Published online before print December 4, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Division of Immunobiology, National Institute for Biological Standards and Control, Herts, United Kingdom
1 Correspondence: Division of Imunobiology, NIBSC, Blanch Lane, South Mimms, Potters Bar, Herts, EN6 3QG, UK. E-mail: ccarter{at}nibsc.ac.uk
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: glycosylation • proteases • deglycosylation • granulopoiesis • bioassay
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Naturally occurring G-CSF occurs as a single, 174 amino acid polypeptide chain with two intramolecular disulfide bonds, one free cysteine at residue 17 and one O-linked carbohydrate chain attached to thr 133 [6 , 7 ]. The carbohydrate residue accounts for less than 4% of the molecular mass and has little impact on the three-dimensional structure of the molecule [8 ]. G-CSF has been cloned [9 ] and is available as recombinant protein as a nonglycosylated (NG) protein (expressed in Escherichia coli) and as a glycosylated protein [expressed in yeast or Chinese hamster ovary (CHO) cells]. It has been reported that the biological activity of NG G-CSF is reduced following incubation with serum, and the glycosylated form of the cytokine is relatively resistant to serum inactivation [10 11 12 ]. However, neither the mechanism of the down-regulation nor the reasons for how the glycosylated form is protected is clear.
A recent study reported that human neutrophil-derived elastase (HNE) destroyed G-CSF and directly antagonized the biological activity of G-CSF in vitro [13 ]. The glycosylation state of the G-CSF used in this study was NG. As the natural cytokine is glycosylated and following claims that the glycosylation state of a molecule can influence its susceptibility to proteolytic degradation, it was of interest to compare the relative susceptibility of NG and glycosylated G-CSF to elastase degradation.
The aims of this study were twofold. The stability of glycosylated G-CSF (expressed in CHO cells) after in vitro elastase treatment was investigated, and the role of the sugar residues in any protection observed was assessed. This was performed using two different G-CSF preparations. The carbohydrate residues were enzymatically removed, resulting in deglycosylated (Dg) G-CSF. A mutant G-CSF (amino acids 17 and 21 substituted), which was NG (termed "diet" G-CSF), expressed in yeast cells, was also used. The effect of elastase on G-CSF was measured using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis (to measure degradation) and by bioassay (to assess the effect on the biological activity of the cytokines). As it has been reported that serum reduces the biological activity of NG G-CSF, an elastase inhibitor was used to investigate if HNE was the component present in serum responsible for the decrease in biological activity of NG G-CSF.
The results presented here demonstrate that glycosylated G-CSF is partly protected from attack by HNE and that the sugar residues are critical for this protection. These results and their in vivo significance are discussed. We also show that HNE is not the serum factor responsible for the reduction in G-CSF activity observed following G-CSF incubation in serum.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Incubations of G-CSF and elastase
In routine experiments, G-CSF (5x106 U/ml; 50 µg/ml in 40 µl vol) was incubated with an equal volume of phosphate-buffered saline (PBS; negative control) or HNE (2 µg/ml) at 37°C for times indicated in the text. The G-CSF/elastase was diluted in media before inclusion in a GNFS-60 bioassay. In some experiments, the concentrations of elastase or G-CSF were varied, as indicated, and in others, a HNE inhibitor (MSACK) was included in the reaction at 100 µg/ml. At the completion of the reaction, samples were placed on ice and analyzed by SDS (15%) gel electrophoresis, silver staining was performed as described previously [15
], and the biological activity of G-CSF was assessed in a G-CSF (GNFS-60) bioassay as described previously [12
]. In some experiments, NG G-CSF (1000 U/ml; 100 ng/ml) was incubated in undiluted heat-inactivated human AB serum (90 µl) for 16 h at 37°C in the presence of an elastase inhibitor (HNE II).
GNFS-60 bioassay
The biological activity of G-CSF was assessed in a G-CSF bioassay as described previously [12
]. Briefly, this assay entails making serial dilutions of the samples to be assayed in 96-well plates (in duplicate). GNFS-60 cells were washed to remove exogenous G-CSF and were added to the samples (104/well). The plates were cultured for 24 h at 37°C and 5% CO2, and proliferation was assessed by measuring the incorporation of 3H-thymidine (0.0185 MBq/well) for the final 4 h of culture. Cells were harvested onto filter mats, and radioactivity was incorporated into the DNA measured by scintillation counting. Results were obtained as log G-CSF concentration versus counts per minute (CPM). However, for clarity, the CPM value obtained from 12.5 IU/ml G-CSF is used. In all cases, the CPM values obtained from elastase-treated G-CSF are expressed as a percentage of the values obtained from control (PBS)-treated G-CSF.
Dg of glycosylated G-CSF
Prior to the Dg reaction, aliquots of glycosylated G-CSF were concentrated using an Ultrafree centrifugal filter device with 10 kDa exclusion (Millipore, Bedford, MA). An aliquot of this concentrated G-CSF was Dg, using a panel of enzymes provided in a Dg kit obtained from Calbiochem, and used according to the manufacturer’s instructions, in a reaction volume of 40 µl. This briefly consisted of adding the G-CSF to buffer and adding the relevant enzymes [1 µl of the following enzymes: endo-
-acetylgalactosaminidase (1.25x10-3 U) and 2-3,6,8,9-neuraminidase (5x10-3 U)]. This reaction was incubated for 4 days at 37°C. Following this reaction, the extent of Dg was assessed by analysis on SDS gel (15% acrylamide) and silver staining, and its migration was compared with NG and glycosylated G-CSF (which migrate at different rates on the gel). The degree of Dg was further analyzed using a digoxygenin (DIG) glycan detection kit (Boehringer Mannheim, Germany) used according to the manufacturer’s instructions (protocol B). This technique briefly consists of suspending G-CSF preparations in sodium acetate buffer and oxidizing aldehyde groups and covalently labeling the molecule with DIG. DIG-linked glycoproteins were detected by analysis by enzyme immunoblotting using a DIG-specific antibody conjugated with alkaline phosphatase. The biological activity of the Dg G-CSF was assessed in a GNFS-60 bioassay so that where possible, equivalent activities of glycosylated and Dg G-CSF were used in experiments.
Statistical analysis
Statistical analysis was performed using a Student’s paired t-test.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A band of 18 kDa (the predicted size for NG G-CSF) was observed in a SDS gel following the incubation of NG G-CSF with PBS. However, following incubation of NG G-CSF with elastase (2 µg/ml), the 18-kDa band was visibly less intense, suggesting that the elastase had degraded the G-CSF (Fig. 1a ). Indeed, smaller protein bands were visible from elastase-treated G-CSF, suggesting fragmentation of the cytokine. The inclusion of a specific HNE inhibitor (elastase inhibitor II) resulted in the 18-kDa band being observed at the appropriate place on the gel. Inclusion of a dimethyl sulfoxide (DMSO) vehicle control had no effect on the elastase degradation of the G-CSF (Fig. 1a) . This suggested that elastase was responsible for the destruction of G-CSF. Glycosylated G-CSF was essentially unaffected by the incubation with elastase (Fig. 1a) , and an 18-kDa band of similar intensity was present following incubation with PBS and elastase.
 View larger version (36K): [in a new window] |
Figure 1. NG G-CSF was degraded by elastase, but glycosylated G-CSF was resistant to elastase degradation. NG and glycosylated G-CSF (5x106 U/ml) were incubated with PBS or HNE (2 µg/ml) for 45 min at 37°C, and samples were analyzed for the presence of G-CSF by SDS gel and silver staining (a), and the biological activity was assessed by bioassay (b). In some experiments, the effect of the inclusion of an elastase inhibitor (100 µg/ml) or DMSO was analyzed by gel electrophoresis (a) or bioassay (c). The results (b and c) are expressed as the % activity compared with the control levels (G-CSF incubated with PBS) and are the mean of three experiments (+1 SE). Statistical analysis was performed by paired t-test; *, significant samples. See Key for abbreviation definitions; Elast, elastase (b and c).
|
60% upon incubation with elastase compared with the values obtained from G-CSF incubated with PBS (Fig. 1b)
. This was found to be a statistically significant reduction in biological activity (P<0.05). The values obtained with glycosylated G-CSF were similar from both PBS and elastase treatments. Inclusion of the HNE inhibitor in the NG G-CSF reaction with elastase showed G-CSF activity at a similar level as that observed with PBS (Fig. 1c)
, and the inclusion of DMSO vehicle control had no effect on the ability of elastase to decrease the biological activity of G-CSF. These results strongly suggested that the incubation of Ng G-CSF with elastase led to reduced biological activity of the cytokine and that glycosylated G-CSF was protected from elastase destruction.
Kinetics of elastase degradation of NG G-CSF
To elucidate the kinetics of elastase activity, NG and glycosylated G-CSF were incubated with PBS or elastase for 15, 30, 60, or 120 min and were analyzed by bioassay and gel electrophoresis. At 15 min, the biological activity obtained from NG G-CSF incubated with elastase was reduced compared with the PBS control (34% of the control activity lost). At 30 min and 1 h, this reduction in activity was statistically significant (68% and 78% of activity lost), and at 2 h, virtually all (95%) biological activity was lost (Fig. 2a
). In contrast, the biological activity of glycosylated G-CSF was not affected by incubation with elastase at any time-point tested. The gel electrophoresis results confirmed that NG G-CSF was degraded by elastase in a time-dependent manner, and glycosylated G-CSF was resistant at all time-points tested (not shown).
 View larger version (33K): [in a new window] |
Figure 2. Elastase degradation of NG G-CSF was time- and concentration-dependent. NG and glycosylated G-CSF (5x106 U/ml) were incubated with PBS or elastase (1 µg/ml), samples were taken at time-points, and the biological activity was analyzed by bioassay (a). In a separate series of experiments, NG and glycosylated (Gly) G-CSF were incubated with PBS or decreasing concentrations of elastase for 2 h (b) or overnight (c), and the activity of GSF was analzyed by bioassay. In all cases, the results are expressed as % activity compared with the control levels (G-CSF incubated with PBS) and are the mean (+1 SE) of three experiments. *, Statistical significance (P<0.05).
|
Using the complimentary techniques of SDS-PAGE (measuring cytokine proteolysis) and bioassay (measuring the biological activity of the cytokine), the cumulative data strongly suggested that NG G-CSF showed degradation and a reduction in biological activity following exposure to elastase. Exposure of glycosylated G-CSF to elastase also degraded the cytokine (leading to decreased biological activity), although this occurred at a much slower rate than that observed with NG G-CSF, implying that the glycosylated molecule was partially protected against elastase activity.
Use of diet and Dg G-CSF
To determine the significance of glycosylation in the resistance of glycosylated G-CSF to elastase degradation, we took advantage of two different preparations of G-CSF. Glycosylated G-CSF was enzymatically Dg, and a mutated form of G-CSF expressed in yeast, but otherwise, NG was used. Both of these preparations were analyzed by SDS-PAGE, where we took advantage of differential migration between NG and glycosylated G-CSF. We found that Dg and diet G-CSF were NG, as they both migrated at a similar speed to NG G-CSF (Fig. 3a
). This was further confirmed using a DIG glycan detection kit, which measures carbohydrate residues present on glycoproteins. Figure 3b
shows a representative example of the DIG analysis of the Dg G-CSF. Although the glycosylated G-CSF is clearly seen, neither NG nor Dg G-CSF reacts in this system, suggesting that the Dg G-CSF is fully Dg. To test the significance of glycosylation on G-CSF molecules, diet and Dg G-CSF were included in elastase degradation assays.
 View larger version (53K): [in a new window] |
Figure 3. Dg and diet G-CSF was NG. Dg G-CSF was generated by enzymatic removal of sugar residues. SDS gel electrophoresis and silver staining analyzed samples of this, along with NG, diet, and glycosylated (Gly). The degree of glycosylation was also assessed by comparing rates of migration in the gel. Samples were also analyzed by DIG analysis (b). Deglyc, Dg.
|
 View larger version (43K): [in a new window] |
Figure 4. Elastase degraded NG and diet G-CSF, but Dg and glycosylated G-CSF were resistant to elastase attack. PBS or elastase (Elast) were incubated with NG, diet, Dg, and glycosylated (Gly) G-CSF (5x106 U/ml) for 2 h, and SDS gel electrophoresis and silver staining assessed the presence of G-CSF (a). Bioassay assessed the biological activity of samples (b). The results are expressed as % activity compared with the control levels (G-CSF incubated with PBS) and are the mean of three experiments (+1 SE). *, Statistically significant reductions in biological activity.
|
 View larger version (18K): [in a new window] |
Figure 5. Diluted G-CSF preparations showed that elastase differentially degraded NG, diet, Dg, and glycosylated (Gly) G-CSF. G-CSF preparations were diluted to 1000 U/ml and were treated with PBS or elastase (overnight), and residual G-CSF activity was assessed by bioassay. Results are expressed as % activity compared with values obtained from the PBS control and are the mean of three experiments (+1 SE). *, Statistically significant reductions in biological activity.
|
In agreement with previous findings, it was observed that incubation of NG G-CSF with serum led to decreased biological activity. The inclusion of the elastase inhibitor had little impact on serum inhibition observed in NG G-CSF (Fig. 6 ). Inclusion of DMSO was also included as a vehicle control, and this also had little impact on serum inactivation of G-CSF. The effect of elastase inhibitor on elastase degradation of NG G-CSF has been demonstrated in an earlier experiment (see Fig. 1c ) to ensure that the elastase inhibitors were active and were being used appropriately.
 View larger version (13K): [in a new window] |
Figure 6. Elastase was not the serum component responsible for the degradation of NG G-CSF. NG G-CSF which (1000 U/ml) was incubated in PBS or human serum (Ser) in the presence of elastase inhibitor (Inh) or DMSO vehicle control for 16 h at 37°C, and residual G-CSF activity was assessed using a GNFS-60 bioassay. Results are expressed as % activity compared with values obtained from the PBS control and are the mean of three experiments (+1 SE). *, Statistically significant reductions in biological activity.
|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
To further elucidate the role of sugar residues on G-CSF protection against elastase degradation, we took advantage of two NG G-CSF preparations (diet G-CSF and enzymatically Dg G-CSF). In elastase degradation studies, the results using diet G-CSF were easy to interpret. As the molecule lacked carbohydrate residues, it was found to be consistently susceptible to elastase degradation to a degree similar to that seen with the NG cytokine, suggesting that glycosylation was a significant factor in protecting the molecule against protease attack.
The situation with the enzymatically Dg G-CSF was a little confusing, as in initial experiments, it seemed to demonstrate resistance to elastase degradation, which contradicted the results obtained using diet G-CSF. This matter was resolved by the observation that the elastase enzyme was inactive in the buffers required for the Dg reaction. The explanation for this was not clear, although it could relate to changes in pH or salt concentrations. As Dg reactions were performed in small volumes, it proved difficult to change the buffer to a more suitable one, and the problem was overcome by using higher dilutions of cytokines in PBS, thereby placing the elastase enzyme in an environment in which it was active. Following this modification, we consistently found that Dg and diet G-CSF were degraded following incubations in elastase. It was noted with interest that the Dg G-CSF was still more resistant to elastase than diet G-CSF. This could relate to the buffer still being less than optimal for elastase activity or to incomplete removal of the sugar residues during the Dg reaction. However, the Dg G-CSF was associated with decreased biological activity of the cytokine following incubation with elastase compared with glycosylated G-CSF. Collectively, this therefore strongly suggested that the sugar residues present on glycosylated G-CSF were critical in its protection against elastase-induced degradation.
These results are important, as the significance of carbohydrate residues on G-CSF in protection against elasatase degradation has not been previously reported. However, glycosylation has been described to increase the stability of a number of molecules [16
, 17
] including the increased resistance to protease attack [18
]. There are also a number of reports suggesting that proteases are capable of cleaving a large number of biologically relevant molecules [19
, 20
]. Although protease enzymes have been shown to degrade a number of cytokines such as tumor necrosis factor (TNF-), interleukin-8 (IL-8), and IL-2 [21
22
23
24
25
], the significance of the role of glycosylation in these cases has not been examined or is not relevant. In the case of IL-6, glycosylation of the cytokine has little impact on its degradation by protease enzymes such as elastase. It has been shown that NG and glycosylated IL-6 are equally susceptible to elastase degradation [26
, 27
]. However, with interferon-
(IFN-), it has been demonstrated that the glycosylation state of the cytokine does have significance in its protection from proteases. N-glycosylation of IFN- has been shown to be critical for its protection against elastase degradation and that glycans at Asn-25 were critical for protease resistance [28
]. It was thought that in the case of IFN-, the sugar residues might cover a large surface area of the IFN- dimer and thus mask the parts of the molecule most susceptible to protease cleavage. This seems to be the most likely scenario with G-CSF, although it should be noted that it is not a highly glycosylated molecule, and the sugar residues are estimated to take up only 4% mass of the molecule. In the case of G-CSF, previous studies suggested that the glycosylation of the molecule may improve the physical stability of the molecule. Furthermore, a recent nuclear magnetic resonance study concluded that the increased stability displayed by glycosylated G-CSF was a result of decreased mobility around the glycosylation site [29
].
It has been recently suggested that human leukocyte elastase, cathepsin G, and proteinase 3 may play a role in extracellular proteolytic processes at sites of inflammation as well as the intralysosomal degradation of cell debris or microorganisms [30
]. Indeed, there has been some interest in the possible role of neutrophil enzymes in the control of cytokine bioactivity through proteolytic interactions. This could be negatively regulating inflammatory cytokines such as IL-6, TNF-, and IL-2 [21
, 22
, 25
26
27
] by causing their destruction or positively regulating cytokines through the cleavage of the active molecule from inactive precursor molecules [31
32
33
34
]. In addition, it has been suggested that neutrophil enzymes can have an enzyme-specific, clear regulatory effect with regard to active C–X–C chemokines [23
].
The fact that glycosylated G-CSF (and presumably, naturally occurring G-CSF) was degraded by HNE, at a much slower rate than NG G-CSF, suggests that elastase may have a role in an in vivo setting. G-CSF has its most significant impact on hematopoiesis in the BM [35 ], where it is likely that it will be effective in the ng/ml range. Although we have demonstrated that high concentrations (5x106 U/ml) of glycosylated G-CSF were resistant to HNE activity during 2-h incubations, lower concentrations of glycosylated G-CSF (1000 U/ml) were degraded during prolonged exposure the enzyme. This suggests that glycosylated G-CSF was degraded by HNE but at a slower rate than occurs with NG G-CSF. It is important to note recent findings demonstrating that the hematopoietic environment in the BM undergoes a dramatic change following treatment with G-CSF. In particular, neutophil elastase and cathepsin G were found to be present at high concentrations and were found to cleave a number of physiologically relevant molecules such as vascular cell adhesion molecule-1, stromal cell-derived factor-1, and its receptor, CXC chemokine receptor 4 [36 37 38 ]. The possibility of elastase degrading glycosylated G-CSF clearly depends on the relative concentrations of cytokine and enzymes as well as other factors, including the length of exposure and the presence of enzymatic inhibitors in the local environment. This underlines the complex nature of the reactions between cytokines and protolyic enzymes [39 ].
It is well documented that the biological activity of NG G-CSF is reduced following exposure to human serum. The second aim of this study was to investigate whether the serum component responsible for the degradation of G-CSF was elastase. The possible role of the carbohydrate residues on G-CSF has been subject to speculation [7
, 40
, 41
]. Previous studies suggested that glycosylated G-CSF but not NG G-CSF was protected from serum inactivation, and preliminary evidence from our laboratory indicates that serum inactivated NG G-CSF by a protease-dependent mechanism. The mechanism of serum-induced inhibition/inactivation has yet to be fully elucidated, although a number of explanations have been proposed, and this question remains a focus of study in our laboratory. Mire-Sluis [12
] suggested G-CSF binding to the serum protein 2M to be responsible for the reduction in biological activity. Other suggestions [10
, 11
] have included protease destruction of G-CSF. As it has now been reported that elastase is able to degrade NG G-CSF, and we have shown that glycosylated G-CSF is resistant to elastase degradation, it is possible that neutrophil-derived elastase is the serum component responsible for the reduction in biological activity of NG G-CSF.
An elastase-specific inhibitor [42
] was used to observe if it would block the reduction in NG G-CSF following incubation with serum. It was a consistent finding in a number of experiments that the elastase inhibitor had little impact on serum inactivation of NG G-CSF. Furthermore, we have also found that elastase and serum were different in their ability to degrade diet G-CSF. It was consistently found that elastase was able to degrade diet G-CSF in a manner similar to NG G-CSF, whereas diet G-CSF was resistant to serum inactivation. This difference in activity between serum and elastase remains a focus of work within our laboratory, and it is strongly suggestive (along with the inhibitor data) that elastase is not the serum component responsible for the degradation of NG G-CSF. This finding is not surprising, as it is very likely that any elastase present in serum would be irreversibly bound by 1-antitrypsin, although this does not exclude other proteases being involved.
In conclusion, this study confirms the previous finding that NG G-CSF was susceptible to elastase degradation that led to decreased biological activity of the cytokine. Glycosylated G-CSF was degraded by HNE, but the enzymatic destruction of this G-CSF occurred at a slower speed. We investigated the potential role of glycosylation in the protection against elastase attack by using a mutant form of G-CSF and also by using enzymatically treated glycsosylated G-CSF, both of which were shown to be NG. We concluded from these experiments that the sugar residues present on glycosylated G-CSF were critical for protection against protease degradation.
Received August 12, 2003; revised October 11, 2003; accepted October 14, 2003.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
by proteolytic enzymes released from stimulated human neutrophils J. Immunol. 147,3862-3893[Abstract]: glycans at Asn-25 are critical for protease resistance Biochem. J. 308,9-14 J. Biol. Chem. 270,23688-23692 in arterioles: cell surface processing of 1st precursor by elastases Mol. Cell. Biol. 10,4596-4602This article has been cited by other articles:
 |
|
 |
|
  H. Mellstedt, D. Niederwieser, and H. Ludwig The challenge of biosimilars Ann. Onc., March 1, 2008; 19(3): 411 - 419. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. M. Goselink, P. S. Hiemstra, P. van Noort, R. M.Y. Barge, R. Willemze, and J.H. F. Falkenburg Cytokine-Dependent Proliferation of Human CD34+ Progenitor Cells in the Absence of Serum Is Suppressed by Their Progeny's Production of Serine Proteinases Stem Cells, February 1, 2006; 24(2): 299 - 306. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
