Article Text


Extended report
Granulocyte colony stimulating factor exacerbates antineutrophil cytoplasmic antibody vasculitis
  1. Simon J Freeley1,
  2. Alice M Coughlan1,
  3. Reena J Popat1,
  4. Deborah K Dunn-Walters2,
  5. Michael G Robson1
  1. 1Medical Research Council (MRC) Centre for Transplantation, King's College London, London, UK
  2. 2Department of Immunobiology, King's College London, London, UK
  1. Correspondence to Dr Michael Robson, King's College London, King's Health Partners, Medical Research Council (MRC) Centre for Transplantation, Guy's Hospital, London SE1 9RT, UK; Michael.robson{at}


Objectives Granulocyte colony stimulating factor (GCSF) is important in mobilising neutrophils from the bone marrow but also has a range of proinflammatory effects. We therefore decided to investigate the role of GCSF in antineutrophil cytoplasmic antibody (ANCA) vasculitis.

Methods We measured GCSF levels in the serum of 38 patients with active ANCA vasculitis compared with 31 age-matched controls, and assessed the effect of GCSF priming on the response of human neutrophils to ANCA. We also examined the effect of exogenous GCSF administration in a murine model of antimyeloperoxidase (anti-MPO) vasculitis, and the effect of GCSF on murine neutrophil activation.

Results The serum levels of GCSF in patients with active ANCA vasculitis were significantly higher than those of age matched healthy controls (mean 38.04 vs 18.35 pg/ml, p<0.001). Furthermore, we demonstrated that GCSF primed human neutrophils in vitro for a respiratory burst in response to anti-MPO ANCA. In an anti-MPO antibody transfer model, mice given GCSF had more crescents (mean 29.1% vs 5.8% per glomerular cross section, p<0.05), more macrophages (mean 3.2 vs 1.2 per glomerular cross-section, p<0.01), higher serum creatines (mean 13.6 vs 8.3 μmol/l, p<0.05) and more haematuria (p<0.05) compared with controls. In vivo administration of GCSF with lipopolysaccharide (LPS), but not LPS alone, led to upregulation of CD11c on murine neutrophils.

Conclusions These data suggest that GCSF, which is raised in patient serum, may play an important role in exacerbating disease in ANCA vasculitis. In addition, GCSF therapy for neutropenia should be used with caution in these patients.

  • Autoimmune Diseases
  • Autoantibodies
  • Inflammation
View Full Text

Statistics from


Antineutrophil cytoplasmic antibody (ANCA) vasculitis is a severe systemic disease which affects joints, lungs, kidneys, skin and other tissues. It occurs most often in older adults with a peak incidence at the age of 65–74 in the UK.1 ANCA were first described in patients with focal necrotising glomerulonephritis in 1982,2 with myeloperoxidase (MPO) and proteinase 3 (PR3) subsequently shown to be the antigenic targets responsible for the perinuclear and cytoplasmic staining patterns, respectively.3 ,4 Seminal observations 20 years ago showed that ANCA can activate cytokine-primed neutrophils, and several studies have confirmed and extended these findings.5 Despite this large body of in vitro work, evidence that ANCA are pathogenic in vivo was only relatively recently obtained. Anti-MPO antibodies raised by immunising MPO-deficient mice with murine MPO caused a focal necrotising crescentic glomerulonephritis when injected into wild type mice.6

Several observations show that granulocyte colony stimulating factor (GCSF) has inflammatory effects in addition to mobilising neutrophils from the bone marrow. The largest body of evidence comes from studies examining the phenotype of neutrophils in the blood of patients or healthy volunteers who have been administered GCSF. Demonstrated effects of in vivo activation with GCSF include enhancement of phagocytosis, superoxide generation, bacterial killing and heightened responses to N-formyl-methionyl-leucyl-phenylalanine (fMLP), zymosan and C5a (reviewed in depth in reference7). Effects on CD11b, CD11c and Fc receptor expression have also been demonstrated. Some studies have also examined the effect of GCSF treatment in vitro using isolated human neutrophils with an increased respiratory burst induced by fMLP shown in some reports8 but not in others,9 and CD18/CD11b upregulation being demonstrated.10 GCSF levels are raised in the serum of patients with rheumatoid arthritis11 and there are also data to show the importance of GCSF in vivo using animal models of arthritis.12 In addition, GCSF-deficient mice were protected from heterologous nephrotoxic nephritis, a model of acute neutrophil-mediated glomerular infammation.13 In view of these proinflammatory effects, we decided to investigate the role of GCSF in ANCA vasculitis.


Measurement of GCSF in serum samples

Serum from patients or controls was stored at −80°C until analysis. It was analysed according to the manufacturer's instructions using an ultrasensitive chemiluminescence assay (Meso Scale Discovery, Gaithersburg, Maryland, USA). All blood samples were taken with informed consent and ethical approval (NRES committee London—London Bridge 09/H084/72).

Human neutrophil assays

Blood from healthy volunteers was taken into lithium heparin tubes and PBMCs separated with Ficoll-Paque (GE Healthcare). The pellet was treated with ammonium chloride buffer to lyse red cells, washed with ice-cold Hank's balanced salt solution (HBSS) and finally resuspended in ice-cold HBSS with 1 mM Hepes. For the dihydrorhodamine 123 (dihydrorhodamine (DHR) 123) assay, neutrophils were resuspended at 2.5×106 cells/ml and loaded with 17 μg/ml DHR123 (Calbiochem, Nottingham, UK). Sodium azide 2 mM and Cytochalasin B (Sigma) 5 μg/ml were added and incubated for 10 min at 37°C. Cells were primed with 2 ng/ml tumour necrosis factor (TNF)α or 50 ng/ml GCSF (both from Peprotech, London, UK) for 15 min at 37°C, followed by stimulation with one of the following monoclonal antibodies: 5 μg/ml anti-PR3, Clone WGM2 (Hycult Biotech, Uden, The Netherlands), 2.5 μg/ml anti-MPO, clone 266.6K2 (IQ Products, Groningen, The Netherlands) or isotype control IgG1, clone 107.3 (Becton Dickinson, Oxford, UK) for 60 min at 37°C. As a positive control, cells were also stimulated with fMLP (Sigma) at 1.25 or 5 μg/ml for 10 min. The reaction was stopped by adding a 30-fold volume of cold HBSS containing 1% bovine serum albumin (BSA) followed by centrifugation and resuspension in a 300 μl volume for analysis. To assess MPO or PR3 expression, neutrophils (prepared as above) were stained for PR3 with FITC-PR3G-2 (Hycult) or for MPO with APC-MPO-7 (DAK0, Cambridge, UK). FITC or APC-MPOC-21 (Biolegend) was used as an isotype control, with median fluorescence intensities subtracted. Flow cytometry was performed using a FACS Canto with FACSDiva software (Becton Dickinson); a minimum of 10 000 events were collected per sample and data were analysed using FlowJo software (Treestar, Ashland, Oregon, USA).

Induction of disease

All animal experiments were performed according to UK home office and local regulations. MPO purification and immunisation of MPO−/− mice to generate anti-MPO serum are described in the online supplementary material. C57BL/6 mice were purchased from Harlan (Bicester, Oxon, UK). Mice were bled from the saphenous vein for assessment of baseline neutrophil numbers and then given daily subcutaneous injections of 6 µg GCSF (Filgastrin, Neupogen, Amgen, Cambridge, UK) which continued until the mice were killed. Five days after starting GCSF, mice were bled again from the saphenous vein prior to intravenous injection with anti-MPO serum via the tail vein. After 1 h, and again after 3 days, the mice were then given an intraperitoneal injection of 10 µg of lipopolysaccharide (LPS) from Escherichia coli, Serotype R515 (Enzo Life Sciences, Exeter, UK). Mice were killed 7 days after anti-MPO injection after having been placed in metabolic cages for urine collection the previous day.

Assessment of disease

Haematuria was analysed by dipstick (Combur-Test from Roche, Sussex, UK). Albuminuria was quantified by ELISA according to the manufacturer's instructions (Bethyl Laboratories, Montgomery, Texas, USA). Serum creatine was measured with electrospray mass spectroscopy as described.14 Kidney was fixed in Bouin's solution, paraffin embedded and stained with periodic acid-Schiff reagent. Crescents were assessed in at least 100 glomerular cross-sections per sample. Kidney was also fixed in phosphate-lysine-periodate prior to freezing in isopentane and cryosectioning. Immunofluorescence staining for CD68 was with clone FA11 (Serotec, Abingdon, Oxon, UK) and DyLight 488 mouse antirat IgG (Jacksons Immunoresearch).

Mobilisation and activation of mouse neutrophils

Blood was taken at times specified above for induction of disease. Total leukocyte counts were obtained after diluting whole blood in Turk's solution to lyse red cells (Merck, Nottingham, UK) and absolute numbers of neutrophils were calculated from percentage of Ly-6G positive cells and total leukocyte numbers. In an additional experiment to assess activation markers, mice were bled from the saphenous vein 4 days after injection with 30 μg human pegylated GCSF (Neulasta, from Amgen, Cambridge, UK). They were then injected intraperitoneally with PBS control or with 10 µg of LPS from E coli, Serotype R515 (Enzo Life Sciences), and bled again from the saphenous vein 2 h later. The following antibodies were used for flow cytometry: AlexaFluor 700 anti-mouse Ly-6G (Clone:1A8), FITC antimouse CD11b (Clone:M1/70), phycoerythrin (PE) antimouse CD11c (Clone:HL3) from Becton Dickinson and Pacific Blue antimouse CD62L (Clone: MEL-14) from Biolegend. Whole blood was stained with the appropriate antibodies in the dark for 20 min at room temperature. Red cells were lysed using BD FACS Lysing Solution (Becton Dickinson) according to the manufacturer's instructions. Flow cytometry was performed on a FACS Canto flow cytometer using FACSDiva software (Becton Dickinson). A minimum of 10 000 events were collected per sample and data were analysed using FlowJo software (Treestar, Ashland, Oregon, USA). Neutrophils were identified as Ly6G+ cells to allow the assessment of expression levels.


Statistics were performed using Graphpad Prism software (Graphpad Software Inc, La Jolla, California, USA) using the following tests. A Student t test was used for figure 1A, and a paired t test for figure 1C. For figure 2, a Student t test was used. In figure 3, a one-way analysis of variance (ANOVA) with Dunnett's post test was used. Data were logarithmically transformed before analysis in some cases. Pearson's correlation coefficient was used for correlations in the murine model.

Figure 1

(A) Granulocyte colony stimulating factor (GCSF) levels in patients with active antineutrophil cytoplasmic antibody (ANCA) vasculitis compared with controls. Clinical details for these patients are shown in online supplementary table 1. (B) Representative histogram of an experiment showing the effect of GCSF priming on the respiratory burst in human neutrophils (measured by the dihydrorhodamine 123 assay) induced by antimyeloperoxidase (anti-MPO) antibody. Each symbol is data from a separate person. (C and D) Data from experiments including eight healthy neutrophil donors, with each symbol corresponding to a different donor. (D) There is a significantly increased anti-MPO response with GCSF. Arbitrary units were obtained from the median fluorescence intensities (MFI) as indicated in the figure, with data normalised for the MFI of unstained, unstimulated cells in each experiment. (D) The response to antiproteinase 3 (PR3) antibody was assessed at the same time with neutrophils from the same donors. There was no significant increase in the response with GCSF priming. In all experiments shown (in C and D) TNFα was used as a positive control, and this primed for a response to MPO-ANCA and PR3-ANCA for all eight donors as expected. (E and F) MPO and PR3 expression after incubation with control or GCSF in three neutrophil donors who were selected on the basis of a strong priming effect of GCSF on the MPO-ANCA response. *p<0.05 ***p<0.001.

Figure 2

Granulocyte colony stimulating factor (GCSF) exacerbates disease in a murine model of antimyeloperoxidase (anti-MPO) vasculitis. (A and B) Peripheral blood neurophil counts at baseline or after 4 days of GCSF or control prior to injection of anti-MPO antibody. (C) Representative flow cytometry plots showing the increase in Ly6G+ cells in peripheral blood after GCSF. (D and E) Seven days after disease induction with anti-MPO antibody and lipopolysaccharide (LPS), mice given GCSF have significantly more disease with increased glomerular crescents and macrophages. (F) Representative light microscopy on periodic acid-Schiff stained sections. (G) Representative immunofluorescence staining showing CD68 macrophages. (H) There was no difference in albuminuria between groups. (I and J) Serum creatine and haematuria were significantly increased in the group of mice given GCSF compared with controls. Each symbol is data from a separate mouse. *p<0.05, **p<0.01. This figure is only reproduced in colour in the online version.

Figure 3

Mouse neutrophil activation in vivo. Mice were treated with granulocyte colony stimulating factor (GCSF) or control for 4 days and then given lipopolysaccharide (LPS) or control with blood taken 2 h later. The groups included were as indicated in the figure, and expression on Ly6G+ neutrophils of CD62L (A and B), CD11b (C and D) and CD11c (E and F) was assessed. Expression levels were obtained after gating on Ly6G+ cells (as shown in figure 2), and we show data from individual mice and a representative flow cytometry plot for each marker. Each symbol is data from a separate mouse. *p<0.05.


GCSF is raised in the serum of patients with active ANCA vasculitis

We first measured GCSF in the serum of 38 patients with active ANCA vasculitis and compared this with 31 age-matched controls. Patients had a mean age of 66.6 and comprised 23 female and 15 male subjects. All patients had renal disease with clinical details shown in online supplementary table 1. The control group comprised serum collected for a study of immune responses from 31 healthy volunteers with a mean age of 69.2 with 18 females and 12 male subjects, with the gender of one serum donor unknown. As shown in figure 1A, GCSF levels were significantly higher in patients with a mean of 38.04 pg/ml compared with 18.35 pg/ml (p<0.001). There was no correlation between GCSF levels and MPO/PR3 positivity, prior steroid treatment, Birmingham vasculitis activity score (BVAS) score or the presence of extra renal disease. This shows that circulating GCSF levels are raised in the acute phase of ANCA vasculitis.

GCSF primes human neutrophils for a response to anti-MPO ANCA

In view of the previous data showing that GCSF could activate human neutrophils and the fact that levels were raised in the serum of patients with active ANCA vasculitis, we considered whether GCSF could prime neutrophils for a response to ANCA. Isolated neutrophils were primed with 50 ng/ml GCSF and the respiratory burst in response to MPO or PR3-ANCA was assessed in a dihydrorhodamine 123 assay. The dose was chosen based on preliminary dose response studies. We assessed eight neutrophil donors and an effect of GCSF priming was seen in some but not all cases. In all experiments, 2 ng/ml TNFα was used as a positive control, and this primed for a response to MPO-ANCA and PR3-ANCA for all eight donors as expected. Overall there was a significantly higher response to MPO-ANCA after GCSF priming than with no priming (p<0.05). A representative histogram is shown in figure 1B with data from all eight experiments in figure 1C. In each of these experiments, the response to PR3 ANCA was assessed in parallel. As shown in figure 1D, GCSF priming did not increase the response to PR3 ANCA and the effect seemed to be specific to MPO-ANCA. GCSF did not increase expression of MPO or PR3 on the neutrophil plasma membrane, as shown in figure 1E,F, although TNFα increased the expression of both MPO and PR3 in all three donors as expected. These data suggest that, at least in anti-MPO vasculitis, increased circulating GCSF levels may exacerbate disease by priming neutrophils for activation by ANCA.

GCSF exacerbates disease in murine anti-MPO vasculitis

To further explore the in vivo relevance of GCSF in ANCA vasculitis, we used a murine model in which crescentic glomerulonephritis is induced by passive transfer of anti-MPO antibody raised in MPO-deficient mice. Mice were given GCSF or control prior to disease induction as described in the methods. As expected, GCSF administration led to an increase in circulating neutrophils as shown in figure 2A,B, with representative FACS plots in figure 2C. Glomerulonephritis was assessed histologically and functionally at day 7 after injection of anti-MPO antibody, and disease was significantly greater in mice given GCSF than in controls. There were significantly more crescents per glomerular cross-section in mice given GCSF with a mean of 29.1 compared with 5.8 in controls (p<0.01) as shown in figure 2D. There were also more CD68 macrophages per glomerular cross-section with a mean of 3.2 compared with 1.2 (p<0.01) in controls as shown in figure 2E. Representative light microscopy and immunofluorescence staining are shown in figure 2F,G. We also assessed functional parameters. There was no significant difference in albuminuria as shown in figure 2H, though levels were above normal levels for adult C57BL/6 (mean obtained from eight mice is shown by the dotted line). Haematuria and serum creatine were both significantly increased (p<0.05 for both) in mice that received GCSF as shown in figure 2I,J. The mean serum creatine was 13.6 μmol/l in GCSF treated mice compared with 8.3 in controls. When data from both groups were combined, there were significant correlations between glomerular macrophages and crescents (r2=0.8336, p=0.0002), macrophages and haematuria (r2=0.7593, p=0.001), and crescents and haematuria (r2=0.6423, p=0.0053). These data show that GCSF administration exacerbates disease in murine ANCA vasculitis, as assessed by both histological and functional parameters.

GCSF activates murine neutrophils in vivo

In order to understand the effect of GCSF administration in the context of our model of anti-MPO vasculitis we treated mice with GCSF starting 4 days prior to LPS administration. All mice then received LPS or phosphate buffered saline (PBS) with neutrophil activation assessed 2 h after LPS administration. The experimental groups were therefore the same as in the anti-MPO model with the addition of a group given control (PBS) instead of GCSF alone. The experiment was designed to assess the combined effect of GCSF and LPS on neutrophil activation, as in the anti-MPO experiment. As shown in figure 3A,B, the neutrophils of mice treated with LPS alone, or GCSF and LPS, shed significantly more CD62L (L-selectin) than neutrophils of control mice (p<0.05 for both). As shown in figure 3C,D, CD11b was upregulated in mice given LPS alone compared with control mice (p<0.05). In mice given GCSF and LPS, there was a trend towards increased CD11b expression which did not reach significance. As shown in figure 3E,F, CD11c (LFA1) was upregulated in mice given GCSF and LPS (p<0.05) but not in mice given only LPS. In mice given only LPS, CD11c levels were no different from that seen in control treated mice, showing that this upregulation was due to the effect of GCSF.


We have shown that GCSF is raised in the serum of patients with active ANCA vasculitis compared with controls. Longitudinal studies will be needed to assess the rate at which GCSF levels fall as patients enter remission, and this will be the subject of future work. A case study previously reported a high GCSF level in one patient with active GPA15 although another paper actually reported lower levels in 19 patients with active ANCA vasculitis compared with six controls, who were not aged matched and probably younger than the patients.16 This study also differed from ours in the use of an ELISA from R&D systems. We tested this assay and found that it was not sensitive enough to measure GCSF in most of our samples, which could also explain the different results. Although we showed that GCSF primed human neutrophils for a response to anti-MPO ANCA we did not demonstrate this for PR3-ANCA. In addition, the murine model used is an anti-MPO model and so we cannot be sure if our findings are relevant only to patients with MPO-ANCA. GCSF did not increase expression of MPO or PR3 on the neutrophil surface, confirming data from a previous report,17 and showing that this was not the mechanism responsible for the priming effect of GCSF seen with MPO-ANCA.

In a murine model of anti-MPO vasculitis, exogenous GCSF administration exacerbated disease. We should note that endogenous GCSF levels would be much lower than those after GCSF administration. Therefore, we have not yet demonstrated that endogenous GCSF is important in this model, and this will be explored in future studies. We acknowledge that part of the mechanism for exacerbation may be the increase in neutrophil numbers but disease was dramatically worsened in experiments with only modest increases in circulating neutrophil numbers. The murine model of anti-MPO vasculitis presents a number of technical challenges, is difficult to establish and usually results in mild disease. Few reports have shown disease severe enough to cause kidney injury with a raised serum creatine, as in the current study. Therefore, the administration of GCSF may be a useful strategy to achieve more severe and robust disease in this model. In order to maximise disease severity, all mice received LPS18 and we have not yet assessed the effect of GCSF in the absence of concurrent LPS administration. The effect of LPS in this model is partly TNFα dependent.18 ,19 However, stimulation of TLR4 on renal cells has been shown to be important both in an anti-MPO model and in other models of glomerulonephritis,20–22 Therefore, multiple mechanisms contribute to the effect of LPS, and it is likely that we would have seen less disease in the absence of LPS both with and without GCSF administration.

There are limited previous data on the effect of GCSF on murine neutrophils. Isolated murine neutrophils incubated with GCSF have been shown to shed CD62L and upregulate CD11b23 but the effect of GCSF administration in vivo was not clear. In addition to the priming effect of GCSF, seen in human neutrophils, we have shown upregulation of CD11c on neutrophils in mice given GCSF. Although LPS alone upregulated CD11b it did not upregulate CD11c and this is in keeping with previous work in which the mechanisms for this were explored.24 Recent data using intravital microscopy have shown that, in the context of a proinflammatory stimulus such as LPS, CD11c (LFA1) is a key molecule in anti-MPO triggered recruitment of neutrophils to glomeruli.25 Therefore, the upregulation of CD11c (LFA1) by GCSF could be playing an important role in pathogenesis, suggesting an additional mechanism for the observed effect on glomerular disease.

There have been case reports of GCSF administration exacerbating disease in patients with ANCA vasculitis.26 Although such reports cannot prove causation, our results give experimental and mechanistic data to support the suggestion that GCSF administration can exacerbate disease in patients. We suggest that exogenous GCSF should be administered with caution in patients with ANCA vasculitis and that in addition endogenous GCSF may be a therapeutic target.


We are grateful to colleagues at Kent and Canterbury, Royal Sussex County, King's College, St Helier and Guy's and St Thomas' hospitals for help obtaining patient serum samples. We thank Susanne Heck in the BRC flow cytometry facility for her expert advice. We thank Neil Dalton and Charles Turner at the Evelina Hospital for serum creatinine measurements. We also acknowledge the support of the MRC Centre for Transplantation.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Contributors SF, AC and RP performed experiments and data analysis. MR performed experiments and data analysis, conceived the study and wrote the paper. DD provided control serum samples. All authors reviewed and approved the manuscript.

  • Funding This work was supported by the Genzyme Renal Innovations Program, Kidney Research UK and the Sir Jules Thorn Charitable Trust. We acknowledge financial support from the Department of Health via the National Institute for Health Research (NIHR) comprehensive Biomedical Research Centre (BRC) award to Guy's & St Thomas' NHS Foundation Trust in partnership with King's College London and King's College Hospital NHS Foundation Trust.

  • Competing interests None.

  • Ethics approval NRES London- London Bridge.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.