Enhanced vulnerability to oxidative stress and induction of inflammatory gene expression in 3‐phosphoglycerate dehydrogenase‐deficient fibroblasts

l‐Serine (l‐Ser) is a necessary precursor for the synthesis of proteins, lipids, glycine, cysteine, d‐serine, and tetrahydrofolate metabolites. Low l‐Ser availability activates stress responses and cell death; however, the underlying molecular mechanisms remain unclear. l‐Ser is synthesized de novo from 3‐phosphoglycerate with 3‐phosphoglycerate dehydrogenase (Phgdh) catalyzing the first reaction step. Here, we show that l‐Ser depletion raises intracellular H2O2 levels and enhances vulnerability to oxidative stress in Phgdh‐deficient mouse embryonic fibroblasts. These changes were associated with reduced total glutathione levels. Moreover, levels of the inflammatory markers thioredoxin‐interacting protein and prostaglandin‐endoperoxide synthase 2 were upregulated under l‐Ser‐depleted conditions; this was suppressed by the addition of N‐acetyl‐l‐cysteine. Thus, intracellular l‐Ser deficiency triggers an inflammatory response via increased oxidative stress, and de novo l‐Ser synthesis suppresses oxidative stress damage and inflammation when the external l‐Ser supply is restricted.

Furthermore, the conversion of L-Ser into Gly participates in the biosynthesis of purines and pyrimidines, by transferring a one-carbon unit to tetrahydrofolate (THF). Our previous in vivo study demonstrated that severe L-Ser deficiency in mice with systemic targeted disruption of Phgdh resulted in intrauterine growth retardation, multiple organ malformation, and embryonic lethality [1,2]. L-Ser biosynthesis defects in humans resulting from Phgdh mutations were identified to be Ser synthesis disorders and Neu-Laxova syndrome, the symptoms of which are characterized by severe fetal growth retardation, microcephaly, and perinatal lethality [3][4][5]. These findings have demonstrated that de novo L-Ser synthesis is essential for embryonic development and survival in mice and humans.
We recently reported that reduced availability of intracellular L-Ser promotes the biosynthesis and accumulation of 1-deoxysphinganine (doxSA) and its metabolites 1-deoxysphingolipids in mouse embryonic fibroblasts (MEFs) lacking functional Phgdh (KO-MEFs) [6]. The condensation of palmitoyl-CoA with L-Ala instead of L-Ser generated doxSA and its biosynthesis were triggered by an increasing ratio (> 3.0) of L-Ala to L-Ser within the cells. doxSA elicited the activation of stress-activated protein kinase/Jun aminoterminal kinase and p38 mitogen-activated protein kinase, resulting in growth arrest and death in KO-MEFs even in the presence of L-Ser [7]. Consistent with these observations, our microarray analysis of L-Ser-depleted KO-MEFs revealed that the activation of a network containing the stress-response-activating transcription factor ATF4-ATF3-DNA damage-inducible transcript 3 (Ddit3) axis was most prominent among the 560 upregulated genes [8], implying that L-Ser deficiency causes metabolic stress in KO-MEFs. However, the causal link between reduced L-Ser availability and vulnerability to stress remains unexplored. Here, we show that L-Ser-depleted KO-MEFs are vulnerable to oxidative stress, which is accompanied by increased expression of thioredoxin-interacting protein (Txnip), a mediator of oxidative stress to inflammation, and the proinflammatory enzyme prostaglandinendoperoxide synthase 2 [Ptgs2; also known as cyclooxygenase (COX) 2]. These findings suggest that L-Ser deficiency leads to an inflammatory response through diminished protection against oxidative stress.

Total glutathione quantification
Knockout-MEFs grown in DMEM with 10% FBS and 10 lgÁmL À1 gentamicin were replated in either L-Ser-supplemented (at 40% cell confluence) or L-Ser-depleted (at 80% cell confluence) conditions for 24 h. Cells maintained under both conditions reached 80% cell confluence and were washed with Dulbecco's phosphate-buffered saline (DPBS) followed by scraping from the dishes with DPBS. The cell suspensions were centrifuged at 1500 g, and each pellet was resuspended and lysed with 80 lL of 10 mM HCl. The lysates were alternately frozen and thawed twice, after which 20 lL of 5% (w/v) 5-sulfosalicylic acid was added. The lysates were centrifuged at 8000 g for 10 min, and the supernatants were used for glutathione (GSH) measurement. The total GSH levels were quantified using the GSSG/GSH Quantification kit (Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer's protocol, on a Multiskan TM FC microplate photometer (Thermo Fisher Scientific).

Measurement of intracellular H 2 O 2 generation
Knockout-MEFs were seeded at 5 9 10 3 -1 9 10 4 cells per well on Clear Fluorescence Black Plates (Greiner Bio-One International GmbH, Frickenhausen, Germany) in 100 lL of complete medium and incubated overnight at 37°C, after which the medium was replaced with EMEM containing 1% FBS with or without L-Ser and incubated for 6 h. To detect endogenous H 2 O 2 within cells, KO-MEFs were washed with DPBS and incubated with 2 mM BES-H 2 O 2 -Ac, a cell-permeable fluorescent probe for H 2 O 2 (Wako Pure Chemical Industries) [9], and Hoechst 33342 (Dojindo Laboratories) for 20 min. Images were acquired using the In Cell Analyzer 1000 (GE Healthcare UK Ltd., Buckinghamshire, UK) using 360-and 492-nm excitation filters, and 460-and 535-nm emission filters, as previously described [10]. The threshold of BES-H 2 O 2 -Ac intensity was set to the point at which approximately 75% of L-Sersupplemented KO-MEFs were negative, and cells were scored as positive or negative using SPOTFIRE DECISIONSITE CLIENT 8.2 software (GE Healthcare Japan, Tokyo, Japan). This software was used to visualize and analyze the results [11,12].

Antioxidant treatments
Knockout-MEFs were cultured in complete medium for 20 h, after which the culture medium was changed to EMEM containing 1% FBS in the absence of L-Ser. N-acetyl-Lcysteine (NAC) was added to the culture medium at concentrations of 1 or 5 mM for 6 h, after which total RNA was extracted and used for qRT-PCR as described above.

Statistical analyses
Data were evaluated using t-tests to analyze differences between two groups. To analyze differences among more than two groups, one-way analysis of variance followed by Dunnett's post hoc test was used. P-values < 0.05 were considered significant. Data are expressed as the means AE standard error. All statistical analyses were performed using

L-Ser deficiency reduced glutathione and increased vulnerability to oxidative stress in Phgdh KO-MEFs
We have previously reported that L-Ser depletion reduces the intracellular levels of Gly, Cys, and L-Ser [2]. As both Gly and Cys are necessary precursors of GSH, we compared the total GSH levels in L-Ser-supplemented and L-Ser-depleted KO-MEFs. Figure 1A shows that intracellular GSH levels were reduced significantly in KO-MEFs under L-Ser-depleted conditions compared to L-Ser-supplemented conditions. We then sought to determine whether endogenous production of intracellular H 2 O 2 was altered in KO-MEFs under L-Ser-depleted conditions using BES-H 2 O 2 -Ac, a cell-permeable fluorescent dye for H 2 O 2 [9]. After 6-h incubation, the percentage of BES-H 2 O 2 positive cells in L-Ser-depleted KO-MEFs was significantly increased compared to the L-Ser-supplemented condition (Fig. 1B). Next, we examined the effect of H 2 O 2 treatment on cell viability and observed that increasing the concentration of H 2 O 2 reduced the viability of both types of MEFs (Fig. 1C). However, KO-MEFs were more vulnerable to H 2 O 2 than WT-MEFs at lower H 2 O 2 concentrations (0.1 and 0.01 lM). These observations indicated that the loss of de novo L-Ser synthesis culminated in enhanced H 2 O 2 generation and vulnerability to its oxidative stress.

L-Ser depletion upregulates Txnip and Ptgs2 expression in KO-MEFs
To verify whether L-Ser deficiency affects the oxidative stress response in KO-MEFs, we focused on Txnip, a multifunctional protein linking oxidative stress to inflammation [12]. Txnip was identified by microarray analysis as being upregulated in L-Ser-depleted KO-MEFs [7,8] and transcriptionally activated by increased H 2 O 2 [13]. First, we compared Txnip mRNA levels in KO-and WT-MEFs under L-Ser-supplemented or L-Ser-depleted conditions. After incubation in L-Serdepleted medium for 6 h, an 8-fold increase in Txnip mRNA was detected in KO-MEFs but not in WT-MEFs ( Fig. 2A). Consistently, Txnip protein expression level in L-Ser-depleted KO-MEFs significantly increased to 1.8-fold higher than that in L-Ser-supplemented KO-MEFs (Fig. 2B). To examine whether Txnip mRNA induction was due to Phgdh deletion, we measured the Txnip mRNA levels in KO-MEFs +Phgdh and in KO-MEFs +Gfp under L-Ser-depleted conditions. Viral transduction of Phgdh, but not Gfp, suppressed Txnip mRNA induction under L-Ser-depleted conditions (Fig. 2C), indicating that loss of Phgdh was primarily responsible for Txnip induction under L-Serdepleted conditions. Time course analysis of Txnip mRNA expression demonstrated a sharp 7-fold increase as early as 2 h after the exposure of L-Serdepleted medium to KO-MEFs, and a significant 3-fold increase 24 h after exposure (Fig. 2D).
Next, we tested whether the expression of Ptgs2, a proinflammatory enzyme, was also induced in L-Ser-depleted KO-MEFs, because Txnip participates in the regulation of Ptgs2 expression [14,15]. We compared Ptgs2 mRNA levels in KO-MEFs and found that after 6-h incubation, a substantial 16-fold increase in Ptgs2 mRNA was detected in the L-Ser-depleted condition compared to the L-Ser-supplemented condition, while WT-MEFs did not show such an increase in L-Ser-depleted conditions (Fig. 2E). Accordingly, a significant increase in Ptgs2 protein was observed in L-Ser-depleted KO-MEFs compared to those in L-Sersupplemented conditions (Fig. 2F). As with Txnip (Fig. 2C), viral transduction of Phgdh cDNA but not Gfp suppressed Ptgs2 mRNA induction (Fig. 2G). Time course analysis of Ptgs2 mRNA expression demonstrated a subtle but significant 1.3-fold increase after 2-h incubation of KO-MEFs in L-Ser-depleted medium, which reached a plateau after 6-h incubation, and retained a 1.4-fold increase even after 24 h (Fig. 2H). These observations indicated that reduced L-Ser availability caused by Phgdh disruption results in the upregulation of both mRNA and protein levels of Txnip and Ptgs2 within 6 h.

Transcriptional activation of Txnip and Ptgs2 is independent of the integrated stress response pathway
To gain insight into the underlying molecular mechanisms by which Txnip expression was upregulated under L-Ser-depleted conditions, we examined whether the integrated stress response (ISR) pathway, which is increased expression of the transcription factor Atf4 [16,17]. As a result of amino acid deprivation, Atf4 target genes are substantially induced [18]. As we have demonstrated that L-Ser depletion in KO-MEFs causes a robust increase in Atf4 protein expression (Sayano et al., manuscript in preparation) and upregulation of several Atf4-target genes [2,7], we sought to determine whether Txnip and Ptgs2 induction was regulated by Atf4 via the ISR pathway. To evaluate the functional involvement of Atf4 in KO-MEF transcription, we generated shRNA-mediated Atf4 knockdown (KD)-KO-MEFs, in which Atf4 protein expression was suppressed by 0.4-fold under L-Ser-depleted conditions compared to mock-treated KO-MEFs (Sayano et al., manuscript in preparation). As shown in Fig. 3A,B, Txnip and Ptgs2 induction in Atf4 KD-KO-MEFs was unchanged compared to mock KO-MEFs under L-Serdepleted conditions, suggesting that Atf4 and the ISR pathway did not play a role in Txnip and Ptgs2 induction in L-Ser-depleted KO-MEFs. The ISR pathway is activated in response to the deficiency of indispensable amino acids, especially L-Leu [16][17][18]. We then investigated whether the depletion of the indispensable amino acid L-Leu could induce Txnip and Ptgs2 expression in KO-MEFs. Figure 3C,D shows that L-Leu depletion elicited significant increases in Txnip and Ptgs2 expression, although the magnitudes of induction were lower than those observed during L-Ser-depleted conditions (Fig. 3C,D). These observations suggest that L-Ser deficiency plays a more profound role in Txnip and Ptgs2 expression than L-Leu deficiency.

Antioxidant addition suppresses Txnip and Ptgs2 expression caused by L-Ser depletion
To clarify the upstream mechanism underlying the upregulation of Txnip and Ptgs2 seen in L-Ser-deficient KO-MEFs, we examined the effects of antioxidant treatment on mRNA expression. The addition of NAC, a GSH precursor, caused significant suppression of mRNA expression of both Txnip and Ptgs2 in KO-MEFs under L-Ser-depleted conditions (Fig. 4A,B). These results suggest that increasing oxidative stress elicited by L-Ser depletion causes the upregulation of Txnip and Ptgs2 in KO-MEFs.

Discussion
This study demonstrated that intracellular L-Ser deficiency caused by Phgdh deletion and external L-Ser depletion elicited increased vulnerability to oxidative and Gly and prevents damage to cellular components against oxidative stress generated by intracellular reactive oxygen species (ROS). Several studies in cancer cells have reported that the intracellular levels of L-Ser, Gly, and GSH decrease after exposure to H 2 O 2 [19], and L-Ser and Gly depletion leads to decreased GSH levels and increased cell death in p53 À/À or p21 À/À HCT116 cells [20]. Our observations in KO-MEFs, indicating that the Phgdh-dependent L-Ser biosynthetic pathway plays a primary role for maintaining the intracellular GSH level and preventing cell death under nutritional stress conditions, are consistent with these reports [21,22]. We demonstrated that L-Ser depletion induced Txnip expression in KO-MEFs. Txnip, also known as vitamin D3-upregulated protein 1, was originally identified as a negative regulator of thioredoxin 1/2 (Trx), a key sensor of cellular redox status that regulates protection against oxidative stress. The Trx-Txnip complex is a critical regulator of intra-and extracellular redox signaling and ROS [23]. Txnip expression is upregulated by ROS [12,24,25] and oxidative stress caused by ischemic-reperfusive injuries [26,27]. This study showed that Txnip expression was induced by L-Ser depletion in KO-MEFs, and this was inhibited by viral transduction of Phgdh cDNA ( Fig. 2A-C). It is well documented that Txnip links oxidative stress to inflammation by activating NLR family pyrin domain containing 3 [11,28,29], and participates in the upregulation of Ptgs2 expression [13,14]. The ISR pathway is an oxidative stress defense mechanism in response to essential amino acid deficiency [16]. This study showed that Txnip induction caused by L-Ser deficiency was not suppressed by Atf4 KD in L-Serdepleted KO-MEFs (Fig. 3A), indicating that Txnip induction is independent of the ISR pathway. Txnip induction by L-Ser depletion was significantly suppressed by the addition of an antioxidant (Fig. 4A). These observations suggest that Txnip expression is increased by reduced L-Ser availability, linking aberrant redox regulation and the induction of an inflammatory response to L-Ser deficiency that is independent of the ISR pathway.
Txnip affects the inflammatory response and cell death signaling by regulating the cellular redox status [30], and loss of Txnip can lead to the proliferation of cancer cells [31]. We previously reported that L-Serdepleted KO-MEFs exhibited cell growth arrest and increased cell death after 96-h incubation, which was associated with diminished mRNA translation and aberrant sphingolipid metabolism [2,6,7]. In addition to GSH, proteins, and lipids, Phgdh expression has been proved to be critical to maintain molecules important for cell proliferation, including reduced nicotinamide adenine dinucleotide phosphate [32], purine nucleotides [33], and THF metabolites [18,33]. Taken together, the present study implies that the loss of de novo L-Ser biosynthesis leads to cell proliferation arrest, followed by oxidative stress and inflammation, which seems to be a more severe cellular consequence compared to the loss of other nonessential amino acids but comparable to the severe phenotypes caused by the genetic deficiency of Gln [34]. We demonstrated that L-Ser deficiency promotes the biosynthesis and accumulation of doxSA, which can activate p38 MAPK in L-Ser-depleted KO-MEFs [7]. As Txnip expression was induced by H 2 O 2 via p38 MAPK activation in human aortic smooth muscle cells [35], further study is needed to clarify whether p38 MAPK regulates Txnip induction in L-Ser-depleted KO-MEFs. These insights might contribute to the elucidation of the pathobiology of patients with Neu-Laxova syndrome/L-Ser deficiency disorders [3][4][5]34] or diseases associated with increased Txnip expression such as diabetes and obesity [23].