Hydrolyzing NAD to NMN and AMP by SelO is Required for Mitochondria Homeostasis

doi:10.21203/rs.3.rs-5137152/v1

Download PDF

Biological Sciences - Article

Hydrolyzing NAD to NMN and AMP by SelO is Required for Mitochondria Homeostasis

https://doi.org/10.21203/rs.3.rs-5137152/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The regulation of mitochondrial nicotinamide adenine dinucleotide (mNAD) is crucial for numerous life processes. However, until now, no metabolic reaction responsible for downregulating mNAD has been identified. Through in silico screening of potential NAD-binding proteins, we discovered a previously unrecognized reaction where NAD+ (or NADH) is hydrolyzed to NMN (or NMNH) and AMP in mitochondria. This reaction is catalyzed by the selenoprotein SelO, with Mn2+ serving as an essential cofactor. Selenocysteine 667 on the SelO protein plays a critical role in its NAD-hydrolyzing activity. SelO directly interacts with the mitochondrial trifunctional complex, and inhibits lipid β-oxidation, where NAD+ is required as an essential cofactor. This reaction is enhanced by a gradual increase in pH, which signals heightened mitochondrial catabolism, and it downregulates matrix pH as a negative feedback mechanism to maintain mitochondrial homeostasis. The NAD-hydrolyzing activity of SelO and its role in lipid oxidation are conserved across distant species, from mammalian cells to bacteria. Furthermore, SelO has been identified as a direct target of scutellarein, a clinically used agent for various ailments. These findings reveal a conserved mechanism for spatiotemporal NAD regulation, and highlight its physiological significance across both prokaryotes and eukaryotes.

Biological sciences/Cell biology

Biological sciences/Molecular biology

Nicotinamide adenine dinucleotide (NAD) is a metabolite that plays a central role in energy metabolism. It exists in two forms as a redox couple: NAD⁺ and NADH. NAD⁺ functions as a cofactor, accepting and shuttling hydride ion (H^-) as NADH from various redox reactions to the electron transport chain (ETC) complex.¹ Subsequently, H⁺is pumped out to the mitochondrial intermembrane space, giving rise to membrane potential gradients for ATP production. Therefore, increased matrix pH also reflects efficient NAD⁺/NADH turnover and energy production.² Additionally, NAD⁺ is used as a cofactor or substrate in various metabolic reactions or protein modifications, playing important roles in regulating various life processes.

NAD is distributed throughout the cell but is highly compartmentalized.³^,⁴ Thus, mitochondrial NAD (mNAD) is regulated independently of other subcellular pools. The homeostasis of mNAD is pivotal for numerous life processes, making both its upregulation and downregulation crucial. Despite being studied for nearly a century, the identification of the first transporter, SLC25A51, responsible for NAD⁺ uptake into mitochondria, occurred only recently.⁵ In contrast to other metabolites, NAD participates in metabolic pathways by transforming between its two forms, but does not inherently consume itself. While the regulation of NAD consumption is equally vital, no metabolic reaction for consuming mNADhas been identified thus far.

It is known that, in mitochondria, NAD⁺ can be passively consumed for protein ADP ribosylation by PARPs or protein deacetylation by SIRTs.¹^,⁶ These protein modifications are strictly controlled by specific signaling, such as DNA damage, which is obviously not a proactive way for mNAD regulation. Moreover, protein modification typically occurs in the picomolar to nanomolar range, making it unlikely that the modification alone could account for mNAD fluctuation in the micromolar to millimolar range within a limited timeframe. Therefore, novel regulatory mechanisms responsible for mNAD fluctuation warrant further exploration.

mNAD is hydrolyzed to NMN and AMP by SelO

Metabolic reaction is initiated by metabolite, as substrate, binding to its enzyme. The potential enzyme responsible for mNAD fluctuationmust be a NAD-binding protein. To explore novel NAD-binding proteins, we have conducted in silico screening employing reverse docking with three tools (Vina⁷, AutoDock 4⁸ and LeDock⁹) against the AlphaFold2-modeled structures¹⁰ for all human proteins (ED Fig. 1a-b, and Supplementary Table 1), while using the scores of known NAD-binding proteins as references for thresholds (ED Fig. 1c). Considering the expression levels and sub-organelle localizations, the preliminary screening yielded 34 mitochondrial candidates (Fig. 1a and ED Fig. 1a). Among them, we prioritized 9 candidates with no known function in metabolic pathways (ED Fig. 1a-d). Interestingly, SELENOO (SelO), a known protein AMPylation enzyme¹¹, emerged as the only protein whose knockdown(KD) resulted in significant change in total NAD amount (the sum of NAD⁺and NADH) (Fig. 1b and ED Fig. 1e). This finding was further confirmed with tetracycline (Tet)-induced KD in stable cell lines (Fig. 1c and ED Fig.1f).

Thus, we purified human SelO protein (ED Fig. 1g) and performed surface plasmon resonance (SPR) analysis, which displayed a well NAD⁺-binding profile (Fig. 1d and ED Fig. 1h). To investigate whether SelO protein directly consumes NAD, we in vitro incubated the protein with NAD⁺ under various reaction conditions. Excitingly, NAD⁺ underwent enzymatic hydrolysis, and the products were identified as NMN and AMP by both LC-MS/MS (Fig. 1e) and HPLC-UV (Fig. 1f), indicating an undiscovered reaction in eukaryotic cells. More importantly, Mn²⁺,but not Mg²⁺,was identified as the essential coenzyme for this reaction (Fig. 1g). The product NMN suppressed this reaction in a dose-dependent manner (Fig. 1h). In addition, NADH can also be hydrolyzed to NMNH and AMP (Fig. 1i), indicating that SelO breaks down both NAD⁺ and NADH.

To directly quantify mNAD in living cells, we used a genetically encoded NAD sensor targeted to mitochondria.¹² This sensor revealed an increase in mNAD, indicated by decreased fluorescence at 488 nm, in response to Tet-induced KD, which was subsequently restored upon SelO reintroduction (Fig. 1j). In addition, the kinetic analysis of SelO-mediated NAD⁺ hydrolysis demonstrated that the reaction closely adheres to Michaelis-Menten kinetics (Fig. 1k). These results revealed an unrecognized reaction in which NAD is hydrolyzed to NMN and AMP by SelO in mitochondria.

SelO downregulates mNAD and lipid metabolism

To explore the direct function of SelO in human cells, we performed a metabolomic analysis by LC-MS/MS in cells shortly after Tet-induced KD, within a timeframe when the gene expression pattern was not notably affected, as indicated by RNA-seq (ED Fig. 2a and Supplementary Table 2). In line, NAD is among the significantly upregulated metabolites (ED Fig. 2b). Moreover, the “Nicotinate and nicotinamide metabolism” ranked at the top in the metabolite set overrepresentation analysis¹³. Notably, the nicotinate and nicotinamide metabolism in downregulated metabolite was the only pathway exceeding the significant threshold (Fig. 2a and Supplementary Table 3). More interestingly, as evident from the volcanic plot (Fig. 2b), lots of the altered metabolites were lipids and lipid-like molecules, which were particularly enriched in downregulated metabolites (Fig. 2c), suggesting an important role of SelO in the regulation of lipid metabolism.

SelO is ubiquitously expressed, with the highest level in human (ED Fig. 2c) and mouse (ED Fig. 2d) liver. Therefore, we generated liver-specific SelO knockout (KO) mice using Alb Cre (ED Fig. 2e). As expected, SelO KO in liver led to an increase in mNAD levels (Fig. 2d). Moreover, the degradation rate of mNAD also decreased in isolated mitochondria from KO liver (Fig. 2e). In addition to endogenous mNAD, we further detected that SelO KO also decreased the degradation rate of exogenously supplemented isotope-labeled NAD⁺ in liver mitochondria (Fig. 2f). More interestingly, the significantly decreased respiratory exchange ratio (RER)in KO mice suggests a greater reliance on lipids as a fuel source, since an RER value of 1.0 indicates carbohydrate usage, while a value of 0.7 indicates lipid usage¹⁴ (Fig. 2g). Upon normalization to body weight, liver KO mice consistently exhibited higher lipid oxidation and lower carbohydrate oxidation (Fig. 2h). It is worth mentioning that whole-body SelO KO mice were also generated (ED Fig. 2f), which consistently exhibited a notable phenotype of decreased RER (ED Fig. 2g) as well as higher lipid oxidation and lower carbohydrate oxidation (ED Fig. 2h). Due to the complicated roles of NAD in different tissues, we only focused on the liver-specific KO mice currently.

In line with enhanced lipid oxidation, the lipid deposition was obviously decreased in the liver of KO mice fed a high fat diet (Fig. 2i and ED Fig. 2i). Moreover, the metabolic intermediates in fatty acid oxidation, specifically the medium to short chain acyl-CoA derivatives, were found to be dramatically increased in mitochondria from KO liver (Fig. 2j). Consistently, lipid accumulation in cultured primary hepatocyte from KO mice was highly suppressed after exposure to palmitic acid (PA)/oleic acid (OA) (Fig. 2k). In addition, lipid accumulation in SelO KD cells using either Tet-induced shRNA (Fig. 2l) or siRNA (ED Fig. 2j) was significantly compromised, and both could be subsequently rescued after reintroduction of exogenous SelO. BODIPY dyes¹⁵ was further used to probe cellular nonpolar lipid, which confirmed the phenotype induced by SelO KD and reintroduction (Fig. 2m).

The RNA-seq analysis of liver transcriptomes revealed around 100 differentially expressed genes upon SelO KO (ED Fig. 2k, and Supplementary Table 4). Gene set enrichment analysis (GSEA)¹⁶showed that TCA cycle and fatty acid β-oxidation were among the most enriched gene sets (Fig. 2n). The upregulation of multiple genes involved in fatty acid β-oxidation by SelO KO was further confirmed by qPCR (Fig. 2o). These results indicated that SelO directly suppresses lipid metabolism, and its KO leads to alterations in the transcriptome to coordinate with the increase in lipid oxidation.

SelO protects mitochondria from metabolic stress

SelO KO in mouse liver increases its mNAD content and prevents hepatic steatosis induced by a high fat diet, suggesting a beneficial effect of SelO inhibition, which is in line with the decreased TC, TG, HDL-C levels in the blood (ED Fig. 3a). Unexpectedly, in the same analysis, the significantly increased alanine aminotransferase (ALT) and aspartate aminotransferase (AST)¹⁷ indicated an impaired liver function in KO mice (Fig. 3a).

To explain this contradiction, we performed electron microscopy (EM) to analyze the morphology of mitochondria in liver section. Mitochondria in KO liver was obviously more fragmented, as evidenced by the increase in numbers and the decrease in sizes, indicating mitochondrial stress, which exhibited a progressive nature, intensifying with age (Fig. 3b). Another interesting finding is the obscured intermembrane space of mitochondria in KO liver, which obviously worsened with age progression (Fig. 3c). The obscured intermembrane space is likely due to the shrinkage of the mitochondrial membranes or partial loss of membrane integrity. To confirm that, we co-stained the outer membrane and matrix makers in primary hepatocytes, showing their colocalization was severely compromised upon SelO KO (Fig. 3d), recapitulating the observation via EM. Consistently, SelO KD in human cells resulted in more fragmented mitochondria (Fig. 3e). Mitochondrial dysfunction is known to be an important cause of liver injury. CD45 staining in liver sections also indicated an increased inflammatory response in KO mice (ED Fig. 3b).

For ATP production in mitochondria, NADH shuttles electrons to the ETC complex, which is accompanied by pumping H⁺ into the intermembrane space, resulting in increased matrix pH. However, overactivation of this process could lead to mitochondrial stress.^18-22 Interestingly, a significantly higher matrix pH in cells shortly after Tet-induced KD was detected using a genetically encoded pH sensor, super ecliptic pHluorin (SEpHluorin)²³, targeted to the mitochondria (ED Fig. 3c). More strikingly, the NAD-hydrolyzing activity of SelO was greatly enhanced by a gradual pH increase (7.2 to 8.2), which mimics the dynamic variation in matrix pH²⁴^,²⁵ (Fig. 3f). This suggests that this reaction serves as a negative feedback mechanism to protect mitochondria from metabolic stress. Supporting this notion, the increased matrix pH in KD cells using either Tet-induced shRNA (Fig. 3g) or siRNA (ED Fig. 3d) was recovered upon SelO reintroduction. In addition, another mitochondria-targeted pH sensor, ratiometric pHluorin²⁶, confirmed the same phenotypes induced by SelO KD and rescue (Fig. 3h). Thus, it is reasonable that sustained SelO KO in mouse liver leads to mitochondrial stress.

SelO regulates catabolism through hydrolyzing NAD

NAD plays a central role in mitochondrial catabolism, and its downregulation by SelO is expected to weakens ETC activity, as evidenced by the decreased matrix pH (Fig. 3). However, SelO has been previously identified as a protein AMPylation enzyme, although only a limited number of proteins AMPylated by SelO have been identified¹¹. In our study, purified SelO protein demonstrated clear self-AMPylating activity in vitro, which was further enhanced by increasing pH levels (ED Fig. 4a). Therefore, whether its AMPylating activity is also involved in regulating catabolism remains to be determined.

To address that, we need to generate a mutated protein that retains the protein-AMPylating activity but has impaired NAD-hydrolyzing function. In the aforementioned in silico screening (Fig. 1a), NAD−SelO binding models were generated using the docking tools Vina, AutoDock 4 and LeDock. It is important to note that each tool can provide multiple binding conformations, and we only focused on the several top-ranked models. Based on these models, some amino acid residues (K176, D188, K191, E199, D229, D359 and D368) were predicted to be in the binding pockets and potentially involved in NAD hydrolysis. The mutant proteins were purified (ED Fig. 4b), and exhibited varying effects on NAD hydrolysis (ED Fig. 4c). The D188A, K191E, and D359A mutations resulted in different levels of compromised NAD-hydrolyzing activity, however, they also showed inhibitory effects on their self-AMPylation (ED Fig. 4d).

All the molecular docking tools we used assumed the SelO protein to be a rigid body, without accounting for the flexibility of the binding pocket. Additionally, the essential factor Mn²⁺was not included. To address this, we employed the newly released AlphaFold3²⁷ to model the interaction between both NAD⁺ and Mn²⁺ with SelO. Interestingly, the AlphaFold3-modeled structure revealed that the selenocysteine 667 (C667-SeH) is positioned near the phosphoanhydride bond of NAD, which needs to be cleaved during hydrolysis (Fig. 4a). According to literature, C667 should adversely affect the AMPylating activity¹¹. Consistent with the AlphaFold3 model, the C667A mutation highly impaired NAD-hydrolyzing activity (Fig. 4b and ED Fig. 4e). Meanwhile, this mutation notably enhanced SelO’s AMPylating activity (Fig. 4c).

In fact, the same binding pocket was identified by other docking tools, though not in the top-ranked model (without Mn²⁺). For instance, among the top 20 Vina-identified conformations, two potential NAD-binding pockets were predicted (ED Fig. 4f). While the highest-scoring conformation binds to an adjacent pocket, 7 conformations, including those ranked No.2 and No.3, were docked to the AlphaFold3-predicted pocket. Here, it is worth mentioning that we replaced the selenocysteine (C667-Se) with cysteine in the recombinant protein, since it was technically challenging to obtain protein containing selenocysteine. The deprotonated thiol group (-S^-) can act as a nucleophile to initiate hydrolysis; by comparison, the ionized selenothiolate (-Se^-) exhibits greater nucleophilicity, due to its larger atomic radius²⁸^,²⁹. Theoretically, native SelO (C667-SeH) should exhibit a stronger NAD-hydrolyzing activity than the recombinant protein in this study. The fairly low but detectable activity of C667A mutant suggests that other sites may also contribute to the hydrolysis. According to AlphaFold3-model, R253 is also located near the phosphoanhydride bond (Fig. 4a). We replaced R253 with glutamic acid (E), which inverted the charge at this position. Interestingly, R253E mutation dramatically enhanced NAD hydrolysis (Fig. 4b), suggesting that the binding pocket indeed includes R253, further supporting the AlphaFold3 model.

Functionally, reintroducing SelO (C667A) into KD cells showed a reduced ability to downregulate NAD compared to WT SelO (Fig. 4d and ED Fig. 4g). Moreover, the increased matrix pH in SelO KD cells was fully restored upon reintroduction of WT SelO, but not C667A SelO (Fig. 4e). Similarly, the impaired lipid accumulation in SelO KD cells was rescued by exogenous WT SelO, but not by C667A SelO (Fig. 4f and ED Fig. 4h). These findings suggest that SelO's NAD-hydrolyzing activity plays a key role in regulating mitochondrial catabolism and cellular lipid accumulation.

SelO KO enhances lipid utilization but leads to mitochondrial stress, which worsens over time. This suggests that moderate or timely, rather than continuous, inhibition of SelO may be more beneficial for overall health. Thus, we designed a throughput strategy to screen for inhibitors of SelO’s NAD-hydrolyzing activity (ED Fig. 4i). Among the candidates, scutellarein and ethyl gallate were further confirmed in vitro (Fig. 4g). Both compounds rapidly and significantly increased cellular NAD within 1 hour, but not in cells without SelO (Fig. 4h). Scutellarein, a natural agent clinically used in various ailments, exhibits a range of effects, including anti‑vascular disease, antitumor, anti‑diabetic and anti‑inflammatory activities.³⁰ It was also reported to ameliorate lipid accumulation in liver.³¹^,³² Using SPR, scutellarein was further proved to directly bind to SelO protein (Fig. 4i) and inhibits its NAD-hydrolyzing activity in a non-competitive manner (Fig. 4j). Functionally, scutellarein reduced lipid accumulation in control but not in SelO KD HepG2 cells exposed to PA/OA; on the other hand, SelO reintroduction recovered the lipid accumulation in KD cells, which was further negated by scutellarein (Fig. 4k). These findings suggest that the therapeutic effect of scutellarein on hepatic steatosis is, at least partially, mediated through targeting SelO.

SelO inhibits β-oxidation across prokaryotes and eukaryotes

It is conceivable that NAD hydrolysis leads to compromised catabolism in mitochondria; however, the primary role of SelO, mainly manifested in lipid oxidation, seems paradoxical. Metabolism is highly orchestrated through the spatial organization of metabolic enzymes, as only local metabolites around the enzymes determine reaction efficiency^33-35. We hypothesized that SelO preferentially degrades NAD around the enzymes involved in lipid oxidation.

Therefore, we performed immunoprecipitation (IP) and mass spectrometry (MS) analysis using lysates from mouse liver to explore SelO-associated proteins. Among the mitochondrial proteins those were highly ranked on the MS list, HADHA and ACAA2 were the precise enzymes involved in neighboring steps of the sequential reactions at the core of the fatty acid β-oxidation cycle (ED Fig. 5a). β-oxidation occurs in four steps, repeating until the fatty acid chain is completely broken down³⁶ (ED Fig. 5b). Therefore, we verified the interactions between SelO and HADHA or ACAA2, as well as all the other isoenzymes of HADHA or ACAA2 in mitochondrial β-oxidation. Indeed, the intracellular HADHA, HADHB, HADH, ACAA2 and ECHS1 are all associated with SelO (Fig. 5a-e), while ECHS1 binds to SelO more strongly than HADH (ED Fig. 5c).

Mitochondrial trifunctional protein (MTP) HADHA and HADHB form a complex that catalyzes three of the four key steps in the β-oxidation of medium to short chain fatty acids³⁶ (Fig. 5f). SelO was associated with this complex but did not influence the affinity between HADHA and HADHB (Fig. 5g and 5h). Moreover, we investigated whether SelO associates with HADHB depending on HADHA. However, HADHA KD severely affected HADHB levels, suggesting that HADHA and HADHB stabilize each other (ED Fig. 5d). This is supported by the evidence that recombinant HADHA/HADHB complex, rather than HADHA alone, can be purified from E. coli lysates(ED Fig. 5e). Moreover, SelO protein was confirmed to directly interact with the purified HADHA/HADHB complex in vitro (Fig. 5i). HADHA catalyzes the conversion of hydroxyacyl-CoA to ketoacyl-CoA, in which NAD⁺ serves as an essential acceptor for electrons (Fig. 5f). Therefore, we measured the activity of HADHA in vitro by quantifying reaction products through LC-MS. The results demonstrated that, not only the forward reaction using NAD⁺as a cofactor (Fig. 5j), but also the reverse reaction using NADH as a cofactor (Fig. 5k) was significantly suppressed in the presence of SelO protein. These observations suggest that SelO preferentially degrades NAD⁺/NADH around the enzyme complex for β-oxidation, leading to suppressed lipid oxidation. This notion was further supported by that either C667A mutation (Fig. 5l) or scutellarein treatment (Fig. 5m) directly negated the inhibitory effect of purified SelO protein on the activity of HADHA in vitro.

SelO is highly conserved from prokaryotes to humans, therefore, we purified its homologous protein in E. coli, ydiU¹¹ (ED Fig. 5f), which also displayed NAD⁺-binding properties (Fig. 5n and ED Fig. 5g). Consistently, ydiU hydrolyzed NAD⁺ (ED Fig. 5h), using Mn²⁺ as an essential cofactor (Fig. 5o). The kinetic parameters of ydiU hydrolyzing NAD⁺closely followed Michaelis-Menton kinetics, but exhibited lower catalytic efficiency than SelO (Fig. 5p). Therefore, whether ydiU also displays a conserved role in β-oxidation in E. coli was investigated. E. coli possesses two sets of trifunctional enzymes for β-oxidation: one set comprises FadA and FadB, primarily for aerobic conditions; the other set comprises yfcX and yfcY, primarily for anaerobic conditions (Fig. 5f).³⁷^,³⁸ The pulldown assays indicated that SelO selectively interacted with fadB (Fig. 5q) but not yfcX (ED Fig. 5i). fadA and fadB proteins were further purified (ED Fig. 5j). It was found that ydiU directly interacted with either fadA or fadB, and the presence of either fadA or fadB did not affect the binding of the other to ydiU (Fig. 5r). Size exclusion chromatography showed that purified ydiU co-eluted with the fadA/fadB complex, suggesting that ydiU was stably assembled into this trifunctional complex (Fig. 5s). To explore the functional consequences, we generated ydiU KO E. coli strain (ΔydiU). ΔydiU proliferated at a similar rate compared to its parental strain when using glucose as the carbon source (Fig. 5t upper). In contrast, ΔydiU proliferated much faster than its parental strain when growing in a nutrient-deficient media using oleic acid as the sole carbon source (Fig. 5t lower), demonstrating an inhibitory role of ydiU on fatty acid metabolism. Taken together, these results indicated that SelO/ydiU is directly assembled into trifunctional complex, inhibiting β-oxidation in both prokaryotes and eukaryotes.

NAD is one of the most attention-grabbing metabolites, serving as a cofactor for hundreds of enzymes, particularly those involved in energy metabolism, redox reactions, and signal transduction. Our initial aim was to explore any uncharacterized NAD-related enzymes in human cells. Our approach heavily relied on structural proteomics; however, less than one-third of human proteins have experimentally determined structures³⁹. Fortunately, AlphaFold2 has provided highly confident structural predictions for all human proteins. Unexpectedly and excitingly, this strategy enabled us to uncover not only an unknown NAD-related enzyme but also a novel reaction within mitochondria. To our knowledge, this is the first metabolic reaction identified in mitochondria from mammalian cells specifically dedicated to break down mNAD. Moreover, the NAD hydrolysis activity, the interactions between SelO/ydiU and MTP, and the biological significance of this hydrolysis reaction, are highly conserved across prokaryotes and eukaryotes (Fig. 6). The same approach could probably be further modified and used to explore other unrecognized metabolic reactions or enzymes.

Traditionally, NAD has been viewed as a beneficial metabolite.¹^,⁶ In line with this, SelO KO promotes mNAD content and lipid catabolism, preventing hepatic steatosis. However, our study provides evidence that uncontrolled mNAD can have adverse effects, highlighting the necessity for spatiotemporal regulation of mNAD at an appropriate level, rather than assuming that more is always better. The highest level of SelO in liver tissue also indicates its special necessity for maintaining homeostasis in hepatic mitochondria.

Moreover, SelO is also present at decent levels in almost all other tissues. Both liver-specific and global KO mice were generated in this study. However, phenotypes are not entirely consistent between them. For example, global KO eventually leads to increased body weight with age, a phenotype noticed but not further repeated as it was beyond the scope of this study. A consistent observation across cellular and animal models is that SelO KO or KD rapidly increases mNAD and enhances fatty acid metabolism. Meanwhile, mitochondrial stress occurs and worsens over time. Given that mitochondria are the primary organelles for fatty acid metabolism, once mitochondrial damage reaches a certain threshold, it is expected to suppress fatty acid metabolism in turn. That may partially explain the contradictory phenotype. On the other hand, NAD has complicated physiological functions that can vary across different organs. Its dysregulation is linked to various human diseases, further emphasizing the need for a precise regulation of mNAD levels. This also suggests the need for more refined strategies to maximize the benefits of NAD intervention or when targeting SelO. This study marks the first example of regulation of cellular catabolism via NAD degradation. Thus, we only focused on its direct functions and physiological significance. Further studies are needed to elucidate its physiological and pathological roles in various other tissues as well as across different stages of life.

Antibodies

Flag tag [PA1-984B, Invitrogen. 1:1000-2000 (WB)]; HA tag [12013819001, Roche 1:5000 (WB)]; Actin [A3854, Sigma-Aldrich, 1:10000 WB]; V5 tag [13202, Cell Signaling, 1:2000 (WB)], CD45 [70257, Cell Signaling, 1:100 (IHC, IF)], SelO [ERP11968, Abcam, 1:2000 WB]; HSP60 [SC-13115, Santa Cruz Biotechnology, 1:25 IF]; TOM20 [A19403, ABclonal, 1:100 IF]; HADHA [10758-1-AP, Protein tech, 1:1000 WB]; HRP-linked goat anti-rabbit [7074, Cell Signaling, 1:10000 (WB)]; HRP-linked goat anti-rabbit [GK500705, GeneTech (Shanghai), ready-to-use diluted solution (IHC)]; HRP-linked horse anti-mouse [7076, Cell Signaling, 1:10000 (WB)]; goat anti rabbit-Alexa Fluor 633 [A21071, Invitrogen, 1:500 (IF)]; goat anti-rabbit Alexa Fluor 488 [A11008, Invitrogen, 1:500 (IF)] Goat anti mouse-Alexa Fluor 488 [A11001, Invitrogen, 1:500 (IF)]

The AMPylation antibody (7C11) was a kind gift from Prof. Aymelt Itzen from Universitätsklinikum Hamburg-Eppendorf.

siRNA

All the siRNAs were purchased from Synbio Technologies.

The sequences of the siRNAs were:

siCont:

sense: UUCUCCGAACGUGUCACGUdTdT

antisense:ACGUGACACGUUCGGAGAAdTdT

siSelO:

sense: CGCUGUUCUUCAGCGGCAAdTdT

antisense: UUGCCGCUGAAGAACAGCGdTdT

siSelO-2 (targeting 3’-UTR):

sense: AGUUGCUGGUUUAUUUAGGAGCCUGdTdT

antisense: CAGGCUCCUAAAUAAACCAGCAACUdTdT

siHADHA:

sense: UGUGGCAGUUGUUCGAAUUAAdTdT

antisense: UUAAUUCGAACAACUGCCACAdTdT

siABHD10:

sense: CCCUGGCUAUCUUUCUUAUAUdTdT

antisense: AUAUAAGAAAGAUAGCCAGGGdTdT

siDAP3:

sense: CCCGAGGAAUUAGCACUUGUUdTdT

antisense: AACAAGUGCUAAUUCCUCGGGdTdT

siSLC25A32:

sense: GCAGCAACAUACCCAUAUCAAdTdT

antisense: UUGAUAUGGGUAUGUUGCUGCdTdT

siFOXRED1:

sense: GCAGUUCUCAUUGCCUGAGAAdTdT

antisense: UUCUCAGGCAAUGAGAACUGCdTdT

siMETTL15:

sense: AGCCAGGCAGAAGCCUUAUUAdTdT

antisense: UAAUAAGGCUUCUGCCUGGCUdTdT

siOSGEOL1:

sense: CCAAGUGACCUCUCAGCAAUUdTdT

antisense: AAUUGCUGAGAGGUCACUUGGdTdT

siLYRM2:

sense: UGAUGAUUACUCAAGGCAAUAdTdT

antisense: UAUUGCCUUGAGUAAUCAUCAdTdT

siSTARD7:

sense: CGGUUGGAAGAAAUGUCAAAUdTdT

antisense: AUUUGACAUUUCUUCCAACCGdTdT

Chemicals and Reagents

NAD⁺, AMP, ADP, and ATP were purchased from Sangon Biotech. NADH, NMN, dibutylammonium acetate (DBAA), and Adenosine 2’,5’-diphosphate were from MACKLIN. NMNH was purchased from Bidepharm. 3-Ketopalmitoyl-CoA was from Toronto Research Chemicals. Oleic acid, palmitic acid, and methyl tert-butyl ether (MTBE) were from HEOWNS. Stearoyl coenzyme A was bought from Shanghai Yuanye Bio-technology Co., Ltd. Isotope labeled NAD⁺ was from Cambridge Isotope Laboratories. Palmitoyl-CoA was from Sigma-Aldrich. Oil Red O was from Solarbio, and BODIPY 493/503 was from MCE. Opti-MEM and DMEM were from Gibco. 100X penicillin-streptomycin, 100X non-essential amino acids, and doxycycline were from Solarbio. Fetal bovine serum was from Gibco.

Cell Lines and Animals

HEK293T, Hepa1-6, HepG2, and HeLa were from ATCC.

The SelO global knockout mice (SelO^-/-) and the SelO floxed (SelO^fl/fl) mice were purchased from Cyagen. The SelO floxed (SelO^fl/fl) mice were crossed with Alb-Cre transgenic mice to generate liver-specific SelO knockout mice. All animals were kept under specific pathogen free (SPF) environment with 12 h light/12 h dark cycle. All animal experiments were approved by the Ethics Committee of Tianjin Medical University.

Plasmids

The plasmid for expression of SUMO-SelO (human) in E. coli was a kind gift from Prof. Vincent S. Tagliabracci from UT Southwestern. The plasmids encoding mitochondrial NAD sensors (186791), mitochondrial localized ratiometric pH sensor pHluorin (163053), and cytosolic super ecliptic pH sensor pHluorin (32001) were purchased from Addgene. The cytosolic super ecliptic pH sensor pHluorin was used as a template for cloning mitochondrial localized super ecliptic pHluorin,

The plenti-GFP-Flag was purchased from OBio Tech and served as the backbone vector for generating all lentiviral plasmids tagged with Flag, as well as plenti-GFP-HA and plenti-GFP-V5, which were used as backbone vectors for generating all the lentiviral plasmids tagged with HA and V5 tag, respectively.

The plasmids below were constructed for this study as described in Table S1:

pet28a-His₆-ydiU-Flag, pLKO-tet-shSelO, plenti-SelO-Flag (mouse), pet28a-fadA-V5, pet28a-fadB-HA, pet28a-yfcX-HA, plenti-HADHB-V5, plenti-HADHA-HA, plenti-ACAA2-Flag, plenti-HADH-Flag, plenti-ECHS1-Flag, plenti-SelO-HA, pet28a-GST-ydiU, pet28a-ACOX1, mito-mCherry-SEpHluorin, pet28a-SUMO-SelO (C667A), pet28a-SUMO-SelO (R253E).

All homemade plasmids were validated by sequencing at Genewiz. For overexpression of SelO protein, we mutated the codon for selenocysteine to cysteine.

In silico Screening of NAD⁺ Binding Proteins

We employed three molecular docking programs, Vina, Autodock 4, and Ledock, to screen potentially uncharacterized NAD⁺ binding proteins in mitochondria. The subcellular localization information for all proteins in the human proteome was retrieved from the UniProt Database. Additionally, we identified known NAD⁺, NADH, NADP⁺, or NADPH binding proteins based on catalytic activity, binding site, and cofactor information retrieved from the UniProt database. Reverse blind docking was used for the evaluation of the binding affinities of each protein to NAD⁺ with AlphaFold2 modeled structures. These structures were downloaded from the AlphaFold2 database, except for the selenium-containing proteins, which were generated in situ using the AlphaFold2 algorithm upon replacing selenocysteine with cysteine in the sequence. This is necessary because the AlphaFold2 algorithm does not recognize residue abbreviation other than the twenty common amino acids, and structures for selenium-containing proteins are absent from the online database. Prior to molecular docking, we trimmed the intrinsically disordered regions in the AlphaFold2 modeled structures and truncated the structures into individual domains using the Leiden graph-clustering method based on the predicted alignment error matrices generated along with each AlphaFold2 modeled structure. The individual domains lacking sufficient topological complexity (alpha helix ⁺ beta strand ≤ 3) were discarded. For evaluation of the binding affinities, the molecular docking programs autonomously determined the binding sites by searching against the whole protein surface. Default parameters were used for all three molecular docking programs, with the exception of Vina, where search exhaustiveness was increased from the default value of 8 to 12. The versions for the docking programs were as follows: Vina: 1.2; Autodock 4: 4.2; Ledock 1.0. We evaluated the binding affinities towards NAD⁺ for all mitochondrial proteins, excluding known NAD⁺/NADH or NADP⁺/NADPH binding proteins, as well as known intracellular or extracellular NAD⁺ binding proteins (not confined to mitochondrial proteins). The average scores for known NAD⁺ binding proteins were used as the cutoffs for the selection of potentially uncharacterized NAD⁺ binding proteins in mitochondria.

HPLC-UV Analysis of the Enzymatic Reaction Products

Recombinantly purified SelO or ydiU were incubated with NAD⁺ or NADH in a buffer consisting of 50 mM Hepes (pH=8.2), 150 mM NaCl, 1 mM MnCl₂, and 1 mM TCEP at 30 ^oC, and then the proteins were removed prior to HPLC analysis. For the reaction involving NAD⁺, the enzyme was removed by acid precipitation, followed by centrifugation to eliminate the precipitation. Subsequently, the solution was neutralized with NaOH before undergoing HPLC analysis. For the reaction involving NADH, enzyme removal was achieved through heat inactivation followed by centrifugation, because NADH is unstable when exposed to strong acid or base. The resulting reaction products were separated using an Agilent 1100 system coupled with an Agilent 1200 diode array detector equipped with deuterium and tungsten lamps for measurement of the absorbance. An HC-C18 column (5 μm, 4.6 X 250 mm) was employed for the separation. The eluents were (A) 100% methanol and (B) an aqueous solution containing 40 mM KH₂PO₄and 60 mM K₂HPO₄, pH=7.0. The separation and subsequent analysis with HPLC-UV were conducted at room temperature. Isocratic elution with 100% B was used for the reaction involving NAD⁺, and gradient elution was used for the reaction involving NADH (solvent gradient: 0-10 min: 100% B; 10-20 min 100%-80% B). The quantification of AMP and NMN was based on absorbance measurements at 260 nm, while the quantification of NMNH relied on absorbance measurements at 340 nm. Note that NMNH does not exhibit absorbance at 260 nm.

Recombinant Protein Purification

Bacterial expressing vectors harboring fadB or fadA were transformed into BL21(DE3) cells. Expression vectors carrying wildtype or mutant ydiU or SUMO-tagged SelO were transformed into BL21(DE3) ΔydiU cells. The ydiU gene was knocked out to prevent potential formation of chimeric heteromultimer between endogenous ydiU and the induced proteins. The bacteria were cultured in LB medium at 37 ^oC until reaching an OD600 of 0.5-0.9. Protein expression was induced by adding 0.5 mM isopropyl-β-D-thiogalactoside and incubating for 12-16 h at 16 ^oC. Following induction, the bacterial cells were harvested and sonicated in lysis buffer (50 mM NaH₂PO₄ (pH 7.0), 0.1% Triton X-100, 10 mM imidazole, 5 mM BME, and 1 mM PMSF). The lysate was then centrifuged at 10,000 g for 10 min. Nucleic acids in the supernatant were precipitated by gradually adding streptomycin sulfate (final concentration w/v 2%), followed by another centrifugation step. The resulting supernatant was loaded onto a cobalt column (TALON). After washing with five bed volume of wash buffer (50 mM NaH₂PO₄ (pH 7.0), 0.1% Triton X-100, 20 mM imidazole, 5 mM BME, and 800 mM NaCl), proteins were eluted from the column using elution buffer (50 mM NaH₂PO₄ (pH 7.0), 250 mM imidazole, 1 mM BME and 300 mM NaCl). The eluate was further concentrated using Amicon Ultra and buffer-exchanged using Sephadex G-25 into a storage buffer containing 50 mM Hepes, 100 mM NaCl, 10% glycerol, and 1 mM DTT.

Measurement of NAD⁺ and Related Nucleotides from in vitro Enzymatic Reactions or Mitochondria using Targeted UPLC-MS/MS

For in vitro enzymatic reactions, the reaction was terminated by the addition of 1 M HCl to adjust the pH to 1-2, followed by neutralization with 1 M NaOH solution. For measurement of NAD⁺ levels in mitochondria, mitochondria were extracted from mouse liver using the Tissue Mitochondria Isolation Kit from Beyotime (C3606). The mitochondria were then suspended in mitochondrial respiration medium MiR05 buffer and incubated for the indicated time, followed by metabolite extraction with 80% methanol and sonication. The reaction mixtures from in vitro reactions or mitochondrial extracts were filtered through a 0.22 μm pore centrifuge tube filter (Costar 8169) prior to analysis on an ultra-performance liquid chromatography system equipped with a BEH T3 column (1.7 μm, 2.1 X 100 mm). The eluents were (A) an aqueous solution containing 0.25 mM di-n-butylamine acetate (DBAA) and (B) acetonitrile containing 3 mM DBAA. The separations were conducted at room temperature using gradient elution (solvent gradient: 0-6.5 min, 0-35% B). Targeted profiling of NAD⁺ and related nucleotides was performed with a hybrid triple quadrupole linear ion trap mass spectrometer (5500 QTRAP, AB SCIEX) equipped with an electrospray ionization source. The mass-to-charge ratios (m/z) were determined in negative mode. The concentrations of the analytes were detected by multireaction monitoring monitoring.

Surface Plasmon Resonance

SPR data were acquired on a Biacore 8K device (GE Healthcare). Recombinantly purified ydiU or SelO proteins were covalently immobilized on a CM5 sensor chip series S. The mobile phase was 10 mM Hepes (pH=7.5), 150 mM NaCl, 0.05% Tween-20. Serial dilutions of NAD⁺ were injected into flow cells at concentrations ranging from 12.5-1600 μM, with a flow rate of 30 μL/min at 25 ^oC. Each binding cycle consisted of an association phase of 120 s and a dissociation phase of 160 s. The maximal response units at each concentration were fitted to a 1:1 binding equation to determine the dissociation constants (K_d).

Quantification of Cellular and Mitochondrial NAD Levels using Colorimetric Methods

Cellular NAD levels (sum of NAD⁺ and NADH) were quantified using the WST-8 colorimetric method with a commercial kit (Beyotime S0175). Cell lysates were prepared using the buffer provided by the kit, and the total NAD levels were assessed by the formation of formazan through catalysis with excess alcohol dehydrogenase. For determination of the mitochondrial NAD levels, mitochondria were first isolated from the cells using a mitochondrial isolation kit (Beyotime C3601). Absorbance readings at 450 nm were taken with the Tecan Spark microplate reader at a slit width of 3.5 nm. Background absorbance at 450 nm was subtracted using the absorbance of a reaction mixture set up in the absence of cell lysate. The protein concentrations of the lysates were determined using the BCA method for the normalization of the cellular NAD levels.

Western Blot

Cells were lysed in PBS containing 0.5% Triton X-100 and a protease inhibitor cocktail by repeated freeze-and-thaw and vortexing with glass beads. Protein samples were separated by SDS-PAGE and transferred to the PVDF membrane. After blocking with 10% fat-free milk, membranes were incubated with primary antibodies overnight at 4 ^oC with gentle shaking, followed by a one-hour incubation with horseradish peroxidase-conjugated secondary antibodies at room temperature. All the primary and secondary antibodies were diluted in 1% fat-free milk in 1XTBST except for those used for the detection of AMPylation. For detection of AMPylation, the primary and secondary antibodies were diluted in protein-free blocking buffer (Pierce, 37585) supplemented with MnCl₂. The PVDF membrane was also blocked with the protein-free blocking buffer for the detection of AMPylation. Blots were developed by chemiluminescence with ECL substrates from NCM Biotech. Immunoblots were visualized using the Tanon 5200 chemiluminescence imaging system.

Stable Cell Line Generation

HEK293T cells were co-transfected with pLKO-shSelO, packaging plasmid psPAX2, envelope plasmid pMD2.G using TRANIT-LT1. The supernatant containing lentivirus particles was collected and filtered through a sterilized 0.8 μm filter to remove floating cells or cell debris. HeLa cells were then infected with the lentivirus in the presence of polybrene and selected with 2 μg/ml puromycin for at least three days. To increase homogeneity within cell lines, monoclonal cell lines were generated by isolating single colonies from culture plates seeded at very low concentrations. The tetracycline-inducible expression of shSelO was verified by Western Blot analysis.

Flag Pulldown Assay

For pulldown with transfected mammalian cells, cells were lysed in PBS containing 0.5% Triton X-100 and a protease inhibitor cocktail by repeated freeze-and-thaw cycles and vortexing with glass beads. For pulldown with transformed E. coli, the bacterial cells were disrupted by sonication. After disrupting the cell pellets (either mammalian cells or bacterial cells), the lysates were centrifuged at 10,000g for 10 min at 4 ^oC, and the supernatant was loaded onto the Flag M2 magnetic beads. After incubation, the beads were washed three times with PBS. The bound proteins were then eluted with PBS containing 0.15 mg/mL 3XFlag peptide. The eluted proteins were either subjected to Western Blot analysis or mass spectrometry analysis. The mass spectrometry analysis, including tryptic digestion, peptide desalting and separation, mass spectra collection, and protein identification through database search, was conducted by the proteomic facility of Tianjin Medical University. An Orbitrap Q-extractive Plus mass spectrometer was used for the mass spectrometry analysis.

Size Exclusion Chromatography

Size exclusion chromatography analysis was carried out on an Agilent 1100 system equipped with an Agilent 1200 diode array detector for absorbance measurement. The column used was Superdex 200 Increase 10/300 GL column. Elution volume was estimated based on UV absorbance at 280 nm and 260 nm. The mobile phase consisted of 50 mM Hepes (pH=7.5) and 150 mM NaCl. The flow rate was 0.5 ml/min. For each analysis, we used 80-100 μl of protein solution at a concentration of approximately 10 μM. The proteins separated by size exclusion chromatography were collected using an automatic sampler, with one fraction collected every one minute. The proteins in each fraction were precipitated by adding three volumes of acetone and then redissolved in 1XLaemmli buffer. The samples were resolved by SDS-PAGE and stained with silver stain for analysis.

Generation of ydiU Knockout E.coli Cells

The ydiU gene was knocked out in BL21(DE3) E. coli using the λ RED recombinase system. First, BL21(DE3) cells were transformed with the helper plasmid pKD46, which encodes the RED recombinase. Subsequently, the transformed E. coli cells were again transformed with a PCR product containing an FRT-flanked kanamycin-resistant gene, with extensions on each end homologous to the sequence immediately before the start codon and after the stop codon of the ydiU gene. This PCR product was amplified from the pKD4 plasmid using the following primers:

ydiU_knockout_fwd:GACGAGAGTAACCGTCTACACTATCAAACAGGAGGATCTGTGTAGGCTGGAGCTGCTTC

ydiU_knockout_rev:AAAACTCAGGCTGGCAAGCTGCTGTTGACCAAGTAGCCTCATATGAATATCCTCCTTAG

Then, the cells were selected with kanamycin for recombination of the resistant gene at the locus of ydiU. The resistant gene was removed by transformation with the pCP20 plasmid, which encodes the FLP recombinase. Both the pKD46 and pCP20 plasmids contain temperature-sensitive replicons, enabling easy removal of the plasmids by raising the growing temperature.

Metabolomic Analysis of Cultured Cells

HeLa cells expressing doxycycline-inducible SelO targeting shRNA were treated with either doxycycline or vehicle for 48 h prior to cell harvesting and counting. Approximately 5X10⁶ cells were used for each replicate. The cells were washed with PBS, and metabolites in the cells were extracted using a mixture of methanol, acetone, and water (2:2:1, v/v/v). The mixture was centrifuged at 14,000 g for 20 min, and the supernatant was then vacuum-dried. For LC-MS analysis, the samples were re-dissolved in a mixture of acetonitrile and water (1:1, v/v). The analysis was performed using a UPLC (1290 Infinity LC) equipped with a BEH Amide column (1.7 μm, 2.1 X100 mm) and coupled to a quadrupole time-of-flight mass spectrometer at Shanghai Applied Protein Technology Co., Ltd. The eluents were (A) an aqueous solution containing 25 mM NH₄OAc and 25 mM NH₃∙H₂O, and (B) acetonitrile. The separations were conducted at room temperature using gradient elution (solvent gradient: 0-0.5 min, 95% B; 0.5-7 min, 96-65% B;7-8 min 65-40% B; 8-9 min 40 % B). Both ESI positive and negative modes were scanned. The data was processed with the XCMS software.

Drug Screening of SelO inhibitors

Drug screening was performed using the Explorer G3 integrated workstation, which includes a Janus G3 workstation equipped with a pin tool for dispensing compounds. Enzyme activity was measured using an Envision 2105 microplate reader, integrated into the system. A kinase inhibitor library from TargetMol was employed for the screening. We quantified the SelO activity based on the amount of AMP produced during NAD⁺ hydrolysis in a given time period. AMP is an inhibitor of firefly luciferase, which catalyzes a chemiluminescent reaction using luciferin and ATP as substrates. The amount of AMP produced was determined by assessing the inhibition of the firefly luciferase-mediated chemiluminescent reaction. Recombinantly purified SelO protein was first added to each well, followed by the addition of test compounds using the pin tool. After a 5-minute incubation, NAD⁺ was added, and the reaction was allowed to proceed for 2 hours. Subsequently, luciferase and its substrates were introduced into each well. Chemiluminescence was measured using the Envision 2105 microplate reader. Relative inhibitory activity was calculated by comparing the results from control wells (without inhibitors) and wells without SelO protein on the same microplate.

RNA-seq

The liver tissues were ground to a fine powder in liquid nitrogen, which was then transferred into TRIzol. Following this, the tissue samples underwent centrifugation at 12,000 g for 5 min at 4 ^oC, and then a mixture of chloroform and isoamyl alcohol (24:1) was added to the resulting supernatant. After another centrifugation at 12,000g for 8 min, the aqueous layer was recovered. Total RNA was precipitated from the aqueous layer and subsequently washed with 75% cold ethanol twice. Library construction and RNA-sequencing were carried out using the BGISEQ-500 platform by BGI-SHENZHEN Co.

Measurement of HADHA/HADHB Activity in vitro

For the forward reaction, HEK293T cells were co-transfected with both HADHA-Flag and HADHB-V5. The MTP complex was then immunopurified by pulldown with Flag M2 magnetic resin and elution with 3XFlag peptide. Palmitoyl-CoA (60 μM) was used as the starting material for the assay and was first converted into trans-hexadec-2-enoyl-CoA with excess recombinantly purified ACOX1 protein in O₂saturated buffer in situ. This is necessary because trans-hexadec-2-enoyl-CoA is not commercially available. After heat inactivation of the ACOX1 protein, and the reaction mixture was centrifuged at 10,000g for 10 min. The supernatant containing trans-hexadec-2-enoyl-CoA was directly used for the assay with HADHA and HADHB. The in vitro reaction was carried out with 250 μM NAD⁺ in the presence or absence of 1 μM recombinantly purified SelO, along with the purified HADHA and HADHB, at 30 ^oC for 2 h. The reaction was terminated by heat inactivation.

For the reverse reaction, HEK293T cells were transfected with HADHA-Flag, and the HADHA protein was immunopurified by pulldown with Flag M2 magnetic resin and elution with 3XFlag peptide, 3-ketopalmitoyl CoA (60 μM) was used as the starting material. The in vitro reaction was carried out with 250 μM NADH in the presence of absence of 1 μM recombinantly purified SelO, along with the purified HADHA, at 30 ^oC for 3 h. The reaction was terminated by heat inactivation.

After halting the reactions, stearoyl-CoA (30 μM) was supplemented into the reaction mixture as an internal control to account for the loss of fatty acyl-CoA molecules in the subsequent steps. The samples were vacuum-dried and then re-dissolved into a mixture of methanol and H₂O (1:1). The re-dissolved samples were filtered through a 0.22 μm pore centrifuge tube filter (Costar 8169) prior to analysis on an ultra-performance liquid chromatography system equipped with a BEH C18 column (1.7 μm, 2.1 X 100 mm) The eluents were (A) an aqueous solution containing 5 mM NH₄OAc and 2.5 mM DBAA and (B) 95% acetonitrile and 5% (A). The separations were conducted at room temperature using gradient elution (solvent gradient: 0-1 min, 2-50% B; 1-8.5 min, 50-98% B; 8.5-13.5 min 98% B). Non-targeted profiling of fatty acyl-CoA was performed with Thermo Orbitrap Exploris 480 equipped with an electrospray ionization source. The mass-to-charge ratios (m/z) were determined in positive mode. Thermo Xcalibur Qual Browser was used for data analysis. The formation of 3-ketopalmitoyl-CoA was used for quantifying the enzymatic activity in the forward reaction, and the formation of 3-hydroxylpalmitoyl-CoA was used for quantifying the enzymatic activity in the reverse reaction.

Measurement of the Hydrolysis of Isotope Labeled NAD⁺ with Isolated Mitochondria

Mitochondria were isolated from liver tissues obtained from wildtype mice or mice with liver-specific knockout of SelO using the Tissue Mitochondria Isolation Kit from Beyotime (C3606). Subsequently, the isolated mitochondria were resuspended in an oxygen saturated solution containing 20 mM NH₄OAc (pH=7.45), 30 mM lactobionic acid, 20 mM taurine, 0.5 mM EGTA, 1 mM MgCl2, 1g/L BSA, and 1% Tween-20. Isotope labeled NAD⁺ (C13 labeling on the ribose linked to nicotinamide) was supplemented into the mixture and incubated at 30 ^oC for 1.5 h. Following incubation, metabolites were extracted with a mixture of methanol and acetone (1:1) using a Bioruptor pico sonication device. The solution was then centrifuged at 10,000 g for 10 min, and the supernatant was recovered and vacuumed dried. Subsequently, the samples were redissolved in ddH₂O and filtered through a 0.22 μm pore centrifuge tube filter (Costar 8169) prior to analysis on an ultra-performance liquid chromatography system equipped with BEH C18 column (1.7 μm, 2.1 X 100 mm). The eluents were (A) an aqueous solution containing 10 mM NH₄OAc (pH=9) (B) acetonitrile. The separations were conducted at room temperature using gradient elution (solvent gradient: 0-2 min, 3% B; 2-4 min, 3-90% B; 4-5 min 90% B). The hydrolysis of isotope labeled NAD⁺ was quantified by the formation of isotope labeled NMNH in the extract. The mass-to-charge ratios (m/z) were determined in positive mode.

Measurement of Blood Biochemical Parameters in Mice

For the measurement of HDL-C, LDL-C, ALT, AST, TC, and TG, blood was withdrawn from mice via cardiac puncture. The mice were anesthetized with isoflurane prior to blood collection. The blood samples were analyzed using an automatic biochemistry and immunoassay analyzer (Mindary SAL9000). The levels of ALT, AST, TC, and TG were determined using colorimetric enzymatic assays that employed lactate dehydrogenase, malate dehydrogenase, cholesterol oxidase, and triglyceride lipase, respectively. The HDL-C and LDL-C levels were determined using direct measurement methods. Ready-to-use reagent kits provided by the analyzer’s manufacturer were used. For measurement of blood glucose, tail blood was collected, and a Sinocare blood glucose meter was used along with the manufacturer-provided blood glucose test strips.

Measurement of Mouse Liver NEFA and TG Levels

Mouse tissues were mechanically homogenized using a bead mill homogenizer in either normal saline (for NEFA) or a 1:1 mixture of heptane and isopropanol (for TG). The resulting mixture was centrifuged at 10,000g for 10 min. The supernatant was collected for the determination of NEFA and TG levels. NEFA levels were assessed using a commercial kit from Nanjing Jiancheng Bioengineering Institute (A042-2-1). The levels were determined using an enzyme-based colorimetric assay employing acetyl-coA synthetase, acetyl-coA oxidase, and peroxidase. TG levels were assessed using a commercial kit from Solarbio (BC0620). The levels were determined using a chemical reaction-based colorimetric assay involving saponification of TG and oxidation of glycerol by periodic acid.

Oil Red Staining for Analysis of Lipid Content

Oil Red Staining was used to assess the levels of neutral lipids accumulated in cultured cells or liver tissues. Cultured cells were grown directly on microscopic slide before fixation with paraformaldehyde. Liver tissues were prefixed with 4% paraformaldehyde and then incubated in 20% and 30% sucrose sequentially for dehydration. The dehydrated tissues were cryosectioned and air-dried on precoated slides. The tissue slide or cell culture was fixed in 4% paraformaldehyde for 20 min at room temperature, followed by a brief wash with water. Subsequently, the slides were incubated with 60% isopropanol for 5 min before staining with Oil Red O working solution at room temperature for 20 min. After Oil Red staining, the slides were washed with water and stained with Hematoxylin. Images of the Oil Red Staining were captured using a Nikon Eclipse Ti microscope.

qPCR

For RNA extraction from liver tissues, the tissues were ground into fine powder in liquid nitrogen, and TRIzol reagent was added to the powder. After centrifugation for 10 min at 10,000g, the supernatant was used for total RNA extraction. For RNA extraction from cultured cells, the media was aspirated, and TRIzol reagent was directly added to the attached single-layer cells. Total RNA was isolated from the TRIzol solution following the manufacturer’s instructions. With the total RNA as template, cDNA was synthesized using a mixture of reverse transcriptase, oligo dT primer, and random primers (Transgene, AT301-03). Sybr Green master mix was used for PCR reactions as per the manufacturer’s instructions. qPCR assays were conducted using a BioRad CFX96 Real-Time PCR detection system. The sequence information for primer pairs used in qPCR is described in Table S2.

Measurement of fatty acyl-CoA in Mitochondria using Targeted UPLC-MS/MS

For measurement of fatty acyl-CoA levels in mitochondria, mitochondria were extracted from mouse liver using the Tissue Mitochondria Isolation Kit from Beyotime (C3606). Isolated mitochondria from approximately 600 mg liver tissues were suspended in 400 μl of 75% methanol. Metabolites were extracted by sonication, followed by centrifugation at 1,000 g for 10 min. The resulting supernatant was mixed with 1 ml of MTBE and incubated at room temperature with vortexing. Subsequently, around 250 μL of H₂O was supplemented to the mixture and allowed to incubate for another 10 min. The mixture was then centrifuged at 12,000 g for 10 min, and the upper organic layer was collected and vacuum-dried for measurement of fatty acyl-CoA. The dried metabolites were redissolved in 80% methanol and filtered through a 0.22 μm pore centrifuge tube filter (Costar 8169) prior to analysis on an ultra-performance liquid chromatography system equipped with a BEH C18 column (1.7 μm, 2.1 X 100 mm). The eluents were (A) acetonitrile and (B) H₂O. The separations were conducted at room temperature using gradient elution (solvent gradient: 0-2 min, 40% A; 2-7.5 min, 40-100 % A; 7.5-8 min 100% A). Targeted profiling of NAD⁺ and related nucleotides was performed with a hybrid triple quadrupole linear ion trap mass spectrometer (5500 QTRAP, AB SCIEX) equipped with an electrospray ionization source. The mass-to-charge ratios (m/z) were determined in positive mode.

Immunohistochemistry Staining of Liver Tissues

Human liver paraffin sections were prepared from freshly harvested tissues, which were fixed in 4% paraformaldehyde overnight and processed for paraffin embedding. The resulting paraffin blocks were sliced into 6 μm sections and mounted onto glass slides. The sections were deparaffinized with xylene, followed by rehydration through sequential immersion in alcohol solution of varying concentrations from 100% to 60%. The antigen unmasking was achieved with 10 mM sodium citrate (pH=6.0) buffer. The slides were then incubated with 3% H₂O₂ to quench endogenous peroxidase activity. Subsequently, the sections were blocked by 10% goat serum at room temperature for 1 h. After draining off the blocking buffer, the slides were incubated with the primary antibody solution containing 0.5% BSA at 4 ^oC overnight, followed by incubation with HRP conjugated secondary antibody solution at room temperature for 30 min. The antibody staining was visualized using DAB substrate solution, and the slides were then counterstained with Hematoxylin for 5 min. After dehydration with alcohol, the immunohistochemistry images were captured usinga Nikon Eclipse Ti microscope.

Fluorescence Microscopy

The fluorescence images for quantification of the mitochondrial NAD⁺ levels were obtained using live cells, the images for CD45 staining were obtained using liver paraffin sections, and the images for mitochondria were obtained using paraformaldehyde fixed cells. The liver parafiin sections were prepared following the same procedures as described in the immunohistochemistry staining section elaborated above, with the exception of using a fluorophore-conjugated secondary antibody instead of an HRP conjugated secondary one. Furthermore, an antifade mounting media (Beyotime, P0126) was applied to the microscope slides in the final step. For fixation of cultured cells, 2.5 % paraformaldehyde was used, followed by blocking and permeabilization with PBS buffer containing 3% BSA and 0.2 % Triton X-100. After permeabilization and blocking, cells were sequentially incubated with the primary and secondary antibody solutions diluted with PBS containing 1% BSA and 0.02% Triton-X 100. All fluorescent images were taken with a Zeiss LSM 800 confocal laser scanning microscope at room temperature using a 63X objective oil len.

Electron Microscopy Analysis of the Morphology of Mitochondria

Cells or tissues were fixed with 2.5% glutaraldehyde solution at 4 ^oC for 12 h. Subsequently, the samples were further fixed with 1% osmium tetroxide and dehydrated using a series of acetone solutions. Following dehydration, the samples were embedded in epoxy resin and then sectioned into 70-90 nm thin slides, which were then mounted on the specimen grid. The samples were then sequentially stained with uranyl acetate and lead citrate, followed by drying with filter paper. Electron microscopic images were captured using a Hitachi-800 Transmission Electron Microscope.

Statistical Analyses

Data were presented as mean ± SD or mean ± SEM as indicated in figure legends. Sample size N was indicated in figure legends. Statistical significance was set at p<0.05. n.s. stands for “not significant”. Significance between two groups was determined using Student’s t-test (unpaired, two-tailed, unequal variance).

Data availability

Source Data are all provided with this paper. Additional information is available from the corresponding author upon reasonable request.

The unprocessed metabolomic data was uploaded to EMBL-EBI MetaboLights under the accession code MTBLS11123. The dataset will be publicly available before the publication. For now, reviewers can use the following link to access the raw data: https://www.jianguoyun.com/p/DVGwyl8QzsecDBiFp9gFIAA.

The unprocessed RNA-seq data for mouse liver was deposited in the NCBI GEO under the accession code GSE277735. The raw data and processed data can now be accessed at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE277735 with the secure token "sletockaltcfler".

Acknowledgements

We would like to thank Dr. Vincent S. Tagliabracci (vincent.tagliabracci@ utsouthwestern.edu) for offering human SelO plasmid of ppSumo-H. sapiens SelO. We would like to thank Dr. Xu Zhang at Tianjin Medical University for performing LC-MS/MS. We would like to thank the Core Facility of Research Center of Basic Medical Sciences at Tianjin Medical University for assistance in immunofluorescence imaging. We also gratefully acknowledge the technical support by the High-performance Computing Platform and High-throughput screening facility of Tianjin Medical University. We thank Jie Shen at Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences for assisting SPR analysis. This work was supported by National Natural Science Foundation of China 82350116 (T.W.), 32070711 (Y.F.), 82073051 (T.W.), 82273088 (Q.Y.).

Author contributions

Teng Zhang and Yuan Fu uncovered the reaction. Xiaofan Jia performed most of the cell and animal experiments. Teng Zhang performed most of the biochemical and protein analyses. Xiaofan Jia and Teng Zhang contributed equally, and the order of authorship between them is determined entirely by the 'ladies first' principle. Yuan Fu performed most of the bioinformatic analysis. Chenxi Yang, Li Wu, Longfei Diao, Yuting Yang and Jie Wu performed some of the experiments. Yeyi Li provided various trivial support for the project. Ting Wang and Yuan Fu conceived and designed the study. Qiujing Yu and Peng Jiang provided conceptual advice. Ting Wang and Yuan Fu supervised the project. Ting Wang and Yuan Fu wrote the manuscript. All authors edited the manuscript.

Competing interests

The authors declare no competing interests.

Additional Information

Additional information includes a Word file for detailed figure legends, an Excel file for all cloning primers and a PDB file for AlphaFold3-modeled structure.

Supplementary information

Supplemental information includes Extended Data Figures 1-5, Extended Data Figure legends 1-5 and Supplementary Tables 1-4.

Correspondence and requests for materials should be addressed to Ting Wang or

Yuan Fu.

Xie, N. et al. NAD(+) metabolism: pathophysiologic mechanisms and therapeutic potential. Signal Transduct Target Ther5, 227, doi:10.1038/s41392-020-00311-7 (2020).
Santo-Domingo, J. & Demaurex, N. Perspectives on: SGP symposium on mitochondrial physiology and medicine: the renaissance of mitochondrial pH. J Gen Physiol139, 415-423, doi:10.1085/jgp.201110767 (2012).
Cambronne, X. A. & Kraus, W. L. Location, Location, Location: Compartmentalization of NAD(+) Synthesis and Functions in Mammalian Cells. Trends Biochem Sci45, 858-873, doi:10.1016/j.tibs.2020.05.010 (2020).
Cohen, M. S. Interplay between compartmentalized NAD(+) synthesis and consumption: a focus on the PARP family. Genes Dev34, 254-262, doi:10.1101/gad.335109.119 (2020).
Luongo, T. S. et al. SLC25A51 is a mammalian mitochondrial NAD(+) transporter. Nature588, 174-179, doi:10.1038/s41586-020-2741-7 (2020).
Covarrubias, A. J., Perrone, R., Grozio, A. & Verdin, E. NAD(+) metabolism and its roles in cellular processes during ageing. Nat Rev Mol Cell Biol22, 119-141, doi:10.1038/s41580-020-00313-x (2021).
Eberhardt, J., Santos-Martins, D., Tillack, A. F. & Forli, S. AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model61, 3891-3898, doi:10.1021/acs.jcim.1c00203 (2021).
Morris, G. M. et al. AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility. J Comput Chem30, 2785-2791, doi:10.1002/jcc.21256 (2009).
Zhang, N. & Zhao, H. Enriching screening libraries with bioactive fragment space. Bioorg Med Chem Lett26, 3594-3597, doi:10.1016/j.bmcl.2016.06.013 (2016).
Varadi, M. et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res50, D439-D444, doi:10.1093/nar/gkab1061 (2022).
Sreelatha, A. et al. Protein AMPylation by an Evolutionarily Conserved Pseudokinase. Cell175, 809-821 e819, doi:10.1016/j.cell.2018.08.046 (2018).
Cambronne, X. A. et al. Biosensor reveals multiple sources for mitochondrial NAD(+). Science352, 1474-1477, doi:10.1126/science.aad5168 (2016).
Pang, Z. et al. MetaboAnalyst 6.0: towards a unified platform for metabolomics data processing, analysis and interpretation. Nucleic Acids Res52, W398-W406, doi:10.1093/nar/gkae253 (2024).
Widmaier, E. P., Raff, H., Strang, K. T. & Vander, A. J. Vander's human physiology : the mechanisms of body function. Fourteenth edition. edn, (McGraw-Hill, 2016).
Loudet, A. & Burgess, K. BODIPY dyes and their derivatives: syntheses and spectroscopic properties. Chem Rev107, 4891-4932, doi:10.1021/cr078381n (2007).
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A102, 15545-15550, doi:10.1073/pnas.0506580102 (2005).
Giannini, E. G., Testa, R. & Savarino, V. Liver enzyme alteration: a guide for clinicians. CMAJ172, 367-379, doi:10.1503/cmaj.1040752 (2005).
Korshunov, S. S., Skulachev, V. P. & Starkov, A. A. High protonic potential actuates a mechanism of production of reactive oxygen species in mitochondria. FEBS Lett416, 15-18, doi:10.1016/s0014-5793(97)01159-9 (1997).
Wu, J. et al. Pancreatic mitochondrial complex I exhibits aberrant hyperactivity in diabetes. Biochem Biophys Rep11, 119-129, doi:10.1016/j.bbrep.2017.07.007 (2017).
Rottenberg, H., Covian, R. & Trumpower, B. L. Membrane potential greatly enhances superoxide generation by the cytochrome bc1 complex reconstituted into phospholipid vesicles. J Biol Chem284, 19203-19210, doi:10.1074/jbc.M109.017376 (2009).
Robb, E. L. et al. Control of mitochondrial superoxide production by reverse electron transport at complex I. J Biol Chem293, 9869-9879, doi:10.1074/jbc.RA118.003647 (2018).
Mor, D. E. & Murphy, C. T. Mitochondrial hyperactivity as a potential therapeutic target in Parkinson's disease. Transl Med Aging4, 117-120, doi:10.1016/j.tma.2020.07.007 (2020).
Koivusalo, M. et al. Amiloride inhibits macropinocytosis by lowering submembranous pH and preventing Rac1 and Cdc42 signaling. J Cell Biol188, 547-563, doi:10.1083/jcb.200908086 (2010).
Poburko, D., Santo-Domingo, J. & Demaurex, N. Dynamic regulation of the mitochondrial proton gradient during cytosolic calcium elevations. J Biol Chem286, 11672-11684, doi:10.1074/jbc.M110.159962 (2011).
Casey, J. R., Grinstein, S. & Orlowski, J. Sensors and regulators of intracellular pH. Nat Rev Mol Cell Biol11, 50-61, doi:10.1038/nrm2820 (2010).
Werley, C. A., Boccardo, S., Rigamonti, A., Hansson, E. M. & Cohen, A. E. Multiplexed Optical Sensors in Arrayed Islands of Cells for multimodal recordings of cellular physiology. Nat Commun11, 3881, doi:10.1038/s41467-020-17607-5 (2020).
Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature630, 493-500, doi:10.1038/s41586-024-07487-w (2024).
Ren, X., Zou, L., Lu, J. & Holmgren, A. Selenocysteine in mammalian thioredoxin reductase and application of ebselen as a therapeutic. Free Radic Biol Med127, 238-247, doi:10.1016/j.freeradbiomed.2018.05.081 (2018).
Reich, H. J. & Hondal, R. J. Why Nature Chose Selenium. ACS Chem Biol11, 821-841, doi:10.1021/acschembio.6b00031 (2016).
Xie, Y. et al. Current advances on the therapeutic potential of scutellarin: an updated review. Nat Prod Bioprospect14, 20, doi:10.1007/s13659-024-00441-3 (2024).
Luan, H. et al. Scutellarin, a modulator of mTOR, attenuates hepatic insulin resistance by regulating hepatocyte lipid metabolism via SREBP-1c suppression. Phytother Res34, 1455-1466, doi:10.1002/ptr.6582 (2020).
Zhang, X. et al. Scutellarin ameliorates hepatic lipid accumulation by enhancing autophagy and suppressing IRE1alpha/XBP1 pathway. Phytother Res36, 433-447, doi:10.1002/ptr.7344 (2022).
Srere, P. A. Complexes of sequential metabolic enzymes. Annu Rev Biochem56, 89-124, doi:10.1146/annurev.bi.56.070187.000513 (1987).
Troncoso-Ponce, M. A., Nikovics, K., Marchive, C., Lepiniec, L. & Baud, S. New insights on the organization and regulation of the fatty acid biosynthetic network in the model higher plant Arabidopsis thaliana. Biochimie120, 3-8, doi:10.1016/j.biochi.2015.05.013 (2016).
Oehler, S. & Muller-Hill, B. High local concentration: a fundamental strategy of life. J Mol Biol395, 242-253, doi:10.1016/j.jmb.2009.10.056 (2010).
Houten, S. M. & Wanders, R. J. A general introduction to the biochemistry of mitochondrial fatty acid beta-oxidation. J Inherit Metab Dis33, 469-477, doi:10.1007/s10545-010-9061-2 (2010).
Campbell, J. W., Morgan-Kiss, R. M. & Cronan, J. E., Jr. A new Escherichia coli metabolic competency: growth on fatty acids by a novel anaerobic beta-oxidation pathway. Mol Microbiol47, 793-805, doi:10.1046/j.1365-2958.2003.03341.x (2003).
Pramanik, A., Pawar, S., Antonian, E. & Schulz, H. Five different enzymatic activities are associated with the multienzyme complex of fatty acid oxidation from Escherichia coli. J Bacteriol137, 469-473, doi:10.1128/jb.137.1.469-473.1979 (1979).
Zhang, S., Zhang, T. & Fu, Y. Proteome-wide structural analysis quantifies structural conservation across distant species. Genome Res33, 1975-1993, doi:10.1101/gr.277771.123 (2023).

There is NO Competing Interest.

Supplementaryinformationprimersforcloning.xlsx
Primers for cloning
AlphaFold3structureSelONADMn2.txt
Related Manuscript File
SupplementaryTable1dockingscores.xlsx
Supplementary Table 1
SupplementaryTable2SelOKDRNAseqdata.xlsx
Supplementary Table 2
SupplementaryTable3SelOKDmetabolosm.xlsx
Supplementary Table 3
SupplementaryTable4RNAseqmouseliverSeloKO.xlsx
Supplementary Table 4
figureS1S5.pdf
Extended Data Figures 1-5
Additionalinformationdetailedfigurelegends.docx
Detailed figure legends 1-5

Download PDF

Version 1

posted

You are reading this latest preprint version

Hydrolyzing NAD to NMN and AMP by SelO is Required for Mitochondria Homeostasis

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Materials and Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1