Autophagy is required for stem cell mobilization by G-CSF

Hematopoietic stem cells (HSCs) maintain life-long blood production.¹  During stress, HSCs must be able to rapidly proliferate and differentiate to regenerate mature blood cells. Granulocyte colony-stimulating factor (G-CSF) is used to mobilize HSCs before transplantation and prevent neutropenia in patients treated with chemotherapy. G-CSF treatment has both direct effects on HSCs, leading to proliferation, and indirect effects on the interactions between HSCs and the bone marrow (BM) stroma.^2,3 

Autophagy encompasses the self-degradation of intracytoplasmic components, including damaged proteins and organelles,^4,5  and is induced by nutrition deprivation, drugs such as rapamycin, and cytokine-induced cellular stress.^6-9  Recent studies have demonstrated that autophagy is required for HSC survival at homeostasis and during aging.^10-12  G-CSF and other inflammatory cytokines induce HSC proliferation and the accumulation of genotoxic stress.¹³  We therefore examined the requirement for autophagy in G-CSF responses.

Wild-type (WT) or Atg5^−/− fetal liver cells were transplanted into lethally irradiated Ptprc^a primary recipients. Animal studies were performed in accordance with QIMR-Berghofer Ethics Protocols A01047M (G.R.H.), A08608M (K.P.A.M.), and A11605M (S.W.L.). LysMCre:Atg7fl/fl mice have been previously described.¹⁴ 

Blood counts were determined with the Hemavet 950FS (Drew Scientific). Antibodies were purchased from BD Biosciences or Biolegend. HSC flow cytometry was performed as previously described.¹⁵ 

A total of 100 µL of blood or 1 × 10⁴ BM cells were plated in M3234 methylcellulose (StemCell Technologies) with stem cell factor, erythropoietin, interleukin-3, and interleukin-6 and counted after 1 week.

In vivo assay, LC3-GFP mice were injected with chloroquine (CQ) (80 mg/kg per day, Sigma-Aldrich) for 3 days before analysis. Cells were flow-sorted and analyzed with ImageStream^X (Amnis). In vitro, cells were cultured in CQ (80 µM) for 6 hours. Images (×60) and statistics were processed by using IDEAS software (version 6.0, Amnis).

Cells were sorted, treated for 6 hours with or without CQ (80 µM) or G-CSF (100 ng/mL, Amgen), and prepared for immunoblotting as described.¹⁶  Antibodies were from Cell Signaling or Sigma-Aldrich.

Mice were injected subcutaneously with recombinant human G-CSF (Amgen) at 10 µg/dose per day for 6 days¹⁷  and analyzed 12 to 24 hours later or with AMD3100 (Sigma-Aldrich) at 5 mg/kg and analyzed 1 or 3 hours later.¹⁸ 

Primary clinical samples were collected from healthy stem cell donors before and after 5 days of G-CSF treatment (10 mcg/kg per day) in accordance with human ethics protocol (Protocol identifier LR 09/10) and the Declaration of Helsinki. Following red cell lysis, neutrophils were stained and underwent fluorescence-activated cell sorting (forward-scatter/side-scatter characteristics and CD14^neg).

RNA extraction and reverse transcription were performed as per manufacturer’s instructions (QIAGEN and Thermo Scientific). Human ATG5 (5′-AGCAACTCTGGATGGGATTG-3′, 5′-AGGTCTTTCAGTCGTTGTCTG-3′) and ATG7 (5′-CTTGGATGTTGGGTTTTGGC-3′, 5′-GAACTCCAATGTTAAGCGAGC-3′) messenger RNA levels were determined from sorted neutrophils using SYBR Green (Invitrogen) and normalized to β2 microglobulin.

Data are shown as mean ± standard error of the mean (SEM). An unpaired 2-tailed Mann-Whitney U test or a paired Student t test was used (GraphPad Prism, version 6.01, software).

To examine the effect of G-CSF administration on autophagy, we treated LC3-GFP mice with G-CSF (10 µg/day subcutaneously, 6 days) and CQ (80 mg/kg per day intraperitoneally, 3 days) to block lysosomal degradation of LC3. LKS⁺ cells were sort-purified from the BM and LC3 punctae quantified by imaging flow cytometry. G-CSF treatment increased LC3-GFP punctae in LKS⁺ cells, demonstrating that G-CSF enhances autophagic activity (Figure 1A). This was a direct effect of G-CSF on LKS⁺ cells because LC3-II protein accumulation was greater after ex vivo G-CSF treatment of purified LKS⁺ cells (Figure 1B). G-CSF administration also increased LC3 punctae in BM neutrophils (Figure 1C). To confirm these findings, we examined paired samples from clinical stem cell donors before and after G-CSF mobilization. Sorted neutrophils from the peripheral blood (PB) of these donors were analyzed for the expression of autophagy-associated genes and autophagic flux (stabilization of LC3 II).⁷  G-CSF mobilization increased transcriptional expression of ATG5 and ATG7 (Figure 1D), which was confirmed at the protein level (Figure 1E). LC3 I-II was variable between samples, possibly explained by variation in the lysosomal degradation of LC3, which confounds the accurate measurement of autophagic flux. Unfortunately, CQ administration to test this was not feasible in healthy donors. Altogether, these data demonstrate increased autophagy in both mouse and human cells after G-CSF mobilization.

We next examined HSC mobilization in mice deficient for autophagy in hematopoietic cells. In Atg5^−/− chimeric mice, G-CSF had a reduced capacity to increase blood white cell and neutrophil counts compared with WT mice (Figure 2A-B); however, there was no difference in BM neutrophils (Figure 2C). Similarly, following G-CSF administration, only the circulating progenitor cells were reduced in Atg5^−/− chimeras compared with WT (Figure 2D-E). Confirming these results, mobilized PB CFUs were reduced in Atg5^−/− chimeras compared with controls; however, no differences were seen in nonmobilized CFUs or BM CFUs between Atg5^−/− and WT (Figure 2F-G). Therefore, although G-CSF induces autophagy in vivo, autophagy is only functionally required upon cellular activation and mobilization to the peripheral circulation.

We therefore examined the effects of AMD3100, a direct antagonist of the CXCR4–CXCL12 interaction that, in contrast to G-CSF, does not directly signal HSCs or induce cell proliferation. Accordingly, there were no differences in mobilized CFUs between WT and Atg5^−/− chimeras, either 1 or 3 hours after AMD3100 (Figure 2F-G; supplemental Figure 1A). Although combined G-CSF and AMD3100 resulted in greater mobilized CFUs in both groups than either agent alone, Atg5^−/− chimeras had fewer mobilized CFUs after combined G-CSF/AMD3100 treatment (supplemental Figure 1A). These findings were confirmed using LysMCre:Atg7^fl/fl mice, which lack autophagy in myelomonocytic cells and committed myeloid progenitors.¹⁹  G-CSF–mobilized CFUs were reduced in LysMCre:Atg7^fl/fl mice compared with Cre negative controls; however, there was no difference in BM cells or PB from nonmobilized mice (Figure 2H-I). We observed fewer monocytes and a trend to decreased white cells and neutrophils after G-CSF treatment (supplemental Figure 1B) and significantly fewer mobilized progenitors in LysMCre:Atg7^fl/fl (supplemental Figure 1C). These data confirm that autophagy is required for efficient G-CSF–induced mobilization.

Finally, to investigate how autophagy regulates G-CSF–induced HSC/progenitor cell mobilization, we examined the cellular responses induced by G-CSF in Atg5^−/− and WT cells. There were no differences observed between WT and Atg5^−/− chimeras in regard to protease expression, degradation of CXCL12, depletion of phagocytic macrophages, or in the expression of cell adhesion molecules (supplemental Figure 1D-G). We therefore tested whether autophagy activation was essential for the survival of mobilized HSC/progenitor cells and neutrophils rather than BM egress per se. We observed increased apoptotic progenitors from Atg5^−/− compared with WT chimeras in the blood after G-CSF treatment; however, only minor differences were seen in BM (Figure 2J-K). Similarly, PB Atg5^−/− neutrophils exhibited a striking increase in apoptosis after G-CSF, but not in WT controls (Figure 2L). These findings demonstrate that autophagy is required for optimal G-CSF–induced mobilization and cell-intrinsic survival after G-CSF therapy.

We have found that the clinically relevant stress mediated by exogenous cytokines such as G-CSF acts directly to increase autophagy in HSCs and neutrophils and that this induction of autophagy is essential for their survival in the periphery after G-CSF mobilization. This role of autophagy in G-CSF responses has clinical relevance in transplantation and in the use of immunomodulatory drugs that modulate autophagy.

The authors thank Mark Smyth, Christian Engwerda, Stephen Sykes, and the members of the MacDonald, Lane, and Hill laboratory for helpful discussion and suggestions; the staff at the QIMR-Berghofer including the animal facility, Grace Chojnowski, Paula Hall, and Melinda Christensen (flow cytometry), Siok Tey for consenting donors, and Antiopi Varelias and Stuart Olver for assisting with blood processing; Noboru Mizushima for providing Atg5^+/− mice; James Harris for providing LysMCre:Atg7^fl/fl mice and for helpful comments; and nurses Elise Sturgeon, Judy Avery, and Justine Leach from the Royal Brisbane and Women’s Hospital for the collection of human samples.

These studies were supported by the University of Queensland International Research Tuition Award and University of Queensland Research Scholarship (L.L.-E.M.), the Novo Nordisk Foundation (F.O.B.), and the National Health and Medical Research Council (NHMRC) (grants 1011787, 1059286, and 1089138). J.D.M. and S.W.L. are Australia Fellows and Queensland Career Development Fellows, G.R.H. is an NHMRC and Queensland Health Senior Clinical Research Fellow, K.P.A.M. is a Cancer Council Queensland Senior Research Fellow, and S.W.L. has received fellowship funding from the Leukaemia Foundation.

Contribution: L.L.-E.M. designed and performed experiments and wrote the manuscript; T.V., K.E.L., R.D.K., B.E.T., and M.L. performed experiments; F.O.B., G.M.B., C.B., and J.D.M. contributed to experimental design and interpretation of data; G.R.H., K.P.A.M., and S.W.L. contributed to experimental design and interpretation of data and cowrote the manuscript; and all authors reviewed and edited the manuscript.

Conflict-of-interest disclosure: G.R.H. has received a consultancy payment from Amgen within the past 2 years. The remaining authors declare no competing financial interests.

Correspondence: Kelli P. A. MacDonald, QIMR Berghofer Medical Research Institute, 300 Herston Rd, Herston, Brisbane 4006, Australia; e-mail: kelli.macdonald@qimrberghofer.edu.au; and Steven W. Lane, QIMR Berghofer Medical Research Institute, 300 Herston Rd, Herston, Brisbane 4006, Australia; e-mail: steven.lane@qimrberghofer.edu.au.