Emerging roles of IL-34 in health and disease

Colony-stimulating factor 1 (CSF-1; M-CSF) was originally defined by its ability to promote macrophage development from bone marrow precursors in vitro >40 yr ago (Stanley et al., 1978). Mice with an inactivating mutation in the Csf1 gene (“osteopetrotic” mice [Csf1^op/op]) display severely reduced tissue macrophages, including osteoclasts, leading to osteopetrosis and other developmental defects (Wiktor-Jedrzejczak et al., 1990). Equally, most macrophages are absent in mice lacking the cognate receptor for CSF-1 (Csf1r^−/− mice), highlighting the importance of CSF-1–mediated CSF-1R signaling for macrophage development (Dai et al., 2002; Erblich et al., 2011). Yet,Csf1r^−/− mice have an overall more severe phenotype than mice lacking Csf1, leading to the proposed existence of an alternative ligand for CSF-1R (Dai et al., 2002). In 2008, IL-34 was identified as the additional ligand (Lin et al., 2008). In the past decade, many studies provided insight into IL-34 biology, but many questions remain unanswered, specifically in terms of its function. Here we review recent findings regarding IL-34, outline the impact of IL-34 on the development of the mononuclear phagocyte system, and discuss the current understanding of the role of IL-34 in pathology.

IL-34 is a secreted homodimeric glycoprotein highly conserved among vertebrates and more conserved than CSF-1 among mammalian and avian species (Garceau et al., 2010; Lin et al., 2008). CSF-1, however, exists in three isoforms, which can be either secreted or membrane bound (Dai et al., 2004; Nandi et al., 2006). The dimers of IL-34 are noncovalently linked, in contrast to the disulfide-linked CSF-1 homodimers (Liu et al., 2012; Ma et al., 2012b; Pandit et al., 1992). Despite binding to the same receptor, IL-34 and CSF-1 do not share sequence homology, but their active regions are similar (Liu et al., 2012;Ma et al., 2012b). In vitro studies suggested that the two ligands could even form heterodimers (Ségaliny et al., 2015a), but whether this occurs in vivo and translates into biological activity remains unknown.

Both ligands bind to the same regions of the tyrosine kinase receptor CSF-1R, leading to phosphorylation and homodimerization of the receptor (Fig. 1). CSF-1 and IL-34 similarly support cell growth and survival of human monocytes and their differentiation into macrophages (Barve et al., 2013; Boulakirba et al., 2018; Chihara et al., 2010; Lin et al., 2008; Wei et al., 2010). Yet, some differences in signal transduction and polarization of macrophages were found. IL-34–stimulated macrophages induced eotaxin-1, while CSF-1–mediated signaling led to higher levels of MCP-1 (Chihara et al., 2010). IL-34 also promoted increased secretion of IL-10 and CCL17 in macrophages in comparison with CSF-1–exposed macrophages (Boulakirba et al., 2018). Interestingly, the two ligands seem to induce different transcriptional profiles in vivo in microglia, the resident macrophages in the brain parenchyma (Kana et al., 2019).

On the other hand, an overlapping function of the two ligands was demonstrated in transgenic mice lacking Csf1 (Csf1^op/op) but expressing Il34 under the control of the CSF-1 promoter. This rescued the developmental defects of Csf1^op/op mice, suggesting that IL-34 can take over the function of CSF-1 if it is expressed in a similar spatiotemporal manner (Wei et al., 2010). In addition, osteoclast deficiency in Csf1^op/op mice was compensated over time by IL-34 in the spleen (Nakamichi et al., 2012), indicating a functional redundancy of IL-34 and CSF-1. Yet, in the absence of either IL-34 or CSF-1, no increase of the respective other ligand was observed, arguing against a compensatory impact (Wei et al., 2010). Thus, future studies may reveal potential differences in downstream CSF-1R signaling pathways triggered by either IL-34 or CSF-1 or whether similar CSF-1R signal transduction is determined by the regional ligand availability and/or abundance. This might also provide insight into the reason for the evolution of two ligands signaling through the same receptor.

CSF-1R is the main receptor for IL-34. While it is virtually expressed by all macrophages, monocytes, myeloid precursors, and a subset of dendritic cells, some other nonhematopoietic cells were also suggested to express CSF-1R (Stanley and Chitu, 2014). Among those were placental trophoblasts, renal proximal tubular epithelial cells (TECs), and colonic epithelial cells (Huynh et al., 2013; Jokhi et al., 1993; Stanley and Chitu, 2014) and some neurons (Clare et al., 2018; Luo et al., 2013; Nandi et al., 2012). However, other studies (Erblich et al., 2011; Kana et al., 2019; Sehgal et al., 2018) could not provide evidence that CSF-1R is indeed expressed outside of the mononuclear phagocyte lineage (reviewed inHume et al., 2019).

Recently, alternative receptors engaging with IL-34 have been proposed: the receptor-type protein-tyrosine phosphatase-ζ (PTP-ζ) and syndecan 1 (CD138; Fig. 1; Nandi et al., 2013; Ségaliny et al., 2015b). Within the central nervous system (CNS), the chondroitin sulfate proteoglycan PTP-ζ is expressed on neural progenitors, neurons, and glial cells and was shown to be involved in various neural functions (e.g., proliferation, migration, differentiation, synaptogenesis, and myelination;Lafont et al., 2009). Whether IL-34 indeed contributes to PTP-ζ signaling in vivo and whether IL-34 has overlapping functions with other PTP-ζ ligands remains to be shown. Furthermore, albeit with low affinity, IL-34 can bind syndecan-1, a type I transmembrane heparin sulfate proteoglycan (Ségaliny et al., 2015b). Syndecan-1 expression correlated with IL-34–induced CSF-1R activation in macrophages in vitro (Ségaliny et al., 2015b). Whether syndecan-1 modulates CSF-1R activities in vivo remains unclear.

Both IL-34 and CSF-1 seem to be broadly expressed in a variety of tissues in the steady state. CSF1 mRNA can be identified in humans and mice in the brain, spleen, mammary gland, testis, kidney, and small intestine and also in human lymph nodes, bone marrow, uterus, and ovary (Chitu and Stanley, 2006; Sherr et al., 1985; Stanley et al., 1983; Yoshida et al., 1990). In addition,Csf1 seems to be expressed in the heart, fat, pancreas, liver, and lung (Tabula Muris Consortium, 2018). At the protein level, CSF-1 was detected in the human heart and the murine lung (Bettina et al., 2016; Hohensinner et al., 2007).

The highest levels of IL-34 are found in the brain and the skin, where it is expressed by neurons and keratinocytes, respectively (Greter et al., 2012; Nandi et al., 2012; Wang et al., 2012). Il34 is further expressed in various tissues, including human and mouse heart, lung, liver, kidney, and testes, and in human intestine, thymus, spleen, colon, prostate, and lymph nodes (Lin et al., 2008;Nakamichi et al., 2012; Tabula Muris Consortium, 2018; Wang et al., 2012, 2016a; Fig. 2 A). While most of these studies reported gene expression, IL-34 protein was described in the healthy murine lung and human skin, kidney, intestine, spleen, and brain (Baek et al., 2015; Bettina et al., 2016; Franzè et al., 2015; Greter et al., 2012; Lin et al., 2008). We also found that IL-34 protein was detectable across various mouse tissues, with highest expression in brain, skin, and lymph nodes (Fig. 2 B), consistent with published mRNA data. In many organs, the cell types producing IL-34, its distribution pattern compared with CSF-1, and its role in the function of tissue-resident macrophages remain to be investigated.

Many tissue macrophages are derived from embryonic precursors and self-maintain throughout adulthood (Ginhoux and Jung, 2014). LCs, the resident mononuclear phagocytes of the epidermis, are mostly derived from fetal liver monocytes (Hoeffel et al., 2012). In contrast, primitive macrophages generated in the yolk sac during early embryonic development give rise to microglia (Ginhoux et al., 2010). Similarly to most other tissue macrophages, the genesis of these two populations also depends on CSF-1R signaling (Ginhoux et al., 2006,2010; Greter et al., 2012). However, while most tissue macrophages rely solely on CSF-1 for their development, CSF-1 appears to only partially contribute to the development of LCs (Cecchini et al., 1994; Dai et al., 2002). Indeed, LC differentiation and maintenance are mostly dependent on IL-34, which is produced in the developing epidermis from embryonic day 17.5 onward. Consistent with this expression pattern, Il34-deficient mice display a reduction of LC precursors and absence of adult LCs (Greter et al., 2012; Wang et al., 2012, 2016a).

Within the brain, expression of CSF-1 and IL-34 is nonoverlapping and spatially distinct (Kana et al., 2019; Nandi et al., 2012). During embryogenesis, IL-34 is restricted to the marginal zone and cortical plate, whereas CSF-1 is expressed within the subventricular zone and the ventricular area. In the neonatal cortex, CSF-1 is detected in layer VI, and IL-34 is detected in layers II–V (Nandi et al., 2012). Despite the presence of IL-34 in the developing brain, microglial development in mice is not dependent on IL-34 but requires CSF-1 (Greter et al., 2012; Kana et al., 2019). Yet, in zebrafish, a further role of IL-34 in the colonization of embryonic macrophages to the anterior head was demonstrated (Wu et al., 2018).

In contrast to development, adult microglia are reduced inIl34-deficient mice in regions correlating with high IL-34 expression, indicating that their maintenance depends on IL-34 (Greter et al., 2012; Wang et al., 2012). IL-34 is mainly produced by neurons, at the highest levels in the cortex, the hippocampus, and the striatum (Greter et al., 2012; Kana et al., 2019;Wang et al., 2012; Wei et al., 2010). In comparison, CSF-1 is predominantly expressed by astrocytes in the cerebellum, and in mice lacking CSF-1, cerebellar microglia were decreased, while in other regions, only moderate or no reduction of microglia was detected (Calvo et al., 1998; Ginhoux et al., 2010; Kana et al., 2019; Wegiel et al., 1998).

In addition to neurons, IL-34 is expressed by ependymal cells lining the ventricular system and in the choroid plexus (Nandi et al., 2012; Wang et al., 2012). Whether this influences the development or function of choroid plexus macrophages requires more studies. In the retina, IL-34 expression was detected in retinal ganglion cells, where it also controls the homeostasis of one subset of retinal microglia residing in the inner plexiform layer (O’Koren et al., 2019). Altogether, while CSF-1 is critical for microglial development, a region-specific expression of IL-34 and CSF-1 regulates microglial homeostasis, phenotype, and transcriptional profile.

The importance of IL-34 for LCs and microglia has been described, but relatively little is known about its role in pathology. Here, we summarize studies showing correlations of IL-34 expression in a variety of diseases and discuss its potential function.

In neuron–microglia cocultures, treatment of microglia with IL-34 triggered the upregulation of TGF-β, insulin-degrading enzyme, and the antioxidant enzyme heme oxygenase 1 (HO-1), which limited the neurotoxic effects of oligomeric amyloid-β 1–42 (Ma et al., 2012a; Mizuno et al., 2011). IL-34 administration to the CNS in vivo promoted microglial accumulation and improved cognitive function in the APP/PS1 mouse model of Alzheimer’s disease (Mizuno et al., 2011), indicating a neuroprotective role of IL-34 via regulation of microglial function. Similarly, administration of IL-34 or CSF-1 reduced microgliosis and neuronal loss after kainic acid–induced excitotoxicity, a model for neurodegeneration (Luo et al., 2013; Wang et al., 2005). Finally, beneficial effects of IL-34 on neuropathology could also be shown in an animal model of prion disease where Il34 deficiency led to accelerated PrP^Sc deposition and shorter survival (Zhu et al., 2016). Yet, terminally ill animals showed similar microglia numbers, proposing that increased susceptibility of Il34-deficient animals might be due to the decreased microglia numbers in steady state or altered functions in the absence of IL-34.

CSF-1R mutations are associated with neurodegenerative diseases, including adult-onset leukoencephalopathy with axonal spheroids and pigmented glia, a hereditary diffuse white matter disorder (Pridans et al., 2013; Rademakers et al., 2011). The underlying cause is the partial loss of CSF-1R signaling on microglia leading to microglial dysfunction. Whether decreased IL-34 or CSF-1–mediated CSF-1R transduction leads to adult-onset leukoencephalopathy with axonal spheroids and pigmented glia pathogenesis and whether mutations affecting either ligand would lead to a similar disorder remain to be shown.

RA is a complex autoimmune disease characterized by synovial inflammation and hyperplasia, bone and cartilage destruction, and systemic disorders affecting different organs (McInnes and Schett, 2011). Many studies revealed an association between RA severity and IL-34 expression. Higher serum levels of IL-34 in RA patients correlated with a variety of RA features and were described as an independent risk factor for radiographic progression in RA (Moon et al., 2013; Tian et al., 2013; Chang et al., 2015; Zhang et al., 2015; Yang et al., 2016; Wang et al., 2018a). Similarly, higher levels of IL-34 in the synovial fluid in RA patients were indicative of increased inflammation in the synovium (Chemel et al., 2012; Hwang et al., 2012; Moon et al., 2013; Tian et al., 2013;Yang et al., 2016) and corresponded to higher histological severity of synovitis, increased hyperplasia, higher leukocyte counts, and increased levels of IL-6 and receptor activator of NF-κB ligand (Chemel et al., 2012; Moon et al., 2013; Wang et al., 2017). Synovial biopsies from RA patients showed that IL-34 is mostly expressed by fibroblast-like synovial cells (FLS) and endothelial cells (Chang et al., 2015; Chemel et al., 2012; Hwang et al., 2012). Moreover, IL-34 production in FLS from RA patients was increased by the proinflammatory cytokines TNF-α and IL-1β and was inhibited by TGF-β1 and BMP-2 (Chemel et al., 2012;Hwang et al., 2012; Wang et al., 2017). In comparison, while CSF-1 is also highly expressed by RA FLS, it is not enhanced upon TNF-α stimulation, indicating that the expression of the two cytokines is regulated by different mechanisms (Hwang et al., 2012). Consequently, serum levels of IL-34 could be used to predict the response to TNF-α antagonist therapy (Ding et al., 2015; Tian et al., 2013; Zhang et al., 2015). In line with these observations, injection of recombinant IL-34 in mice aggravated the disease severity of collagen-induced arthritis and promoted an increase in IL-6 and TNF-α (Zhang et al., 2018). Similarly, administration of CSF-1 exacerbated disease in models of arthritis (Bischof et al., 2000; Campbell et al., 2000). On the other hand, blocking CSF-1R in mice had a protective role in collagen-induced arthritis and was more effective than a TNF-α antagonist (Garcia et al., 2016). Nevertheless, it is not clear whether this effect was mediated only by neutralization of CSF-1R signaling or by the general depletion of macrophages upon CSF-1R blockade (Sauter et al., 2014). Altogether, IL-34 is induced by proinflammatory cytokines to sustain an inflammatory cascade promoting joint pathology. Future studies should further elucidate the precise mechanisms of IL-34 action in vivo and establish its possible role as a diagnostic or therapeutic factor.

SLE is a chronic autoimmune disease affecting different organs (Tsokos, 2011). Increased serum levels of IL-34 in SLE patients correlated positively with SLE disease activity index, titers of anti–double-stranded DNA antibodies, and C-reactive protein and negatively with levels of C3 in serum (Wang et al., 2016b; Xie et al., 2018). IL-34 was also found to be an independent risk factor for the development of lupus nephritis, an SLE-caused kidney inflammation with high morbidity and mortality (Almaani et al., 2017; Cheng et al., 2019). Both IL-34 and CSF-1 are expressed in TECs of the kidney in both SLE patients and the MRL-Fas^lpr mouse model of lupus nephritis.Il34-deficient MRL-Fas^lpr mice showed decreased kidney pathology, reduced signs of multiorgan failure, decreased infiltration of monocytes and lymphocytes in the kidney, and lower intrarenal production of chemokines and proinflammatory cytokines. Thus, IL-34 appears to promote lupus nephritis pathogenesis by supporting the accumulation of macrophages in the kidney (Wada et al., 2019).

IL-34 and CSF-1 expression in TECs was also increased in another model of kidney pathology: ischemia–reperfusion acute kidney injury that can progress to chronic kidney disease (Baek et al., 2015). Similarly to lupus nephritis, IL-34 promoted kidney pathology and TEC destruction, as well as proliferation of kidney macrophages and myeloid cell infiltration. IL-34 deficiency resulted in significant reduction in kidney atrophy and myeloid cell accumulation (Baek et al., 2015). Overall, inflammation-induced IL-34 in the kidney promotes expansion and recruitment of myeloid cells to drive kidney destruction. Interestingly, CSF-1 was suggested to promote repair in models of acute kidney injury (Alikhan et al., 2011; Menke et al., 2009; Zhang et al., 2012), indicating opposing roles for CSF-1 and IL-34. How these two CSF-1R ligands can mediate opposing functions, however, remains to be defined.

Il34 mRNA was up-regulated in the salivary glands of patients with Sjögren’s syndrome, a chronic autoimmune disorder of the exocrine glands (Ciccia et al., 2013; Fox, 2005). IL-34 correlated with the expression of proinflammatory cytokines TNF-α, IL-1β, IL-17, and IL-23p19 and with infiltration of inflammatory monocytes (Ciccia et al., 2013). In patients with Crohn’s disease or ulcerative colitis, the two major forms of inflammatory bowel disease, IL-34 (and also CSF-1) was up-regulated in the inflamed colon epithelium through TNF-α–induced NF-κB signaling (Franzè et al., 2015; Zwicker et al., 2015).

In nonalcoholic fatty liver disease, levels of CSF-1 and IL-34 were highly increased in the liver and the serum of patients with nonalcoholic fatty liver disease. IL-34 concentration increased with fibrosis progression and could be used as an independent diagnostic marker for fibrosis (Shoji et al., 2016). Likewise, IL-34 was expressed by hepatocytes in liver fibrosis induced by hepatitis B or C infection and also correlated with increasing liver fibrosis (Preisser et al., 2014; Wang et al., 2018b). Consistent with these findings in the clinic, in vitro IL-34–differentiated monocytes up-regulated profibrotic factors and promoted collagen I production in hepatic stellate cells, supporting a pathogenic function for IL-34 in promoting fibrotic liver diseases (Preisser et al., 2014).

Altogether, increased levels of IL-34 correlate with many inflammatory, autoimmune, metabolic, and cardiovascular disorders (Chang et al., 2014; Fan et al., 2016; Li et al., 2012; Tao et al., 2017;Xi et al., 2018; Zorena et al., 2016). Yet, clear evidence for a function of IL-34 in many of these disorders is lacking.

In contrast to the inflammatory diseases described above, acute skin inflammation (induced by UV light or in a mouse model of atopic dermatitis) results in down-regulation of IL-34 and increased expression of CSF-1 in the epidermis. This leads to a partial repopulation of LCs inIl34-deficient mice, suggesting that in inflammation, CSF-1 can replace IL-34 in supporting LC differentiation (Ginhoux et al., 2006; Greter et al., 2012; Wang et al., 2012, 2016a). Consistent with the findings in animal models, in patients with atopic dermatitis, IL-34 expression was reduced compared with skin from healthy volunteers (Esaki et al., 2015). Thus, in the skin, IL-34 appears to be a homeostatic cytokine, which is down-regulated in acute inflammation. Experimental systems designed to elevate local levels of IL-34 in skin could limit skin inflammation and thus result in clinical benefit.

Tumor-associated macrophages (TAMs) are implicated in tumor growth, metastasis formation, angiogenesis, and immune suppression/activation. Expression of CSF-1R as well as CSF-1 often correlates with poor prognosis in various types of cancer (Behnes et al., 2014; Espinosa et al., 2014; Kawamura et al., 2009; Liu et al., 2016;Richardsen et al., 2015; Scholl et al., 1994). Thus, the CSF-1R axis, which is involved in regulating the function of TAMs, represents an attractive target for immunotherapeutic strategies, and several approaches inhibiting CSF-1R are currently under clinical development (Cannarile et al., 2017; Kumari et al., 2018;Peyraud et al., 2017).

Regarding IL-34, emerging data also demonstrate its correlation with poor prognosis in several cancers, such as osteosarcoma and lung and breast cancer (Fig. 3; Baghdadi et al., 2018; Wang et al., 2015). Expressed by various types of cancer cells and further induced upon chemotherapy treatment (Baghdadi et al., 2016; Baud’Huin et al., 2010; Cioce et al., 2014; Franzè et al., 2017; Wang et al., 2015), IL-34 binds to CSF-1R on mononuclear phagocytes, which play distinct roles during different stages of tumor progression (Baud’Huin et al., 2010; Noy and Pollard, 2014; Ségaliny et al., 2015c; Zins et al., 2018). IL-34 contributes to new vessel formation and promotes tumor progression and metastasis via accumulation of TAMs (Ségaliny et al., 2015c), as was shown for CSF-1 (Curry et al., 2008; Lin et al., 2001). Embryonic stem cell–derived IL-34 was shown to polarize bone marrow–derived macrophages into macrophages that phenotypically and functionally resemble TAMs, showing increased levels of Arg-1, Tie-2, and TNF-α, and which contributed to angiogenesis and teratoma progression (Chen et al., 2014).

Another study showed that IL-34 produced by hepatocellular carcinoma cells increased TGF-α production by TAMs, which in turn inhibited the negative regulator of IL-34, miR-28-5p. This feedback loop was suggested to modulate hepatocellular carcinoma growth and metastasis and to promote tumor progression. Moreover, patients with low miR-28-5p expression, high IL-34 levels, and high numbers of TAMs had a poor prognosis with shorter overall survival (Zhou et al., 2016).

A potential involvement of IL-34 in immunotherapy resistance was demonstrated in a patient with metastatic melanoma. Resistance to anti–programmed cell death protein 1 immunotherapy was acquired concomitant with enhanced expression of IL-34 and increased frequencies of CD163⁺ macrophages (Han et al., 2018). A similar observation was made for CSF-1 (Neubert et al., 2018). Altogether, while a direct role of IL-34 on TAMs and in tumor progression and immunotherapy resistance still has to be demonstrated, it is a promising potential candidate for novel therapeutic strategies.

Yet, both IL-34 and CSF-1 were also shown to negatively correlate with tumor progression. CSF-1R blockade or anti–CSF-1 treatment promoted metastasis formation in several animal models (Beffinger et al., 2018; Swierczak et al., 2014). Likewise, lower expression of IL-34 was found in a cohort of patients with non–small cell lung cancer and was related to poor survival (Wang et al., 2015). In addition, high expression of IL-34 also seemed to be favorable in some cases of hematologic malignancies, head and neck cancer, and breast and colorectal cancers (Mizuno et al., 2009; Uhlen et al., 2017; Wang et al., 2015). This suggests that the function of IL-34 in cancer may be multidimensional and can vary across different types of tumors and tissues.

IL-34 is a fairly new member of the family of “myelopoietic” cytokines. Originally described as a nonredundant growth factor specifically regulating LC development and microglial homeostasis, recent advances suggest that IL-34 is also expressed in a host of tissues other than skin and brain, in both the steady state and inflammation. However, whether IL-34 supports the generation, maintenance, or function of the resident macrophage compartment in these various organs requires further studies. In addition, the distribution pattern of IL-34 compared with CSF-1 in a given tissue might dictate the regulation of different macrophage subsets and their functions.

In pathological conditions, IL-34 seems to be involved in an array of diseases ranging from inflammation and autoimmunity to cancer. Nonetheless, while many human studies predominantly show a correlation between disease progression and IL-34 expression, very little is known about the actual function of IL-34. Additionally, whether IL-34 is beneficial or detrimental in a given pathology appears to be disease dependent, adding a layer of complexity to uncovering the biological roles of this cytokine. Thus, further research will reveal whether IL-34 indeed sustains an inflammatory disease, has proinflammatory or immunoregulatory properties, or is simply a hallmark of inflammation. This information will provide the necessary insight into whether IL-34 can serve as a predictive biomarker or even be exploited as a therapeutic target. Due to experimental design limitations, in many studies, including the ones discussed here, CSF-1 and IL-34 were not analyzed simultaneously. Thus, a further challenge is to carefully dissect similarities and differences between CSF-1 and IL-34 signaling and to identify the regulation and activities of both cytokines and their spatiotemporal expression pattern across tissues in pathological conditions. In cancer, for example, specific targeting of either CSF-1 or IL-34 instead of CSF-1R blockage, which eliminates most tissue macrophages, could be advantageous and could offer new perspectives for therapeutic interventions. Recent technologies (such as mass cytometry, multiparameter flow cytometric analysis, and single-cell RNA sequencing) combined with animal models and conditional mutagenesis should further help to advance our understanding of the interplay between IL-34, macrophages, and the immune system and could point toward novel disease mechanisms.

This work was supported by the Swiss National Science Foundation (PP00P3_170626 and BSSG10_155832 to M. Greter).

Author contributions: I. Lelios, D. Cansever, S.G. Utz, W. Mildenberger, S.A. Stifter, and M. Greter wrote the manuscript.