Biphasic regulation of glutamine consumption by WNT during osteoblast differentiation

Osteoblasts are the primary bone forming cell responsible for producing and secreting type I collagen and other proteins that comprise the bone matrix. A constant supply of amino acids is required to maintain high rates of protein and matrix synthesis associated with bone anabolism. To fulfill this demand, osteoblasts must maximize the production or acquisition of amino acids. Indeed, recent evidence links amino acid uptake and metabolism to osteoblast function (Elefteriou et al., 2006; Karner et al., 2015; Rached et al., 2010; Yu et al., 2019). However, little is known about the transporters mediating amino acid uptake, nor their regulation during bone development.

The WNT family of secreted glycoproteins are critical regulators of osteoblast differentiation (Babij et al., 2003; Bennett et al., 2005; Day et al., 2005; Gong et al., 2001; Hu et al., 2005; Rodda and McMahon, 2006; Tu et al., 2007). WNTs activate multiple intracellular signaling cascades to induce osteoblast differentiation and modulate osteoblast activity. In the canonical pathway, WNT regulates the stability of the transcriptional coactivator β-catenin (CTNNB1), a critical regulator of osteoblast specification and differentiation (Day et al., 2005; Hu et al., 2005; MacDonald and He, 2012; Rodda and McMahon, 2006). In the absence of WNT, β-catenin is phosphorylated by the β-catenin destruction complex and targeted for proteasomal degradation. WNT stimulation inhibits the destruction complex, resulting in stabilization of β-catenin, which can then translocate into the nucleus and induce expression of target genes (for example, Tcf7) (Angers and Moon, 2009; Clevers, 2006; Clevers and Nusse, 2012; Roose et al., 1999). WNTs can also regulate osteoblast differentiation independently of β-catenin through the serine/threonine kinase mechanistic target of rapamycin (mTOR) (Chen et al., 2014; Inoki et al., 2006; Karner et al., 2015). Indeed, mTORC1 is critical for preosteoblasts to increase protein synthesis and differentiate into mature osteoblasts (Chen and Long, 2015; Fitter et al., 2017; Karner et al., 2017; Lim et al., 2016; Xian et al., 2012). We previously discovered that WNT stimulates glutamine uptake and catabolism necessary for osteoblast differentiation and bone formation (Karner et al., 2016, 2015). Mechanistically, it is not clear how WNT stimulates glutamine uptake in osteoblasts.

Glutamine transport is facilitated by a diverse array of membrane-tethered amino acid transporters categorized into discrete transport systems based on substrate specificity, kinetics and ion and pH dependence (Pochini et al., 2014). Glutamine uptake can occur in a Na⁺-dependent or -independent manner mediated by Systems ASC, A, γ(+)-L and N, or System L, respectively (Biltz et al., 1983; Bode, 2001; Jacob et al., 1986; Mackenzie et al., 2003; Tamarappoo et al., 1992, 1997; Taylor et al., 1992). Na⁺-dependent transport is subdivided by the strict requirement of Na⁺ [Systems ASC, A and γ(+)-L] or the substitution of Li⁺ for Na⁺ (System N). Finally, inhibitors can further subdivide Na⁺-dependent transporters, with System A being sensitive to 2-(methylamino)isobutyric acid (MeAIB) and System ASC to γ-glutamyl-p-nitroanilide (GPNA) (Esslinger et al., 2005; Freeman et al., 1999; Mackenzie et al., 2003). Although a great deal is known about the functional characteristics of these transporter systems, little is known about the specific transporters mediating glutamine uptake or their regulation in osteoblasts.

Here, we describe the biphasic regulation of glutamine uptake in response to WNT. WNT rapidly stimulates glutamine uptake that is sustained throughout osteoblast differentiation. Mechanistically, WNT activates β-catenin to rapidly stimulate glutamine uptake through Slc7a7, whereas mTORC1 regulates basal glutamine uptake through Slc1a5. These data highlight previously unknown roles for the amino acid transporters encoded by Slc7a7 and Slc1a5 and their regulation by WNT during osteoblast differentiation.

We previously determined that WNT stimulates glutamine uptake associated with osteoblast differentiation (Karner et al., 2015). In order to understand how WNT regulates glutamine consumption, we first evaluated the kinetics of glutamine uptake in ST2 cells, a bone marrow-derived cell line that undergoes osteoblast differentiation in response to WNT (Bennett et al., 2005; Kang et al., 2007; Karner et al., 2016, 2015; Otsuka et al., 1999; Tu et al., 2007) (Fig. 1A). ST2 cells were treated with recombinant WNT3a (rWNT3a) for up to 96 h, and glutamine uptake was quantified using L-(2,3,4-³H)-glutamine. We observed a rapid and sustained increase in glutamine consumption beginning as rapidly as 6 h after rWNT3a stimulation (Fig. 1B). Glutamine consumption continued to increase throughout the course of the experiment, even 96 h after rWNT3a stimulation (Fig. 1B). Basal glutamine transport occurred primarily in a Na⁺-dependent manner that could not be rescued by Li⁺ or inhibited by the amino acid analog MeAIB in ST2 cells (Fig. 1C). These data indicate that glutamine uptake occurs primarily in a Na⁺-dependent manner characteristic of the amino acid transport systems System ASC and System γ(+)-L (Fig. 1D). Likewise, WNT-stimulated glutamine uptake in ST2 cells was Na⁺ dependent and insensitive to either MeAIB or Li⁺ rescue (Fig. 1C). These data indicate that WNT-induced glutamine uptake is not the result of enhanced activity of other transport systems, rather it results from increased System ASC and/or γ(+)-L activity (Fig. 1D) (Jacob et al., 1986; Tamarappoo et al., 1997; Taylor et al., 1992).

To determine how WNT signaling regulated glutamine uptake, we first evaluated the mRNA expression of genes encoding glutamine transporters in ST2 cells. To do this, we analyzed our previously generated RNAseq dataset of ST2 cells treated with rWNT3a for up to 72 h (Karner et al., 2016). In unstimulated ST2 cells, the System ASC transporter alanine serine cysteine transporter 2 (ASCT2, encoded by Slc1a5) was the highest expressed glutamine transporter in either System ASC or γ(+)-L (Table 1). Conversely, the System γ(+)-L transporter γ(+)-L transporter 1 [γ(+)-LAT1, encoded by Slc7a7] was expressed at low levels in unstimulated ST2 cells (Table 1). It is important to note that ASCT2 and γ(+)-LAT1 are the only members of their respective transport systems that accept glutamine as a substrate. Despite being expressed at lower levels in unstimulated ST2 cells, Slc7a7 was significantly increased by WNT treatment at 6, 24 and 72 h (Table 1). On the other hand, expression of Slc1a5 was not significantly increased until after 72 h WNT treatment (Table 1). We confirmed the temporal regulation of both Slc1a5 and Slc7a7 using reverse transcription qPCR (RT-qPCR). Both Slc7a7 and Slc1a5 were significantly induced by WNT signaling, albeit with differing kinetics (Fig. 2A). Slc7a7 was induced rapidly, with peak induction occurring within 6–24 h of WNT stimulation. By 96 h, Slc7a7 had returned to baseline expression (Fig. 2A). Conversely, Slc1a5 was significantly induced beginning 72 h after WNT stimulation (Fig. 2A). These data demonstrate that WNT regulates two amino acid transporters with differing kinetics in ST2 cells.

We next sought to determine the necessity of these amino acid transporters for glutamine uptake. We first evaluated the role of Slc1a5 in glutamine uptake because it was highly expressed in ST2 cells (Table 1). To do this, we used GPNA, a specific inhibitor of Slc1a5/ASCT2 (Esslinger et al., 2005). GPNA treatment reduced basal glutamine uptake by 77% indicating the majority of glutamine uptake is mediated by Slc1a5/ASCT2 (Fig. 2B). We next sought to determine the effects on WNT-induced glutamine uptake. Interestingly, despite significantly reducing basal glutamine uptake, glutamine consumption was stimulated equally by WNT in the presence or absence of GPNA (Fig. 2B,C). To confirm the specificity of GPNA, we targeted Slc1a5 using shRNA. This approach significantly reduced both ASCT2 protein and basal glutamine uptake, similar to the effects of GPNA treatment (Fig. 2D,E). Similar to the results of the GPNA experiment, glutamine uptake was stimulated equally by WNT in both control and ASCT2-knockdown cells (Fig. 2E,F). These data indicate that Slc1a5/ASCT2 mediates the majority of glutamine uptake in ST2 cells but is not responsible for the acute increase in glutamine consumption in response to WNT. We next sought to determine whether Slc7a7 mediates glutamine uptake in ST2 cells. To do this, we targeted Slc7a7 using shRNA. This approach effectively reduced γ(+)-LAT1 protein levels (Fig. 2G). Slc7a7 knockdown minimally affected basal glutamine uptake but completely abrogated increased glutamine consumption in response to WNT stimulation (Fig. 2H,I). Collectively these data indicate WNT signaling rapidly increases Slc7a7 expression to increase glutamine uptake whereas Slc1a5 is responsible for the majority of glutamine uptake in ST2 cells.

We next sought to understand how WNT signaling regulated Slc7a7 expression. The rapid induction kinetics of Slc7a7 expression suggested it may be regulated directly by β-catenin. Inhibition of LRP5 and LRP6 using recombinant DKK1 (rDKK1) reduced WNT-induced β-catenin stabilization and prevented the induction of alkaline phosphatase (Fig. 3A). Moreover, DKK1 treatment prevented WNT induction of known β-catenin target genes (e.g. Tcf7) as well as of Slc7a7 (Fig. 3B,C). Expression of a dominant-negative form of TCF4 that is unable to interact with β-catenin also prevented induction of Slc7a7, suggesting that Slc7a7 is a β-catenin target gene (data not shown). To test the role of β-catenin directly, we targeted the gene encoding β-catenin (Catnnb1) using shRNA (Fig. 3D). β-catenin knockdown prevented osteoblast differentiation, as shown by inhibition of alkaline phosphatase induction (Fig. 3D). Importantly, β-catenin knockdown significantly reduced the basal expression of Slc7a7 and prevented the induction of Slc7a7 in response to WNT treatment (Fig. 3E). Moreover, β-catenin knockdown reduced WNT-induced glutamine uptake without affecting Slc1a5 expression (Fig. 3F and data not shown). These data indicate that β-catenin is necessary for the induction of Slc7a7 and acute glutamine consumption in response to WNT. We next sought to determine the sufficiency of β-catenin by inhibiting GSK3β using LiCl (Klein and Melton, 1996). LiCl treatment stabilized β-catenin and induced both alkaline phosphatase staining and Tcf7 mRNA expression similar to WNT treatment (Fig. 3G,H). Moreover, LiCl stimulated both Slc7a7 mRNA and γ(+)-Lat1 protein expression and increased glutamine uptake similar to WNT treatment (Fig. 3G–I). Importantly, LiCl did not increase Slc1a5 mRNA expression (Fig. 3H). These data indicate that β-catenin-dependent WNT signaling rapidly stimulates glutamine uptake via transcriptional upregulation of Slc7a7.

We next evaluated the regulation of Slc1a5 expression by WNT. Unlike Slc7a7 and other canonical WNT target genes, which are rapidly induced by WNT, Slc1a5 mRNA levels were not significantly increased until 72 h after WNT treatment (Fig. 2A and Table 1). Indeed, we observed no change in Slc1a5 expression in β-catenin-knockdown cells or in response to LiCl treatment, suggesting Slc1a5 is not regulated by canonical WNT–β-catenin signaling (Fig. 3H and data not shown). Rather, the induction of Slc1a5 followed a timecourse reminiscent of protein anabolism genes induced by WNT through secondary activation of the transcription factor ATF4 (Karner et al., 2015). We have previously determined that WNT signals through mTORC1 to activate ATF4 and protein anabolism genes during osteoblast differentiation (Karner et al., 2015). Here, we observed a significant increase in ATF4 protein expression after 72 h of WNT treatment (Fig. 4A). Inhibition of mTORC1 signaling using rapamycin prevented ATF4 protein induction by WNT (Fig. 4A). Importantly, rapamycin treatment reduced ASCT2 expression in both unstimulated and WNT treated ST2 cells (Fig. 4A). Consistent with decreased ASCT2 expression, rapamycin significantly reduced glutamine uptake under both basal conditions and in response to WNT stimulation (Fig. 4B). Despite a significant reduction in overall glutamine uptake, WNT was able to stimulate glutamine consumption in the presence of rapamycin (Fig. 4B,C). This is likely due to normal induction of Slc7a7, because rapamycin treatment did not affect WNT induction of Slc7a7 despite completely abrogating Slc1a5 induction by WNT (Fig. 4D,E). These data indicate that mTORC1 activity is required for both glutamine uptake and the normal expression of both ATF4 and ASCT2. We next sought to determine whether ATF4 activity is required for Slc1a5 induction by WNT. Atf4 knockdown using shRNA reduced both ATF4 protein expression and Slc1a5 mRNA expression in ST2 cells (Fig. 4F,G). Moreover, Atf4 expression was necessary for Slc1a5 induction by WNT, because Atf4 knockdown completely prevented the induction of Slc1a5 by WNT (Fig. 4F,G). Likewise, Atf4 knockdown significantly reduced basal glutamine uptake (Fig. 4H). Similar to the rapamycin treatment, Atf4 knockdown did not affect either WNT-induced glutamine uptake or Slc7a7 induction by WNT (Fig. 4I,J). Collectively, these data indicate that ATF4 is critical for the majority of glutamine uptake by regulating Slc1a5 mRNA expression downstream of mTORC1.

We next sought to determine the role of these glutamine transporters during WNT-induced osteoblast differentiation. Knockdown of Slc7a7 using shRNA reduced the induction of both early osteoblast marker genes (exemplified by Akp2, also known as Alpl) and terminal osteoblast marker genes (exemplified by Bglap) and prevented matrix mineralization in response to WNT (Fig. 5A–D). Interestingly, Slc1a5 knockdown did not affect early osteoblast marker gene induction (e.g. Akp2) but specifically reduced the induction of terminal osteoblast marker genes (e.g. Bglap) and prevented matrix mineralization in response to WNT (Fig. 5E–H). These data indicate that both Slc7a7 and Slc1a5 are required for WNT-induced osteoblast differentiation in ST2 cells.

Increased matrix synthesis associated with bone formation increases the demand for amino acids. It stands to reason that osteoblasts must increase amino acid production or acquisition to meet this biosynthetic demand. Here, we describe the intricate regulation of glutamine acquisition by WNT signaling in an osteoblast progenitor cell line. We identified genes encoding two amino acid transporters, Slc1a5 and Slc7a7, responsible for the majority of glutamine uptake in ST2 cells undergoing osteoblast differentiation in response to WNT. Slc1a5 encodes the primary glutamine transporter in ST2 cells whereas Slc7a7 mediates WNT-stimulated glutamine uptake associated with osteoblast differentiation. These transporters appear to work in concert to provide sufficient glutamine, and likely other amino acids, to initiate and sustain osteoblast differentiation and matrix production in vitro.

Osteoblast differentiation is characterized by rapid proliferation of osteoblast progenitors followed by differentiation into mature matrix-producing osteoblasts. This process is associated with increased glutamine consumption and metabolism (Fig. 1) (Karner et al., 2015; Yu et al., 2019). Blocking glutamine consumption (Fig. 5) or metabolism prevents osteoblast differentiation and matrix formation (Brown et al., 2011; Karner et al., 2015; Yu et al., 2019). It is not clear why osteoblasts have such an acute requirement for glutamine to facilitate these processes. Osteoblast differentiation is associated with altered energetic and biosynthetic demands (Guntur et al., 2014; Karner and Long, 2017; Riddle and Clemens, 2017). For example, proliferating cells require nucleotides and must duplicate their cell mass in order to divide (Hosios et al., 2016). Similarly, differentiation and matrix production is associated with increased protein synthesis and secretion, which may increase reactive oxygen species and oxidative stress detrimental to osteoblast differentiation (Almeida et al., 2007; Mody et al., 2001). Glutamine is a multifunctional amino acid uniquely suited to fulfill these anabolic demands. For example, glutamine-derived nitrogen is important for the de novo synthesis of both nucleotides and amino acids. Similarly, glutamine-derived carbon is used for the synthesis of both amino acids and the antioxidant glutathione. Finally, glutamine-derived α-ketoglutarate can provide energy to fulfill energetic demands through entry into the TCA cycle (Newsholme et al., 2003). Thus, osteoblasts likely utilize glutamine disparately during differentiation to fulfill distinct metabolic purposes.

WNT is a potent regulator of osteoblast differentiation and bone formation (Cui et al., 2011; Hill et al., 2005). One of the earliest events associated with WNT-induced osteoblast differentiation is increased glutamine consumption (Fig. 1; Karner et al., 2015). It is interesting to note that glutamine uptake is not only rapidly increased but also sustained throughout the differentiation process. Here, we identified two glutamine transporters encoded by Slc1a5 and Slc7a7 that mediate glutamine uptake in ST2 cells. Slc1a5 is critical for the majority of glutamine uptake (Fig. 2). Conversely, Slc7a7 is responsible for mediating the acute increase in glutamine uptake in response to WNT. Importantly, the transcription of these transporters is regulated disparately by WNT. First, expression of Slc7a7 is upregulated within 6 h, whereas Slc1a5 expression is not increased until 72 h after WNT stimulation. Mechanistically, β-catenin is both necessary and sufficient for the rapid induction of Slc7a7 expression and the acute increase in glutamine uptake in response to WNT. Conversely, β-catenin is dispensable for Slc1a5 expression and basal glutamine consumption in ST2 cells (Fig. 3 and data not shown). Rather, Slc1a5 expression is regulated downstream of mTORC1 and ATF4 in ST2 cells (Fig. 4). We previously determined that WNT induces a GCN2-dependent integrated stress response (ISR) through mTORC1 (Karner et al., 2015). We show here that the ISR transcriptional effector ATF4 is critical for both basal and WNT-induced Slc1a5 expression and glutamine uptake. This is consistent with recent data demonstrating that Slc1a5 is directly regulated by ATF4 (Han et al., 2013; Hu et al., 2020).

Collectively, these data support a biphasic model in which WNT signaling regulates glutamine consumption via canonical and noncanonical pathways to facilitate osteoblast differentiation. Interestingly, inhibiting Slc7a7, but not Slc1a5, inhibited Akp2 induction, whereas inhibition of either transporter inhibited terminal osteoblast differentiation and matrix mineralization. The precise mechanism underlying this discrepancy is not clear; however, it is important to note that in addition to glutamine, ASCT2 can transport alanine, serine and asparagine. Conversely, γ(+)-Lat1 mediates the influx of ornithine, arginine and lysine in exchange for the efflux of cationic amino acids (Chillaron et al., 1996; Pfeiffer et al., 1999; Torrents et al., 1999), which may be contributing to the disparate early osteoblast differentiation phenotypes observed. It will be important to elucidate the substrates of these amino acid transporters in osteoblast progenitors and determine the precise role they play during WNT-induced osteoblast differentiation.

The bone marrow-derived cell line, ST2 (RRID: CVCL_2205) was plated at 40,000 cells/ml in α-MEM (GIBCO) supplemented with 10% FBS (Invitrogen). In WNT treatment experiments, 25 ng/ml WNT3a (TIME Bioscience) or vehicle control (0.1% BSA in PBS) was supplemented in the α-MEM. For LiCl treatment, 20 mM LiCl or vehicle control (20 mM NaCl) was added in the growth medium. When inhibitors were used, cells were pretreated with corresponding inhibitors for 30 min before any other treatment. In indicated experiments, growth medium was supplemented with 100 nM rapamycin (Sigma-Aldrich), 250 ng/ml DKK1 (R&D) or 0.03 mM GPNA (MP Biomedicals), or the respective vehicle (DMSO, 0.1% BSA in PBS or 1 M HCl for Rapamycin, DKK1 and GPNA, respectively) as a control. For WNT3a-induced mineralization, cells were treated with WNT3a for 72 h followed by osteogenic medium [α-MEM supplemented with 50 mg/ml ascorbic acid (Sigma-Aldrich) and 10 mM β-glycerophosphate (Sigma-Aldrich)] for 6 days. Osteoblast differentiation was assayed by visualizing alkaline phosphatase activity using 5-bromo-4-chloro-3′-indolyl-phosphate/Nitro Blue tetrazolium (BCIP/NPT; Katagiri et al., 1994) or von Kossa staining of deposited calcium phosphate (Rungby et al., 1993). The alkaline phosphatase assay was performed 24 h after WNT stimulation. Von Kossa staining was performed 6 days after addition of osteogenic medium.

Lentiviral vectors were obtained from the shRNA consortium at Washington University School of Medicine. All knockdown results were confirmed using two or more unrelated shRNA constructs. The shRNA sequence listed first was shown in the results. The lentiviral vector pLKOpuro (RRID: Addgene_8453) was modified to express shRNAs targeting Slc1a5 (5′-CCTGTAGAGTTCTCTACCCTT-3′, 5′-GCAGTGTTCATCGCACAACTA-3′), Slc7a7 (5′-GCTACATGTTTCAGACTTCAT-3′, 5′-GCCATCTGTATGGTTCATG-3′), Atf4 (5′-CCAGAGCATTCCTTTAGTTTA-3′, 5′-CCTCTAGTCCAAGAGACTAAT-3′), Catnnb1 (5′-GCGTTATCAAACCCTAGCCTT-3′, 5′-CCATCACAGATGTTGAAACAT-3′), or either RFP (5′-ACAACAGCCACAACGTCTATA-3′) or LacZ (5′-GCGATCGTAATCACCCGAGTG-3′) as a negative control. The shRNA-expressing lentiviral vector was co-transfected in 293T cells with the plasmids pMD2.g (RRID: Addgene_12259) and psPax2 (RRID: Addgene_12260). Virus-containing medium was collected and filtered. ST2 cells were infected for 24 h and recovered for 24 h in regular medium prior to further treatment.

Total RNA was isolated using an RNAeasy kit with on-column DNase treatment (Qiagen). 500 ng of total RNA was reverse transcribed using the Iscript cDNA Synthesis kit (Bio-Rad). Reactions were set up in technical and biological triplicate in a 96-well format on an ABI Quantstudio 3 using SYBR Green chemistry (SsoAdvanced; Bio-Rad). The PCR conditions were 95°C for 3 min followed by 35 cycles of 95°C for 10 s and 60°C for 30 s. Gene expression was normalized to 18S rRNA, and relative expression was calculated using the 2^−(ΔΔCt) method. Primers were used at 0.1 µM, and their sequences are listed in Table S1. PCR efficiency was optimized, and melting curve analyses of products were performed to ensure reaction specificity.

ST2 cells were scraped in lysis buffer containing 50 mM Tris (pH 7.4), 15 mM NaCl, 0.5% NP-40, 0.1% SDS and 0.1% sodium deoxycholate with a protease and phosphatase inhibitor tablet (Roche). Protein concentration was quantified using the BCA method (Pierce). Proteins (20 µg) were resolved on 12% polyacrylamide gels and transferred onto Immuno-Blot PVDF membrane (Bio-Rad). Proteins were detected using the following specific antibodies: anti-ASCT2 (RRID: AB_10891440), anti-Slc7a7 (RRID: AB_2576546), anti-α-tubulin (RRID: AB_2619646), anti-ATF4 (RRID: AB_2058752), anti-β-actin (RRID: AB_330288), anti-phospho-S6 (Ser240/244) (RRID: AB_10694233), anti-S6 (RRID: AB_331355) and anti-β-catenin (RRID: AB_634603). Further antibody information is provided in Table S2. The membranes were blocked for 1 h at room temperature in 5% milk powder in TBS with 0.1% Tween (TBST) and then incubated at 4°C with the primary antibody overnight. Membranes were washed three times with TBST and further incubated with anti-rabbit IgG, HRP-linked antibody (RRID: AB_2099233) in 5% milk (in TBST) for 1 h at room temperature. All blots were developed using enhanced chemiluminescence (Clarity Substrate Kit, Bio-Rad). Each experiment was repeated with a minimum of three independently prepared protein samples.

Statistical significance was determined by one-way ANOVA or by an unpaired two-tailed Student's t-test.

The authors would like to thank all members of the Karner laboratory for critical comments on the manuscript.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.251645.reviewer-comments.pdf