Hindawi Oxidative Medicine and Cellular Longevity Volume 2021, Article ID 5534241, 15 pages https://doi.org/10.1155/2021/5534241

Research Article Circulating Metabolomic Analysis following Cecal Ligation and Puncture in Young and Aged Mice Reveals Age-Associated Temporal Shifts in Nicotinamide and /Histamine Metabolic Pathways

Anthony Cyr ,1 Lauryn Kohut ,1,2 Lauran Chambers ,1 Sladjana Stratimirovic ,1,2 and Brian Zuckerbraun 1,2

1Department of Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA 2Surgical Service Line, VA Pittsburgh Healthcare System, Pittsburgh, PA, USA

Correspondence should be addressed to Brian Zuckerbraun; [email protected]

Received 23 February 2021; Accepted 13 August 2021; Published 3 September 2021

Academic Editor: Cristina Angeloni

Copyright © 2021 Anthony Cyr et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Introduction. Aged individuals are at higher risk for morbidity and mortality following acute stressors than similarly stressed young people. Evaluation of age-associated metabolic changes could lead to the identification of specific therapeutic targets to improve outcomes from acute stressors, such as infections, in the elderly. We thus compared the plasma metabolomes of both young and old mice following cecal ligation and puncture (CLP), an accepted model of acute infection and stress. Methods. Young (9-17 wks) and aged (78-96 wks) male C57bl/6 mice were subjected to a retro-orbital bleed and two-week recovery prior to sham surgery (laparotomy alone) or CLP. Animals were sacrificed at 4 h, 8 h, or 12 h following intervention, and plasma was isolated from blood for subsequent analysis. Metabolomic analysis of samples were performed (Metabolon; Durham, NC). Results. Aged animals demonstrated greater intraprocedural mortality than young (30.2% vs. 17.4%, χ2p =0:0004), confirming enhanced frailty. Principal component analysis and partial-least squares discriminant analysis of 566 metabolites demonstrated distinct metabolomic shifts following sham surgery or CLP in both young and aged animals. Identification of metabolites of interest using a consensus statistical approach revealed that both the histidine/histamine pathway and the nicotinamide pathway have significant age-associated alterations following CLP. Conclusions. The application of untargeted plasma metabolomics identified key pathways underpinning metabolomic responses to CLP in both young and aged animals. Ultimately, these data provide a robust foundation for future mechanistic studies that may assist in improving outcomes in frail patients in response to acute stressors such as infection, trauma, or surgery.

1. Introduction for diseases that are specifically associated with aging and for the altered physiologic responses to disease processes. The medical advances of the past century, including the Trauma, sepsis, or acute critical illness pose a particular identification of antibiotics and the discovery and imple- diagnostic and therapeutic challenge, as these are associated mentation of vaccine programs, have led to a significant with increased morbidity and mortality among the elderly increase in life expectancy worldwide. In turn, the number [2–4]. Much of our understanding of these poor outcomes of individuals aged >65 years is expected to surpass 1.5 bil- has come from statistical analysis of frailty scores in the lion by 2050 [1]. This vast segment of the population elderly patients presenting with emergency conditions, and requires special attention from the medical community, both historical studies have been limited in scope [2–4]. More 2 Oxidative Medicine and Cellular Longevity recently, a large-scale study among patients at the Veterans 2. Materials and Methods Affairs Hospital Network affirms that mortality is broadly increased across all surgical risk strata among the most frail 2.1. Animal Studies. Animal protocols were approved by the patients, even in elective and optimized situations [5]. Col- University of Pittsburgh Institutional Animal Care and Use lectively, these observations underscore the critical burden Committee. The experiments were performed in adherence that an aging and increasingly frail populace pose to socie- to the US National Institutes of Health guidelines on the ty—both for the patients themselves who experience higher use of laboratory animals. C57BL/6 mice ages 11-18 weeks risks and for the healthcare systems that provide complex (young) or 86-96 weeks (aged) were utilized for the experi- care. mental protocol. For retro-orbital blood sample collection, Aging itself is a progressive decline in functional capacity intraperitoneal injection of a ketamine-xylosine mix was resulting from a combination of degenerative processes in used for procedural sedation, and blood samples were har- cellular metabolism, with multiple consensus hallmark fea- vested using heparinized capillary tubing. Animals were tures [6–8]. DNA repair, epigenetic signaling, and protein observed during immediate postprocedural recovery and then homeostatic mechanisms become broadly deficient with allowed to recover fully for a period of two weeks prior to ran- fi aging across the phylogenetic landscape, leading to broad domization to either immediate sacri ce, sham surgical inter- alterations in mitochondrial function and cellular bioener- vention consisting of a laparotomy and closure in layers, or getics. Modern high-throughput assays and systems biology cecal ligation and puncture done as previously described tools have enabled the use of broad omics-based approaches [17]. Antibiotics were not administered. At 0 h, 4 h, 8 h, and fi fl to provide insight into these aging pathways [9]. In particu- 12 h postprocedurally, animals were sacri ced by iso urane lar, metabolomic studies, which involve the identification of asphyxiation and cervical dislocation, and blood was immedi- small (MW < 1500 Da) metabolites from isolated samples, ately collected via cardiac puncture for analysis. These time- have proven to have significant value [10]. As these small points were chosen based on data from biometric molecules represent the end result of layered epigenomic, monitoring with implantable sensors following CLP, which transcriptomic, and proteomic regulation, metabolite pro- underscored that initial physiologic changes take place within files provide an integrative view of complex biological sys- this time period following the initial CLP insult [17]. tems. In humans, studies utilizing metabolomic methods to 2.2. Sample Preparation. Following collection, all blood sam- help delineate age-associated patterns and patient subgroups ples were stored briefly on ice in heparinized 1.5 mL micro- in sepsis [11], pneumonia [12], and trauma [13] have proven centrifuge tubes. Samples were spun at 500 g for 15 minutes to be of significant value in highlighting potentially novel to isolate the plasma fraction, which was distributed into avenues for both basic research and potential intervention. multiple aliquots of 150-200 μL prior to flash-freezing in liq- However, human studies remain costly and time-consum- ° uid nitrogen followed by long-term storage at -80 C. ing, and murine models remain a mainstay for evaluation of central metabolic pathways following insult. Several 2.3. Measurement of Plasma AST/ALT. Circulating groups have investigated the metabolomic signature of aging AST/ALT was measured using the Liver Panel chemistry in mice [14–16]. Notable among these studies, Tomás-Loba fi set and associated Heska analysis platform (Heska, CO, et al. identi ed that metabolite signatures in aged mice cor- USA). Serial dilutions were performed of plasma samples related with biochemical aging as measured by telomere to ensure appropriate reads within the linear range of the shortening rather than purely chronological aging [16], detector, as specified by manufacturer’s instructions. demonstrating the value of metabolomic profiling in exam- ining the effects of aging on disease. 2.4. Untargeted Metabolomics. Plasma metabolomics was One of the key limitations of the murine analyses to date accomplished by Metabolon (Durham, NC, USA, http:// is the lack of evaluation of physiologic insult in young vs. www.metabolon.com) using a multiplatform system encom- aged animals. Based on prior murine metabolomic studies, passing tandem ultrahigh-performance liquid we hypothesized that aging would be associated with distinct chromatography-mass spectrometry as well as tandem gas metabolomic trajectories following physiologic stress. To chromatography-mass spectrometry systems with different address this, and to bridge the gap between murine model- column ionization parameters. Peak calling, compound ing and the human cohort data, we performed plasma meta- identification, and batch-effect normalization accounting bolomic profiling of both young and aged mice at various for day-to-day interrun variability for individual metabolites time points following a standard cecal ligation and puncture were accomplished according to Metabolon’s proprietary (CLP) or sham surgical intervention. These data provide a methodology and internal library of standards. A total of robust signature both of the baseline metabolomic differ- 834 metabolites were identifiable, of which 566 had specific ences between young and aged mice and several differences KEGG or HMDB identifier labels. in metabolite responses to surgical stressors and sepsis. In particular, we identified significant signature differences in 2.5. Metabolite Normalization and Statistical Evaluation. the nicotinamide metabolite axis as well as in histidine and Statistical analysis as described in the text was undertaken carnosine biology, which provide a clear path using MetaboAnalyst (http://www.metaboanalyst.ca), an forward for future investigation and potential intervention open-source suite of metabolomics analysis tools maintained to medically treat some of the biological sequelae of aging by the Xia Lab at McGill University [18, 19]. Raw peak area in the acute care setting. data was imported, normalized to appropriate pooled Oxidative Medicine and Cellular Longevity 3

Plasma Plasma Plasma isolation and isolation and isolationand storage storage storage

Baseline sacrifice 14-day 11-18 weeks recovery Retro-orbital Control surgery 0 h, 4 h, 8 h, 12 h bleed (Sham laparotomy) Sacrifice timepoints

Cecal ligation 0 h, 4 h, 8 h, 12 h and puncture 86-96 weeks Sacrifice timepoints (a) 150 Intraprocedural survival ALT AST 600 500 Group: p = 0.0033 Group: p = 0.0002 n = 109 p p Time: = 0.0082 400 Time: < 0.0001 100 Age × Time: Age × Time: ⁎ n = 86 33 (30.2%) 400 p = 0.1826 p < 0.0001 300 15 (17.4%) ⁎ ⁎

200 ALT (U/L) 50 AST (U/L)

Number of mice 200 71 (82.6%) 76 (69.7%) 100

0 0 0 Young Aged ROB 0 h 4 h 8 h 12 h ROB 0 h 4 h 8 h 12 h X2 p = 0.0004 Young (Sham) Aged (Sham) Young (Sham) Aged (Sham) Deceased Young (CLP) Aged (CLP) Young (CLP) Aged (CLP) ⁎ Alive p < 0.05, Young CLP vs. Aged CLP ⁎p < 0.05, Young CLP vs. Aged CLP (b) (c) (d)

Figure 1: Summary of the experimental protocol and establishment of aged mouse frailty. (a) Graphical representation of the experimental design, including the sampling strategies and the various experimental groups included for analysis. (b) Description of the intraprocedural survival, demonstrating statistically significantly enhanced mortality for the aged animals in the protocol compared to the young. aminotransferase (c) and aspartate aminotransferase (d) levels in the plasma of animals at various timepoints, demonstrating increased levels of liver damage in the aged animals throughout the protocol when compared to the young animals.

control data, mean-centered, and scaled to the standard 3. Results deviation of the metabolite across all samples prior to down- stream statistical analysis, which included PCA, partial least- 3.1. Aged Mice Are More Frail than Young. Male C57BL/6 squares discriminant analysis (PLS-DA), empiric Bayesian mice,agedeitherbetween11and18weeks(young)or86and analysis of microarray (EBAM), and random forest methods. 96 weeks (aged), were utilized for all studies according to the Metabolites were rank ordered by importance by taking the experimental protocol described graphically in Figure 1(a). To median rank order of the results of these individual methods populate all experimental groups, a total of n =86young and for downstream visualization purposes. The top 100 metab- n = 109 aged animals were required, with aged animals suffer- olites for individual comparisons as described in the text ing statistically significantly elevated intraexperimental mortal- were subjected to pathway analysis within MetaboAnalyst, ity, calculated as the number of animals who did not survive using the list of 566 input variables as a reference dataset to the predetermined experimental timepoint (30.2% vs. to evaluate for pathway enrichment and statistical signifi- 17.4%, χ2, p =0:0004, Figure 1(b)). As an additional adjunct cance. Prism 8.4.2 (GraphPad Software, LLC, San Diego, measureoffrailty,weexaminedforliverinjuryusingplasma CA, USA) was used to plot individual metabolite data. alanine aminotransferase (ALT, Figure 1(c)) and aspartate ami- Two-way ANOVA was used to compare time trajectories notransferase (ALT, Figure 1(d)) using aliquots of plasma at the of individual metabolites across different subgroups (young various isolated timepoints. For both ALT and AST, there were sham, young CLP, aged sham, and aged CLP). Post hoc mul- clear differences in ALT/AST trajectory over time, with aged tiple comparison testing was accomplished in Prism using animals demonstrating increased enzyme release in response Tukey’s correction, and we specifically reported the compar- to CLP compared to young animals. Taken together, these isons between aged CLP and young CLP and aged sham and results suggest that aged animals have increased frailty to CLP aged CLP. compared to young animals. 4 Oxidative Medicine and Cellular Longevity

PLS-DA scores plot Feature identification workflow 15 Normalized autoscaled 10 metabolite data

5

0

t-Test/ANOVA –5

Component 2 (4.8%) –10

–15 PCA PLS-DA EBAM Random loadings VIP scores output forest –20 –15 –10 –5 0510 Component 1 (10.2 %)

Young Aged Median rank order metabolites of importance

Consensus metabolites for pathway analysis

(a) (b) 566 metabolite profile Top 100 Young vs. Old

X2

p = 0.77

29.33% 166 Amino acid 27.00% 27 Amino acid 4.42% 25 Carbohydrate 2.00% 2 Carbohydrate 4.77% 27 Cofactors and vitamins 7.00% 7 Cofactors and vitamins 1.41% 8 Energy 1.00% 1 Energy 41.87% 237 Lipid 46.00% 46 Lipid 6.71% 38 Nucleotide 4.00% 4 Nucleotide 3.89% 22 Peptide 5.00% 5 Peptide 7.60% 43 Xenobiotics 8.00% 8 Xenobiotics (c)

Figure 2: Continued. Oxidative Medicine and Cellular Longevity 5

−6 Aged Young

4-Guanidinobutanoic acid 1-Methylidine Ansarine 3-Methylihistidine 1-Methylhistamine L-Hisidine Methylimidazoleacetic acid Pi-Methylimidazoleacetic acid Ethylmalenic acid Isobutyryl-L-carnitine Taurine (S)C(S)S-S-Methylcysteine sulfoxide Indolelactic acid −4 Indoxyl sulfate L- Amino acids Indole-3-propionic acid Argininic acid N-Acetylaspartylglutamic acid N-Carboxyethyl-L-aminobutyric acid 4-Hydroxy-L- 2-Hydroxy-3-glutamic acid 5-Hydroxylysine N(6)-Methyllysine Ne, Ne dimethyllysine N-Formayl-L- 4-Hydroxyproline Prolylhydroxyproline –2 Carbohydrates 1,5-Anhydrosorbitol L-Arabinose

1-Methylnicotinamide N1-Methyl-4-pyridone-3-carboxamide Quinolinic acid Panotothenic acid Cofactors Alpha-Tocopherol /Vitamins Vitamin A Gulonic acid Energy

Caramide (d18:1/18:0) N-Oleoylethanolamine Olaoylcarnitine − Butyrylcarnitine Malonic acid Sebacic acid Undecanedioic acid 3-Hydroxyoctanoic acid LysoPC(18:1(9Z)) LysoPC(18:0) PO(14;0/20:4(5Z,8Z,11Z,14Z)) PC16:0/18:0) PC(18:0/18:0) PC(18:0/18:1(9Z)) PC(18:1(9Z)/22:6(4Z,7Z,13Z,16Z,19Z)) Metabolome pathway map PC(18:2(9Z, 12Z)/20:4(5Z,8Z,11Z,14Z) P1[18:1(9Z)/20:4(5Z,8Z,11Z,14Z)) Trimethylamine N-oxide 0 PE(O-18:1(1Z)/20:4(5Z.8Z,11Z,14Z)) SM(d18:1/14:0) 2.5 Alpha-linolenic DG(16:0/18:2(9Z,12Z)/0:0) DG(18:2(9Z,12Z)/18:2(9Z,12Z)/0:0) acid metabolism DG(18:2(9Z,12Z)/18:3(6Z,9Z,12Z)/0:0) Histidine DG(18:2(9Z,12Z)/18:3(9Z,12Z,15Z)/0:0) DG(18:2(9Z,12Z)/120:4(5Z,8Z,11Z,14Z)/0:0) metabolism 9.10-DHOME Lipids 12,13-DHOME Retinol 5-Hydroxyhexanoic acid 2.0 Docosadienoate (22:2n6) Docosatrienocic acid Pentose and metabolism Gemma-Linolenic acid Stearidonic acid Nonadecanoic acid glucuronate 2 Arachidic acid MG(0:0/18:2(9Z.12Z)/0:0 interconversions MG(18:3(6Z.9Z12Z)/0:0/0:0) 1.5 PC(18:2(9Z,12Z)/1/:3(9Z,12Z,15Z)) PE(16:0/18:2(9Z:12Z) PE 16:0/22:6(4Z,7Z,10Z,13Z,16Z19Z) PE(18:0/18:2(9Z,12Z)) Nicotinamide

PE(18:0/22:6(4Z,7Z,10Z,13Z,16Z,19Z)) –log(p) PE(18:1(9Z)/18:2(9Z,12Z)) metabolism PE(18:2(9Z,12Z)/18:2(9Z,12Z)) PI(16:0/18:2(9Z,12Z)) 1.0 PE(P-16:0/18:2(9Z,12Z)) SM(d18:1/22:0) GPI anchor Orotidine 4 5-Methylcytidine Nucleotides 5-Methyldecoxycytidine biosynthesis ymidine 0.5 Isoleucyl-Hydroxyproline Leucyl-Hydroxyproline Gamma-Glutamylalanine Peptides Gamma-Glutamylglycine Gamma-Glutamylglycine

3-(3-Hydroxyphenyl)propanoic acid Desaminotyrosine 0.0 Erythritol Ergothioneine Xenobiotics Betonicine 0.0 0.1 0.2 0.3 0.4 betaine 6 Phenylpropionylglycine Cinnamoylglycine Pathway impact score (d) (e)

Figure 2: Comparison of baseline metabolomic status between young and aged animals. (a) PLS-DA plot of the metabolomic data from retro- orbital bleed samples from both aged (red) and young (green) animals, demonstrating distinct grouping and suggesting a clear difference at the metabolome level between young and aged animals at baseline. (b) Schematic representation of the statistical analysis algorithm used to identify the top 100 metabolites that best differentiate between young and aged animals at baseline. (c) Distribution of metabolite families among the original 566 metabolites in the dataset compared to the top 100 metabolites capable of differentiating young from aged. (d) Complete heatmap representing the top 100 metabolites, organized according to metabolite family. (e) Metabolome pathway map demonstrating specific metabolic pathway enrichment among the top 100 metabolites compared to the original 566 in the dataset.

3.2. Young and Aged Mice Have Distinct Metabolomes at Figure S1. Initial attention was focused on identifying the Baseline. Plasma samples isolated according to protocol were major differentiating metabolites distinguishing the young sent to Metabolon (Durham, NC, USA) for metabolite from aged animals at baseline using the retro-orbital detection and annotation (see the Materials and Methods plasma isolates (Figure 2). Partial least squares section for further details). A total of 834 discrete metabo- discriminant analysis (PLS-DA) was used to visually lites were identifiable across the dataset; 566 of which had identify broad metabolomic differences (Figure 2(a)), with identifiable Human Metabolome Database (HMDB) or clear delineation between aged and young animals. A Kyoto Encyclopedia of Genes and Genomes (KEGG, http:// consensus methodology was used to specifically stratify www.kegg.jp) accession numbers [20]. Raw metabolite peak metabolites of interest; thereafter, integrating the areas provided by Metabolon were analyzed using the open contributions of principal component analysis (PCA), PLS- source MetaboAnalyst package available at http://www DA, empiric Bayesian analysis of microarray (EBAM), and .metaboanalyst.ca [18]. A flowchart illustrating data man- random forest (RF) approaches to generate a consensus agement strategies and representative output is in rank ordering of metabolites (Figure 2(b)). This did not 6 Oxidative Medicine and Cellular Longevity

PCA scores plot: Young CLP PCA scores plot: Young Sham

4 h 0 h 4 h 20 Sham CLP CLP 20 0 h Sham

0 0 8 h 12 h Sham CLP Pre-bleed 12 h Pre-bleedleed PC2 (11.1%)

PC2 (11.1%) Sham –20 –20

8 h –40 –40 CCLP

–20 –10 01020 30 –20 –10 0 10 20 30 PC1 (20.7%) PC1 (20.7%) Pre-bleed 8 h Sham 0 h Sham 12 h Sham Pre-Bleed 8 h CLP 4 h Sham 0 h CLP 12 h CLP 4 h CLP (a) (b) Top 100 Young Metabolome pathway map CLP-sham 3.0 Steroid biosynthesis

2.5 Nicotinamide metabolism 2.0

Alpha-linolenic Linoleic acid 1.5 acid metabolism metabolism –log(p) X2 p = 0.0497, compared 1.0 Sulfur to complete metabolite profile metabolism

30.00% 30 Amino acid 0.5 9.00% 9 Carbohydrate 8.00% 8 Cofactors/vitamins 0.0 1.00% 1 Energy 38.00% 38 Lipid 0.0 0.2 0.4 0.6 0.8 1.0 10.00% 10 Nucleotide Pathway impact score 2.00% 2 Peptide 2.00% 2 Xenobiotics (c) (d)

Figure 3: Continued. Oxidative Medicine and Cellular Longevity 7

CLP Sham Pre-bleed 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h

N1-Acetylspermidine Imidazolelactic acid Methylimidazoleacetic acid N-Acetylhistidine Glutarylcarnitine 2-Hydroxy-3-methylbutyric acid 3-Methylcrotonylglycine Hydroxyisocaproic acid L-Cysteinylglycine disulfide Phenylacetic acid N-Acetyl-L- Amino acids Urea Imidazolepropionic acid (E)-Urocanic acid 3-Sulfinoalanine Phenyllactic acid N-Acetylputrescine Indolelactic acid 1-Pyrroline-5-carboxylic acid 4-Hydroxy-L-glutamic acid L-reonine Histamine 3-Amino-2-piperidone N-Acetylhistamine N-Acetyl-L-aspartic acid Indoxyl sulfate Indole-3-propionic acid yroxine L-Targinine

N-Acetylneuraminic acid N-Glycolylneuraminic acid Sucrose Carbohydratess N-Acetyl-D-glucosamine Raffinose Sorbitol Ribitol Sedoheptulose

Bilirubin N1-Methyl-2-pyridone-5-carboxamide Cofactors N1-Methyl-4-pyridone-3-carboxamide Nicotinate D-ribonucleoside 1-Methylnicotinamide /Vitamins Niacinamide Pantothenic acid 4-Amino-5-hydroxymethyl-2-methylpyrimidine Energy 2-Methylcitric acid Ceramide (d18:1/18:0) SM (d18:0/18:0) SM (d18:0/22:0) Arachidonyl carnitine DL-2-Aminooctanoic acid 3-Hydroxydodecanoic acid 3-Hydroxyoctanoic acid 3-Hydroxycapric acid (R)-2-Hydroxyhexadecanoic acid Myristoleic acid Nonadeca-10(Z)-enoic acid 7Z,10Z-Hexadecadienoic acid 5,8-Tetradecadienoic acid Linoleic acid Gamma-Linolenic acid Eicosadienoic acid Stearidonic acid Myristic acid Pentadecanoic acid 5-Dodecenoic acid Dodecanoic acid Undecylenic acid SM(d18:1/20:0) SM(d18:1/22:1(13Z)) Lipids SM(d18:1/24:1(15Z)) Cholesterol Beta-Sitosterol Campesterol Oleamide Palmitic amide Octadecanamide PC(14:0/20:4(5Z,8Z,11Z,14Z)) PC(16:0/18:0) PI(16:0/18:2(9Z,12Z)) PI(16:0/20:4(5Z,8Z,11Z,14Z)) Choline PE(P-16:0/18:1(9Z)) PE(P-16:0/18:2(9Z,12Z))

7-Methylguanine Beta-Alanine Pseudouridine Ribothymidine Nucleotides Uridine Allantoin Orotic acid ymine 5,6-Dihydrouridine N2,N2-Dimethylguanosine

Peptides Glycylleucine gamma-Glutamylserine Xenobiotics Sulfate Indole-3-methyl acetate

−2 −1 012 (e)

Figure 3: Identification of the metabolic pathways that drive the response to CLP in young animals. PCA plots of the young sham (a) and CLP (b) metabolomes. Sham-treated animals demonstrated a return to baseline over time, while CLP-treated animals did not. (c) Profile of the top 100 metabolites that differentiate between CLP and sham interventions in young mice over time. (d) Pathway analysis comparing enrichment of the top 100 metabolites compared to the 566 metabolites originally contained in the dataset. (e) Heatmap analysis demonstrating group average values of the top 100 metabolites at all CLP and sham surgical timepoints in young animals, illustrating broad differences in metabolomic responses following CLP compared to sham intervention. produce a significant enrichment among any of the major 3.3. CLP or Sham Surgery Induces Distinct Metabolomic categorical metabolite superfamilies (χ2, p =0:77, Differences in Young Animals. We next isolated the meta- Figure 2(c)). A detailed heatmap visualization of the top bolomic signatures associated with CLP or sham surgery 100 metabolites, broken down by metabolite superfamily, specifically in the young population. For sham, the meta- is seen in Figure 2(d). As expected based on selection bolomic response was to initially migrate from the retro- criteria, metabolites visualized in this fashion segregate orbital bleed baseline (4 h and 8 h), but by 12 h, there was a into those that are predominantly enhanced in aged or notable colocalization with the retro-orbital bleed metabo- young groups. To further break down key differences, we lome (Figure 3(a)). In contrast, CLP produced a persistent employed the pathway analysis algorithms available in shift away from the baseline, with 4 h, 8 h, and 12 h time- MetaboAnalyst (Figure 2(e)). Here, we demonstrate that points all remaining distinct from the retro-orbital bleed key differences between aged and young mice involve samples (Figure 3(b)). Using the same consensus approach lipid biology (GPI anchor biosynthesis, α-linolenic acid described above, we isolated the top 100 metabolites as fea- metabolism), carbohydrate metabolism (pentose and tures of interest. This generated a statistically significantly glucuronate interconversions), cofactor and vitamin biology different distribution of metabolites within superfamilies (nicotinamide and retinol metabolism), and amino acid compared to the original dataset (χ2, p =0:05, Figure 3(c)). regulatory pathways (histidine metabolism). Detailed Key metabolite subpathways identified with pathway analysis information from this analysis is compiled in Figure S2. tools included, as with the young vs. aged comparison, 8 Oxidative Medicine and Cellular Longevity nicotinamide metabolism and α-linolenic acid metabolism. 3.6. Histidine Metabolism Critically Distinguishes Aged from Pathways distinct to the young CLP-sham analysis included Young Metabolomic Responses to Surgical Stress. Using the steroid biosynthesis, linoleic acid metabolism, and sulfur raw data, the elements of the histidine metabolism axis over metabolism (Figure 3(d)). A detailed heatmap view demon- the time course of the experiment grouped by age and surgi- strates mild differences between CLP and sham among the cal intervention was graphed. A total of 11 metabolites from young animals over different timepoints, particularly within the annotated KEGG pathway (http://www.kegg.jp) were the lipid superfamily, whereby a significant induction of cer- represented in the dataset and are organized according tain lipid moieties dominates the 4 h-8 h sham timepoint to biochemical relationships in Figure 5. Statistical analy- prior to a return to retro-orbital bleed baseline values sis was accomplished using two-way ANOVA, and the (Figure 3(e)). different types of statistical significance (by group [G], by time series [T], or by interaction of group and time 3.4. CLP and Sham Induce Significant Metabolomic Changes [I]) are noted. In addition, individual comparisons focus- in Aged Animals. Next, we examined the metabolomic ing on specifically the differences between young and effects of CLP or sham on the aged animal groups. In con- aged animals undergoing CLP, or differences between trast to the young animals, sham intervention in the aged aged sham and CLP, were also accomplished. Arranging animals produced a differentiation from the retro-orbital the data in this fashion demonstrated several clear pat- bleed metabolome that persisted throughout the assayed terns of metabolite trajectories. For some metabolites duration (Figure 4(a)). Similarly, CLP induced large devia- such as urocanic acid and histamine, CLP and sham tions away from retro-orbital bleed samples, with the magni- intervention caused significant suppression without sub- tude increasing over the time course (Figure 4(b)). The stantial variation in levels among groups. However, for pattern of 100 top metabolites isolated using consensus many of the other metabolites, including carnosine, approaches was not statistically significantly different from anserine, 1-methylhistamine, , the original metabolite pool (χ2, p =0:06, Figure 4(c)). Sig- imidazoleacetic acid, and methylimidazole acetic acid, a nificant metabolic pathways isolated in the aged population distinct increase was seen in the aged animals following included histidine metabolism and nicotinamide metabo- CLP compared to either aged sham or young CLP, sug- lism, both of which overlap with analyses demonstrated gesting these metabolites could be age-associated markers above. Additional pathways unique to the aged population of surgical sepsis. Still, other metabolites, including L- include pantothenate/CoA metabolism and taurine metabo- histidine and L-glutamate, had patterns that were less lism (Figure 4(d)). Heatmap evaluation of the average meta- apparent, without clear distinction among groups at post- bolomic responses across all groups for the top 100 surgical timepoints. consensus metabolites demonstrates more robust differences between CLP and sham surgery than seen in the young ani- 3.7. The Circulating Nicotinamide Metabolome Is mals (Figure 4(e)). Notably, multiple amino acid derivatives Significantly Altered following CLP in Aged Mice. Based on and cofactors demonstrate significant enrichment at late the initial evaluation, individual metabolites in the nicotin- CLP timepoints, whereas a sham intervention yields a more amide metabolic pathway, as referenced from http://www robust induction of key lipids at late timepoints. .kegg.jp, were graphed (Figure 6). As above, these features were assessed both by two-way ANOVA and by individual 3.5. Selection of Critical Metabolic Pathways for Further multiple comparison testing approaches; notably, in this Evaluation. Having established key differences between subset of data, all feature metabolites were significantly young and aged animals at baseline, as well as critical different both in terms of group comparison, time series, pathways associated with the response to surgical stress and interaction of group and time. Several key features in both young and aged animals, temporal trends within are apparent on evaluation of the nicotinamide metabolite critical pathways across the entire dataset were evaluated. data. Of note, nicotinamide itself decreases over time, but Tabulated outputs from MetaboAnalyst pathway analysis has different early response patterns between young and tools were compared, identifying overlapping metabolic aged animals, with the young animals demonstrating pathways among the three comparator groups (Figure S2). higher circulating levels of nicotinamide following CLP Nicotinamide metabolism was the sole pathway represented than sham or aged animals. However, for the majority of among all three comparisons, while histidine metabolism the downstream metabolites (nicotinamide riboside, 1- was significantly enriched both for the retro-orbital bleed methylnicotinamide, N1-Methyl-4-pyridone-3-carboxa- young versus aged comparison and the aged CLP-sham mide, and N1-methyl-2-pyridone-5-carboxamide), the comparison. Notably, both α-linolenic acid metabolism aged CLP animals had the most robust induction, com- and linoleic acid metabolism were identified in the young pared to either the young CLP or the aged sham opera- versus aged retro-orbital bleed group and the young tions. Markedly, this pattern held for several detected CLP-sham comparison; however, on closer examination nicotinamide precursor molecules, including nicotinate of these pathway structures, these were not informative D-ribonucleoside and quinolinate. Taken together, these and largely incompletely represented by the raw data data demonstrate that there are significant differential (data not shown). Thus, detailed statistical analysis on responses to CLP between young and aged mice and sug- both histidine metabolism and nicotinamide metabolism gest broadly that redox metabolism is altered in the aged was performed. response to surgical sepsis. Oxidative Medicine and Cellular Longevity 9

PCA scores plot: aged sham PCA scores plot: Aged CLP

4 h 8 h CLP CLP 20 4 h Pre-bleed 20 Sham Pre-bleed 12 h CLP

0 h 0 0 Sham 0 h 8 h CLP PC2 (14%)

Sham PC2 (14 %) –20 –20

12 h ShamSham –40 –40

–20 0 20 40 0 20 40 PC1 (20.8%) –20 PC1 (20.8 %) Pre-bleed 8 h Sham Pre-bleed 8 h CLP 12 h Sham 0 h Sham 0 h CLP 12 h CLP 4 h Sham 4 h CLP (a) (b) Top 100 Aged Metabolome pathway map CLP-sham 4.0

Histidine metabolism

3.0

Nicotinamide metabolism

2.0 X2 –log(p) Pantothenate/CoA p = 0.06, compared metabolism Taurineme to complete metabolite profile tabolism 39.00% 39 Amino acid 1.0 5.00% 5 Carbohydrate 8.00% 8 Cofactors/vitamins 29.00% 29 Lipid 10.00% 10 Nucleotide 0.0 2.00% 2 Peptide 7.00% 7 Xenobiotics 0.0 0.1 0.2 0.3 0.4 0.5 Pathway impact score (c) (d)

Figure 4: Continued. 10 Oxidative Medicine and Cellular Longevity

CLP Sham Pre-Bleed 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h

Hydroxylated lecithin N-Carboxyethyl-g-aminobutyric acid N-Acetylglutamine 3-Methylhistidine Imidazolelactic acid Methylimidazoleacetic acid Pi-Methylimidazoleacetic acid N-Acetylhistidine 2-Hydroxy-3-methylbutyric acid Glutarylcarnitine Phenylacetic acid N1-Acetylspermidine N-Acetylputrescine 4-Acetamidobutanoic acid 5-Hydroxyindoleacetic acid N-Acetyl-L-tyrosine Urea L-Asparagine L-Aspartic acid Amino acids N-Acetyl-L-alanine N-Acetyl-L-aspartic acid L-Glutamic acid 1-Pyrroline-5-carboxylic acid Betaine Urocanic acid Histamine 1-Methylhistamine Imidazoleacetic acid (E)-Urocanic acid Alpha-ketoisovaleric acid Hydroxyisocaproic acid L- 2,3-Dihydroxy-2-methylbutanoic acid Hypotaurine 3-Sulfinoalanine L- Indole-3-propionic acid 3-Amino-2-piperidone

N-Acetylneuraminic acid Erythronic acid N-Glycolylneuraminic acid Carbohydrates Sucrose D-Ribose

reonic acid Biotin N1-Methyl-2-pyridone-5-carboxamide N1-Methyl-4-pyridone-3-carboxamide Cofactors/Vitamins Oxalic acid Nicotinamideriboside NicotinamideN-oxide 1-Methylnicotinamide

Stearoylcarnitine trans-2-Dodecenoylcarnitine L-Acetylcarnitine Malonylcarnitine DL-2-Aminooctanoic acid 3-Hydroxyadipic acid 3-Methyladipic acid Tetradecanedioic acid 4-Deoxyerythronic acid 4-Deoxythreonic acid 12,13-DHOME 3-Hydroxysuberic acid 3-Hydroxydodecanoic acid 3-Hydroxyoctanoic acid 3-Hydroxycapric acid Lipids 16-Hydroxyhexadecanoic acid Myristoleic acid 7Z,10Z-Hexadecadienoic acid 5,8-Tetradecadienoic acid Docosapentaenoic acid Myristic acid 5-Dodecenoic acid Undecylenic acid (S)-3,4-Dihydroxybutyric acid Levoinositol Capric acid Oleamide Palmiticamide Octadecanamide 1-Methylinosine 1-Methyladenine 7-Methylguanine Cytidine 5,6-Dihydrouridine Nucleotides Pseudouridine Uridine Allantoin Ribothymidine Uracil Peptidess Phenylacetylglycine gamma-Glutamyltryptophan 2-Keto-3-deoxy-D-gluconic acid Erythritol (R)-2,3-Dihydroxy-isovalerate Betonicine Xenobiotic Salicylic acid 2,6-Dihydroxybenzoic acid Prolinebetaine

−2 −1 012 (e)

Figure 4: Identification of the metabolic pathways that drive the response to CLP in aged animals. PCA plots of the aged sham (a) and CLP (b) metabolomes. Sham-treated animals demonstrated a slow return to baseline over time, though not to the same degree seen in young animals. On the other hand, aged CLP-treated animals experienced progressive derangement of the metabolome over time. (c) Profile of the top 100 metabolites that differentiate between CLP and sham interventions in aged mice over time. (d) Pathway analysis comparing enrichment of the top 100 metabolites compared to the 566 metabolites originally contained in the dataset. (e) Heatmap analysis demonstrating group average values of the top 100 metabolites at all CLP and sham surgical timepoints in aged animals, illustrating broad differences in metabolomic responses following CLP compared to sham intervention.

4. Discussion strate clear baseline metabolomic differences between young and aged mice. These differences became more pronounced Aging is a complex series of degenerative processes that following surgical stress (sham operation) or surgical stress results in the degradation of cellular homeostatic mecha- combined with infectious insult (CLP), underscoring the nisms and loss of metabolic, structural, and functional contribution of aging to metabolomic homeostasis. Further- capacities that significantly impact the response to stress more, some metabolic pathways were found through statisti- and illness. The data presented in this manuscript demon- cal analysis to contribute more to the observed metabolomic Oxidative Medicine and Cellular Longevity 11 alterations than others, highlighting important potential tar- mic shift away from both the retro-orbital bleed controls gets for the mitigation of age-associated adverse outcomes and the 0 h sacrifice, with an apparent movement back to following surgical intervention. baseline by the 12-hour timepoint for both young and aged The experimental design was chosen to provide maximal animals (Figures 3(a) and 4(a)). Interestingly, this return to insight into multiple different metabolic trajectories. This is the 0 h/retro-orbital bleed baseline was more marked in the reflected in the specific analyses reported here, including ini- young animals, suggesting a more rapid recovery from sham tial comparison of young vs. aged animals, evaluation of surgery than seen in the aged animals. Collectively, these global metabolomic trajectories from baseline in both young data underscore that the biological stress of surgical inter- and aged animals individually, and in the final evaluation of vention is a significant insult to metabolomic equilibrium pathways of overlapping significance between the young and is more pronounced with aging—a finding that has been and aged animals. Critically, for the evaluation of each of routinely demonstrated in human cohorts. In contrast, the these specific pathways, we utilized a consensus-based more pronounced stress of CLP induced a persistent meta- methodology combining multiple statistical analysis tools bolomic shift in both young and aged animals, with a more for the identification of key metabolites. This approach was marked deviation from the control metabolome seen over selected to mitigate biases in individual feature identification time among the aged animals (Figures 3(a) and 4(b)). These approaches, and to limit the possibility that any specific data demonstrate clear metabolomic shifts following both methodology (RF, EBAM, PLS-DA, or PCA) would overfit sham surgery and CLP. For both the young and aged groups, the data. In particular, the use of the PCA loading scores statistically significant metabolite enrichment profiles were among the data selection criteria provided an important noted after application of the feature selection algorithms counterweight to the supervised methods, as PCA loadings (Figures 3(c) and 4(c)), suggesting more narrow pathway are agnostic to group assignment and are only sensitive to alterations following these respective insults. Importantly, the total variance of the system. This allowed for the genera- when evaluating these top 100 metabolites visually using tion of a metabolite feature list incorporating both the broad heatmaps, some broad visual differences are apparent. Nota- structure of the data as well as key group-defining metabo- bly, in the sham surgery groups, the aged cohort appears to lites. This approach generated a robust profile of the broad have persistent enrichment of several key lipids across time, metabolomic differences for each of the distinct experimental whereas in the young sham intervention, there appears to be comparisons. a transient enrichment followed by a return to a more “base- In the initial comparison, the baseline differences line” pattern (Figures 3(e) and 4(e)). This effect is less pro- between young and aged animals from the retro-orbital nounced for the CLP groups, with substantial variation in bleed samples taken at the outset of the experimental proto- the patterns identified therein. In the context of the broad col were evaluated. The data, summarized in Figure 2, dem- metabolomic shifts identified on the 2D-PCA plots, this onstrates that no specific subclass among the top 100 may be contributing to a delayed “injury recovery phase” metabolites significantly enriched above the distribution of for the sham surgical interventions in the aged mice com- the original dataset following the feature identification work- pared to the young. Further studies with longer term time- flow (Figure 2(c)). As multiple means to identify metabolites points are required to better characterize this trajectory. based on group identification were used (PLS-DA, EBAM, In addition to the broad metabolome-wide trajectories RF), these 100 selected metabolites provided excellent dis- identified among the groups noted above, key pathway crimination between young and aged animals when dis- enrichment results were examined for each of the main anal- played as a heatmap (Figure 2(d)). Notably, the findings yses (young vs. aged baseline, young CLP/sham, and aged here are concordant with many of the findings in prior CLP/sham) and overlapping pathways of interest were iso- studies, specifically relating to the identification of lipids lated across the three groups (Figure S2, Table S1). Within and amino acids that strongly discriminate between aged the histidine metabolism pathway, a key notable finding and young mice [14, 16]. The pathway analysis also was was the elevation of histidine derivatives carnosine and concordant with prior studies, demonstrating significant anserine predominantly in the aged CLP group, when enrichment of key antioxidant and redox pathways (nico- compared either to aged sham or young CLP. Carnosine, a tinamide metabolism, pentose interconversions) as well as dipeptide comprise of β-alanine and L-histidine, is an lipid-specific signaling pathways (retinol, linolenic acid bioactive molecule with pleiotrophic function in different metabolism). tissue compartments, including antioxidant scavenging, pH Additionally, these data examined the temporal changes buffering, metal chelation, inhibition of advanced glycation in the metabolome following surgical sepsis (CLP) or sham end product formation, and inhibition of advanced operation in both young and aged animals. These data pro- lipoxidation end product formation (Alexander A. [21]). vide significant insight into some key differences underpin- These functions are integrated in several animal studies ning differential responses to surgical stress in aged vs. that have focused on its potential protective effects in young animals. A key difference is the broad metabolomic ischemia-reperfusion injuries. This protective effect has landscape shifts that occur in both young and aged animals been observed in the murine brain, liver, kidney, and testis and how these differ over time. PCA-based analysis and 2- and initially was suspected to be associated with the dimensional plotting demonstrated distinct temporal pat- antioxidant capacity of carnosine [22–25]. However, it has terns for sham surgical intervention and CLP in both age been proposed that carnosine may also function via cohorts. Notably, sham surgery demonstrated a metabolo- histaminergic signaling cascades as a reservoir for 12 Oxidative Medicine and Cellular Longevity

L-Histidine Formiminoglutamic⁎⁎ acid 1.4 10 Urocanic acid 10 ⁎ ⁎⁎ ⁎ 8 8 ⁎ ⁎ 1.2 4-imidazolone-5- 6 6 1.0 propanoate 4 4 Peak area Peak area Peak area 0.8 2 2

0.6 0 0 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h

6 Anserine

Pre-bleed Pre-bleed ⁎⁎ Pre-bleed GTI T ⁎ GTI

4 L-Glutamate 6 Carnosine 2.0 ⁎⁎ ⁎ ** ⁎ * ⁎⁎ Peak area 2 ⁎ ⁎ 1.5 4 ⁎ 0 1.0 0 h 4 h 8 h 12 h Peak area Histidine 2 Peak area 0.5

and histamine Pre-bleed GTI 0 0.0 pathway 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h Pre-bleed GTI Pre-bleed TI Imidazole-4- Multiple comparison testing acetaldehyde Imidazoleacetic acid 2.0 1-methylhistidine 2.0 ⁎⁎ ⁎p < 0.05 Aged CLP vs. Young CLP ⁎ ⁎ ⁎⁎ 1.5 1.5 ⁎⁎ p < 0.05 Aged CLP vs. Aged sham ⁎ 1.0 1.0 Peak area Peak area Methylimidazole Two-way ANOVA statistics 0.5 0.5 acetaldehyde 0.0 p 0.0 G = Experimental group < 0.05 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h T = Time p < 0.05 Pre-bleed I = Interaction group × Time p < 0.05 Pre-bleed GT GTI Methylimidazoleacetic acid Aged (CLP) 6 ⁎ Histamine 1-methylhistamine 3 20 ⁎⁎ Aged (Sham) ⁎ ⁎ ⁎⁎ ⁎⁎ 15 ⁎ Young (CLP) 4 2 Young (Sham) 10 Peak area Peak area 2 Peak area 1 5

0 0 0 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h Pre-bleed Pre-bleed T Pre-bleed GTI GTI

Figure 5: Detailed view of the histidine-histamine metabolic pathway. Schematic representation of the histidine-histamine pathway organized according to KEGG pathway diagrams, with individual values plotted for each of the four experimental groups (young sham, young CLP, aged sham, and aged CLP). Data was graphed for all available metabolites in the pathway, and undetected metabolites were left as labeled boxes. Statistical testing was accomplished using two-way ANOVA, and significance in this analysis was denoted in the right lower corner of each metabolite graph using G (group), T (timepoint), or I (interaction). Post hoc multiple comparisons testing was done to evaluate the differences between both aged CLP and aged sham as well as aged CLP and young CLP, as described in the methods.

histdine/histamine, as several studies have demonstrated The other key pathway that emerged from statistical that subantioxidant doses of carnosine supplementation evaluation of the metabolite data involved nicotinamide during renal ischemia-reperfusion injury modeling in both biology (Figure 6). Here, the data was less complete, due to renal and cerebral perfusion were able to induce a renal the inability of the utilized metabolomic platform to identify protective effect [26]. In rat sepsis models, carnosine has plasma NAD, NADP, or nicotinamide D-ribonucleotide, also been shown to have protective effects both on liver which are well-known key regulators of redox homeostasis function tests, markers of liver oxidative damage, and renal and key drivers of oxidative metabolism. One hypothesis function [27–29]. Critically, carnosine also is suspected to that this is due to the intracellular compartmentalization have functions in modulating lifespan. Several groups have required to limit degradation of these labile redox species demonstrated that carnosine has variable effects on life and suspect that any of these molecules in free circulation span increases in murine models of accelerated aging [30, would be rapidly degraded or otherwise consumed by circu- 31] as well as in cell senescence assays [32], though the lating redox enzymes. However, the detectable molecules in specific mechanisms underpinning this are not fully this pathway included several interesting species, including elucidated yet. The present data, which demonstrate an nicotinamide riboside, which is considered to be an orally enhanced circulating carnosine signature in aged mice with bioavailable NAD precursor vitamin capable of increasing CLP (Figure 5), suggest an age-associated endogenous NAD levels in both humans and mice [33]. Critically, nico- carnosine response to injury that merits further tinamide riboside has been shown in experimental models investigation for potential therapeutic leverage. To our of murine sepsis to limit both pulmonary and cardiac toxic- knowledge, no current data links carnosine with the aged ity, likely through NAD+/SIRT1 signaling, suggesting that response to sepsis, which makes this a remarkable target this may be a key factor influencing outcomes in our CLP for future study. model [34]. Curiously, in the aged CLP animals, there was Oxidative Medicine and Cellular Longevity 13

Quinolinate⁎⁎ 10 ⁎⁎⁎ ⁎ 8 6 Deamino NAD Nicotinate D-ribonucleotide 4 NAD Peak area ⁎ 2 0 0 h 4 h 8 h 12 h Pre-bleed

Nicotinate D-ribonucleoside 6 ⁎⁎ NADP Nicotinamide D-ribonucleotide ⁎ 4 ⁎⁎ Nicotinate

Peak area 2

0 0 h 4 h 8 h 12 h Pre-bleed

Nicotinamide Nicotinamide riboside Trigonelline 3 10 2.0 ⁎ ⁎⁎ ⁎ ⁎ ⁎ Nicotinamide 8 1.5 2 ⁎⁎ pathways 6 ⁎ 1.0 4 ⁎

Peak area 1 Peak area Peak area 2 0.5 0 0 0.0 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h Multiple comparison testing ⁎p < 0.05 Aged CLP vs. Young CLP Pre-bleed Pre-bleed Pre-Bbleed ⁎⁎p < 0.05 Aged CLP vs. Aged Sham

1-methylnicotinamide N1-methyl-4-pyridone N1-methyl-2-pyridone Two-way ANOVA Statistics 8 10 15 -3-carboxamide -5-carboxamide All graphs statistically significant ⁎⁎ 8 ⁎⁎ ⁎⁎ ⁎⁎ 6 ⁎ ⁎ ⁎ ⁎⁎⁎ by group, time and interaction at 10 ⁎ 6 p < 0.0005 ⁎⁎ ⁎ 4 ⁎ ⁎ ⁎⁎ 4 ⁎ Peak area Peak area Peak area ⁎ 5 2 2 Aged (CLP) 0 0 0 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h 0 h 4 h 8 h 12 h Aged (Sham)

Prebleed Young (CLP) Pre-bleed Pre-bleed Young (Sham)

Figure 6: Detailed view of the nicotinate-nicotinamide metabolic pathway. Schematic representation of the nicotinate-nicotinamide pathway organized according to KEGG pathway diagrams, with individual values plotted for each of the four experimental groups (young sham, young CLP, aged sham, and aged CLP). As previously, data was graphed for all available metabolites in the pathway, and undetected metabolites were left as labeled boxes. Statistical testing was accomplished using two-way ANOVA, and significance in this analysis was denoted in the right lower corner of each metabolite graph using G (group), T (timepoint), or I (interaction). Post hoc multiple comparisons testing was done to evaluate the differences between both aged CLP and aged sham as well as aged CLP and young CLP, as described in the methods. a statistically significant elevation in the circulating nico- rat models [36]. Taken together, these findings suggest that tinamide riboside compared to either aged sham or young standard supplementation of orally bioavailable NAD deriv- CLP or sham animals. However, evaluation of the down- atives in the current model system may not improve func- stream metabolites in the pathway suggests that this spike tional outcomes, as nicotinamide riboside is already is potentially not associated with an increase in NAD or elevated. Further, they suggest that study of the downstream NADP, but instead with enhanced degradation of NAD transformation products of NAD and its associated related and its derivatives, as we see similar increases in both 1- metabolites may be playing a role in the aged murine methylnicotinamide and its metabolic products (Figure 6, response to CLP. While we hypothesize that this may be asso- bottom panels). Critically, these metabolites are known to ciated with increased tissue fragility and spillage of intracellu- be associated with signaling pathways involved in transition- larly quarantined NAD pools in the aged animals, these ing from carbohydrate to fat energy metabolism [35], as well findings require further experimental investigation to deter- as the pathophysiology of pulmonary artery hypertension in mine specifically how the nicotinamide-NAD axis may be 14 Oxidative Medicine and Cellular Longevity altered in the aging organism in sepsis and how it may be Conflicts of Interest appropriately leveraged for clinical benefit. In particular, fi fl evaluation of the NAD salvage pathways, which have known The authors report no signi cant con icts of interest associ- implications in both lifespan and healthspan in various ated with this work. organisms, will be critical moving forward [37]. The work presented here, while providing intriguing evi- Authors’ Contributions dence of circulating age-associated metabolic signatures in a murine CLP model, is not without limitations. The experi- AC, LK, and BZ conceived the experimental design and mental design sought to develop a time course post-CLP that analysis strategies. AC, LK, LC, and SS all assisted with the would capture a metabolic shift in the animals, based on animal work, execution of the experimental protocol, and telemetry data suggesting a biometric inflection point in logistical considerations. AC performed statistical analysis the 5-10-hour range following CLP in young C57Bl/6 ani- and figure generation. AC and BZ contributed to the draft- mals [17]. The PCA visualization of individual subgroups ing of the manuscript. demonstrates that the postintervention timepoints (0 h, 4 h, 8 h, and 12 h) likely are adequate for capturing post-CLP metabolomic changes, but that more granularity would Acknowledgments fi potentially be important for more speci c biochemical path- This work was supported by the Department of Veteran’s way interrogation. This is particularly true given the Affairs grant 5I01BX003924-02 (BZ) and by the American increased intraprocedural mortality in the aged animal pop- College of Surgeons Resident Research Award (AC). ulation, which suggests a more rapid deterioration following surgical intervention. This study was also initially conceived to be a one-to-one series with each animal providing its own Supplementary Materials control data. The goal of this was to provide data about indi- vidual metabolic trajectories, as opposed to comparisons of Table S1: summary of pathway analysis output from Meta- point measurements of the circulating metabolome, in boAnalyst utilized to identify pathway enrichment among young vs. aged animals following surgical intervention or the top 100 metabolites for each of the listed comparisons. septic insult. Unfortunately, due to unforseen logistical cir- Of these, the nicotinate and nicotinamide metabolism and cumstances, it was not possible to progress with that initial histidine metabolism pathways were selected for down- design, but maintained the experimental protocol such that stream evaluation based on their presence in multiple com- every animal entered into the experimental protocol under- parisons as well as the number of represented metabolites went the same retro-orbital bleed prior to sham or CLP. in the pathway. Supplemental Figure S1: schematic of nor- fl fl Nonetheless, evaluation of these data as pooled populations malization and data processing work ow. (a) Work ow for remained robust. Finally, while the nicotinamide-NAD data reduction from the 834 discrete initial metabolites to pathway remains an interesting target for future evaluation, the 566 annotated metabolites, followed by the approaches this study also illustrates the limitations of the circulating used for data normalization and evaluation. (b) Graphical metabolome for identification of labile redox-sensitive spe- output from MetaboAnalyst demonstrating normalization cies. While this study provides compelling foundational evi- approach. Supplemental Figure S2: Venn diagram of over- dence that the NAD axis is altered in aged animals following lapping metabolic pathways among the three individual sta- CLP, it is clearly not definitive in determining the mecha- tistical comparisons. These included the young CLP/sham nisms at play. only, the aged CLP/sham only, and the young/aged baseline comparisons. Two pathways (histidine-histamine and nico- tinate-nicotinamide) were chosen for further downstream 5. Conclusions evaluation based on the representative number of metabo- lites as well as their appearance in the metabolite features In summary, this manuscript presents a large-scale evalua- following statistical analysis. (Supplementary Materials) tion of the circulating plasma metabolome in both young and aged mice following CLP or sham surgical interventions. References Several key metabolic pathways of interest were identified [1] N. Poulose and R. Raju, “Aging and injury: alterations in cellu- that differentiate between the young and aged response to ” fi lar energetics and organ function, Aging and Disease, vol. 5, surgical insult and subsequent infection providing signi - no. 2, pp. 101–108, 2014. cant foundational evidence for future study, including both [2] K. F. Desserud, T. Veen, and K. Søreide, “Emergency general mechanistic evaluation of the aged response to acute disease surgery in the geriatric patient,” The British Journal of Surgery, and potential therapeutic development. vol. 103, no. 2, pp. e52–e61, 2016. [3] P. Rhee, B. Joseph, V. Pandit et al., “Increasing trauma deaths in the United States,” Annals of Surgery, vol. 260, no. 1, pp. 13– Data Availability 21, 2014. [4] M. Rose, H. Pan, M. R. Levinson, and M. Staples, “Can frailty The data that support the findings of this study are available predict complicated care needs and length of stay?,” Internal from the corresponding author upon request. Medicine Journal, vol. 44, no. 8, pp. 800–805, 2014. Oxidative Medicine and Cellular Longevity 15

[5] M. C. Shinall, S. Arya, A. Youk et al., “Association of preoper- sion liver injury in rats,” European Journal of Pharmacology, ative patient frailty and operative stress with postoperative vol. 572, no. 1, pp. 61–68, 2007. ” mortality, JAMA Surgery, vol. 155, no. 1, p. e194620, 2020. [24] T. Fujii, M. Takaoka, T. Muraoka et al., “Preventive effect of L- [6] G. Cornelissen and K. Otsuka, “Chronobiology of aging: a carnosine on ischemia/reperfusion-induced acute renal failure mini-review,” Gerontology, vol. 63, no. 2, pp. 118–128, 2017. in rats,” European Journal of Pharmacology, vol. 474, no. 2–3, [7] C. López-Otín, M. A. Blasco, L. Partridge, M. Serrano, and pp. 261–267, 2003. G. Kroemer, “The hallmarks of aging,” Cell, vol. 153, no. 6, [25] C. Pekcetin, M. Kiray, B. U. Ergur et al., “Carnosine attenuates pp. 1194–1217, 2013. oxidative stress and apoptosis in transient cerebral ischemia in [8] J. Vijg and B. K. Kennedy, “The essence of aging,” Gerontology, rats,” Acta Biologica Hungarica, vol. 60, no. 2, pp. 137–148, vol. 62, no. 4, pp. 381–385, 2016. 2009. [9] J. S. Lorusso, O. A. Sviderskiy, and V. M. Labunskyy, “Emerg- [26] H. Kurata, T. Fujii, H. Tsutsui et al., “Renoprotective effects of ing omics approaches in aging research,” Antioxidants and l-carnosine on ischemia/reperfusion-induced renal injury in Redox Signaling, vol. 29, no. 10, pp. 985–1002, 2018. rats,” The Journal of Pharmacology and Experimental Thera- [10] R. Sharma and A. Ramanathan, “The aging metabolome—bio- peutics, vol. 319, no. 2, pp. 640–647, 2006. markers to hub metabolites,” Proteomics, vol. 20, no. 5–6, [27] S. Sahin and D. Burukoglu Donmez, “Effects of carnosine p. 1800407, 2020. (beta-alanyl-L-histidine) in an experimental rat model of acute [11] D. Xu, S. Liao, P. Li et al., “Metabolomics coupled with tran- kidney injury due to septic shock,” Medical Science Monitor : scriptomics approach deciphering age relevance in sepsis,” International Medical Journal of Experimental and Clinical Aging and Disease, vol. 10, no. 4, pp. 854–870, 2019. Research, vol. 24, pp. 305–316, 2018. [12] C. W. Seymour, S. Yende, M. J. Scott et al., “Metabolomics in [28] S. Sahin, S. Oter, D. Burukoğlu, and E. Sutken, “The effects of pneumonia and sepsis: an analysis of the GenIMS cohort carnosine in an experimental rat model of septic shock,” Med- study,” Intensive Care Medicine, vol. 39, no. 8, pp. 1423– ical Science Monitor Basic Research, vol. 19, pp. 54–61, 2013. 1434, 2013. [29] C. Sun, Q. Wu, X. Zhang, Q. He, and H. Zhao, “Mechanistic [13] A. Cyr, Y. Zhong, S. E. Reis et al., “Analysis of the plasma evaluation of the protective effect of carnosine on acute lung metabolome after trauma, novel circulating sphingolipid sig- injury in sepsis rats,” Pharmacology, vol. 100, no. 5–6, natures, and in-hospital outcomes,” Journal of the American pp. 292–300, 2017. – College of Surgeons, vol. 232, no. 3, pp. 276 287.e1, 2021. [30] A. A. Boldyrev, M. O. Yuneva, E. V. Sorokina et al., “Antioxi- [14] R. H. Houtkooper, C. Argmann, S. M. Houten et al., “The met- dant systems in tissues of senescence accelerated mice,” Bio- abolic footprint of aging in mice,” Scientific Reports, vol. 1, chemistry, vol. 66, no. 10, pp. 1157–1163, 2001. no. 1, 2011. [31] S. L. Stvolinskii, T. N. Fedorova, M. O. Yuneva, and A. A. Bol- “ [15] C. Seo, Y.-H. Hwang, Y. Kim et al., Metabolomic study of dyrev, “Protective effect of carnosine on Cu, Zn-superoxide aging in mouse plasma by gas chromatography-mass spec- dismutase during impaired oxidative metabolism in the brain ” – trometry, Journal of Chromatography B, vol. 1025, pp. 1 6, in vivo,” Bulletin of Experimental Biology and Medicine, 2016. vol. 135, no. 2, pp. 130–132, 2003. [16] A. Tomás-Loba, B. Bernardes de Jesus, J. M. Mato, and M. A. [32] L. Shao, Q.-H. Li, and Z. Tan, “L-carnosine reduces telomere “ ” Blasco, A metabolic signature predicts biological age in mice, damage and shortening rate in cultured normal fibroblasts,” – Aging Cell, vol. 12, no. 1, pp. 93 101, 2013. Biochemical and Biophysical Research Communications, [17] A. J. Lewis, D. Yuan, X. Zhang, D. C. Angus, M. R. Rosengart, vol. 324, no. 2, pp. 931–936, 2004. “ fi and C. W. Seymour, Use of biotelemetry to de ne physiology- [33] S. A. J. Trammell, M. S. Schmidt, B. J. Weidemann et al., “Nic- based deterioration thresholds in a murine cecal ligation and otinamide riboside is uniquely and orally bioavailable in mice puncture model of sepsis,” Critical Care Medicine, vol. 44, ” – – and humans, Nature Communications, vol. 7, no. 1, pp. 1 14, no. 6, pp. e420 e431, 2016. 2016. [18] J. Chong, D. S. Wishart, and J. Xia, “Using MetaboAnalyst 4.0 [34] G. Hong, D. Zheng, L. Zhang et al., “Administration of nico- for comprehensive and integrative metabolomics data analy- tinamide riboside prevents oxidative stress and organ injury sis,” Metabolites, vol. 68, no. 1, p. e86, 2019. in sepsis,” Free Radical Biology and Medicine, vol. 123, “ [19] Z. Pang, J. Chong, S. Li, and J. Xia, MetaboAnalystR 3.0: pp. 125–137, 2018. toward an optimized workflow for global metabolomics,” [35] K. Ström, D. Morales-Alamo, F. Ottosson et al., “N1-methylni- Metabolites, vol. 10, no. 5, p. 186, 2020. cotinamide is a signalling molecule produced in skeletal mus- [20] D. S. Wishart, Y. D. Feunang, A. Marcu et al., “HMDB 4.0: the ” fi ” cle coordinating energy metabolism, Scienti c Reports, vol. 8, human metabolome database for 2018, Nucleic Acids no. 1, p. 3016, 2018. Research, vol. 46, no. D1, pp. D608–D617, 2018. [36] A. Fedorowicz, Ł. Mateuszuk, G. Kopec et al., “Activation of [21] A. A. Boldyrev, G. Aldini, and W. Derave, “Physiology and the nicotinamide N-methyltransferase (NNMT)-1-methylni- pathophysiology of carnosine,” Physiological Reviews, vol. 93, cotinamide (MNA) pathway in pulmonary hypertension,” no. 4, pp. 1803–1845, 2013. Respiratory Research, vol. 17, no. 1, pp. 108–113, 2016. [22] L. Abbasoğlu, E. B. Kalaz, M. Soluk-Tekkeşin, V. Olgaç, [37] R. M. Anderson, K. J. Bitterman, J. G. Wood et al., “Manipula- S. Doğru-Abbasoğlu, and M. Uysal, “Beneficial effects of tau- tion of a nuclear NAD+ salvage pathway delays aging without rine and carnosine in experimental ischemia/reperfusion altering steady-state NAD+ levels∗,” Journal of Biological injury in testis,” Pediatric Surgery International, vol. 28, Chemistry, vol. 277, no. 21, pp. 18881–18890, 2002. no. 11, pp. 1125–1131, 2012. [23] A. A. Fouad, M. A.-A. El-Rehany, and H. K. Maghraby, “The hepatoprotective effect of carnosine against ischemia/reperfu-