Netrin-1 functions as a suppressor of bone morphogenetic protein (BMP) signaling

Netrin-1 is a secreted protein that is well known for its involvement in axonal guidance during embryonic development and as an enhancer of cancer cell metastasis. Despite extensive efforts, the molecular mechanisms behind many of the physiological functions of netrin-1 have remained elusive. Here, we show that netrin-1 functions as a suppressor of bone morphogenetic protein (BMP) signaling in various cellular systems, including a mutually inhibitory interaction with the BMP-promoting function of leucine-rich repeats and immunoglobulin-like domains (LRIG) proteins. The BMP inhibitory function of netrin-1 in mouse embryonic fibroblasts was dependent on the netrin receptor neogenin, with the expression level regulated by both netrin-1 and LRIG proteins. Our results reveal a previously unrecognized function of netrin-1 that may help to explain several of the developmental, physiological, and cancer-promoting functions of netrins at the signal transduction level.

www.nature.com/scientificreports/ DCC, neogenin, UNC5A, UNC5B, UNC5C, and UNC5D function as 'dependence receptors' , i.e., receptors that initiate apoptotic signaling in the absence of ligands 20,21 . Intriguingly, the molecular mechanisms behind many of the developmental, physiological, and tumor-promoting functions of netrin-1 remain poorly characterized. Given the many prominent developmental and physiological functions of netrin-1, it is important to elucidate the molecular mechanisms by which these functions are executed. Here, we show that netrin-1 negatively regulates bone morphogenetic protein (BMP) signaling in different cellular contexts. This discovery might help to explain many previous observations regarding the physiological and pathological functions attributed to this pleiotropic protein.
Phospho Smad1/5 immunofluorescence assay. Cells were seeded at a density of 3,000 cells per well in a 96-well microtiter plate (#655090, Greiner Bio-One International GmbH, Monroe, NC, USA) coated with 0.1% bovine gelatin. After an overnight incubation, the cells were starved in serum-free medium for 1 h followed by the induction of BMP signaling using different concentrations of BMP4, BMP6, or BMP9. After an hour of BMP treatment, the cells were washed with phosphate-buffered saline (PBS) and fixed with 4% formaldehyde. For the cross-titration studies, Ntn1 −/− , Lrig-null MEFs with Tet-On inducible LRIG1-FLAG or LRIG3-FLAG alleles were used. The cells were treated with different concentrations of doxycycline overnight. The next day, the cells were serum-starved for an hour followed by the addition of 5 ng/ml BMP4 with different amounts of netrin-1. After 1 h of incubation, the cells were washed with PBS and fixed with 4% formaldehyde. To investigate the kinetics of netrin-1 inhibition, 0.25 μg/ml netrin-1 was added to wild-type MEFs at different timepoints before, with, or after the addition of BMP4. Sixty minutes after the addition of BMP4, the cells were washed with PBS and fixed with 4% formaldehyde. Then, the pSmad assay was conducted as described previously 25 .
RNA extraction and quantitative RT-PCR. RNA extraction was carried out using a PureLink RNA Mini Kit (Invitrogen, Fisher Scientific GTF AB) with PureLink DNase (Invitrogen, Fisher Scientific GTF AB) treatment. A TaqMan assay was used to quantify the gene expression of Id1 (Mm00775963_g1, Fisher Scientific GTF AB, #4331182), and normalization was performed using the reference gene RN18S as an internal control as described previously 26 . Data acquisition was performed using a CFX96 system C1000 thermal cycler (Bio-Rad Laboratories AB).  26 were transfected using Fugene6 transfection reagent (Promega Biotech AB, Nacka, Sweden), as previously described 26 , using the CRISPR/Cas9 HDR-mediated knockout kit KN311288RB (Ntn1) or KN310902RB (Neo1) from OriGene Technologies (BioNordika Sweden AB, Solna, Sweden) and a donor plasmid with a Blasticidin S resistance gene together with either a CRISPR/ Cas9 plasmid guide-RNA for the Ntn1 or Neo1 gene or a scramble control plasmid. Forty-eight hours after trans- www.nature.com/scientificreports/ fection, 10 μg/ml blasticidin S (Gibco, Fisher Scientific GTF AB) was added to the cells. After two weeks of selection with blasticidin S, the resistant MEFs were cloned, yielding three Ntn1 knockout MEF clones together with four scramble wild-type control MEF clones and four Neo1 knockout MEF clones together with four scramble wild-type control MEF clones. The Ntn1 knockout MEF clones were then transduced with adenovirus carrying either Cre recombinase or control to create Lrig-null MEFs as previously described 26 .
Western blot analysis. Cells were seeded in 6-or 12-well plates the day before stimulation with netrin-1 and BMP4. On the day of stimulation, the cells were starved in FBS-free media for an hour prior to stimulation with 10 ng/ml BMP4 and 0.25 μg/ml netrin-1. After one hour of stimulation, the cells were washed with PBS and lysed with lysis buffer (Invitrogen, Fisher Scientific GTF AB) containing a protease inhibitor cocktail (Roche Diagnostics Scandinavia AB, Bromma, Sweden) and the phosphatase inhibitor PhosSTOP (Roche Diagnostics Scandinavia AB). The cell lysates were kept on ice for 30 min and centrifuged at 20,800×g for 10 min at 4 °C. The clear supernatant was subjected to SDS-PAGE using 3-8% Tris-acetate gels (#EA03752, Invitrogen, Fisher Scientific GTF AB) followed by transfer onto nitrocellulose or PVDF membranes (Bio-Rad Laboratories AB, Solna, Sweden). The membranes were blocked using Odyssey blocking buffer (LI-COR Biosciences GmbH, Bad Homburg, Germany). The blocked membranes were then incubated with the primary antibodies at room temperature for an hour. Thereafter, the membranes were washed three times with TBST (20 mM Tris, 150 mM NaCl, 0.1% Tween 20), incubated with the fluorescent secondary antibodies for an hour at room temperature and then washed again three times with TBST. The fluorescent bands were analyzed using the Odyssey CLx imaging system (LI-COR Biosciences GmbH).
Adipogenesis assay. The adipogenesis assay has been described elsewhere 26 . Briefly, the MEFs were treated with an adipogenic cocktail with or without 0.3 μg/ml netrin-1 for 9 days. Then, the cells were fixed, and the adipocytes were stained using Oil Red O (Sigma-Aldrich Sweden AB). The Oil Red O-stained cells were quantified in a Spectramax i3x plate reader (Molecular Devices, San Jose, CA, USA) using Softmax Pro 7 software (Molecular Devices) as described elsewhere 26 .
Alkaline phosphatase activity assay. For the alkaline phosphatase (ALP) activity assay, 3,000 ATDC5 cells were seeded per well in a 96-well microtiter plate and cultured overnight. The next day, the culture medium was replaced with DMEM:Ham′s F12 (1:1) containing 2 mM glutamine and supplemented with 2% FBS, 2 µg/ml heparin (Sigma Aldrich Sweden AB, #H3149), and 50 μg/ml gentamicin. BMP4, netrin-1, or noggin was added at different concentrations as indicated in the Results section. After 72 h of incubation, the cells were washed with PBS, and the ALP activity assay was carried out using a colorimetric kit (Abcam, Sweden, #ab83369) according to the manufacturer's guidelines.

Statistical analyses.
All the data are presented as the averages of three or more independent experiments with standard deviations. Statistical comparisons were performed using a two-sided Student's t-test or a one-way analysis of variance (ANOVA) with multiple comparisons using Dunnett's test (GraphPad Prism, version 5.0), and a p-value of less than 0.05 was considered statistically significant.
To investigate the kinetics of netrin inhibition, netrin-1 was added at different time points before, with, or after the initiation of BMP stimulation (Fig. 1d). There was a stronger inhibition the earlier netrin-1 was added to the cells; however, the addition of netrin-1 after the initiation of BMP4 stimulation also inhibited signaling. To investigate whether the observed inhibition of Smad1/5 phosphorylation translated into altered gene expression, BMP4-induced Id1 expression was assessed through qRT-PCR (Fig. 1e). Obviously, the BMP4-induced expression of Id1 was also inhibited by netrin-1.

Netrin-1 suppresses BMP signaling in various cell types.
To investigate the generality of the observed netrin-1-induced suppression of BMP signaling, a panel of mammalian cell lines consisting of an MEF line together with twelve human cell lines was evaluated to determine their sensitivity to netrin-1-mediated inhibition. The human cell lines were chosen to represent different tissues of origin. The cells were treated with BMP4 in the presence or absence of netrin-1 for 60 min, and the pSMAD1/5 levels were analyzed by Western blot (Fig. 2). MEFs and HEK293 cells were also treated with netrin-1 alone (Fig. 2a); however, because treatment with netrin-1 alone had no effect on pSmad1/5 levels in MEFs or HEK293 cells, 'netrin-1 only' treatment was not assessed in the other cell lines (Fig. 2b). The pre-B acute lymphocytic leukemia cell lines 697 and RCH-ACV did not respond to BMP4 treatment with increased pSMAD1/5 levels and thus could not be evaluated for possible inhibition by netrin-1. Of the eleven cell lines that responded to BMP4, eight were significantly inhibited by netrin-1, including the MEF line, the ovarian cancer cell line Ovcar-3, the malignant melanoma cell line A375, the breast cancer cell line MDA-MB-415, the human embryonic kidney cell line HEK293, the glioblastoma cell lines TB101 and TB107, and the colorectal cancer cell line HCT116 (Fig. 2). Thus, netrin-1 inhibited BMP signaling in a general manner, affecting mouse and human cells of various tissues of origin.

Netrin-1-induced inhibition of BMP signaling in MEFs is dependent on neogenin.
To address the role of neogenin in netrin-1-induced inhibition of BMP signaling in MEFs, Neo1 was inactivated through CRISPR/Cas-9 (Fig. S5). The resulting Neo1 −/− MEFs showed reduced sensitivity to low concentrations of BMP4 (Fig. 3a). The Lrig1, Lrig3, and netrin-1 levels were unaltered in the Neo1-deficient MEFs (Fig. S5a-d). Intriguingly, the exogenous addition of netrin-1 to Neo1 −/− MEFs did not affect their pSmad1/5 response to low or high concentrations of BMP4 (Fig. 3a). Neogenin levels are increased in Ntn1-deficient mice 27 . Similarly, our Ntn1 −/− MEFs showed increased neogenin levels, as monitored by immunoblotting (Fig. 3b). Lrig-null MEFs showed a less pronounced increase in neogenin levels after Ntn1 ablation than Lrig wild-type MEFs (Fig. 3b). To investigate whether the neogenin levels were regulated by BMP signaling per se, MEFs were treated with a BMP4 concentration that induces a robust pSmad1/5 response (10 ng/ml for 60 min) followed by immunoblotting. Clearly, this BMP4 stimulus did not affect the apparent neogenin levels, as determined through immunoblotting (Fig. S5e). The exogenous addition of recombinant netrin-1 to Ntn1 −/− MEFs resulted in a dose-dependent  www.nature.com/scientificreports/ (Fig. 3c) and time-dependent (Fig. 3d) decrease in neogenin levels in both Lrig wild-type and Lrig-null MEFs. The induced expression of human LRIG1 or LRIG3 but not LRIG2 in Ntn1 wild-type MEFs resulted in increased neogenin levels (Fig. 3e-g). This result was recapitulated in Ntn1 −/− MEFs with regard to the effect of LRIG1 but not LRIG3; i.e., the Ntn1 −/− MEFs showed significantly enhanced neogenin expression in response to the induced expression of LRIG1 ( Fig. S6a) but not in response to the induced expression of LRIG3 (Fig. S6b).

Netrin-1 suppresses adipogenesis and chondrogenesis in vitro.
To investigate whether netrin-1 could regulate physiological processes in which BMP signaling plays important roles, we chose to study adipogenesis and chondrogenesis in vitro because these processes have been shown to depend on BMP signaling, albeit to our knowledge, the role of netrin-1 has not been addressed previously. Clearly, exogenously added recombinant netrin-1 was able to inhibit adipogenesis of MEFs as assessed through Oil Red O staining, both in the presence and in the absence of exogenously added BMP4 (Fig. 4a-e). Of note, also in the absence of added BMP4, the cell culture medium contained low levels of BMPs that are present in FBS, which is enough to drive basal adipogenesis 26,37,38 . Also chondrogenesis of BMP4-stimulated ATDC5 cells as assessed with an alkaline phosphatase activity assay was inhibited by netrin-1 (Figs. 4f and S8). As a control, the well-characterized BMP inhibitor noggin also inhibited BMP4-induced chondrogenesis of ATDC5 cells (Fig. S8b).

Discussion
Despite the large number of investigations addressing the functions of netrins, to the best of our knowledge, a role of netrins in the regulation of BMP signaling has not been proposed previously. Our discovery showing that netrin-1 negatively regulated BMP signaling in a variety of cell types may explain many previous observations regarding the developmental and physiological functions of this protein.
Based on our discovery that netrin-1 inhibited BMP signaling, including via a mutually inhibitory interaction with the BMP-promoting activities of LRIG1 and LRIG3, together with previous work on inner ear development 5,6,28 , we propose that the molecular function of netrin-1, as well as the molecular functions of Lrig1 and Lrig3, in semicircular canal formation is to fine-tune BMP signaling. Similarly, the observed inhibition of adipogenesis and chondrogenesis in vitro is consistent with a function of netrin-1 in modulating BMP signaling during cell differentiation. Taken together, our results suggest that the regulation of BMP signaling might be a prominent feature of the developmental and physiological functions of netrin-1 and, possibly, other netrins.
Interestingly, we observed a commonality in the specificities of the BMP signal-modulating activities of netrin-1 and LRIG proteins. That is, netrin-1 suppressed the signaling that was induced by BMP4 and BMP6 but not the signaling that was induced by BMP9; these results reflect the BMP-promoting functions of Lrig proteins, which extend to BMP4 and BMP6 but not to BMP9 26 . This commonality in BMP ligand specificity may indicate that netrin-1 and LRIG proteins act through a common molecular machinery. However, our observation that netrin-1 and LRIG proteins also regulated BMP signaling in the absence of their mutually inhibitory counterparts suggests that netrin-1 and LRIG proteins function autonomously of each other. To resolve the underlying mechanisms, it will be important to elucidate the composition of the molecular machinery involved.
Regarding the netrin receptor(s) involved, we found that neogenin was required for the BMP inhibitory function of netrin-1 in MEFs. Neogenin is a well characterized receptor for netrin-1 as well as a coreceptor, together with the repulsive guidance molecules, for BMPs 29,30 . Hence, neogenin is well poised to integrate netrin and BMP signaling. Although the effect of netrin-1 on BMP signaling in MEFs was strictly dependent on neogenin, a role for other netrin receptors in the regulation of BMP signaling in MEFs or other cell types cannot be ruled out. For example, the neogenin homolog DCC is not expressed in MEFs (unpublished observation) but may play a role similar to neogenin in other cell types. UNC5B has been shown to activate BMP signaling during osteogenic differentiation 31 , although a possible regulatory role of netrins in this context was not investigated. Another study showed that two out of three netrin receptors (neogenin, Unc5b, and/or A2b) have to be inactivated to rescue MC3T3-E1 cells from netrin-1-induced inhibition of BMP4-induced osteoblast differentiation 7 , further supporting a connection between netrin-1 and BMP signaling as well as demonstrating a functional redundancy among the netrin receptors. Future studies will reveal the roles of different netrin receptors in the regulation of BMP signaling. In parallel to its inhibition of BMP signaling, netrin-1 downregulated neogenin protein expression levels both in a genetic manner, as evident from the upregulation of neogenin seen in Ntn1 −/− cells, as well as Table 1. The coefficient of determination (r 2 ) for the BMP4 response (pSmad1/5 level) against the LRIG1-FLAG expression levels at different netrin-1 concentrations as shown in Fig. 1g.  Table 2. The coefficient of determination (r 2 ) for the BMP4 response (pSmad1/5 level) against the LRIG3-FLAG expression levels at different netrin-1 concentrations as shown in Fig. 1h. www.nature.com/scientificreports/  www.nature.com/scientificreports/  www.nature.com/scientificreports/ biochemically, as seen in the cellular response to recombinant netrin-1 protein. These observations are consistent with previous reports showing that Ntn1 −/− mice display increased neogenin and DCC levels 27 and that exogenous netrin-1 downregulates the DCC levels in embryonic cortical neurons 32 . In contrast, LRIG proteins upregulated the neogenin levels in parallel with their enhancement of BMP signaling. This is consistent with a previous report that showed a physical interaction between Lrig proteins and neogenin, which was associated with the protection of neogenin from proteolytic cleavage by ADAM17 33 . To what degree the netrin-1-and LRIG-mediated effects on neogenin expression levels represent causes or consequences of signal regulation remains to be investigated. In this regard, it is intriguing that all three Lrig proteins (Lrig1-3) seem to protect neogenin from proteolytical degradation 33 , whereas only LRIG1 and LRIG3, but not LRIG2, promote BMP signaling 26 . Several of the netrin receptors, including DCC, neogenin, UNC5A, UNC5B, UNC5C, and UNC5D, have been proposed to function as so-called 'dependence receptors' , i.e., receptors that mediate pro-apoptotic signals in the absence of ligand. Here, we propose an alternative hypothesis to account for some of the published observations pertaining to this concept. We suggest that in the absence of their netrin ligands, dependence receptors may promote basal BMP signaling, as has been shown for neogenin and UNC-40 30,34,35 , rather than exerting 'ligand-independent' signaling. Constitutive low-level BMP signaling is possible because different BMPs are widely expressed in mammalian tissues and are also present in FBS-containing cell culture media 36,37 . Then if, as shown here, netrin-1 inhibits BMP signaling at low BMP concentrations, we propose that some of the cellular effects caused by netrin-1 may be due to its negative regulation of BMP signaling rather than, or in addition to, its interruption of the intrinsic ligand-independent signaling of dependence receptors.
Netrin-1 has also been reported to be overexpressed and to promote cancer in many other contexts. We noticed an intriguing reciprocal relationship between the reported effects of netrin-1 and BMP signaling on several types of cancer cells. For example, although netrin-1 promotes the development, spread, metastasis, and stemness of breast cancer, colorectal cancer, and glioma cells, BMP signaling reciprocally suppresses these cancer cell types and processes, at least under some circumstances [38][39][40][41][42] . Thus, our results presented herein, which reveal a previously unrecognized function of netrin-1 as a suppressor of BMP signaling, are consistent with a tumor-promoting function of netrin-1 that is achieved via the suppression of BMP signaling. This function may warrant further investigation.
In conclusion, our demonstration that netrin-1 regulates BMP signaling via its receptor neogenin sheds new light on previous observations and opens up new areas for research. The urgent questions to address include the elucidation of the detailed molecular mechanisms involved and the delineation of the physiological processes that are directed by this regulatory mechanism.