., Kenney,W.C., Philo,J.S., Morris,C.F. and Arakawa,T. (1996) Arch. Biochem. Biophys., 332, 41–46.[CrossRef][Web of Science][Medline]
Zhang,J., Cousens,L.S., Barr,P. and Sprang,S.R. (1991) Proc. Natl Acad. Sci. USA, 88, 3440–3450.
Zhu,X., Hsu,B.T. and Rees,D.C. (1993) Structure, 1, 27–34.[Medline]
Zhu,X., Komiya,H., Chirino,A., Faham,S., Fox,G.M., Arakawa,T., Hsu,B.T. and Rees,D.G. (1991) Science, 251, 90–93.
Received January 28, 2005; revised December 16, 2005; accepted December 19, 2005.
Edited by Andreas Kungl
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. O. Yildirim, M. Veith, T. Rausch, B. Muller, P. Kilb, L. S. Van Winkle, and H. Fehrenbach Keratinocyte growth factor protects against Clara cell injury induced by naphthalene Eur. Respir. J., September 1, 2008; 32(3): 694 - 704. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Sadovski, T. Kuchenbuch, C. Ruppert, A. Fehrenbach, M. Hirschburger, W. Padberg, A. Gunther, J. M. Hohlfeld, H. Fehrenbach, and V. Grau Keratinocyte growth factor prevents intra-alveolar oedema in experimental lung isografts Eur. Respir. J., January 1, 2008; 31(1): 21 - 28. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. N. Gomperts, J. A. Belperio, M. C. Fishbein, M. P. Keane, M. D. Burdick, and R. M. Strieter Keratinocyte Growth Factor Improves Repair in the Injured Tracheal Epithelium Am. J. Respir. Cell Mol. Biol., July 1, 2007; 37(1): 48 - 56. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||