G-CSFR expression on neurons has been demonstrated few years ago , and G-CSFR expression by neuronal and non-neuronal cells has recently been identified in numerous regions of the brain [3, 22]. Besides other cell types, it was lately found to be expressed by neuronal populations that are subject to cell death in neurodegenerative diseases, e.g. motoneurons [[24, 25], own unpublished observation] or dopaminergic neurons in the substantia nigra [22, 43].
Similar to the expression pattern of the erythropoietin receptor (EPOR), another hematopoietic cytokine with neuroprotective action, we show for the first time constitutive expression of the G-CSFR on the somata of RGCs. In contrast to the EPOR  we could not find G-CSFR expression on the dendrites of the RGCs in vivo. Our finding that the G-CSFR is expressed by RGCs prompted us to investigate possible neuroprotective effects of G-CSF on RGCs. We demonstrate profound neuroprotection against axotomy-induced RGC death through G-CSF. 40 μg/kg BW s.c. lead to a significant attenuation of RGC cell loss after optic nerve transection even when G-CSF was given after lesioning. The mode of RGC death in the model we used was shown to homogenously exhibit classical features of apoptosis [26–31]. Thus, based on the large body of evidence showing almost exclusively apoptotic cell death in this paradigm, the substantial reduction in RGC death can be explained by an anti-apoptotic activity of G-CSF. An impact of G-CSF on other confounding parameters, e.g. on the retrograde transport of Fluorogold, is theoretically possible. However, retrograde Fluorogold labelling of RGCs is already completed two days after axotomy , while rats were processed 14 days after axotomy in our experiments. Moreover, Fluorogold was proven to be a reliable marker for RGC survival that correlates well with markers for apoptosis, also under treatment with protective neurotrophins or cytokines in several studies [30, 33]. Finally, our data from purified RGC primary cultures further support an anti-apoptotic effect of G-CSF independent from retrograde axonal dye transport.
It has been postulated that G-CSF may confer an improved outcome in animal stroke models by bone marrow stem cell mobilization and migration to the lesion site, followed by neuronal differentiation . Additional systemic mechanisms could be postulated, as G-CSF has prominent systemic anti-inflammatory properties  that might contribute to its neuroprotective action in this model. In the experimental autoimmune encephalitis (EAE) model of multiple sclerosis, G-CSF was found to reduce T-cell-recruitment to the CNS and protected from further inflammation . Moreover, G-CSF could be neuroprotective by indirect mechanisms, e.g. inducing the release of neurotrophic factors from glial cells . However, G-CSF might also act directly on neural cells after systemic application as it passes the intact blood brain barrier . Furthermore, we [, this study] and others [3, 11] found the G-CSFR expressed on neuronal target cells, and neuroprotection was described in cultures of neuronal cell lines free of non-neuronal cells, suggesting a direct protective action of G-CSF within the nervous system [3, 22]. Our data presented here show that local application of G-CSF to the retina is sufficient to induce neuroprotection in vivo. Moreover, the results from our in vitro experiments using primary cultures of immunopurified RGCs argue in favour of a direct inhibition of neuron-specific apoptotic pathways by activation of neuronal G-CSFR. Nevertheless, this does not exclude an additional contribution of the above mentioned indirect mechanisms, e.g. glial cell-derived neurotrophic factors. Similarly, potential systemic effects that might partially contribute to G-CSF neuroprotection beyond its local activity can not be excluded at this point. However, the finding of substantial CNS neuroprotection after local injection of G-CSF could become of clinical relevance if, in order to avoid potential unwanted systemic effects of G-CSF, intracerebral or intrathecal administration will be tested in future treatment studies for neurodegenerative diseases.
In contrast to most other neurotrophic molecules, G-CSF is already used in clinical practice for many years to treat neutropenic patients, e.g. after chemotherapy or in cases of severe congenital neutropenia. Especially the latter application, which often requires G-CSF treatment for many years, yielded valuable safety data showing that also long-term treatment with G-CSF is principally possible . Thus, G-CSF could be immediately transferred to clinical studies. Nevertheless, a safety record for the chronic treatment of neurodegenerative diseases with G-CSF, also with respective dose-finding studies, will have to be performed in phase IIa/b studies.
As already outlined in the introduction, cell death in our optic nerve transection model displays the properties of classical neuronal apoptosis. Therefore, it provides an elegant opportunity to study potential new treatment concepts for neurodegenerative diseases in an in vivo model. In addition, our model replicates important steps in the pathologic course of glaucoma, because axonal lesions due to increased ocular tension are thought to induce retrograde RGC death and vision loss [48–50]. Therefore, besides its potential usefulness for the treatment of neurodegenerative diseases, G-CSF may also be a powerful add-on drug to stop RGC degeneration and to stop or slow down vision loss, especially since lowering the intraocular pressure does not necessarily halt the continuous degeneration of RGCs. Further, autoantibody-induced RGC apoptosis is a possible mechanism of autoimmune retinopathy in cancer patients and may represent another potential therapeutic application for G-CSF in ophthalmology .