RESEARCH ARTICLE

MCSF orchestrates branching morphogenesis in developingsubmandibular gland tissueGulsan Ara Sathi1, Mahmoud Farahat1,2, Emilio Satoshi Hara1, Hiroaki Taketa3, Hitoshi Nagatsuka4,Takuo Kuboki2 and Takuya Matsumoto1,*

ABSTRACTThe importance of macrophages in tissue development andregeneration has been strongly emphasized. However, the specificroles of macrophage colony-stimulating factor (MCSF), the keyregulator of macrophage differentiation, in glandular tissuedevelopment have been unexplored. Here, we disclose newmacrophage-independent roles of MCSF in tissue development.We initially found that MCSF is markedly upregulated at embryonicday (E)13.5, at a stage preceding the colonization of macrophages (atE15.5), in mouse submandibular gland (SMG) tissue. Surprisingly,MCSF-induced branching morphogenesis was based on a directeffect on epithelial cells, as well as indirectly, by modulating theexpression of major growth factors of SMG growth, FGF7 and FGF10,via the phosphoinositide 3-kinase (PI3K) pathway. Additionally, giventhe importance of neurons in SMG organogenesis, we found thatMCSF-induced SMG growth was associated with regulation ofneurturin expression and neuronal network development duringearly SMGdevelopment in an in vitro organogenesismodel as well asin vivo. These results indicate that MCSF plays pleiotropic roles and isan important regulator of early SMG morphogenesis.

KEY WORDS: 3D culture, Branching morphogenesis, Fibroblastgrowth factor, Macrophage colony-stimulating factor, Neurturin,Salivary gland

INTRODUCTIONDuring development, the submandibular gland (SMG) tissueundergoes unique morphological changes (termed branchingmorphogenesis) that share similarities with changes seen in othertissues, including the lungs, kidneys, mammary glands andpancreas. Branching morphogenesis proceeds via spatiotemporalinteractions among a variety of cell types including epithelial andmesenchymal cells, neurons and immune cells (Gumbiner, 1992).These cellular interactions involve both biophysical andbiochemical dimensions, including direct cell–cell or cell–extracellular matrix (ECM) contacts (Walker et al., 2008), andautocrine or paracrine effects of growth factors, such as epidermalgrowth factor (EGF), fibroblast growth factor (FGF) andtransforming growth factor β (TGF-β), that induce differentiation

of cells as well as changes in ECM properties and composition (e.g.in collagen, glycoprotein, fibronectin and laminin) (Kadoya et al.,1995; Nakanishi et al., 1986; Sakai et al., 2003; Yang and Young,2008). Extracellular signals activate intracellular signaling pathways[e.g. RhoA and Rho-associated protein kinase, mitogen-associatedprotein kinase, phosphoinositide 3-kinase (PI3K), and wingless(WNT) signaling pathways] (Porazinski et al., 2015) that furtherinduce specific gene expression and protein synthesis, and/orpromote cell polarity, movement and tissue self-assembly.Eventually, epithelial bud formation, tissue clefting andbranching, and concomitant differentiation and maturation ofepithelial cells into acinar and ductal cells occurs (Patel et al.,2011; Steinberg et al., 2005).

Recent studies have highlighted the importance of immune cells,more specifically that of macrophages, in tissue development andgrowth, as well as in tissue regeneration (Cecchini et al., 1994; Joneset al., 2013). Macrophages have been detected during branchingmorphogenesis of glandular-like tissues and it has been shown thatthey can direct tissue growth (Cecchini et al., 1994; Jones et al.,2013). For instance, during normal mammary development,macrophages directly associate with terminal end buds, lining anddirecting the developing duct. Comparably, at early stages ofpancreatic development, macrophages colonize the duct–isletinterface, and their numbers increase steadily as the organdevelopment proceeds (Geutskens et al., 2005; Ingman et al.,2006). In the absence of macrophages, terminal buds in themammary gland are atrophic and poorly branched, and pancreaticislet cells show abnormal morphology (Banaei-Bouchareb et al.,2004; Gouon-Evans et al., 2000; Shibata et al., 2001; Van Nguyenand Pollard, 2002). These studies have indicated that macrophages,mainly theM2 type, play critical roles in tissue development, as wellas in the clearance of apoptotic cells and unnecessary matrix(Davies et al., 2013; Shibata et al., 2001).

In this context, macrophage colony stimulating factor (MCSF,also known as CSF1) has been shown to be the key regulator of M2macrophage differentiation (Fleetwood et al., 2007; Pyonteck et al.,2013). Nevertheless, since most of previous studies have mainlyfocused on macrophage colonization using mice presenting Csf1(the MCSF-encoding gene) gene deletion (i.e. Csf1op/Csf1op mice),the roles of macrophages on tissue development could have beenoveremphasized, and the exact effect of MCSF could have beenunderestimated. Therefore, we hypothesized that MCSF could havea macrophage-independent function in SMG branching and growth.Indeed, by applying both physical and chemical stimuli, we foundthat MCSF acts as a regulator of SMG development by modulatingthe expression of the major growth factors involved in SMG growth,FGF7 and FGF10, via the phosphoinositide 3-kinase (PI3K)pathway. Additionally, we found that MCSF promoted neuronalnetwork development by regulating the levels of the neurotrophicfactor neurturin (NRTN) during early SMG development.Received 25 August 2016; Accepted 21 February 2017

1Department of Biomaterials, Okayama University, 2-5-1 Shikata-Cho, Okayama700-8558, Japan. 2Department of Oral Rehabilitation and Regenerative Medicine,OkayamaUniversity, 2-5-1 Shikata-Cho, Okayama 700-8558, Japan. 3Center for theDevelopment of Medical and Health Care Education, Okayama University, 2-5-1Shikata-Cho, Okayama 700-8558, Japan. 4Department of Oral Pathology andMedicine, Okayama University, 2-5-1 Shikata-Cho, Okayama 700-8558, Japan.

*Author for correspondence (tmatsu@md.okayama-u.ac.jp)

T.M., 0000-0002-9804-4786

1559

© 2017. Published by The Company of Biologists Ltd | Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

RESULTSExpression of MCSF precedes tissue colonization ofmacrophages in early SMG developmentWe first analyzed the MCSF expression pattern during SMGdevelopment in embryonic day (E)12.5 to E15.5 mouse SMGtissues by immunohistochemistry and western blotting. AlthoughMCSF was nearly undetectable at E12.5, MCSF expression levelsincreased markedly in the mesenchyme around the epithelial buds atE13.5 (Fig. 1A; Fig. S1A). Additionally, in order to understand

which cells express the receptor for MCSF (CSF1R, also known asMCSFR and CD115), we also analyzed the expression profile ofCSF1R, and found it to be mainly present in the mesenchyme inE13 or E13.5, and both in the epithelium and mesenchyme inE14.5 and E15.5 SMGs (Fig. S1B).

Next, we analyzed the presence of macrophages byimmunohistochemical detection for the macrophage marker F4/80(Schulz et al., 2012) cells in E13.5 and E15.5 SMGs. As shown inFig. 1B, there were only a few F4/80+ cells at E13.5, but their

Fig. 1. Identification of MCSFexpression and macrophages inearly SMG development. (A) MCSFprotein expression (red) was assessedby immunohistochemical analysis ofthe SMG at E12.5 to E15.5. Epitheliumwas stained with lectin from Arachishypogaea (PNA) (green). (B) Whole-mount immunohistochemistry of SMGepithelium (E-cadherin, red) andmacrophages (F4/80, green) at E13.5and E15.5. Ba1 and Ba2 show highermagnification images of the areahighlighted inside the square in Ba(E13.5) and Bb (E15.5), respectively.(C) Analysis of iNOSII and ARG1 (red)expression in F4/80+ macrophages(green) by immunohistochemistry ofSMGs at E13.5 and E15.5. (D) Graphdepicting the percentage of the F4/80+

and ARG1+ cells, or F4/80+ and iNOSII+

double-positive cells in native SMGtissue at E15.5. Results are mean±s.d.(n=4). **P≤0.01, unpaired Student’st-test. Scale bars: 100 µm.

1560

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

number increased drastically at E15.5. These cells were located inthe mesenchyme close to epithelial buds (Fig. 1B). In an attempt todetermine the macrophage subtype, we probed for the expression ofinducible nitric oxide synthase II (iNOSII, also known as NOS-2)and arginase 1 (ARG1) as markers for M1 and M2 macrophages,respectively (Hesse et al., 2001; Pyonteck et al., 2013). AlthoughiNOSII-positive cells were rare, we could observe a large number ofARG1-positive cells, indicative of an M2 macrophage phenotype(Fig. 1C,D). Taken together, these data show that MCSF expressionprecedes the appearance of macrophages in the early stages ofSMG development, and suggest that MCSF could be inducing thedifferentiation or migration of macrophages in the SMG tissue ina subsequent step.

MCSF enhances SMG growthGiven that the peak in MCSF expression (E13.5) was earlier thanthat of macrophage numbers (E15.5), we hypothesized that MCSFcould regulate SMG growth via macrophage-independentmechanisms or by modulating the function of other cell types. Inan attempt to clarify the effects of MCSF on SMG growth andbranching, we performed a series of ex vivo tissue culture studiesusing track-etched membranes (which induce SMG growth) and ahydrogel culture system. Since the mechanical properties of theSMG tissue change over the course of development due to theaccumulation and cross-linking of secreted extracellular molecules(Ingber, 2006), we had previously fabricated hydrogels that differedin terms of mechanical stiffness for ex vivo SMG culture, anddemonstrated that stiff gels attenuated, whereas soft gels enhanced,SMG growth (Miyajima et al., 2011).Therefore, we herein cultured E12.5 SMGs (at a stage earlier than

the peak in MCSF levels in vivo) on a stiff hydrogel (Young’smodulus=184 kPa) as a negative control, and analyzed the effect ofdifferent concentrations of MCSF (5, 10, 20 and 50 ng/ml) on SMGgrowth for up to 72 h. As shown in Fig. 2A,B, SMG branching wasenhanced in a dose-dependent manner up to 20 ng/ml of MCSF.However, an inhibitory effect was observed at 50 ng/ml ascompared to 5, 10 or 20 ng/ml concentrations, suggesting theoccurrence of a regulatory feedback mechanism. Additionally, asexpected, MCSF treatment (20 ng/ml) induced a notable increase inthe number of F4/80+ macrophages in SMG tissue after 72 h ofculture (Fig. S2).To confirm the effects of MCSF on SMG growth, we cultured

SMGs on track-etched membranes as a positive control, to allowtissues to grow, and blockedMCSF function with antibodies againstMCSF or CSF1R. As shown in Fig. 2C,D, SMGs showed normalbud and cleft formation, comparable to native tissue growth, whencultured with control medium on a track-etched membrane.However, SMG development was strongly suppressed by thesupplementation with MCSF or CSF1R antibodies (Fig. 2C,D).Taken together, these results suggest that MCSF has important rolesfor not only the attraction or differentiation of macrophages, but alsoas a chemokine that promotes cell differentiation.

MCSF regulates FGF7 and FGF10 expression via PI3KsignalingTo understand the mechanisms underlying the regulation of SMGgrowth and branching by MCSF, we analyzed the interactionbetweenMCSF and factors that are essential for SMG formation andgrowth, namely FGF7 and FGF10, which modulate epithelialgrowth and ductal elongation, respectively (Steinberg et al., 2005).Interestingly, MCSF enhanced the expression of both growthfactors, especially that of FGF10 (Fig. 3A–C). Consistent with

previous results, neutralization of FGF7 and FGF10 with theirspecific antibodies significantly suppressed SMG growth (Taketaet al., 2015), which was, however, partially restored by furtherMCSF supplementation (Fig. 3D,E). This rescue effect of MCSFsupplementation was statistically significant in the case of FGF10,which is in accordance with the fact that MCSF could induce theFGF10 levels more prominently. Interestingly, in an oppositeexperimental design, FGF7 or FGF10 supplementation could notrescue the inhibition of SMG growth induced by the blockade ofMCSF function with anti-MCSF antibody (Fig. 3F,G). Takentogether, these results indicate that MCSF works as an upstreamfactor of FGF7 and FGF10.

Additionally, since MCSF could partially rescue the inhibitoryeffects of FGF7 and FGF10 neutralization, MCSF could possiblystimulate other factors in the mesenchyme that can replace thefunction of FGF7 and FGF10, or stimulate the epithelial cellsdirectly. In fact, when we applied MCSF directly on epitheliumrudiments, we observed that MCSF markedly enhanced SMGepithelium growth and elongation (Fig. 3H,I). These results stronglysupport the concept of a macrophage-independent function ofMCSF on SMG growth.

To further understand the intracellular pathways involved inMCSF-induced upregulation of FGF7 and FGF10, we investigatedthe activation of PI3K signaling (p85 subunit, also known asPIK3R1), as this pathway has been reported to be downstream ofCSF1R (Yavropoulou and Yovos, 2008) and is also implicated inthe regulation of FGF expression (Eswarakumar et al., 2005; Larsenet al., 2003). As shown in Fig. 4A,B, MCSF induced activation ofPI3K phosphorylation in whole SMG samples. By contrast, MCSF-induced SMG growth was completely suppressed by treatment withLY294002, a PI3K-specific inhibitor (Vlahos et al., 1994) (Fig. 4C,D).The specificity of the chemical inhibitors is a concern in inhibitionstudies. Based on a previous report (Larsen et al., 2003), LY294002and Wortmannin are the major PI3K inhibitors, and significantlyinhibit the SMG branching morphogenesis. The LY294002inhibitory effect, however, was nontoxic and reversible, and thusit was selected in the experiments. As shown in Fig. 4E,F, FGF7 andFGF10 expression was downregulated by LY294002 even in thepresence of MCSF. Interestingly, after LY294002-treated SMGsamples were washed to remove the inhibitor, and the medium wasreplaced with MCSF-supplemented medium, the SMG growth wascompletely restored (Fig. 4G,H).

Taken together, these results demonstrate that MCSF-inducedFGF expression and SMG growth is dependent on the activation ofPI3K signaling.

MCSF regulates neuronal outgrowth via NRTNSince MCSF could be stimulating factors other than FGFs, wehypothesized that MCSF could also have a stimulatory effect onneurons, based on previous reports showing that MCSF inducesproliferation of microglia and Purkinje cells (Murase and Hayashi,1998) and that neurons play a critical role in SMG organogenesis(Knox et al., 2010). Therefore, we cultured SMGs on stiff hydrogelswith or without MCSF supplementation and analyzed neuronaltissue growth. In the absence of MCSF, neuronal growth wasstrongly inhibited and epithelial bud formation was consequentlyreduced. In contrast, MCSF treatment significantly enhancedneuronal growth, which prompted normal SMG development(Fig. 5A).

Previous reports showed that NRTN is an important neurotrophicfactor (Knox et al., 2013; Liu et al., 2007) and that Nrtn−/− micepresent atrophic innervation of salivary glands (Heuckeroth et al.,

1561

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

1999). Therefore, we investigated whether MCSF regulates neuralgrowth by inducing the expression of NRTN. As shown in Fig. 5B,MCSF strongly enhanced NRTN levels in SMG tissue. Todetermine whether MCSF and NRTN action are interdependent,

we cultured SMGs on stiff hydrogels with NRTN alone or incombination with anti-MCSF antibody, and performed the reverseexperiments with MCSF and anti-NRTN antibody. NRTNstimulation enhanced SMG growth and branching, but

Fig. 2. Effect of MCSF on SMG growth. SMG explants were isolated from E12.5 mice. (A,B) SMGs were cultured on 184 kPa stiff hydrogels in the presence ofdifferent concentrations of recombinant MCSF for 72 h. The optimal MCSF concentration for SMG growth was 20 ng/ml, as confirmed by the quantitativeanalysis (normalized to the value in control; denoted bud ratio) of bud number (B). (C,D) The inhibitory action of anti-CSF1R or anti-MCSF antibodies wasevaluated by culturing SMGs on track-etched membranes for 72 h. (D) Quantitative analysis of bud number showing that SMG growth was inhibited by theneutralization of either MCSF or its receptor with anti-MCSF and anti-CSF1R antibodies. For B and D, results are mean±s.d. Terminal end bud numbers wereobtained from 5–7 SMGs per group, and each experiment was repeated at least three times. **P≤0.01 (ANOVA with Scheffe’s F test). Scale bars: 100 µm.

1562

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

Fig. 3. MCSF regulates FGF7 and FGF10 expression during SMGmorphogenesis. (A) Western blot analysis of FGF7 and FGF10 in SMG explants culturedon stiff hydrogel (184 kPa) for 72 h, with or without MCSF supplementation. The same β-actin loading control is also shown in Figs 3A and 4A because thesame lysates were probed for the indicated proteins. (B,C) Quantitative analysis of relative intensity of protein bands normalized to β-actin (n=3). **P≤0.01(unpaired Student’s t-test). (D,E) Effect of MCSFon the growth of SMGs cultured on a track-etchedmembrane for 72 h, with concomitant neutralization of FGF7 orFGF10 with the specific antibodies. (E) Quantitative analysis of the number of buds (normalized to the value in control; denoted bud ratio). The bud numberdecreasedwith anti-FGF7 or anti-FGF10 antibody treatment, which was partially rescued byMCSF supplementation. (F,G) SMGswere cultured on a track-etchedmembrane for 72 h and the effect of FGF7 and FGF10 was evaluated upon MCSF neutralization. (G) Quantitative analysis of bud number. MCSF induced asignificant decrease in bud number, which could not be rescued with either FGF7 or FGF10. (H) Epithelium rudiments were cultured inside the Matrigel with orwithout MCSF supplementation for 72 h. (I) Quantitative analysis of bud number showing that MCSF directly stimulates the growth and elongation of theSMG epithelium. Results are in B, C, E, G and I are mean±s.d. For E,G (n=7) and I (n=5), terminal end bud numbers were obtained and each experiment wasrepeated at least three times. *P≤0.05, **P≤0.01 (ANOVA with Scheffe’s F test). Scale bars: 100 µm.

1563

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

Fig. 4. MCSFmodulates FGF expression via the PI3K signaling pathway. (A) Western blot analysis of total and phosphorylated PI3K (P-PI3K) levels in SMGexplants cultured on stiff hydrogel (184 kPa) with or without MCSF supplementation. The same β-actin loading control is also shown in Figs 3A and 5Bbecause the same lysates were probed for the indicated proteins. (B) Quantitative analysis of relative intensity of phosphorylated-to-total protein levels normalizedto β-actin levels (n=3). **P≤0.01, unpaired Student’s t-test. (C) SMGs were cultured on a track-etched membrane with LY294002 alone or in combination withMCSF. (D) Quantitative analysis of bud number (normalized to the value in control; denoted bud ratio). LY294002 suppressed MCSF-induced SMG growth.(E,F) SMGs were cultured on track-etched membrane with MCSF, LY294002 or LY294002+MCSF supplementation. Whole protein was extracted after 72 h andsubjected to western blot analysis of FGF7 and FGF10 expression. β-actin was used as a control. Results show that LY294002 inhibited FGF7 and FGF10expression even under MCSF stimulation (n=3). (G,H) SMGs were cultured on a track-etched membrane with LY294002 alone or combined with MCSF. After36 h, tissue samples were washed and the medium was replaced with control medium or MCSF-supplemented medium. After removal of LY294002, MCSFrestored and enhanced SMG growth compared to the unstimulated control group. Results are in B,D, F and H are mean±s.d. In D and H, terminal end budnumbers were obtained from 5–7 SMGs per group, and each experiment was repeated at least three times. **P≤0.01 (ANOVAwith Scheffe’s F test). Scale bars:100 µm.

1564

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

concomitant blockade of MCSF remarkably suppressed the NRTNeffect, thereby inhibiting neuronal outgrowth. In contrast, MCSFpromoted branching morphogenesis and the formation of a neuronalnetwork, even in the presence of anti-NRTN antibody (Fig. 5C–E).Furthermore, we mechanically isolated the parasympatheticganglion (PSG) and cultured the PSG on track-etched membrane,and found that MCSF supplementation can enhance neurite growth(Fig. S3). Collectively, these results indicate that MCSF plays acritical role in promoting neuronal growth, in part by inducing theNRTN expression in developing SMGs. Additionally, MCSFworksas an upstream factor of NRTN to promote neuronal growth.

MCSF is required for SMGgrowth and in vitro tissue synthesisSince MCSF plays pleiotropic roles in different cell types duringembryo development,Csf1op/Csf1op mice are not suitable models tounderstand the exact effects of MCSF on initial SMG development,

which takes place at ∼E11.5. Therefore, we administered anti-MCSF antibody to pregnant mice by intraperitoneal injectionstarting at E11 for 4 consecutive days (until E14) to neutralizeMCSF at early stages of SMG development. Phenotypic analysisshowed that there was no difference in body size between embryosisolated from pregnant mice treated with anti-MCSF antibody orinjected with phosphate-buffered saline (PBS). However, SMGsfrom MCSF-neutralized embryos were notably atrophic (Fig. 6A).Histological analysis revealed a loose mesenchymal tissue alongwith a disconnected intercellular space and lack of a tight interactionbetween cells in the epithelium and mesenchyme, a defectiveneuronal network, and fewer epithelial buds and F4/80+ cells(Fig. 6B–D). The expression of FGF7, FGF10 and NRTN was alsodecreased (Fig. S4), ratifying the results of the organ explantexperiments (Figs 3,5). These results confirm that MCSF isessential for SMG development and acts by directly regulating

Fig. 5. MCSF controls neuronal growth. (A) SMGs cultured on a stiff hydrogel (184 kPa) without MCSF showed inhibition of neuronal network and whole SMGgrowth, which were restored byMCSF supplementation. (B)Western blot analysis of NRTN expression in SMGexplants cultured on a stiff hydrogel with or withoutMCSF supplementation. The same β-actin loading control is also shown in Figs 3A and 4A because the same lysates were probed for the indicated proteins.Quantitative analysis of the relative intensity of protein bands normalized to β-actin (n=3). *P≤0.05 (unpaired Student’s t-test). (C) Effect of MCSF on the growth ofSMGs cultured on a stiff hydrogel was evaluated after blockade of NRTN and vice-versa. (D) Quantitative analysis of bud number (normalized to the value incontrol; denoted bud ratio). MCSF rescued the inhibition of SMG growth induced by anti-NRTN antibody. However, blocking of MCSF dramatically suppressedSMG growth, which was not rescued by concomitant NRTN application. *P≤0.05, **P≤0.01 (ANOVA with Scheffe’s F test). Terminal end bud numbers wereobtained from 5–7 SMGs per group, and each experiment was repeated at least three times. (E) MCSF promoted the formation of a neuronal network even in thepresence of anti-NRTN antibody, as determined by immunohistochemistry (red, β-III-tubulin; green, F4/80). Results are in B and D are mean±s.d. Scale bars:100 µm.

1565

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

FGF7 and FGF10 levels, as well as neuronal network formation andmacrophage differentiation.Finally, we used an in vitro SMG tissue synthesis method (Wei

et al., 2005) to confirm the effects of MCSF on the early stages ofepithelial bud formation of SMG development. In this experiment,epithelial and mesenchymal cells were first dissociated from wholeSMG tissue (E13), and subsequently co-cultured in Matrigel. In thecontrol group, self-assembly and epithelial bud formation wereobserved within 48 h of culture. After 72 h, numerous buds weredetected along with a complex network of neurons. However, whenthe cells were cultured with anti-MCSF antibody, bud formationwas suppressed and the distribution of neurons within the tissue wasalmost completely abolished (Fig. 7).

DISCUSSIONOrganogenesis, including SMG branching morphogenesis, involvesa complex and coordinated sequence of events that regulate thegrowth, proliferation, differentiation, migration and apoptosis ofepithelial and mesenchymal cells. These events are mediated byspecific and time-dependent activation of genes and intracellularsignaling pathways in response to developmental cues. Theimportance of immune cells in tissue development and regeneration

has been revealed by recent studies (Lilla and Werb, 2010; Reed andSchwertfeger, 2010), especially the trophic and scavenging roles ofmacrophages (Gouon-Evans et al., 2000). In the present study, weshowed thatM2macrophages accumulated in developing SMG tissueat E15.5, which was preceded by an upregulation of MCSF andCSF1R levels at E13.5, suggesting that MCSF could be the maintrigger for macrophage polarization during SMG tissue development.More importantly, we demonstrated for the first time that MCSFregulated early SMG development by modulating the expression ofFGF7 and FGF10, and neural network development.

FGFs are critical for SMG formation and development; inparticular, the temporal regulation of FGF7 and FGF10 in branchingmorphogenesis is a determinant of SMG formation and growth(Ohuchi et al., 2000; Ornitz and Itoh, 2015). The PI3K pathway hasbeen reported to be involved in FGF7 and FGF10 signaltransduction (Eswarakumar et al., 2005, 2005; Larsen et al.,2003). Moreover, previous data have also shown that PI3K isdownstream of MCSF and CSF1R (Yavropoulou and Yovos, 2008),and regulates cell proliferation. Thus, consistent with these reports,our results indicate that MCSF binding to CSF1R and subsequentactivation of PI3K pathway could be the major mechanismassociated with MCSF-induced FGF7 and FGF10 expression.

Fig. 6. Neutralization of MCSF in vivo inhibits SMG growth. (A) SMGs from embryos isolated from pregnant mice treated with anti-MCSF antibody wereatrophic, compared to control embryos from PBS-injected pregnant mice. Images are representative of three independent experiments. (B) Histological analysisof SMGs revealed a loose epithelial–mesenchymal tissue in MCSF-neutralized embryos. (C) A defective neuronal network (red, β-III-tubulin) and fewer F4/80+

macrophages (green) were detected in MCSF-neutralized embryos relative to controls. (D) Quantitative analysis showing reduced number of buds (n=6) andF4/80+ cells (n=4) in SMGs isolated from MCSF-neutralized embryos, compared to control embryos (mean±s.d.). **P≤0.01 (unpaired Student’s t-test). Scalebars: 100 µm.

1566

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

Alternatively, MCSF or CSF1R could act as a co-receptor for theFGF–FGFR interaction and thereby regulate FGF7 and FGF10function. Since the major receptors for FGF7 and FGF10,FGFR2IIIb (FGFR2b) and FGFR1b respectively, are expressedthroughout the epithelium (Steinberg et al., 2005), a co-receptorwould probably be necessary to allow FGF to specifically stimulatethe proliferation of SMG cells during branching (Steinberg et al.,2005). In this context, a previous study reported that the mitogenicactivity of FGF10 is mainly stimulated, whereas that of FGF7 isinhibited by the proteoglycan heparan sulfate (Igarashi et al., 1998),suggesting that the effects of FGF7 and FGF10 could be regulatedby heparan sulfate. In fact, heparan sulfate regulates the activity ofFGFs by acting as a co-receptor and enhancing FGF–FGFR bindingaffinity (Patel et al., 2007; Knelson et al., 2014). Since the secretedform ofMCSF is also a glycoprotein (Stanley et al., 1994), it may bepossible that MCSF also acts as a co-receptor in the FGF–FGFRinteraction and thus regulates FGF7 and FGF10 function at the cellmembrane level.MCSF has potent neuroprotective effects following auditory

injury and in neuroinflammation (Smith et al., 2013; Yagihashiet al., 2005). We found thatMCSF induces the expression of NRTN,which can explain at least in part the MCSF-induced formation of aneuronal network surrounding SMG tissue. Our findings indicatethat NRTN-induced neuronal growth or neuroprotection actsdownstream of MCSF, since blocking of NRTN did notcompletely inhibit MCSF-induced SMG growth. On the otherhand, blocking MCSF completely suppressed SMG growth even inthe presence of NRTN. It is therefore possible that MCSF-inducedneuroprotection could involve other neuroprotective or neurotrophicagents. Alternatively, MCSF could also have a NRTN-independenteffect on neurons, and promote their growth, in a similar manner totheir effects on microglia and Purkinje cells.Previous studies showed that, due to alternative mRNA splicing

and post-translational modifications, MCSF can either be secretedas a glycoprotein into the circulation or be expressed on the cellsurface as a membrane-spanning glycoprotein (Stanley et al., 1994).Interestingly, transgenic expression of cell surface MCSF rescuedmost of the defects observed in Csf1op/Csf1op mice, includinggrowth retardation and defective tooth eruption (Dai et al., 2004),indicating that this is the major isoform responsible for regulatingtissue development. Nevertheless, although this report described a

restoration of F4/80+ macrophage populations in salivary glands to∼80% of normal levels upon expression of a full-length MCSF-1transgene (Dai et al., 2004), the importance of these two isoforms inthe process of branching morphogenesis was not examined in detail.Further investigation may clarify the different mechanisms of actionof the two MCSF isoforms in SMG branching morphogenesis andin the regulation of FGF and NRTN expression.

Although much has been learned from the study of individualsignaling pathways that regulate SMG growth and morphogenesis,the interaction between these pathways and how they fit into ahierarchical signaling system has remained obscure. Here, wedemonstrated that MCSF plays crucial roles in SMG development.The proposed model for the MCSF role in SMG growth is thatMCSF controls branching morphogenesis in both a direct andindirect manner. MCSF binds to the CSF1R receptor present in theepithelium and controls the epithelium growth directly, indicatingthat it has a macrophage-independent role. Additionally, whenMCSF binds with the receptor present in mesenchymal cells, wesuggest that it can promote the expression of FGFs via the PI3Kpathway; MCSF also promoted neuronal network growth indeveloping SMG by inducing the expression of NRTN. Finally,since a similar process of branching morphogenesis occurs in otherglandular tissues, MCSF could also play important and direct rolesin epithelial branching in the lungs, pancreas and mammary glands.

MATERIALS AND METHODSAnimal experimentsPregnant ICR mice were purchased from Charles River Laboratories(Yokohama, Japan). Animal procedures strictly adhered to the Guidelinesfor Animal Experiments of Okayama University and were carried out withthe approval of the Animal Use and Care Committee of Okayama University(OKU-2013033).

Preparation of alginate hydrogel sheetsThe preparation of alginate hydrogel sheets has been previously described(Miyajima et al., 2011). Briefly, sodium alginate solution (4% w/w; WakoPure Chemical Industries, Osaka, Japan) was poured into a mold made froma porous alumina plate; this was soaked in calcium chloride solution (5%w/w)(Nacalai Tesque, Kyoto, Japan) for 1.5 h. The porous aluminamold effectivelyprevented any unexpected shrinkage of the formed gel because the molduniformly supplied calcium chloride solution, a cross-linking agent. Theobtained alginate hydrogel sheet was washed with ethanol and Milli-Q

Fig. 7. MCSF is necessary for epithelial differentiation and budding in in vitro SMG tissue synthesis. SMGs were isolated from E13 embryos and singlecells were dissociated by enzymatic digestion. Single cells were then pelleted and co-cultured in Matrigel for self-assembly. (A) Results showed that after72 h, blockade of MCSF completely abrogated the bud formation and neuronal innervation (red, Tubb3). Scale bar: 100 µm. (B) Graph shows the reduction in thebud number (normalized to the value in control; denoted bud ratio) upon anti-MCSF treatment during the 3-day culture period (mean±s.d.; n=5). **P≤0.01(unpaired Student’s t-test).

1567

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

ultrapure water (Millipore, Billerica, MA) and stored in Dulbecco’s modifiedEagle’s medium with nutrient mixture F-12 (DMEM/F12; Wako PureChemical Industries) supplemented with 1% penicillin-streptomycin (NacalaiTesque) (DMEM/F12/PS) for at least 24 h. The sheet was then cut into smallpieces (10×10×1.5 mm) for organ culture.

SMG culture on alginate hydrogel sheet and track-etchedmembraneSMG tissues extracted from E12.5 ICRmouse embryos were placed directlyon the alginate hydrogel sheets with a mechanical stiffness of 184 kPa asnegative controls, or on Whatman Nuclepore track-etched membranes(13 mm diameter, 0.1 μm pore size) as positive controls (GE Healthcare,Little Chalfont, UK). The SMG culture was carried out in DMEM/F12/PS at37°C in a humidified atmosphere of 5% CO2 and 95% air. The medium wasreplenished every 24 h. Recombinant MCSF (5, 10, 20 or 50 ng/ml),recombinant NRTN (1 ng/ml), recombinant FGF7 (100 ng/ml),recombinant FGF10 (500 ng/ml) or goat anti-NRTN antibody (25 μg/ml;cat. no AF477) (all from R&D Systems, Minneapolis, MN); goat anti-MCSF (25 µg/ml; cat. no sc-1324), rabbit anti-CSF1R (10 µg/ml; cat. nosc-692), or rabbit anti-FGF7 (1 µg/ml; cat. no sc-7882) antibody (all fromSanta Cruz Biotechnology, Santa Cruz, CA); rabbit anti-FGF10 antibody(1 µg/ml; cat. no ABN44, Millipore); or LY294002 hydrochloride (25 μM)(Sigma-Aldrich, St. Louis, MO) were added to the medium depending onthe experiment. SMG growth and morphological changes were detected bylight microscopy (TE-2000; Nikon, Tokyo, Japan). The number of buds indeveloping SMGs was counted at 0, 36 or 72 h, or at 0, 24, 48 or 72 h usingImage J software (National Institutes of Health, Bethesda, MD). Terminalend bud numbers were obtained from 5–7 SMGs per group, and eachexperiment was repeated at least three times.

SMG epithelium rudiment and epithelial–mesenchymal cellaggregates cultured in MatrigelThe SMG was isolated from E13 mice. To obtain the epithelium rudiment,the SMG was treated with 4 U/ml Dispase (Roche, Basel, Switzerland) for5 min at room temperature and then separated into epithelial andmesenchymal tissues using fine forceps. Rudiments were placed insidetheMatrigel (BDBiosciences, Bedford, MA) and incubated at 37°C for 72 hin DMEM/F12 containing 10% fetal bovine serum (FBS) with or without20 ng/ml MCSF supplementation. To obtain the epithelial andmesenchymal cell aggregates, whole glands were enzymatically digestedfor 10 min using collagenase type I (50 U/ml) (Worthington BiochemicalCorporation, Lakewood, NJ) followed by digestion with 0.25% trypsin-EDTA (Sigma-Aldrich) for 5 min at 37°C on a rotary shaker. Tissues weredissociated by gentle pipetting, and single cells were obtained by passing thecell suspension through a 70-µm nylon strainer (BD Falcon, Durham, NC).In vitro branching morphogenesis was reconstituted by placing theepithelial–mesenchymal cell aggregates in Matrigel and incubating at37°C for 72 h in DMEM/F12 containing 10% FBS, 20 ng/ml EGF (R&DSystems) and 100 ng/ml FGF7 (R&D Systems) with or without 25 µg/mlanti-MCSF antibody. The number of buds per mm2 in obtained images wascounted at 0, 24, 48 and 72 h using ImageJ software.

ImmunohistochemistryFor immunohistochemical analysis, whole-mount SMGs or epithelialrudiments and mesenchyme alone were fixed with 4% PFA for 20 min atroom temperature andwashedwith PBS (Takara Bio, Otsu, Japan) containing1% bovine albumin serum (Nacalai Tesque) and 0.1% Triton X-100 (Sigma-Aldrich) (PBSX). Tissue samples were blocked with Blocking One Histo(Nacalai Tesque) and incubatedwith the following primary antibodies dilutedin PBSX: rat anti-F4/80 (1:100; cat. no MCA497, AbD Serotec, Kidlington,UK), rabbit anti-E-cadherin (1:500; cat. no ab76055, Abcam, Cambridge,UK), mouse anti-β-III-tubulin (Tubb3; 1:1000; cat. no MAB1195, R&DSystems), rabbit anti-CSF1R (1:500; cat. no sc-692), rabbit anti-ARG1(1:500; cat. no sc-101199) and goat anti-MCSF (1:150; cat. no sc-1324) (allfrom Santa Cruz Biotechnology), and rabbit anti-iNOSII (1:500; cat. noABN26, Millipore), as well as fluorescein isothiocyanate-conjugated lectinfrom Arachis hypogaea (1:200) (Sigma-Aldrich). Antibody binding was

detected with Alexa Fluor-conjugated secondary antibodies (LifeTechnologies, Grand Island, NY), and specimens were imaged by confocalmicroscopy (C1; Nikon).

Western blot analysisFor western blot analysis, samples were separated by 10% SDS-PAGE andtransferred to polyvinylidene difluoride membranes using a Mini Trans-Blotsystem (Bio-Rad,Hercules, CA).Western blottingwas performed according tostandard procedures using primary antibodies against the following proteins:FGF7 (1:1000; cat. no sc-7882) and MCSF (1:500; cat. no sc-1324) (SantaCruz Biotechnology); FGF10 (1:2000; cat. no ABN44, Millipore); NRTN(1:1000; cat. no AF477, R&D Systems); PI3K p85 (1:1000; cat. no 4292) andphospho-PI3K p85 (1:1000; cat. no 4228) (Cell Signaling Technology, MA).After incubation with horseradish peroxidase (HRP)-conjugated secondaryantibodies, protein bands were visualized using Luminata Forte Western HRPsubstrate (Millipore) and a charge-coupled device-type imager (Image QuantLAS 4000 mini; GE Healthcare) according to the manufacturers’ instructions.The blots shown in the figures are representative of three experimental repeats.

In vivo neutralization of MCSFThe procedure for the and amount of the antibody administration was asdescribed previously (Wei et al., 2005) but with modifications. Anti-MCSFantibody (R&D Systems) (1.5 µg/g of body weight/day) diluted in PBS wasadministered twice daily for 4 consecutive days to timed pregnant ICR mice(E11–E14)by intraperitoneal injection. PBS (200 µl)was injectedas anegativecontrol. At day E14.5, SMG samples were collected, fixed, and processed foranalysis. The experiment was repeated three times with similar results.

Statistical analysisAll SMG data were obtained and mean values with standard deviationswere calculated. Statistical significance was taken as P<0.05 as determinedwith a Student’s t-test or one-way ANOVA, with Scheffe’s F test, whenappropriate.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsG.A.S. and T.M. designed the experiments and analyzed the data. G.A.S., M.F.,E.S.H., H.T., H.N., T.K., and T.M. performed the experiments. G.A.S., E.S.H., andT.M. wrote the paper and prepared the figures, and approved the final manuscript.

FundingThis work was supported by the Japan Society for the Promotion of Science (grantJP24106508 and JP26106718).

Supplementary informationSupplementary information available online athttp://jcs.biologists.org/lookup/doi/10.1242/jcs.196907.supplemental

ReferencesBanaei-Bouchareb, L., Gouon-Evans, V., Samara-Boustani, D., Castellotti,

M. C., Czernichow, P., Pollard, J. W. and Polak, M. (2004). Insulin cell mass isaltered in Csf1op/Csf1op macrophage-deficient mice. J. Leukoc. Biol. 76,359-367.

Cecchini, M. G., Dominguez,M. G.,Mocci, S., Wetterwald, A., Felix, R., Fleisch, H.,Chisholm, O., Hofstetter, W., Pollard, J. W. and Stanley, E. R. (1994). Role ofcolonystimulating factor-1 in the establishment and regulation of tissuemacrophagesduring postnatal development of the mouse. Development 120, 1357-1372.

Dai, X.-M., Zong, X.-H., Sylvestre, V. and Stanley, E. R. (2004). Incompleterestoration of colony-stimulating factor 1 (CSF-1) function in CSF-1-deficientCsf1op/Csf1op mice by transgenic expression of cell surface CSF-1. Blood 103,1114-1123.

Davies, L. C., Jenkins, S. J., Allen, J. E. and Taylor, P. R. (2013). Tissue-residentmacrophages. Nat. Immunol. 14, 986-995.

Eswarakumar, V. P., Lax, I. and Schlessinger, J. (2005). Cellular signaling byfibroblast growth factor receptors. Cytokine Growth Factor Rev. 16, 139-149.

Fleetwood, A. J., Lawrence, T., Hamilton, J. A. and Cook, A. D. (2007).Granulocyte-Macrophage Colony-Stimulating Factor (CSF) and MacrophageCSF-dependent macrophage phenotypes display differences in cytokine profilesand transcription factor activities: implications for CSF blockade in inflammation.J. Immunol. 178, 5245-5252.

1568

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience

Geutskens, S. B., Otonkoski, T., Pulkkinen, M.-A., Drexhage, H. A. and Leenen,P. J. M. (2005). Macrophages in themurine pancreas and their involvement in fetalendocrine development in vitro. J. Leukoc. Biol. 78, 845-852.

Gouon-Evans, V., Rothenberg, M. E. and Pollard, J. W. (2000). Postnatalmammary gland development requires macrophages and eosinophils.Development 127, 2269-2282.

Gumbiner, B. M. (1992). Epithelial morphogenesis. Cell 69, 385-387.Hesse, M., Modolell, M., La Flamme, A. C., Schito, M., Fuentes, J. M., Cheever,A. W., Pearce, E. J. and Wynn, T. A. (2001). Differential Regulation of nitric oxidesynthase-2 and arginase-1 by type 1/type 2 cytokines in vivo: granulomatouspathology is shaped by the pattern of L-arginine metabolism. J. Immunol. 167,6533-6544.

Heuckeroth, R.O., Enomoto, H., Grider, J.R., Golden, J.P., Hanke, J.A.,Jackman, A., Molliver, D.C., Bardgett, M.E., Snider, W.D. and Johnson, E.M. Jr, Milbrandt, J. (1999). Gene targeting reveals a critical role for neurturin inthe development and maintenance of enteric, sensory, and parasympatheticneurons. Neuron. 22, 253–263.

Igarashi, M., Finch, P. W. and Aaronson, S. A. (1998). Characterization ofrecombinant human fibroblast growth factor (FGF)-10 reveals functional similaritieswith keratinocyte growth factor (FGF-7). J. Biol. Chem. 273, 13230-13235.

Ingber, D. E. (2006). Mechanical control of tissue morphogenesis duringembryological development. Int. J. Dev. Biol. 50, 255-266.

Ingman, W. V., Wyckoff, J., Gouon-Evans, V., Condeelis, J. and Pollard, J. W.(2006). Macrophages promote collagen fibrillogenesis around terminal end budsof the developing mammary gland. Dev. Dyn. 235, 3222-3229.

Jones, C. V., Williams, T. M., Walker, K. A., Dickinson, H., Sakkal, S., Rumballe,B. A., Little, M. H., Jenkin, G. and Ricardo, S. D. (2013). M2 macrophagepolarisation is associated with alveolar formation during postnatal lungdevelopment. Respir. Res. 14, 41.

Kadoya, Y., Kadoya, K., Durbeej, M., Holmvall, K., Sorokin, L. and Ekblom, P.(1995). Antibodies against domain E3 of laminin-1 and integrin α6 subunit perturbbranching epithelial morphogenesis of submandibular gland, but by differentmodes. J. Cell Biol. 129, 521-534.

Knelson, E. H., Nee, J. C. and Blobe, G. C. (2014). Heparan sulfate signaling incancer. Trends Biochem Sci. 39, 277-288.

Knox, S. M., Lombaert, I. M. A., Reed, X., Vitale-Cross, L., Gutkind, J. S. andHoffman, M. P. (2010). Parasympathetic innervation maintains epithelialprogenitor cells during salivary organogenesis. Science 329, 1645-1647.

Knox, S. M., Lombaert, I. M. A., Haddox, C. L., Abrams, S. R., Cotrim, A.,Wilson,A. J. and Hoffman, M. P. (2013). Parasympathetic stimulation improves epithelialorgan regeneration. Nat. Commun. 4, 1494.

Larsen, M., Hoffman, M. P., Sakai, T., Neibaur, J. C., Mitchell, J. M. and Yamada,K. M. (2003). Role of PI 3-kinase and PIP3 in submandibular gland branchingmorphogenesis. Dev. Biol. 255, 178-191.

Lilla, J. N. andWerb, Z. (2010).Mast cells contribute to the stromalmicroenvironmentin mammary gland branching morphogenesis. Dev. Biol. 337, 124-133.

Liu, W.-G., Lu, G.-Q., Li, B. and Chen, S.-D. (2007). Dopaminergic neuroprotectionby neurturin-expressing c17.2 neural stem cells in a rat model of Parkinson’sdisease. Park. Relat. Disord. 13, 77-88.

Miyajima, H., Matsumoto, T., Sakai, T., Yamaguchi, S., An, S. H., Abe, M.,Wakisaka, S., Lee, K. Y., Egusa, H. and Imazato, S. (2011). Hydrogel-basedbiomimetic environment for in vitro modulation of branching morphogenesis.Biomaterials 32, 6754-6763.

Murase, S. and Hayashi, Y. (1998). Expression pattern and neurotrophic role of thec-fms proto-oncogene M-CSF receptor in rodent Purkinje cells. J. Neurosci. 18,10481-10492.

Nakanishi, Y., Sugiura, F., Kishi, J.-I. and Hayakawa, T. (1986). Collagenaseinhibitor stimulates cleft formation during early morphogenesis of mouse salivarygland. Dev. Biol. 113, 201-206.

Ohuchi, H., Hori, Y., Yamasaki, M., Harada, H., Sekine, K., Kato, S. and Itoh, N.(2000). FGF10 acts as a major ligand for FGF receptor 2 IIIb in mouse multi-organdevelopment. Biochem. Biophys. Res. Commun. 277, 643-649.

Ornitz, D. M. and Itoh, N. (2015). The Fibroblast Growth Factor signaling pathway.Wiley Interdiscip. Rev. Dev. Biol. 4, 215-266.

Patel, V. N., Knox, S. M., Likar, K. M., Lathrop, C. A., Hossain, R., Eftekhari, S.,Whitelock, J. M., Elkin, M., Vlodavsky, I. and Hoffman, M. P. (2007).Heparanase cleavage of perlecan heparan sulfate modulates FGF10 activityduring ex vivo submandibular gland branching morphogenesis. Development134, 4177-4186.

Patel, N., Sharpe, P. T. and Miletich, I. (2011). Coordination of epithelial branchingand salivary gland lumen formation by Wnt and FGF signals. Dev. Biol. 358,156-167.

Porazinski, S., Wang, H., Asaoka, Y., Behrndt, M., Miyamoto, T., Morita, H.,Hata, S., Sasaki, T., Krens, S. F. G., Osada, Y. et al. (2015). YAP is essential fortissue tension to ensure vertebrate 3D body shape. Nature 521, 217-221.

Pyonteck, S. M., Akkari, L., Schuhmacher, A. J., Bowman, R. L., Sevenich, L.,Quail, D. F., Olson, O. C., Quick, M. L., Huse, J. T., Teijeiro, V. et al. (2013).CSF-1R inhibition alters macrophage polarization and blocks glioma progression.Nat. Med. 19, 1264-1272.

Reed, J. R. and Schwertfeger, K. L. (2010). Immune cell location and functionduring post-natal mammary gland development. J. Mammary Gland Biol.Neoplasia 15, 329-339.

Sakai, T., Larsen, M. and Yamada, K. M. (2003). Fibronectin requirement inbranching morphogenesis. Nature 423, 876-881.

Schulz, C., Gomez Perdiguero, E., Chorro, L., Szabo-Rogers, H., Cagnard, N.,Kierdorf, K., Prinz, M.,Wu, B., Jacobsen, S. E.W., Pollard, J.W. et al. (2012). Alineage of myeloid cells independent of Myb and hematopoietic stem cells.Science 336, 86-90.

Shibata, Y., Zsengeller, Z., Otake, K., Palaniyar, N. and Trapnell, B. C. (2001).Alveolar macrophage deficiency in osteopetrotic mice deficient in macrophagecolony-stimulating factor is spontaneously corrected with age and associated withmatrix metalloproteinase expression and emphysema. Blood 98, 2845-2852.

Smith, A. M., Gibbons, H. M., Oldfield, R. L., Bergin, P. M., Mee, E. W., Curtis,M. A., Faull, R. L. M. and Dragunow, M. (2013). M-CSF increases proliferationand phagocytosis while modulating receptor and transcription factor expression inadult human microglia. J. Neuroinflamm. 10, 859.

Stanley, E. R., Berg, K. L., Einstein, D. B., Lee, P. S. and Yeung, Y. G. (1994). Thebiology and action of colony stimulating factor-1. Stem Cells 12 Suppl. 1, 15-24;discussion 25.

Steinberg, Z., Myers, C., Heim, V. M., Lathrop, C. A., Rebustini, I. T., Stewart,J. S., Larsen, M. and Hoffman, M. P. (2005). FGFR2b signaling regulates ex vivosubmandibular gland epithelial cell proliferation and branching morphogenesis.Development 132, 1223-1234.

Taketa, H., Sathi, G. A., Farahat,M., Rahman, K. A., Sakai, T., Hirano, Y., Kuboki,T., Torii, Y. andMatsumoto, T. (2015). Peptide-modified substrate for modulatinggland tissue growth and morphology in vitro. Sci. Rep. 5, 11468.

Van Nguyen, A. and Pollard, J. W. (2002). Colony stimulating factor-1 is required torecruit macrophages into the mammary gland to facilitate mammary ductaloutgrowth. Dev. Biol. 247, 11-25.

Vlahos, C. J., Matter, W. F., Hui, K. Y. and Brown, R. F. (1994). A specific inhibitorof phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY294002). J. Biol. Chem. 269, 5241-5248.

Walker, J. L., Menko, A. S., Khalil, S., Rebustini, I., Hoffman, M. P., Kreidberg,J. A. and Kukuruzinska, M. A. (2008). Diverse roles of E-cadherin in themorphogenesis of the submandibular gland: insights into the formation of acinarand ductal structures. Dev. Dyn. 237, 3128-3141.

Wei, S., Lightwood, D., Ladyman, H., Cross, S., Neale, H., Griffiths, M., Adams,R., Marshall, D., Lawson, A., McKnight, A. J. et al. (2005). Modulation of CSF-1-regulated post-natal development with anti-CSF-1 antibody. Immunobiology 210,109-119.

Yagihashi, A., Sekiya, T. and Suzuki, S. (2005). Macrophage colony stimulatingfactor (M-CSF) protects spiral ganglion neurons following auditory nerve injury:Morphological and functional evidence. Exp. Neurol. 192, 167-177.

Yang, T.-L. and Young, T.-H. (2008). The enhancement of submandibular glandbranch formation on chitosan membranes. Biomaterials 29, 2501-2508.

Yavropoulou, M. P. and Yovos, J. G. (2008). Osteoclastogenesis–currentknowledge and future perspectives. J. Musculoskelet. Neuronal Interact. 8,204-216.

1569

RESEARCH ARTICLE Journal of Cell Science (2017) 130, 1559-1569 doi:10.1242/jcs.196907

Journal

ofCe

llScience