SLC29A1 and SLC29A2 are human nicotinamide cell membrane transporters

Abstract

Nicotinamide (NAM), a main precursor of NAD+, is essential for cellular fuel respiration, energy production, and other cellular processes. Transporters for other precursors of NAD+ such as nicotinic acid and nicotinamide mononucleotide (NMN) have been identified, but the cellular transporter of nicotinamide has not been elucidated. Here, we demonstrate that equilibrative nucleoside transporter 1 and 2 (ENT1 and 2, encoded by SLC29A1 and 2) drive cellular nicotinamide uptake and establish nicotinamide metabolism homeostasis. In addition, ENT1/2 exhibits a strong capacity to change the cellular metabolite composition and the transcript, especially those related to nicotinamide. We further observe that ENT1/2 regulates cellular respiration and senescence, contributing by altering the NAD+ pool level and mitochondrial status. Changes to cellular respiration, mitochondrial status and senescence by ENT1/2 knockdown are reversed by NMN supplementation. Together, ENT1 and ENT2 act as both cellular nicotinamide-level keepers and nicotinamide biological regulators through their NAM transport functions.

Nicotinamide is an NAD+ precursor with crucial roles in cellular respiration and energy production. Here, the authors show that equilibrative nucleoside transporters 1 and 2 (ENT1 and ENT2), encoded by the genes SLC29A1 and SLC29A2, function in nicotinamide uptake into cells.

Introduction

Nicotinamide (NAM), an active form of vitamin B3, is widely studied and utilized for its antiaging, metabolic regulation, organ protective and reparative effects^1,2. NAM participates in a variety of intracellular and intercellular processes and regulates cellular metabolism, stress and immune responses, all of which are molecular mechanisms of its physiological and biochemical functions³. NAM is an important precursor of nicotinamide adenine dinucleotide (NAD+), also called coenzyme I, and NAD+ plays a key role in cellular metabolic processes, including glycolysis, oxidative phosphorylation, and ATP generation⁴. In addition to acting as an electron receptor, NAD+ is an essential substrate of multiple signaling enzymes, including sirtuins (SIRTs), poly ADP-ribose polymerases, and cyclic ADP-ribose synthases, which are involved in metabolic configuration, stress resistance, mitochondrial biogenesis, DNA repair, transcription regulation and other activities⁵. The functions of these enzymes depend on the cellular NAD+ pool. The salvage synthesis pathway, which involves a multistep enzymatic reaction originating from NAM in the cytoplasm, is the main source of cellular NAD+ in mammalian cells, implying that NAM supplementation directly affects the level and function of NAD+^6,7 and even cellular behavior and fate. Therefore, it is critical to elucidate how NAM enters cells.

Despite the gradual elucidation of the physiological effects, cellular metabolism, and signal regulation roles of NAM explored over the past century⁸, the mechanism of cellular uptake of NAM has never been uncovered. Based on the structure of NAM, it is unable to enter cells without transporters⁹; however, it is not clear which transporters are involved. Considering the significance of NAM for cell survival, potential NAM transporters must be ubiquitous. Equilibrative nucleoside transporter 1 and 2 (ENT1 and ENT2, encoded by the SLC29A1/2 gene), which are expressed in all tissues¹⁰, could be candidates, and existing evidence connects ENT1/2 with NAM. First, NAM and nicotinamide ribose have structures similar to those of adenine and adenosine, respectively, and adenine and adenosine are known ENT1/2 substrates. Second, as a common antiaging agent, NAM is also a biomarker of senescence, and both NAM and NAD+ levels in different tissues decrease with age^11,12. Interestingly, both ENT1 and ENT2 expression levels in multiple tissues also decline as aging progresses^13,14. Third, weakened energy metabolism is a feature of cellular senescence, which is characterized by mitochondrial dysfunction and reduced oxidative phosphorylation efficiency^6,15; thus, potential NAM transporters are likely involved in fuel respiration. Notably, tissues with greater energy requirements, such as skeletal muscle and the heart¹⁶, have higher ENT1/2 expression than other tissues. ENT1 and ENT2 expression is downregulated in several diseases, such as metabolic-associated fatty liver disease, inflammatory bowel disease¹⁷, and renal fibrotic disease^18,19, all of which are highly correlated with abnormal cellular fuel respiration. Thus, the existing evidence suggests that ENT1 and ENT2 are potential NAM transporters, but this has not yet been confirmed.

Here, we identified ENT1 and ENT2 as human NAM cell membrane transporters. ENT1/2 not only contributed to maintaining NAM homeostasis but also regulated NAM biological roles. We revealed that ENT1 or ENT2 knockdown reduced cellular uptake of NAM and subsequently caused reduction in NAM and even NAD+ levels, which eventually resulted in metabolic and transcriptional signatures that highly overlapped with those of NAD+ deficiency. Moreover, we found that ENT1/2 knockdown inhibited fuel respiration and promoted senescence, and the underlying mechanism involved ENT1/2-mediated influence on mitochondrial status via intervention in NAM-NAD+ homeostasis. In addition, nicotinamide mononucleotide (NMN) supplementation rescued ENT1/2 knockdown-induced above damages by recovering NAD+ level. Furthermore, ENT1/2 overexpression promoted fuel respiration and prevented cellular senescence, while NAM supplementation potentiated these effects.

Results

ENT1/2 drove cellular NAM uptake

Due to the vital role of NAM in cellular activity²⁰, potential NAM transporters should be expressed in all types of cells. We considered the transporters of essential substances (such as amino acids, nucleotides, and sugars) as candidates, among which nucleoside transporters were most suitable due to the similar structure of NAM to that of adenine and the similar structure of nicotinamide ribose to that of adenosine (Fig. 1a). Considering that ENT1 and ENT2 are widely expressed in multiple organs¹⁰, we deduced that ENT1 and ENT2 were probably the nicotinamide transporters. Therefore, we compared the cellular uptake of d₄-NAM in the absence or presence of ENT1/2 inhibitors in human cardiomyocyte AC16 cells and human mesenchymal stromal cells (MSCs). As shown in Fig. 1b, c, unlabeled NAM decreased cellular d₄-NAM uptake by 35% or 39%, indicating the key role of transporters in the transmembrane transport of NAM. S-(4-Nitrobenzyl)-6-thioinosine (NBTI), a typical inhibitor of ENTs, inhibited ENT1 only at low concentrations and inhibited both ENT1 and ENT2 at high concentrations²¹. We found that 1 μM NBTI decreased d₄-NAM uptake to 51%/68% of that in the vehicle control, while 100 μM NBTI reduced d₄-NAM accumulation to 16%/40% of that in the absence of inhibitor in AC16 cells or MSCs, suggesting that both ENT1 and ENT2 contributed to NAM transport. In addition, adenosine, a known ENT1/2 endogenous substrate²² also inhibited NAM uptake.

Fig. 1. ENT1/2 mediated the transport of NAM across the cell membrane.

a Structural similarity between nicotinamide and adenine, nicotinamide ribose and adenosine. b, c Accumulation of d₄-NAM in AC16 cells (b) and MSCs (c) with or without inhibitors; NBTI, S-(4-Nitrobenzyl)−6-thioinosine; ADE, adenosine; NAM, normal nicotinamide without isotope labeling (n = 3). d, e d₄-NAM uptake in AC16 cells (d) and MSC (e) following transfection with non-targeting siRNA (siNC), siRNA targeting SLC29A1(siENT1), SLC29A2(siENT2) or both SLC29A1 and SLC29A2(siENT1 + siENT2) (n = 3). f d₄-NAM uptake in CHO cells overexpressing SLC29A1(ENT1), SLC29A2(ENT2) or a vector control (vector) (n = 3). g, h Close-up view of the binding pocket residues in close proximity to docked NAM within hENT1 (g) and hENT2 (h). i, j d₄-NAM uptake by wild-type or point mutant of ENT1 (i) or ENT2 (j) point mutants (n = 3). n represents biologically independent replicates. Data are represented as mean ± standard error of mean (s.e.m.). P values were calculated using one-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test for (b, c) Holm–Sidak’s multiple comparisons test for (d, e, i, j) and two-way ANOVA followed by Holm–Sidak’s multiple comparisons test for (f).

To further confirm that ENT1 and ENT2 drive NAM transmembrane transport, both AC16 cells and MSCs were transfected with specific siRNA targeting the SLC29A1 gene, the SLC29A2 gene or co-transfected with the two siRNAs, respectively. Negative control siRNA was added to make sure the total siRNA transfection amount is the same for each group, and d₄-NAM uptake was evaluated. When ENT1 or ENT2 was significantly knocked down (Supplementary Fig. 1a–f) separately, d₄-NAM uptake was found to decrease to approximately half the levels. While the uptake decreased to 24.2% or 30.7% in AC16 Cells or MSCs, respectively, with ENT1 and ENT2 double knockdown, lower than ENT1 or ENT2 knockdown individually (Fig. 1d, e). Therefore, ENT1 and ENT2 contributed to at least approximately 70% of the transport of NAM across cell membrane, which meant ENT1 and ENT2 could be the main NAM transporters in the two cell lines we used.

To compare the capacity of ENT1/2 in NAM uptake, we performed d₄-NAM uptake in CHO cells after ENT1 or ENT2 overexpression with the same vector and equal amounts of human SLC29A1 and SLC29A2 plasmid transfection (Supplementary Fig. 1g, h). Our results showed that ENT1 or ENT2 overexpression promoted cellular d₄-NAM uptake. Specifically, CHO cells with ENT1 or ENT2 overexpression had 4.31-fold or 2.32-fold greater uptake of NAM than those transfected with the empty vector plasmid, respectively, at 100 μM of d₄-NAM (close to the human plasma NAM concentration)^23–25 (Fig. 1f). The above evidence supported that ENT1 had a stronger NAM uptake capacity. Furthermore, considering the higher expression of ENT1 than ENT2 in almost all tissues (Supplementary Fig. 1i, j), ENT1 might play a more important role than ENT2 in NAM transport.

To obtain a deeper understanding of the molecular mechanism of transport, ligand docking was performed. The reported structure of human ENT1²⁶ and the structure of human ENT2 from AlphaFold 2²⁷ were used. The results clearly showed favorable interactions between NAM and residues N30, Q180, G181, F337, D341, and N407 of ENT1 (Fig. 1g) and residues N30, Q167, F337, D341, and N407 of ENT2 (Fig. 1h), which were the same docking sites for the known substrate adenosine (Supplementary Fig. 1k, l) and the inhibitor NBTI²⁶. Hence, the docking results helped to explain the inhibition of NAM by NBTI. As shown in Supplementary Data 1, ENT1 had a higher affinity than ENT2 according to ligand docking, which was also consistent with the uptake curve. To examine the functional importance of the residues comprising the substrate binding pocket, we measured the transport activities of hENT1 and hENT2 mutants by a cellular d₄-NAM uptake assay. Considering the 46% homology in the amino acid sequence of ENT1 and ENT2^28,29, along with their functional complementarity, we believed that the binding sites of NAM with these two transporters are more likely to be located on homologous sites. Therefore, based on ligand docking results, we selected N30, D341, and N407 sites for point mutations, with the D341 also being a binding site for classical inhibitors. The selected amino acid at the designate sites were mutated to alanine, and the results showed (Fig. 1i, j) that point mutations at these three sites significantly reduced the transport capacity of NAM, either by ENT1 or ENT2, when CHO cells were transfected with wild type or point mutation plasmid.

Based on the above results, we deduced that both ENT1 and ENT2 contributed to cellular NAM uptake, with ENT1 making a predominant contribution.

ENT1/2 set cellular NAM homeostasis

Because ENT1 and ENT2 mediate cellular NAM uptake, the role of ENT1/2 in NAM homeostasis was investigated. We compared the cellular composition between the negative control (NC; treated with nontargeting siRNA, siNC) and ENT1 or ENT2-knockdown groups of both AC16 cells and MSCs via nontargeted metabolomics, and the results were further verified by targeted metabolomics and LC-MS/MS (Fig. 2a). Both principal component analysis (PCA, Fig. 2b) and partial least squares-discriminant analysis (PLS-DA, Supplementary Fig. 2a, b) confirmed that the ENT1/2-knockdown group and NC group clearly separated from each other, indicating that ENT1 or ENT2 knockdown changed cellular metabolites. We performed function analysis to convert the MS peaks to pathways using the MetaboAnalyst 5.0 platform³⁰ before compound annotation³¹ to reduce interfering factors during data processing. Function analysis showed that the NAM metabolism pathway was significantly altered when ENT1 was knocked down in both AC16 cells and MSCs (Supplementary Fig. 2c–f), implying that ENT1 was involved in NAM homeostasis. Then, compound annotation (fold change, FC > 1.5 & P value < 0.05) according to the HMDB³² was performed for further analysis. As the lists of metabolites significantly changed (Supplementary Data 2 for AC16 cells, Supplementary Data 3 for MSCs), the levels of adenosine, a well-known substrate of ENT1 and ENT2, were reduced, implying that ENT1/2 was effectively knocked down; concomitantly, the levels of both NAM and NAD+ were expectedly decreased (Fig. 2c, d), proving the contribution of ENT1/2 to NAM homeostasis. Then, enrichment analysis showed changes in multiple pathways, including the nicotinate and NAM, pyridine, and amino acid metabolism pathways; among these, the nicotinate and NAM metabolism pathways exhibited enrichment (Fig. 2e, f), greatly overlapping with the findings in NAM deficiency-associated disorders³³. The above results implied that ENT1/2 knockdown was strongly correlated with the absence of NAM and NAD+.

Fig. 2. Influence of ENT1/2 on cellular NAM metabolism.

a Lysates from cells with ENT1/2-knockdown were prepared for metabolomics analysis and assessment of the NAM/NAD+ levels. b Principal component analysis (PCA) of metabolomics data reveals distinct profiles between the control and ENT1/2-knockdown groups in AC16 cells and MSCs (n = 4). c, d Volcano plots illustrate significantly altered metabolomic peaks in ENT1/2-knockdown AC16 cells (c) and MSCs (d) (n = 4). The x-axis represents the log2 fold change (log2FC), while the y-axis depicts the negative log10 P value (-log10Pval). Welch’s two-side t test was employed to determine P value. e, f Metabolic pathways analysis revealed significant alteration in both ENT1/2-knockdown AC16 cells (e) and MSCs (f) (n = 4). Dot color indicates the level of significance (corresponding to the y axis), while dot size reflects the impact on pathway (as aligned with the x axis). g Heatmap display the relative intensities of cellular NAM, NMN, NAD+, and NA in ENT1/2 knockdown AC16 cells and MSCs based on chromatography-mass spectrometry data normalized to reference substance (n = 4). h, i Concentrations of cellular NAM and NAD+ were quantified in AC16 cells (h) and MSCs (i) with individual or both ENT1/2 knockdown using LC-MS/MS (n = 3). j, k Relative transcript levels for NAMPT, NMNAT1, NMNAT3, SIRT1, SIRT3, PARP1, CD38, and CD157 were assessed in ENT1/2-knockdown AC16 cells (j) and MSC (k) (n = 4). l–o Western blot images of NAMPT, SIRT1 and CD38 in AC16 cells (l, n) and MSC (m, o), along with the quantification of these result (n = 4). n represents biologically independent replicates. Data are represented as mean ± s.e.m. P values were calculated using one-way analysis of variance (ANOVA) followed by Holm–Sidak’s multiple comparisons test for (h and i), and Dunnett’s multiple comparisons test for (j, k, n and o). Gene mRNA and protein expression levels were normalized to β-ACTIN.

NAD+ salvage synthesis and NAM core recycling form a closed loop, which is also at the core of NAM metabolism⁵. Therefore, we focused on three compounds in the core, including NAM, NMN, and NAD+. Targeted metabolomics showed that the relative intensities of NAM, NMN, NAD+, and even nicotinic acid were dramatically decreased with ENT1/2 knockdown in AC16 cells and MSCs (Fig. 2g), which indicated that ENT1/2 were positively correlated with the levels of cellular NAM and its main derivatives. Then, LC‒MS/MS proved that ENT1/2 knockdown reduced the concentration of not only of NAM but also its active form NAD+, with both cellular NAM and NAD+ levels in the ENT1/2 double knockdown group being lower than those in the individual ENT1 or ENT2 knockdown groups (Fig. 2h, i).

ENT1/2 knockdown blocked NAM metabolism, which was probably due to an insufficient supply of key precursor NAM rather than mutual conversion blockade under the regulation of related enzymes. To clarify this hypothesis, the mRNA expression of related genes and the protein levels of key genes in the NAD+ salvage pathway and NAM core recycling pathway were investigated upon ENT1/2 knockdown. We observed upregulation of NMNAT3 (mRNA), CD38 (mRNA and protein), and CD157 (mRNA) and downregulation of SIRT1 (protein), SIRT3, and PARP1 (mRNA) in AC16 cells with ENT1/2 knockdown (Fig. 2j, k). Similarly, NAMPT (mRNA and protein) and CD38 (protein) levels were increased, while SIRT1 (mRNA and protein) levels were reduced, in MSCs with ENT1/2 knockdown (Fig. 2l–o). In summary, ENT1/2 knockdown in cells might promote the decomposition of extracellular NAD+, synthesis of intracellular NAD+, and inhibition of NAM core recycling, which is regarded as negative feedback regulation under NAD+ deficiency^12,34.

Taken together, the findings indicated that ENT1/2 set cellular NAM homeostasis by maintaining NAM and NAD+ levels through NAM transport function.

ENT1/2 knockdown shifted the cellular fate toward an NAD+-deficient state

We speculated that ENT1/2 could influence cellular behavior and fate because of the contribution of ENT1/2 to NAM homeostasis and the vital roles of NAM in multiple cellular processes. First, the top 100 genes (Supplementary Data 4) correlated with SLC29A1 and SLC29A2 were mined with GEPIA 2³⁵, and the genes were subjected to enrichment analysis with Metascape³⁶. As expected, NAD+ metabolism was strongly correlated with SLC29A1 expression (Fig. 3a), and biological oxidation and energy metabolism and NAD+-regulated biological processes were also observed. Although SLC29A2-associated genes were not directly enriched in NAD+ metabolism, the top 3 enrichment pathways were all related to mitochondria (Fig. 3b), a deeply NAD+-influenced cell organelle³⁷. And some genes on these pathways showed a high correlation with SLC29A1 (Fig. 3c) and SLC29A2 (Fig. 3d), such as NMNAT3, SIRT3, SIRT5, SDHAF4, etc. Hence, both SLC29A1 and SLC29A2 showed obvious relevance to NAD+-related genes, which indicated that ENT1/2 could regulate some functions of NAD+ at the gene level.

Fig. 3. Effects of ENT1/2 knockdown on cellular transcriptome.

a–d Enrichment analysis of the top 100 genes coessential with SLC29A1 (a,c) and SLC29A2 (b,d), respectively, in GEPIA2 database. e–h PCA analysis and corresponding heatmaps of differentially expressed gene in AC16 cells (e, g) and MSCs (f, h), following ENT1/2-knockdown (n = 3). i, j Altered NAD+ related pathways in KEGG for AC16 cells (i) and MSCs (j) with ENT1/2-knockdown (n = 3). The significance of the enriched pathways was determined by right-tailed Fisher’s exact test followed by Benjamini–Hochberg multiple testing adjustment. k, l Venn diagram illustrating the overlap between gene set from NAMPT knockout (as reported in literature), SLC29A1 knockdown and SLC29A2 knockdown in MSCs. m, n Enrichment analyses identifying commonly upregulated (m) and downregulated (n) DEGs across NAMPT knockout (literature reported), SLC29A1 knockdown and SLC29A2 knockdown in MSCs. n represents biologically independent replicates.

Then, RNA sequencing (RNA-seq) of AC16 cells and MSCs with ENT1 or ENT2 knockdown was conducted. PCA (Fig. 3e, f) and PLS-DA (Supplementary Fig. 3a, b) confirmed that the negative control (NC) group and ENT1- or ENT2-knockdown group clustered separately with each other in AC16 cells or MSCs, which indicated that the transcript profile had changed. Heatmaps of differentially expressed genes (DEGs) revealed a dramatic distinction between the transcriptomes of the siNC group and ENT1 or ENT2 knockdown. A total of 770/874 genes were differentially expressed between the NC group and ENT1/2 knockdown of AC16 cells (Fig. 3g), and up to 7785/8121 genes were differentially expressed in MSCs (Fig. 3h), indicating alteration of the transcriptional signature. In AC16 cells, enriched metabolic pathways according to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database had the most DEGs with ENT1/2 knockdown (Supplementary Fig. 3c, d). In addition, many DEGs were enriched in environmental information processing, including calcium signaling³⁸, cAMP signaling³⁹, PI3K-Akt signaling⁴⁰, Ras signaling⁴¹, and MAPK signaling⁴² (Supplementary Fig. 3g, h), which have a certain relationship with NAD+. Some KEGG pathways related to NAD+ are shown in Fig. 3i. Consistent with the observation in AC16 cells, the metabolic pathway also had the most DEGs when ENT1/2 was knocked down in MSCs (supplementary Fig. 3e, f), and many environmental information processing pathways related to NAD+ were also observed (supplementary Fig. 3i, j). Interestingly, ENT1/2 knockdown changed several cellular process pathways in MSCs, such as the cell cycle, cell senescence, autophagy, ferroptosis, etc., which indicated that ENT1/2 could dictate the fate of MSCs. Similarly, some representative enriched pathways upon ENT1/2 knockdown in MSCs are shown in Fig. 3j. However, the top-ranked Gene Ontology category for ENT1/2-knockdown-related DEGs was the metabolic pathway, in which most genes were downregulated, regardless of whether AC16 cells or MSCs were used. This suggested that the metabolic state weakened with ENT1/2 knockdown, becoming similar to the state after mitochondrial NAD+ deficiency⁴³. Notably, we observed that cellular senescence and cell cycle pathways changed after ENT1/2 knockdown in MSCs, which indicated that ENT1/2 knockdown could promote cellular senescence when the NAD+ level was generally reduced⁴.

The above results revealed that the transcriptome with ENT1/2 knockdown could correspond to NAD+ deficiency. Therefore, we compared DEGs with ENT1/2 knockdown and DEGs with NAMPT knockout^44,45, a known NAD+-deficient state. There were shared DEGs among the ENT1-knockdown (SLC29A1-KD), ENT2-knockdown (SLC29A2-KD), and NAMPT-knockout (NAMPT-KO) groups in both AC16 cells (Supplementary Fig. 3k, l) and MSCs. When very large numbers of DEGs were shared among ENT1/2-knockdown in MSCs and NAMPT-knockout (Fig. 3k, l), the shared DEGs were analyzed for enrichment with Metascape³⁶. Shared upregulated DEGs were enriched in the regulation of defense response, P53 downstream pathway, positive regulation of programmed cell death, and MAPK signaling pathway (Fig. 3m), which were regarded as participating in negative feedback regulation due to NAD+ deficiency. The shared downregulated DEGs were almost all enriched in metabolic-related pathways (Fig. 3n), including nucleobase-containing small molecular metabolic process, generation of precursor metabolites and energy, mitochondrial organization, respiratory electron transport chain, and metabolism of vitamins and cofactors. Thus, ENT1/2 knockdown weakened cellular metabolism, and the fuel respiration reduction was most prominent.

Collectively, the findings indicated that ENT1/2 knockdown shifted the cellular transcriptional signature toward the NAD+-deficient state. Thus, ENT1/2 influenced cellular behavior and fate at the transcript level on the basis of NAM transport function.

ENT1/2 contributed to energy metabolic configuration

As RNA-seq showed, most DEGs after ENT1 or ENT2 knockdown were from metabolic pathway, and fuel respiration plays a central role in the overall metabolic network. Oxidative phosphorylation (OXPHOS) is one of the cell’s main pathways for energy generation, and the OXPHOS level depends on the NAD+ pool level. For example, NAD+ deficiency caused by NAMPT activity loss impairs mitochondrial respiration and reduces energy generation⁴⁴. Thus, the contribution of ENT1/2 to fuel respiration was investigated. Similar to most cells, AC16 cells obtain energy from OXPHOS; whereas glycolysis is the main energy source of MSCs, similar to the Warburg effect⁴⁶. Thus, AC16 cells and MSCs served as the subjects of our research due to their different energy acquisition ways. ATP generation is a direct reflection of energy metabolism; thus, the total cellular ATP level was evaluated in both AC16 cells and MSCs with ENT1/2 knockdown. Interestingly, AC16 cells had a lower total cellular ATP level with ENT1/2 knockdown (Fig. 4a), but no change was found in MSCs (Fig. 4b). Hence, ENT1/2 mainly contributed to OXPHOS in energy metabolism due to the metabolic character difference between the two kinds of cells. Then, we further investigated the cellular oxygen consumption rate (OCR) and found that ENT1/2 knockdown impaired basal and maximal respiration in both AC16 cells (Fig. 4c) and MSCs (Fig. 4d). The cells with ENT1/2 knockdown consumed less oxygen (Fig. 4e, f), indicating that ENT1/2 knockdown impaired mitochondrial respiration capacity to undermine OXPHOS in cells with different modes of energy metabolism. Gene set enrichment analysis (GSEA) also revealed that ENT1/2 knockdown was negatively related to the OXPHOS pathway in both AC16 cells (Fig. 4g) and MSCs (Fig. 4h). As a carrier of OXPHOS, the electron transport chain on mitochondria directly influences cellular respiration capacity, and we observed that the transcript levels of some electron transport chain genes changed in AC16 cells (Supplementary Fig. 4a) and MSCs (Supplementary Fig. 4b) with ENT1/2 knockdown. Thus, ENT1/2 contributed to the energy metabolic configuration by intervening in mitochondrial respiration and OXPHOS.

Fig. 4. Impact of ENT1/2 on metabolic energy configuration through modulation of oxidative phosphorylation.

a, b Total ATP levels in AC16 cells (a) and MSCs (b) with ENT1/2 knockdown for 24 h, normalized to cellular protein content (n = 3). c, d Oxygen consumption rates (OCR) in AC16 cells (c) and MSCs (d) with ENT1/2 knockdown (n = 8). Basal respiration was measured prior to any additional treatments, while ATP-linked respiration and maximal respiration were assessed after the addition of ATPase inhibitor oligomycin (1.5 μM) and uncoupling reagent FCCP (1 μM for AC16 cells, and 0.5 μM for MSCs), respectively. Subsequently, Rotenone (0.5 μM) and antimycin A (0.5 μM) were added to completely inhibit mitochondrial oxygen consumption for calculating non-mitochondrial respiration. e, f Oxygen consumption in AC16 cells (e) and MSCs (f) with ENT1/2 knockdown was calculated by integrating the area under OCR-time curve and subtracting the non-mitochondrial respiration (n = 8). g, h GSEA plots displaying enrichment profiles of OXPHOS genes in AC16 cells (g) and MSCs (h) with ENT1/2 knockdown (n = 3). i, n Total cellular ATP concentration (i, j) (n = 3), OCR (k, l) (n = 8) and oxygen consumption (m, n) (n = 8) in AC16 cells (i, k, m) and MSCs (j, l, n) with ENT1/2 overexpressed, and additional NAM (500 μM) for 24 h. o–t Total cellular ATP concentration (o, p) (n = 3), OCR (q, r) (n = 6) and oxygen consumption (s, t) (n = 6) in AC16 cells (o–s) and MSCs (p, r, t) following siRNA-mediated knockdown of ENT1 or ENT2, with additional NMN (250 μM) for 24 h. n represents biologically independent replicates. Data are represented as mean ± s.e.m. P values were calculated using one-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test for (a, b, e and f) and two-way ANOVA followed by Holm–Sidak’s multiple comparisons test for (i, j, m–p, s, and t).

To clarify whether the contribution of ENT1/2 to the energy metabolic configuration is associated with the NAM transport role, we compared the ATP level and OCR in AC16 cells and MSCs without or with ENT1/2 overexpression in the presence of additional NAM (500 μM). As expected, overexpression of either ENT1 or ENT2 and supplementation with NAM increased cellular ATP levels, and a synergistic effect on the ATP level increase was exhibited when overexpression and supplementation coexisted in both AC16 cells (Fig. 4i) and MSCs (Fig. 4j). Consistent with the ATP level, ENT1/2 overexpression improved cellular respiration capacity, which was further enhanced by additional NAM supplementation in both AC16 cells (Fig. 4k) and MSCs (Fig. 4l). Oxygen consumption also exhibited the same results (Fig. 4m, n), which indicated that ENT1/2 intervened in fuel respiration through NAM transport.

To demonstrate that the effect of ENT1 or ENT2 knockdown on ATP and OXPHOS is indeed a consequence of reduced NAD+ and not reduced purine nucleotides, the ATP level and OCR were investigated after NMN supplementation, which recovered the cellular NAD+ level decreased by ENT1 or ENT2 knockdown (Supplementary Fig. 4c, d). Both AC16 cells and MSCs treated with 250 μM NMN for 24 h after specific siRNA targeting SLC29A1 or SLC29A2 transfection. We found that NMN rescued the decreased ATP level in AC16 cells (Fig. 4o), and increased the ATP level in MSCs (Fig. 4p). Additionally, NMN also recovered the cellular respiration (including basal and maximal respiration) impaired by ENT1 or ENT2 knockdown in both AC16 cells (Fig. 4q) and MSCs (Fig. 4r), and the oxygen consumption was also recuperated by NMN (Fig. 4s, t).

In short, ENT1/2 contributed to the energy metabolic configuration by intervening with OXPHOS, in which NAM transport by ENT1/2 played a key role.

ENT1/2 deficiency impaired mitochondria

Mitochondria, which serve as cellular energy factories, play a core role in fuel respiration, particularly in OXPHOS⁴⁷. Hence, we investigated whether ENT1/2 induced an energy metabolic configuration by affecting mitochondria. First, the mitochondrial membrane potential, a sensitive biomarker reflecting mitochondrial status, was measured based on JC-1 staining⁴⁸. There were more JC-1 monomers (green) and fewer JC-1 aggregates (red) upon ENT1/2 knockdown in both AC16 cells (Fig. 5a) and MSCs (Fig. 5b), clearly indicating that the mitochondrial membrane potential was reduced. This results was consistent with the in situ quantitative results of JC-1 aggregates (Fig. 5c, d). Mitochondria are highly dynamic organelles, and mitochondrial dynamics (the balance of fusion and fission) enables mitochondria to form highly dynamic regulatory networks for energy production⁴⁹. Then, the transcript levels of key genes related to mitochondrial dynamics (DRP1/FIS1 for fission and OPA1/MFN1/MFN2 for fusion) were studied. ENT1 or ENT2 knockdown changed the transcript level of some fission/fusion genes in both AC16 cells (Fig. 5e) and MSCs (Fig. 5f), indicating that ENT1/2 knockdown disturbed mitochondrial dynamics homeostasis. Imbalanced mitochondrial dynamics are directly reflected in mitochondrial mass and morphology⁴⁹; thus, mitochondrial mass was determined by flow cytometry with a MitoTracker Deep Red label. No significant differences were found in NC or ENT1/2-knockdown AC16 cells (Fig. 5g, h). In contrast, MSCs (Fig. 5i, j) with ENT1/2 knockdown induced an increase in mitochondrial mass, which is a typical feature of mitochondrial damage^50,51. In addition, the mitochondrial morphology of MSCs with ENT1/2 knockdown changed; specifically, fragmentation and the formation of spherical mitochondria (from filaments to globules) were observed (Fig. 5x), which were associated with OXPHOS impairment, mtDNA depletion and ROS production^49,52,53. As shown in section 4, we demonstrated that ENT1/2 deficiency induced weaker fuel respiration by impairing OXPHOS. Some essential subunits of the electron transport chain are encoded by mitochondrial DNA (mtDNA). Therefore, mtDNA depletion directly damages the OXPHOS process, resulting in mitochondrial dysfunction^54,55. As represented by ND1, the copy number of mtDNA showed no changes in AC16 cells (Fig. 5k) but was significantly reduced in MSCs (Fig. 5l) with ENT1/2 knockdown. Subsequently, the transcript level of mtDNA showed no change in AC16 cells (Supplementary Fig. 5c) but dramatically decreased in MSCs with ENT1/2 knockdown (Fig. 5m). We found no apparent change in the expression of the OXPHOS complex in AC16 cells (Fig. 5n). However, the protein levels of Complex II(HDSB), III(UQCRC1), IV(MTCO2) and V(ATP5A1) were reduced in MSCs with ENT1 or ENT2 knockdown (Fig. 5o). The decreased mtDNA copy number, mitochondrial gene transcript level and OXPHOS complex protein level in MSCs suggested that mitochondrial biogenesis was impaired by ENT1/2 knockdown.

Fig. 5. Impact of ENT1/2 deficiency on mitochondria.

a–d Representative images and quantification of mitochondrial membrane potential (JC-1) of AC16 cells (a, c) and MSCs (b, d) following ENT1/2 knockdown, indicating mitochondrial membrane potential; aggregates (red) denote high potential, while monomers(green) indicated low potential. Scale bars represent 100 μm (n = 3). e, f mRNA expression level of mitochondrial dynamics-related genes (DRP1, FIS1 for fission; OPA1, MFN1/2 for fusion) in AC16 cells (e) and MSCs (f) with ENT1/2 knockdown (n = 4). g–j Quantification of mitochondrial content using MitoTracker Deep Red fluorescence in AC16 cell (g, h) and MSCs (i, j) with ENT1/2 knockdown, respectively, measured by flow cytometry Mean fluorescence intensities (MFI) is presented, (n = 3). k, l Mitochondrial DNA (mtDNA) copy numbers in AC16 cells (k) and MSCs (l) with ENT1/2 knockdown, assessed by mtDNA/nDNA ratios using mt-ND1 (mtDNA) and HK2 (nDNA) as reference markers (n = 4). m Heatmap visualization of mitochondrial genome transcript profiles in MSCs with ENT1/2 knockdown (n = 3). n, o Protein levels of oxidative phosphorylation complexes I-V(CI-CV) evaluated by western blot in AC16 cells (n) and MSCs (o) with ENT1 or ENT2 knockdown (n = 4). p–u Analysis of mitochondrial biogenesis-associated gene expression (PGC1α, TFAM, and NRF2) at transcript and protein level in AC16 cells (p–r) and MSCs (s–u) with ENT1/2 knockdown (n = 4). v, w Fluorescence micrographs displaying JC-1 staining in AC16 cells (v) and MSCs (w) following ENT1/2 knockdown with/without extra NMN (250 μM), indicating mitochondrial membrane potential; aggregates (red) denote high potential, while monomers(green) indicated low potential. Scale bars represent 100 μm, (n = 3). x High-resolution images showing mitochondrial morphology in MSCs stained with MitoTracker Deep Red, Scale bars measure 1 μm, (n = 3). n represents biologically independent replicates. Data are represented as mean ± s.e.m. P values were calculated using one-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test. Gene mRNA and protein expression levels were normalized to β-ACTIN.

Mitochondrial biogenesis, which increases mitochondrial content and promotes mtDNA replication, is important to maintain mitochondrial function⁵⁶. The peroxisome proliferator-activated receptor gamma coactivator 1 alpha (PGC1α)-mitochondrial transcription factor A (TFAM) axis plays a key regulatory role in mitochondrial biogenesis⁵⁷. PGC1α is at the core of the regulatory network and can be regulated in multiple ways⁵⁸; it is generally regulated by the NAD+-SIRT1 axis⁵⁹. In addition, NRF2 protects mitochondria by regulating PGC1α⁶⁰. Interestingly, ENT1/2 knockdown did not impair mitochondrial biogenesis, and the transcript (Fig. 5p) and protein (Fig. 5q, r) in AC16 cells, which indicated that the upregulation of NRF2 was probably a key mechanism. Next, ENT1/2 knockdown slightly decreased the transcript levels of PGC1α and TFAM in MSCs (Fig. 5s); thus, PGC1α transcript regulation played limited roles in the ENT1/2-knockdown induced impairment of mitochondrial biogenesis. However, ENT1/2 knockdown significantly decreased the protein levels of PGC1α and TFAM in MSCs (Fig. 5t, u). Since SIRT1 stabilizes PGC1α through deacetylation⁶¹ and since reductions in SIRT1 protein levels were regulated by ENT1/2 NAD+ (Supplementary Fig. 5d), the NAM transport function of ENT1/2 contributed to mitochondrial biogenesis regulation.

To demonstrate that the effect of ENT1/2 knockdown on mitochondrial damage is indeed a consequence of reduced NAD+ and not reduced purine nucleotides, the mitochondrial membrane potential and morphology were investigated after NMN supplementation (250 μM). And we found that NMN recovered the decreased mitochondrial membrane potential in AC16 cells (Fig. 5v) and MSCs (Fig. 5w). In addition, NMN supplementation rescued the mitochondrial filaments (Fig. 5x). Considering that mitochondrial morphology is a key indicator reflecting mitochondrial status, NMN supplementation recovered the mitochondrial damage induced by ENT1 or ENT2 knockdown in MSCs.

Collectively, the evidence indicated that ENT1/2 deficiency impaired mitochondria to varying degrees in AC16 cells and MSCs. MSCs showed more severe mitochondrial damage, including decreased membrane potential, imbalanced mitochondrial dynamics, increased mass, fragmented morphology, decreased OXPHOS complex abundance and weakened biogenesis. Benefiting from the protective effect of NRF2^62,63 (Supplementary Fig. 5e), AC16 cells suffered less mitochondrial damage, but ENT1/2 knockdown still induced decreased membrane potential and an imbalance in mitochondrial dynamics. Hence, ENT1/2 deficiency impaired mitochondria and thereby reduced OXPHOS and weakened fuel respiration.

ENT1/2 intervened in cellular senescence progress

NAM⁶⁴ and its derivatives (NR⁶⁵ and NMN⁶⁶), common NAD+ booster, are believed to have an antiaging effect by regulating some NAD+-dependent antiaging enzymes. Hence, the roles of the NAM transporters ENT1 and ENT2 in cellular senescence progression were investigated. MSCs are excellent models for studying cellular senescence because MSCs spontaneously undergo senescence during culture and passaging⁶⁷ (Supplementary Fig. 6a), while immortalized AC16 cells are not suitable for senescence research. Therefore, we studied the contribution of ENT1/2 to senescence only in MSCs. As already mentioned, the RNA-seq results showed that DEGs were significantly enriched in the cellular senescence pathway when ENT1 or ENT2 was knocked down, and GSEA revealed that ENT1 (Fig. 6a) or ENT2 (Fig. 6b) knockdown aggravated cellular senescence. SA-β-gal staining confirmed that ENT1/2 knockdown induced cellular senescence (Fig. 6c). The increased γH2AX level (Fig. 6d and Supplementary Fig. 6c, d) revealed that ENT1/2 deficiency also induced DNA damage. As shown in Fig. 6e, ENT1/2 knockdown changed the transcript levels of genes related to cellular senescence, which manifested as higher transcript levels of CDKN1A and CDKN2A, cellular senescence biomarkers⁶⁸, and a lower transcript level of SIRT1, an antiaging biomarker, in which the transcript levels of the CDKN1A were further verified by qPCR (Supplementary Fig. 6e). Furthermore, ENT1/2 knockdown increased the protein levels of cellular senescence biomarkers, such as P21 (encoded by CDKN1A), P16 (encoded by CDKN2A), and P53 (encoded by TP53) (Fig. 6f, g), and decreased SIRT1 protein levels (Fig. 2m, o). The above evidence proved that ENT1/2 played a crucial role in the cellular senescence progress.

Fig. 6. Effects of ENT1/2 modulation on cellular senescence in MSCs.

a, b GSEA plots demonstrated the enrichment of cellular senescence-related genes in MSCs with ENT1 (a) or ENT2 (b) knockdown (n = 3). c, d Micrographs depicting SA-β-gal staining (c) and γH2AX immunofluorescence (d) with ENT1/2 knockdown (n = 3). e, h Heatmaps illustrate DEGs linked to cellular senescence (e) and autophagy (h) in MSCs with ENT1/2 knockdown (n = 3). f, g Western blot images (f) and densitometric analysis (g) of senescence biomarker P16, P21, and P53 (n = 4). i, j Images (i) and quantification (j) of western blot for autophagy markers P62 and LC3B (n = 4). k, l Representative images of SA-β-gal staining (k) and γH2AX immunofluorescence (l) in MSCs overexpressing ENT1/2 with additional treatment of NAM (500 μM). m, n Micrographs showcasing SA-β-gal staining (m) and γH2AX immunofluorescence (n) with ENT1/2 knockdown and supplemented with NMN (250 μM) (n = 3). o Research summary. Scale bars of (c, k, m) represent 100 μm, (d, l, n) represent 50 μm. n represents biologically independent replicates. Data are represented as mean ± s.e.m. P values were calculated using one-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test for (g and j). Gene mRNA and protein expression levels were normalized to β-ACTIN.

Because mitochondrial damage contributes to cellular senescence progress⁶⁹, the induction of mitochondrial impairment by ENT1/2 deficiency (Fig. 5) might be an important mechanism to promote senescence. Generally, mitochondrial damage is accompanied by increased production of reactive oxygen species (ROS), which induce DNA damage and cellular senescence⁶⁹. Here, increased ROS levels were observed in MSCs with ENT1/2 knockdown (Supplementary Fig. 6f, g). As a normal pathway of attrition and turnover of intracellular organelles and other structures, autophagy protects damaged cells⁷⁰, whereas mitophagy selectively removes excess or damaged mitochondria⁷¹. Hence, reduced autophagy is also regarded as a key reason for and characteristic of cellular senescence. ENT1/2 knockdown inhibited autophagy (Fig. 6h) and, to a lesser extent, mitophagy (Supplementary Fig. 6h) in MSCs. The downregulation of LC3B and P62 protein expression convincingly proved that ENT1/2 knockdown interfered with autophagy (Fig. 6i, j).

To clarify the contribution of ENT1/2 to senescence prevention through the NAM transport role, we compared the senescence progress in MSCs without or with ENT1/2 overexpression in the presence of additional NAM (500 μM). ENT1/2 overexpression reduced the stained area of SA-β-gal (Fig. 6k) and decreased the expression of γH2AX (Fig. 6l), which indicated that a higher level of ENT1/2 shifted cells toward a younger state. In addition, extra NAM supplementation advanced the antisenescence role of ENT1/2, which revealed that the NAM transport function of ENT1/2 plays an important role in senescence prevention.

To demonstrate that the effect of ENT1/2 knockdown on cellular senescence is indeed a consequence of reduced NAD+ and not reduced purine nucleotides, SA-β-gal staining and γH2AX immunofluorescence were investigated after NMN supplementation (250 μM). As the Fig. 6m, n, NMN supplementation rescued the more SA-β-gal staining (Fig. 6m) and increased γH2AX level (Fig. 6n) induced by ENT1 or ENT2 knockdown.

In conclusion, ENT1/2 deficiency promoted cellular senescence progress, and the underlying mechanism included impairment of mitochondria, increases in ROS and reductions in autophagy. ENT1/2 participates in cellular senescence progression through NAM transport.

Discussion

Uncovering the mechanism of NAM transmembrane transport will be helpful for understanding the biological function and clinical utilization of NAM. Here, we primarily identified both ENT1 and ENT2 as NAM transporters and further elucidated the contributions of these transporters to the physiological effects of NAM. In addition to cell membrane transfer of NAM, ENT1/2 also contributed to maintaining NAM homeostasis and the cellular NAD+ pool level. Therefore, ENT1/2 can intervene in the function of NAM and NAD+ by influencing their intracellular levels. Our studies demonstrated that ENT1/2 knockdown weakened fuel respiration and promoted senescence. The underlying mechanism was that ENT1/2 deficiency induced a decrease in the NAD+ pool level, interfered with the SIRT1-PGC1α pathway, subsequently decreased mitochondrial biogenesis, increased ROS levels, decreased autophagy levels, and finally resulted in mitochondrial dysfunction. In addition, NMN supplementation, recovered the cellular NAD+ level, and subsequently reversed the cellular respiration, mitochondrial damage and cellular senescence (Fig. 6o). Although these results are consistent with nicotinamide being carried by SLC29A1 and SLC29A2, experiments with recombinant protein to exclude any confounding cell biology have not been performed.

The antiaging benefits of NAD+-boosting comprise neurodegeneration protection, increased neovascularization, insulin resistance prevention, muscle atrophy reduction, and enhanced liver/brain/pancreatic function⁶. In addition, NAD+-boosting is effective for the treatment of some human diseases, such as obesity, Alzheimer’s disease, and cerebral ischemia. Compared with other NAD+-boosting strategies, such as the use of CD38/PARP/SARM inhibitors and NAMPT activators, NAD+ precursor supplementation is safer and convenient⁷². However, the transmembrane efficiency of NAD+ precursors influence NAD+-boosting effects, in which transporters of these precursors play a key role. For example, SLC22A7⁹ and SLC12A8⁷³ transport nicotinic acid and NMN, respectively, while SLC25A51²⁰ and SLC25A47⁷⁴ can transport NAD+ into mitochondria to exert biological roles. Our finding that ENT1 and ENT2 are NAM transporters is the another piece of the puzzle that represents the metabolic network of NAM and NAD+. Choosing a more efficient NAD+-boosting strategy on the basis of NAD+ precursor transporter expression in different states enable better promotion of the health and lifespan-extending effects of NAM. For instance, due to low SLC12A8 expression but high SLC29A1 expression, NAM is more suitable than NMN in muscle and heart tissue¹⁶. In addition, considering that SLC12A8 expression does not decrease during aging¹³, NMN has a better antiaging effect than NAM. Hence, the findings of the present research will aid in the development of more effective NAD+-boosting strategies.

Considering the similar substrate spectra among nucleoside transporters, we explored whether other nucleoside transporters participate in the transport of NAM. Since ENT3 is localized mainly in intracellular membrane systems, such as mitochondria and lysosomes^75,76, it was excluded. Regarding ENT4, although the accumulation studies in ENT4(SLC29A4)-knockdown cells and ENT4-stably-transfected cells indicated that ENT4 could also mediate NAM transport (Supplementary Fig. 7a–j), ENT4 is expressed mainly in the brain and heart⁷⁷, thus, the contribution of ENT4 to NAM transport is limited. In addition to ENTs, concentrative nucleoside transporters (CNTs, encoded by SLC28A genes), which are Na⁺-dependent⁷⁸, also belong to the nucleoside transporter family. However, our results revealed that the cellular accumulation of NAM did not show a difference in Na⁺ or Na⁺-free medium (Supplementary Fig. 7k). Therefore, CNTs do not mediate NAM uptake.

Adenosine, a typical substrate of ENT1/2, acts as a basic unit for nucleic acids and a component of the biological energy currency ATP. Adenosine interacts with cells by activating adenosine receptors (a class of G-protein coupled receptors) or interfering with the homeostasis of the intracellular nucleotide pool, in which the former pathway plays a major role. This implies that the extracellular adenosine concentration has a considerable influence on the roles of adenosine⁷⁹. Extracellular adenosine is derived mainly from ATP hydrolysis under the action of ectonucleoside triphosphate diphosphohydrolase 1 and ecto-5′-nucleotidase, and ENT1/2 induces adenosine influx to regulate adenosine levels⁷⁹. Indeed, adenosine can partially promote ATP generation⁸⁰ and plays a senescence-inhibiting role via adenosine A(2A) receptor activation⁸¹, indicating that lower ENT1/2 expression would maintain higher extracellular adenosine levels to promote fuel respiration and prevent senescence. The above speculation has been further confirmed by NMN rescue experiments. Hence, intervention in the homeostasis of NAM rather than adenosine is the underlying mechanism by which ENT1/2 contributes to fuel respiration regulation and cellular senescence prevention.

In conclusion, we have identified ENT1 and ENT2 as NAM cellular membrane uptake transporters. We have also elucidated the contributions of these transporters to fuel respiration and senescence, which are well-known NAM biological effects. Our study will aid in the development of more efficient NAD+-boosting strategies and improve the antiaging, metabolism-regulating and organ-protecting/organ-repairing effects of NAM.

Methods

Cell culture

AC16 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (VivaCell, C3113-0500) supplemented with 10% fetal bovine serum (Gibco, 10099-141C) and 1% penicillin-streptomycin (Biosharp, BL505A). CHO cells were cultured in DMEM/F-12 (VivaCell, C3130-0500) containing 10% fetal bovine serum and 1% penicillin-streptomycin. MSCs were cultured in MEM Alpha (Gibco, C12571500BT) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin.

RNAi/plasmid transfection

siRNAs targeting human SLC29A1, SLC29A2, and non-targeting control (siNC) were obtained from Sangon Biotech (Shanghai, China) and resuspended at 20 μM in DEPC buffer. The sense and antisense sequences of siRNA oligonucleotides are provided in Supplementary Data 5. The pcDNA3.1-hSLC29A1 expression plasmid was constructed in our laboratory⁸², while the pcDNA3.1-hSLC29A2 expression plasmid, and the point mutant of hSLC29A1 or hSLC29A2 plasmid were purchased from ViGene Biosciences Inc. (Jinan, China). JetPRIME versatile DNA/siRNA transfection reagent (Polyplus) was used to transfect the siRNA/plasmid into cells. Cells were grown in complete media and incubated for 24 h after treatment, with the medium being replaced 4 h post-transfection if plasmids were transfected. For double knockdown experiments, both AC16 cells and MSCs were transfected with specific siRNAs targeting the SLC29A1 gene, the SLC29A2 gene, or co-transfected with both siRNAs. Negative control siRNA was added to ensure the total siRNA transfection amount was consistent across all groups.

Analysis of gene expression

RNA was isolated using the AxyPrep Multisource Total RNA Miniprep Kit (Axygen, AP-MN-MS-RNA-250) and cDNA was synthesized with HiScript II Q RT SuperMix (Vazyme, R222-01). As described previously, quantitative PCR (qPCR) analysis was performed on a QuantStudio3 system (ThermoFisher) with TB Green Premix Ex Taq (Takara, RR420A, using the following condition: 95 °C/30 s followed by 40 cycles at 95 °C/5 s and 60 °C/30 s)⁸³. Human β-ACTIN was used as a housekeeping control, and the 2^−ΔΔCt method was applied to calculate relative mRNA expression levels. The sequences of primer for qPCR were provided in Supplementary Data 6.

Lysis and western blot

Cells were washed with ice-cold PBS and lysed directly in RIPA lysis buffer (Beyotime, P0013B) containing PMSF (Beyotime, ST506). Proteins were mixed with loading buffer (Sangon, C508320) and then boiled at 100 °C for 5 min. Protein samples were electrophoresed on Mini-PROTEAN Tetra System (Bio-Rad) and transferred to 0.45 μm PVDF membranes (Millipore, IPVH00010). The membranes were blocked with 5% BSA in Tris Buffered Saline (pH 7.6) containing 0.1% (v/v) Tween-20 (TBST), and incubated with primary antibodies overnight. Information regarding the primary antibodies is provided in Supplementary Data 7. Membranes were then washed with TBST three times (5 min each) and incubated with HRP-conjugated secondary antibodies at room temperature for 2 h. After being washed with TBST another three times (5 min each), signals were visualized by chemiluminescent detection using an enhanced chemiluminescence western blot detection system (Bio-Rad).

Cellular d₄-NAM accumulation

The cellular accumulation of d₄-NAM was evaluated following a previously reported method⁸⁴. Cells or siRNA/plasmid-transfected cells were pre-incubated with Krebs-Ringer-Henseleit buffer (KRH) at 37 °C for 5 min. 200 μL of KRH containing d₄-NAM, with or without inhibitors was added to initiate cellular uptake. The uptake was carried out at 37 °C for 5 min and terminated by rapidly removing the incubation buffer rapidly. Then, the cells were quickly rinsed three times with ice-cold KRH and lysed with 100 μL of 0.01% sodium dodecyl sulfate. After centrifugation at 13,000 × g for 15 min at 4 °C, 80 μL of supernatant was mixed with 160 μL acetonitrile containing the internal standard (50 nM Valacyclovir) for 5 min, then the mixture was centrifuged at 13,000 × g for 15 min, and the supernatant was analyzed by LC–MS/MS (Agilent, 1290/6460 systems). The chromatographic separation was performed on an Agilent ZORBAX SB-Aq (2.1 × 100 mm, 3.5 μm, 861753-914) at 30 °C with gradient elution (0–2 min, 20 % B; 2–2.1 min, 20–90% B; 2.1–3 min, 90% B; 3–4 min, 90–20% B; 4–6 min, 20 % B) at 0.2 mL/min, where mobile phase A and B were 0.1 % (v/v) formic acid in water containing 10 mM ammonium formate and 0.1 % (v/v) formic acid in acetonitrile, respectively. Mass spectrometric analysis was performed using an electrospray ionization source in positive ion mode. Quantification was achieved through multiple reaction monitoring mode at m/z transitions of 127.1→84.1 for d₄-NAM and 325.1→152.1 for valacyclovir. The fragmentor voltage was set to 170 V for d₄-NAM and 120 V for valacyclovir, with collision energies of 24 eV and 12 eV, respectively. The method was validated according to Food and Drug Administration guidelines and satisfied specificity, precision, recovery, matrix effect and accuracy were demonstrated (Supplementary Fig. 8).

Ligand docking

All ligand docking was performed using AutoDock Vina. The crystal structure of human ENT1, with its binding sites for the substrate adenosine and the inhibitor NBTI, was utilized. Additionally, the predicted structure of ENT2 from AlphaFold was employed for ligand docking due to the absence of its crystal structure. The homology between the gene sequences of SLC29A1 and SLC29A2 suggests that the binding site of SLC29A1 can serve as a reference for SLC29A2. NAM was docked into adenosine binding site following the established protocol⁸⁵.

Sample preparation and data analysis for metabolomics study

AC16 cells or MSCs were transfected with siNC, siENT1, or siENT2 for 24 h, then cells were collected by trypsin, and washed with ice-cold PBS. Metabolites were extracted from these cells with 90:10 acetonitrile: water, and extracts were analyzed by Ultimate 3000/Q Exactive (Thermo) UHPLC/Quadrupole-Electric Field Orbitrap High-Resolution Mass Spectrometry System. The LC separation was performed on an Acquilty HSS T3 column (2.1 mm × 100 mm, 1.8 μm particle size, Waters, 186003539). Solvent A was 0.1% formic acid aqueous, and solvent B was acetonitrile. The gradient was set as follows: 0 min, 2% B; 2 min 2% B; 21 min, 100% B; 25 min, 2% B. Besides, the flow rate was 0.3 mL/min, column temperature was 40 °C, and injection volume was 5 μL. The mass spectrometer was operated in positive ion mode for the detection of metabolites. The mass spectrometer operated in positive ion mode for metabolite detection, with the following main parameters: a scan range of m/z 70–1000, a spray voltage of 3.5 kV, and a capillary temperature of 350 °C. Product ion data were acquired in MS/MS mode at collision energies of 10, 20, and 40 eV. Raw LC-MS data were converted to mzXML format using the XCMS Plus platform. Functional analysis was performed using MetaboAnalyst 5.0 before metabolite identification to provide pathway information with minimal human interference. PCA maps and volcano plots were generated using OmicStudio⁸⁶. Pathway and enrichment analyses were also conducted using MetaboAnalyst with default settings for metabolites that showed significant changes in cells. Comparative analysis of standard compounds and their secondary mass spectrometry peaks for compound identification was shown in Supplementary Fig. 9.

Cellular NAM and NAD+ measurement by LC-MS/MS

Cells were washed twice with ice-cold PBS and immediately quenched by adding 500 μL of 50% methanol–50% water (containing d₄-NAM as an internal standard). The cell lysate was then harvested using a cell scraper (Corning, 3010), followed by the addition of 500 μL of chloroform. The mixture was vortexed for 10 s, and then centrifuged at 13,000 × g for 10 min at 4 °C, then the aqueous phase (upper phase) was collected into a new tube, and the process was repeated. Next, the aqueous phase was dried by centrifugal concentrator (LABCONCO, CentriVap), and reconstituted in water. And NAM and NAD+ were quantified on an LC-MS/MS system (Agilent, 1290/6460 systems), and the chromatographic separation was performed on an Acquity HSS T3 column (2.1 mm×50 mm, 1.8 μm particle size, Waters, 0225392111) at 30 °C with gradient elution (0–1.5 min, 5%–10% B; 1.5–4 min, 10%–90%; 4–5 min, 90% B; 5–5.1 min, 90% B-5 % B; 5.1–7 min 5% B) at a flow rate of 0.2 mL/min, where mobile phase A and B were water containing 5 mM ammonium formate and methanol, respectively. Quantification was obtained using multiple reaction monitoring mode at m/z transitions of 664.1→136.0 for NAD+, 123.1→80.1 for NAM and 127.1→84.1 for d₄-NAM. Fragmentor voltage was set at 140 V, 120 V, and 120 V, and the corresponding collision energy was 22, 20, and 12 eV for NAD+, NAM, and d₄-NAM, respectively. The method was validated according to Food and Drug Administration guidelines and satisfied specificity, precision, recovery, matrix effect and accuracy were demonstrated (Supplementary Fig. 10).

Participatory transcriptomics

The RNA libraries were sequenced on the Illumina Novaseq 6000 platform by LC Bio Technology CO., Ltd (Hangzhou, China). Total RNA was isolated from the AC16 cells or MSCs transfected with siNC, siENT1, or siENT2 for 24 h using TRIzol (Thermo, Cat. No. 15596018). Sequencing libraries were constructed following established protocols after sample quality checks. High-throughput sequencing was performed using the Illumina paired-end RNA-seq approach, generating millions of 2 × 150 bp paired-end reads. Low-quality reads were filtered using Cutadapt 1.9, followed by alignment to the human reference genome with HISAT2. Gene mapping was performed using StringTie 2.1.6. Differential gene expression analysis was conducted using OmicStudio tools. Heatmaps were generated with TBtools⁸⁷.

ATP and oxygen consumption rate determination

Cellular ATP concentration was quantified using an ATP assay kit (Beyotime, Cat. No. S0026) following the manufacturer’s instructions, utilizing chemiluminescence and measured with a microplate reader (BioTek, Synergy H1).

AC16 cells or MSCs were transfected with siNC, siENT1 or siENT2 for 12 h, then cells (20,000 cells per well) were re-plated into a well of a Seahorse XF 96-well culture plate (Agilent, 101085-004) and cultured for another 12 h, then the cell media was changed to XF DMEM Medium (Agilent, Cat. No. 103575-100) containing 10 mM glucose, 1 mM sodium pyruvate and 2 mM l-glutamine and incubated for 1 h at 37 °C without CO₂. Cell respiration was performed on a Seahorse XF96 Analyzer and with a Seahorse XF Cell Mito Stress Test Kit (Agilent, 103015-100). Cells were treated with 1.5 μM oligomycin, 1 μM FCCP, and 0.5 μM rotenone/antimycin A. Basal respiration was defined as the respiration before drug addiction, and maximal respiration was defined as the peak respiration following oligomycin and FCCP treatment. OCR was normalized to cell protein concentration.

NMN supplementation

The cells were incubated in a medium containing siRNA and a transfection reagent for 12 h, then replaced to medium with 250 μM NMN for an additional 24 h.

mtDNA copy number

Total cell DNA was isolated using the TIANamp Genomic DNA Kit (TIANGEN, DP304-03) and subsequently analyzed by qPCR (Applied Biosystems, 7500 Fast). The mtDNA/nDNA ratios were calculated as the relative ratios of mt-ND1 (mtDNA) to HK2 (nDNA).

Mitochondria membrane potential

Cells were loaded with JC-1 (Beyotime, C2006) to measure mitochondrial membrane potential according to the manufacturer’s protocol. To minimize the impact of additional handling on the mitochondrial membrane potential, the fluorescence intensity was detected in situ immediately after probe loading and fluorescence images were captured using an inverted fluorescence microscope (Nikon,Eclipse Ti-S). In this assay, JC-1 monomers emit green fluorescence, while JC-1 aggregates emit red fluorescence; a shift from red to green fluorescence indicates a decrease in mitochondrial membrane potential. For quantitative analysis, fluorescence intensity data was collected using excitation 525 nm, and emission 590 nm for JC-1 aggregates by Microplate reader (BioTek, Synergy H1).

Mitochondrial mass and morphology assay

Cells were loaded with MitoTracker Deep Red (ThermoFisher, M22426) to measure mitochondrial mass. After loading, the cells were collected by trypsinization, resuspended in complete media, and washed twice with PBS, filtered, and subjected to flow cytometry analysis. Flow cytometry data were recorded on Novocyte (Agilent, Novocyte TM) for 10,000 events. To measure mitochondrial content in intact cells, the MitoTracker Deep Red fluorescence intensity data were collected using excitation 640 nm, and emission 660 nm.

MSCs were seeded onto cell climbing slices for subsequent procedures. Cells were loaded with 100 nM MitoTracker Deep Red for 15 min in culture without FBS, and washed three times with PBS. MSCs were fixed with 4% formaldehyde for 15 min at room temperature, and permeabilization by 0.1% Triton X-100 in PBS for another 30 min. The nuclei were stained with 1 μg/mL DAPI for 15 min. After the anti-fluorescence quenching agent was added dropwise to the cell climbing slide, it was sealed with nail polish and a coverslip, and stored at 4 °C. Images were captured using confocal fluorescence microscope (Olympus, FV3000).

SA-β-gal staining

Cellular senescence was assessed using the Senescence-Associated β-Galactosidase (SA-β-Gal) Stain Kit (Solarbio, G1580) following the protocol provided by the manufacturer. Images were captured using inverted microscope (Nikon, Eclipse Ti-S).

Immunofluorescence staining

MSCs were seeded in cell culture slides for subsequent procedures. The cells were fixed with 4% formaldehyde for 15 min at room temperature, and permeabilized with 0.1% Triton X-100 in PBS for another 30 min. Then 5% FBS in PBS was used for the blocking step. The cells were washed three times with PBS, each wash lasting 3 minutes. After that, cells were incubated with the Phospho-Histone H2AX-S139 primary antibody (1:250) (AP0099, ABclonal) overnight at 4 °C and subsequently stained with the ABflo-594-conjugated secondary antibody for 1 h at room temperature. DAPI (1 μg/mL) was used to stain the nuclei for 15 min. After applying an anti-fade reagent, the slides were sealed with nail polish and cover glass, then stored at 4 °C. Images were captured using fluorescence microscope (Olympus, BX63).

Cellular ROS level determination

To minimize the impact of the experimental procedures, ROS were measured in situ. The cellular ROS levels were assessed using a ROS assay kit (Beyotime, S0033S) according to the manufacturer’s instructions. Cells were seeded in a 24-well plate for fluorescence imaging and in a 96-well plate for fluorescence intensity measurement. Images were captured by fluorescence microscope (Olympus, BX63). And fluorescence intensity data were collected using excitation 488 nm, and emission 525 nm for ROS by Microplate reader (BioTek, Synergy H1).

Statistical analysis

All studies were designed to generate groups of equal size, with experiments randomized and no data excluded. Outliers were included in data analysis and presentation. Group size refers to the number of independent assays with three or more replicates. Data obtained from qPCR and western blot analyses were normalized to the mean of control groups. Data were represented as the mean ± s.e.m. Differences between two groups were assessed using an unpaired two-tailed Student’s t-test for parametric data with a Gaussian distribution. For multiple-group comparisons, P values were calculated using one-way or two-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test or Holm–Sidak’s multiple comparisons test. Similar results were obtained from at least three independent repeat.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.Chen., L.Yuan., H.Zhou., and H.Jiang., conceived and designed the overall study. M.Chen., L.Yuan., B.Chen., H.Chang., H.Zhou., and H.Jiang. contributed to the development of the hypothesis and experimental approaches. M.Chen., L.Yuan., B.Chen., H.Zhang., Z.Chen., J.Kong., Y.Yi., M.Bai., M.Dong. performed and analyzed experiments. All authors contributed to the interpretation of the experiment. The manuscript was written by M.Chen., L.Yuan., J.Luo., and edited by H.Jiang., and H.Zhou. All authors reviewed the manuscript.

Peer review

Peer review information

Nature Communications thanks Eduardo Chini and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support this study are available in the Supplementary information file. Uncropped western blot images are also available in the Supplementary information. The RNA-sequencing raw data have been uploaded to NCBIs’ Gene Expression Omnibus (GEO) database and can be downloaded with GEO accession number GSE251776. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-56402-y.