2D DIGE proteomic analysis reveals fasting‐induced protein remodeling through organ‐specific transcription factor(s) in mice

Overnight fasting is a routine procedure before surgery in clinical settings. Intermittent fasting is the most common diet/fitness trend implemented for weight loss and the treatment of lifestyle‐related diseases. In either setting, the effects not directly related to parameters of interest, either beneficial or harmful, are often ignored. We previously demonstrated differential activation of cellular adaptive responses in 13 atrophied/nonatrophied organs of fasted mice by quantitative PCR analysis of gene expression. Here, we investigated 2‐day fasting‐induced protein remodeling in six major mouse organs (liver, kidney, thymus, spleen, brain, and testis) using two‐dimensional difference gel electrophoresis (2D DIGE) proteomics as an alternative means to examine systemic adaptive responses. Quantitative analysis of protein expression followed by protein identification using matrix‐assisted laser desorption ionization–time‐of‐flight mass spectrometry (MALDI‐TOFMS) revealed that the expression levels of 72, 26, and 14 proteins were significantly up‐ or downregulated in the highly atrophied liver, thymus, and spleen, respectively, and the expression levels of 32 proteins were up‐ or downregulated in the mildly atrophied kidney. Conversely, there were no significant protein expression changes in the nonatrophied organs, brain and testis. Upstream regulator analysis highlighted transcriptional regulation by peroxisome proliferator‐activated receptor alpha (PPARα) in the liver and kidney and by tumor protein/suppressor p53 (TP53) in the thymus, spleen, and liver. These results imply of the existence of both common and distinct adaptive responses between major mouse organs, which involve transcriptional regulation of specific protein expression upon short‐term fasting. Our data may be valuable in understanding systemic transcriptional regulation upon fasting in experimental animals.

Overnight fasting is a routine procedure before surgery in clinical settings. Intermittent fasting is the most common diet/fitness trend implemented for weight loss and the treatment of lifestyle-related diseases. In either setting, the effects not directly related to parameters of interest, either beneficial or harmful, are often ignored. We previously demonstrated differential activation of cellular adaptive responses in 13 atrophied/nonatrophied organs of fasted mice by quantitative PCR analysis of gene expression. Here, we investigated 2-day fasting-induced protein remodeling in six major mouse organs (liver, kidney, thymus, spleen, brain, and testis) using two-dimensional difference gel electrophoresis (2D DIGE) proteomics as an alternative means to examine systemic adaptive responses. Quantitative analysis of protein expression followed by protein identification using matrixassisted laser desorption ionization-time-of-flight mass spectrometry (MALDI-TOFMS) revealed that the expression levels of 72, 26, and 14 proteins were significantly up-or downregulated in the highly atrophied liver, thymus, and spleen, respectively, and the expression levels of 32 proteins were up-or downregulated in the mildly atrophied kidney. Conversely, there were no significant protein expression changes in the nonatrophied organs, brain and testis. Upstream regulator analysis highlighted transcriptional regulation by peroxisome proliferator-activated receptor alpha (PPARa) in the liver and kidney and by tumor protein/suppressor p53 (TP53) in the thymus, spleen, and liver. These results imply of the existence of both common and distinct adaptive responses between major mouse organs, which involve transcriptional regulation of specific protein expression upon short-term fasting. Our data may be valuable in understanding systemic transcriptional regulation upon fasting in experimental animals. rheumatoid arthritis, and seizures [1,2]. Moreover, in current clinical settings and basic research using animals, overnight fasting is a routine procedure before surgical operations although its influences not directly related to organs of interest or investigated parameters are often ignored [3].
Maintaining adequate blood levels of glucose is prerequisite for energy metabolism in glucose-requiring organs/cells including the brain, kidney, testis, and red blood cells. Upon food deprivation, declining blood glucose levels induce rapid secretion of glucagon and decreased release of insulin, thereby activating hepatic glycogenolysis although hepatic glycogen becomes quickly (~24 h) depleted. If fasting continues, peripheral organs switch as the primary energy source from glucose to fatty acids that are released from triacylglycerol droplets in adipose tissues. However, some organs/ cells are unable to utilize fatty acids as an energy source, and thus, the liver produces ketone bodies from fatty acids so that such organs/cells can use them as a secondary energy source and save glucose. Meanwhile, gluconeogenesis from glucogenic amino acids of protein origin and ketogenesis from ketogenic amino acids takes place to maintain blood glucose and energy sources, respectively. Several lines of evidence suggest that all such biochemical adaptation to fasting is 'transcriptionally regulated' in the liver. The transcriptional factor/ nuclear receptor proliferator peroxisome proliferatoractivated receptor alpha (PPARa) has been shown to primarily mediate adaptive responses to fasting in the liver [4][5][6][7]. In addition, tumor protein/suppressor p53 (TP53) has been shown recently to increase via posttranscriptional regulation in the liver upon fasting, thereby mediating amino acid catabolism and gluconeogenesis [8]. However, such transcriptional regulation upon fasting has not been described in nonhepatic organs. Moreover, proteomic studies on the nonhepatic organs during fasting have been unexpectedly limited [9,10] although the impacts of fasting are often apparent in the nonhepatic organs as we reported fasting-induced cardioprotection in mice [11].
In this study, we investigated the possible sources of organ-specific transcriptional regulation upon fasting using two-dimensional difference gel electrophoresis (2D DIGE) proteomic approach. Although LC-MS/ MS became the mainstream for such proteomic analysis in recent decades, conventional 2D DIGE continues to be an important technology that enables rapid and direct visualization of thousands of proteins and their quantitative analyses [12][13][14][15][16][17][18]. Here, we report novel transcriptional regulation in nonhepatic organs including kidney, thymus, and spleen upon fasting for 2 days.

Materials and methods
Animals C57BL/6J mice were purchased from Japan SLC (Shizuoka, Japan). Eight-week-old male mice were group-housed (4 mice per 470-cm 2 cage) in an air-conditioned room (24°C) kept on a 12-h dark/light (8 pm-8 am) cycle, and allowed free access to water and a CE-2 standard dry rodent diet (Clea Japan, Tokyo, Japan). In fasting experiments, mice were deprived of the diet for 1 or 2 days between 2 pm and 2 pm (hereinafter referred to as F1 and F2 mice, respectively). Ad libitum-fed (AL) mice were analyzed as controls. After anesthetization by isoflurane inhalation, blood was collected through the heart and the liver, kidney, thymus, spleen, brain, and testis were quickly dissected out, snap-frozen by liquid nitrogen, and stored at À80°C. All animal procedures conformed to the Guide for the Care and Use of Laboratory Animals, 8 th Edition published by the US National Research Council, and were approved by the Animal Care Committees of Keio University (No. 09187- [4][5][6]) or Showa Pharmaceutical University (No. P-2016-10).

2D DIGE
Each organ aliquot (50-100 mg) was homogenized (4100 r.p.m., 30 s 9 3, 4°C) in ice-cold urea buffer (7 M urea, 2 M thiourea, 4% CHAPS, 65 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride, and 1 mM sodium orthovanadate) using a Micro Smash MS-100R Beads Cell Disrupter (Tomy, Tokyo, Japan) and 5-mm-diameter zirconia beads (Tomy). Homogenates were centrifuged at 16 000 g for 5 min at 4°C, and then, the supernatants were centrifuged at 20 000 g for 25 min at 4°C. Protein concentrations of the resultant supernatants were determined using a Bio-Rad Protein Assay and bovine serum albumin as a standard. All reagents used in this study were of analytical grades from Wako (Tokyo, Japan) or Sigma-Aldrich unless otherwise noted. 2D DIGE was performed as described previously [12][13][14]. Twenty-five lg of protein (adjusted to pH 8.5 by adding 40 mM Tris/HCl [pH 8.5]) was labeled with 200 pmol of CyDye (Cy2, Cy3, or Cy5 minimal dye fluor [GE Healthcare]) for 30 min at 4°C in the dark. A pool, to be used for calibration between the gels, was generated from equal protein amounts of all eight samples (n = 4 each for AL and F2 mice). The reaction was stopped by adding 0.5 lL of 10 mM lysine. Labeled samples were mixed, and DTT and immobilized pH gradient (IPG) buffer (final 1% each) were added for 10 min at 4°C in the dark. The samples were subjected to isoelectric focusing (IEF) in an Immobiline DryStrip (18 cm, pH 3-10 NL [nonlinear], GE Healthcare) that was rehydrated for 20 h in rehydration buffer (7 M urea, 2 M thiourea, 2% Triton X-100, 13 mM DTT, 2.5 mM acetic acid, 1% IPG buffer, and 5 p.p.m. bromophenol blue) at 20°C, using a CoolPhoreStar IPG-IEF Type-PX system (Anatech, Tokyo, Japan). Once IEF was completed, the strips were equilibrated for 30 min in equilibration buffer (50 mM Tris/HCl [pH 6.8], 6 M urea, 2% SDS, 30% [v/ v] glycerol, 65 mM DTT, and 5 p.p.m. bromophenol blue), followed by in alkylating buffer (equilibration buffer with 4.5% iodoacetamide instead of DTT) for an additional 15 min. The strips were sealed on the top of 12.5% PAGE gels (140 9 140 9 1 mm; Perfect NT Gel S from DRC, Tokyo, Japan) using 0.5% low-melting-point agarose in Tris/glycine electrophoresis buffer. The second dimension of protein separation was performed at a constant 200 V using an ERICA-S high-speed electrophoresis system (DRC). A total of four gels (for the comparisons between AL (n = 4) and F2 (n = 4)) were scanned at once for Cy2/ Cy3/Cy5 fluorescence using a Typhoon Trio image scanner (GE Healthcare), and obtained images were integratively analyzed using DeCyder 2D ver. 6.5 differential analysis software (GE Healthcare).

Silver staining
For protein identification using MALDI-TOFMS, each tissue homogenate sample (100-150 lg) was subjected to 2D PAGE (mentioned above) without CyDye labeling. To get better resolutions, some samples were separated on larger 2D systems using longer strips (24 cm, pH 3-10 NL [nonlinear]) and larger PAGE gels (257 9 200 9 1 mm; Perfect NT Gel W from DRC). After electrophoresis, the gel was stained using a Silver Stain MS Kit (Wako).

MALDI-TOFMS analysis of trypsin digests
Gel pieces were excised from silver-stained gels, destained with a mixture of destaining solutions A and B (Wako), washed twice with deionized water and four times with 50 mM ammonium bicarbonate (NH 4 HCO 3 ):acetonitrile (1 : 1), dehydrated once with acetonitrile, twice alternately rehydrated with 100 mM NH 4 HCO 3 and dehydrated with acetonitrile, and dried by vacuum centrifugation. Protein samples in the gels were digested in 10 lL of trypsin solution (0.1 lg of Trypsin Gold, Mass Spectrometry Grade [Promega] and 0.01% ProteaseMAX Surfactant, Trypsin Enhancer [Promega] in 25 mM NH 4 HCO 3 ) by incubating at 50°C for 1 h. Trypsin digests were mixed with 3 lL of 2% trifluoroacetic acid, and 1 lL of samples was spotted onto a lFocus MALDI plate (900 lm, 384 circles, Hudson Surface Technology [Old Tappan, NJ, USA]) with an equal volume of matrix solution, containing 10 mM a-cyano-4-hydroxycinnamic acid in 1% trifluoroacetic acid/50% acetonitrile. Positive ion mass spectra were obtained using an AXIMA-CFR plus (Shimadzu, Kyoto, Japan) in a reflectron mode. MS spectra were acquired over a mass range of 700-4000 m/z and calibrated using Peptide calibration standards (~1000-3200 Da; Bruker Daltonics, Yokohama, Japan).

Database search for protein identification/ clarification and upstream regulator analysis
Proteins were identified by matching the peptide mass fingerprint with the Swiss-Prot protein database using the MAS-COT search engine (Matrix Science, http://www.matrixscie nce.com). Database searches were carried out using the following parameters: taxonomy, Mus musculus; enzyme, trypsin; and allowing 1 missed cleavage. Carbamidomethylation was selected as a fixed modification, and the oxidation of methionine was allowed as a variable. The peptide mass tolerance was set at 0.5 Da, and the significance threshold was set at P < 0.05 probability based values on Mowse Scores (≥ 55). Protein classification by its biological process involved and its molecular function was carried out using the PANTHER (Protein ANalysis THrough Evolutionary Relationships) clarification system (http://www.pantherdb. org/), which is supported by a research grant from the National Institute of General Medical Sciences [Grant GM081084] and maintained by the group led by Paul D. Thomas at the University of Southern California. Upstream regulator analysis was performed using Ingenuity Pathway Analysis (IPA) software (Qiagen).

Statistical analysis
Data are expressed as mean AE SD (n: sample numbers). Statistical analysis was performed using one-way ANOVA with Tukey's multiple comparison test with Prism ver. 5.0c software (GraphPad, La Jolla, CA, USA); P < 0.05 denoted a significant difference.

Protein expression changes in the liver
We first estimated the duration required to obtain organ proteomic responses by fasting using serum biochemistry. One-day (water-only) fasting was sufficient to maintain minimal levels of glucose, insulin, Cpeptide 2, leptin, and resistin; the levels generally matched those in F2 mice, but, in contrast, the accumulation of ketone bodies and GIP was much more apparent in F2 mice (Fig. 1). Our exploratory 2D DIGE analyses did not find apparent alterations in hepatic protein expression in F1 mice (data not shown), and additional (e.g., 3-day) fasting (that may cause acute > 25% body weight loss) was not allowed for ethical reasons in our university. Therefore, we investigated global protein expression in various mouse organs after 2-day fasting.

Protein expression changes in the kidney
Two-day fasting induced 18.6% kidney weight loss with no or mild activation of protein degradation systems in our previous RT-PCR analysis [20]. Proteomic analysis revealed a total of 1633 spots in the representative 2D gel (Fig. 3A), of which 45 (2.76%, red circles) and 44 (2.69%, green circles) spots were > 1.1fold up-and downregulated, respectively (Fig. 3B). We identified only 12 significantly upregulated proteins in the kidney of F2 mice, which included Pck1 (spot 1), Acox1 (spot 3), and apolipoprotein A-I (Apoa1, spot 4), just like in the liver ( Table 2 and Table S1 [Sheet A and B]). We also identified 20 significantly downregulated proteins in the kidney; most of them (except calreticulin [Calr (spot 15)] and endoplasmin [Hsp90b1 (heat shock protein 90 kDa beta member 1, spot 16)]) were not apparently changed in the liver ( Table 2 and  Table S1 [Sheet A and B]). Nevertheless, the upstream regulator analysis revealed PPARa as the second highest scoring upstream regulator after ATF6 (activating transcriptional factor 6) and followed by KLF15 (Kr€ uppel-like factor 15) ( Fig. 3C and Table S2 [Sheet B]).

Protein expression changes in the thymus and spleen
Two-day fasting induced marked weight loss in both the thymus and spleen (54.7% and 41.2%, respectively) although protein degradation systems were only found to be highly activated in the thymus but not in the spleen [20]. In a representative 2D gel, among a total of 1874 spots detected in the thymus (Fig. 4A), 54 (2.88%, red circles) and 67 (3.57%, green circles) spots were found to be up-or downregulated, respectively (Fig. 4B). We identified 10 and 16 significantly up-and downregulated proteins in the thymus, respectively; the former includes several structural proteins such as keratin, collagen, annexin, actin, and moesin ( Table 3 and Table S1 [Sheet C]). Upstream regulators included MYCN (v-myc myelocytomatosis viral-related oncogene, neuroblastoma derived [avian]), TP53, and huntingtin (HTT) in this order ( Fig. 4C and Table S2 [Sheet C]). In a representative 2D gel, among a total of 1861 spots detected in the spleen (Fig. 5A), 42 (2.26%, red circles) and 34 (1.83%, green circles) spots were found to be up-or downregulated, respectively (Fig. 5B). We could identify only five and nine significantly up-and downregulated proteins in the spleen, respectively ( Table 4 and Table S1 [Sheet D]), although the upstream regulators included TP53 with the highest score, followed by nuclear factor of NFKBIA (kappa light polypeptide gene enhancer in B-cell inhibitor, alpha) and RARB (retinoic acid receptor, beta) ( Fig. 5C and Table S2 [Sheet D]).
When differentially expressed proteins (upon fasting) were classified by the biological processes involved and molecular functions using PANTHER software, both classification patterns were quite similar between the liver and kidney and between the thymus and spleen (Fig. 6).

Protein expression changes in the brain and testis
Two-day fasting induced only 2.7% and 9.2% weight losses in the brain and testis, respectively [20]. No apparent protein degradation systems were found to be activated in either organ in our previous RT-PCR analysis [20]. In representative 2D gels, among a total of 2189 spots detected in the brain (Fig. 7A), 360 (16.4%, red circles) and 337 (15.4%, green circles) spots were found to be up-or downregulated, respectively (Fig. 7B), and among a total of 2301 spots detected in the testis (Fig. 7C), 184 (8.0%, red circles) and 249 (10.8%, green circles) spots were found to be up-or downregulated, respectively (Fig. 7D). However, the DeCyder comparative analysis of four AL and four F2 brain (and testis) samples revealed no consistent alterations in protein expression.

Discussion
This study investigated global protein expression changes in six major mouse organs upon 2-day fasting. After 1-or 2-day fasting, blood levels of glucose, insulin, and C-peptide 2 (another component of proinsulin [insulin + C-peptide 2] that has its own physiological properties [25]) were significantly reduced compared with the levels in AL mice (67.1%, 61.3%, and 66.3%, respectively [ Fig. 1]). Moreover, the levels of two adipose-derived peptide hormones, leptin (the  pleiotropic hormone of satiation signals/energy expenditure [26,27]) and resistin (the hormone that may 'resist' insulin actions [28]), were decreased to the same extent in F1 and F2 mice (Fig. 1). In contrast, ketone bodies and GIP became more accumulated in F2 mice ( Fig. 1). High accumulation of ketone bodies may    reflect a progressive systemic energy shift from glucose to ketone bodies. GIP is a polypeptide inhibitor of gastric acid secretion and acts as an affective promotor of insulin secretion [29]; therefore, its elevation could be a delayed compensatory action against low plasma levels of insulin (Fig. 1). These data indicated somewhat altered systemic conditions between F1 and F2 mice. We identified 72, 26, 14, and 32 proteins that were significantly up-or downregulated in the liver, thymus, spleen, and kidney of F2 mice, respectively, but significant expression changes were not found in F1 mice (data not shown). Previous studies demonstrated the pivotal roles of PPARa in mediating adaptive responses to fasting in the liver (Fig. 2C and Table S2 [Sheet A]), as observed in the present study; PPARa positively regulates gluconeogenesis, peroxisomal/mitochondrial b-oxidation, fatty acid transport, and ketogenesis and also negatively regulates glycolysis, amino acid metabolism, and inflammation, by binding to specific nucleotide sequences known as peroxisome proliferator response elements (PPREs) in the promoter regions of target genes [23,24]. PPARa-knockout mice have been shown to display several dysregulated responses such as severe hypoglycemia, hypoketonemia, elevated plasma free fatty acid levels, and fatty liver upon fasting [4,5]. Accordingly, we found hepatic upregulation of enzymes involved in gluconeogenesis (Pck1 and Pc), lipid b-oxidation (Acox1), ketogenesis (Hmgcs2), and downregulation of the enzymes involved in fatty acid synthesis (Fasn) and glycogenolysis (Pygl) (Fig. 2C, Table 1, and Table S1 [Sheet A]). Although all genes encoding these 'upregulated' proteins have been shown to contain PPREs in their promoter regions [30][31][32][33][34] and thus can be activated directly by ligand-bound PPARa, the suppression of Fasn and Pygl expression could be caused by a secondary mechanism such as low plasma levels of insulin [35,36].
PPARa is activated by both endogenous and synthetic ligands; the former includes long-chain polyunsaturated fatty acids and eicosanoids such as leukotriene B 4 and the latter includes fibrates such as fenofibrate, bezafibrate, and clinofibrate, the drugs for the treatment of hypertriglyceridemia [24]. Endogenous ligand activation of PPARa could occur in other PPARa-expressing organs such as kidney and heart [37] because large amounts of free fatty acids enter the systemic circulation during fasting. Indeed, we found PPARa regulation of nine proteins in the kidney of F2 mice (Fig 3C and   Fasting-regulated proteins in the liver, kidney, thymus, and spleen are categorized by 'biological process' or 'molecular function' using PANTHER software. regulators, respectively, but their P-values were much (7-8 orders) higher than PPARa (Fig. 2C and Table S2 [Sheet A]). Expression of PPARc was rather restricted to adipose tissue and the immune systems, and its hepatic expression was shown to be extremely low compared with PPARa in adult rats [37].
The highest scoring regulator in the kidney of F2 mice was ATF6, which could regulate the expression of five proteins (Fig. 3C): Calr, Hsp90b1, Pck1 (spot 1), Acox1 (spot 3), and Cpt2 (carnitine O-palmitoyltransferase 2, mitochondrial; spot 6; an enzyme involved in acyl transfer across the mitochondrial inner membrane for b-oxidation in the matrix) (Fig. 3A,B, Table 2, and  Table S2 [Sheet B]). Although the upregulation of Pck1/Acox1/Cpt2 was common between liver and kidney, the two major organs for both gluconeogenesis and b-oxidation (Figs. 2C and 3C), the downregulation of Calr and Hsp90b1 was rather kidney-specific, which may place ATF6 upstream of PPARa in IPA analysis (Table S2 [Sheet B]). ATF6 is an endoplasmic reticulum stress-regulated transmembrane transcriptional factor that is activated by its proteolytic cleavage with site 1 and site 2 proteases; the resultant cytosolic portion translocates to the nucleus, binds to ER stress response elements, and induces ER stressresponsive genes [38]. Recent evidence indicates that the interruption of hepatocellular autophagy attenuates the ATF6-mediated unfolded protein response [39]; thus, conversely, renal autophagy during fasting might induce ATF6 activation. In addition, KLF15, a member of the Kr€ uppel-like family of transcriptional factors, has been shown to regulate gluconeogenesis and KLF15-deficient mice displayed severe hypoglycemia after overnight fasting [40]; accordingly, 2day fasting induced KLF15 in two gluconeogenic organs, the liver (71st highest score) and the kidney (third highest score) (Table S2 [Sheet A and B]).
The other major finding of this study was TP53mediated transcriptional regulation in the thymus, spleen, and liver; TP53 was listed as the second, first, and fourth highest scoring upstream regulator, respectively (Figs. 4C and 5C; Table S2 [Sheet A, C, and D]). TP53 has been described as 'the guardian of the genome' because it regulates 'thousands [41]' of target genes to prevent genome mutation and is encoded by the most frequently mutated gene in human cancer; however, TP53 also regulates multiple cellular responses including autophagy [42], inflammation, pluripotency, and energy metabolism [43,44]. A recent study mentioned that fasting robustly increases (stabilizes) TP53 in the mouse liver via hepatocyte autonomous and AMP-activated protein kinase-dependent posttranscriptional mechanisms, thereby regulating gluconeogenesis and amino acid catabolism [8]. In addition, TP53-deleted mice became hypoglycemic and showed defective utilization of hepatic amino acids upon fasting [8]. Of note, TP53 regulated somewhat different sets of genes in the thymus and spleen.  (Fig. 5C, Table 4, and Table S1 [Sheet D]). In the spleen, nearly half of the proteins with altered expression during fasting were TP53 target gene proteins, which may place TP53 at the top of the upstream regulator lists (Table S2 [Sheet D]). Furthermore, TP53 regulation of 12 proteins (among 19 proteins identified) was liver-specific (Table S2 [Sheet A]). Fasting/calorie restriction has been also shown to reduce age-related diseases including cancer [45]; and therefore, fastinginduced TP53 regulation could be involved in such systemic tumor suppression.
MYCN and HTT were listed as the first and third highest scoring upstream regulators in the thymus (Fig. 4C), where NFKBIA and RARB were as the second and third highest scoring upstream regulators in the spleen (Fig. 5C). However, they only regulated only 6, 7, 5, and 3 proteins, respectively, in the IPA analysis, and their physiological roles await further investigations. In conclusion, this proteomic study revealed protein remodeling in response to fasting in the mouse liver, kidney, thymus, and spleen that could be transcriptionally regulated by PPARa and/or TP53.
These findings could open new perspectives to understating the systemic effects of single fasting in animal experiments.