KLF10 Integrates Circadian Timing and Sugar Signaling to Coordinate Hepatic Metabolism

bioRxiv preprint doi: https://doi.org/10.1101/2020.12.22.423999; this version posted December 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 KLF10 integrates circadian timing and sugar

2 signaling to coordinate hepatic metabolism

3 Anthony A. Ruberto,1,4 Aline Gréchez-Cassiau,1 Sophie Guérin,1 Luc Martin,1 4 Johana S. Revel,2 Mohamed Mehiri,2 Malayannan Subramaniam,3 Franck 5 Delaunay,1 Michèle Teboul1*

6 1Université Côte d’Azur, CNRS, Inserm, iBV, 06108 Nice, France ; 2Université Côte 7 d’Azur, CNRS, Institut de Chimie de Nice, UMR 7272, 06108 Nice, France; 3Department 8 of Biochemistry and Molecular Biology, Mayo Clinic, Rochester, MN 55905, USA; 9 4Present address: Department of Parasites and Insect Vectors, Institut Pasteur, 75015 10 Paris, France.

11 For correspondence: [email protected]

14 Abstract The mammalian circadian timing system and metabolism are highly interconnected,

15 and disruption of this coupling is associated with negative health outcomes. Krüppel-like factors 16 (KLFs) are transcription factors that govern metabolic homeostasis in various organs. Many KLFs 17 show a circadian expression in the liver. Here, we show that the loss of the clock-controlled KLF10 18 in hepatocytes results in extensive reprogramming of the mouse liver circadian transcriptome, 19 which in turn, alters the temporal coordination of pathways associated with energy metabolism. We 20 also show that glucose and fructose induce Klf10, which helps mitigate glucose intolerance and 21 hepatic steatosis in mice challenged with a sugar beverage. Functional genomics further reveal 22 that KLF10 target genes are primarily involved in central carbon metabolism. Together, these 23 findings show that in the liver, KLF10 integrates circadian timing and sugar metabolism related 24 signaling, and serves as a transcriptional brake that protects against the deleterious effects of 25 increased sugar consumption.

27 Introduction

28 The mammalian circadian timing system aligns most biological processes with the Earth’s light/dark 29 (LD) cycle to ensure optimal coordination of physiology and behavior over the course of the day 30 (Bass and Lazar, 2016). At the organism level, circadian clocks are organized hierarchically: at the 31 top a central pacemaker located in the suprachiasmatic nuclei of the hypothalamus that receives 32 the light input and in turn coordinates peripheral clocks via internal synchronizers including 33 glucocorticoids and body temperature (Saini et al., 2011). The molecular mechanism underlying 34 circadian clocks is an oscillatory gene network present in virtually all cells which translates the 35 external time information into optimally phased rhythms in chromatin remodeling, gene expression, 36 posttranslational modification and metabolites production (Dyar et al., 2018; Mauvoisin and 37 Gachon, 2019; Rijo-Ferreira and Takahashi, 2014; Robles et al., 2017, 2014; Weidemann et al., 38 2018; Zhang et al., 2014). 39 The extensive circadian regulation of processes linked to metabolic homeostasis, as well as 40 the mechanisms by which cellular metabolism feeds back to core clock components, have been 41 highlighted by functional genomics studies (Panda, 2016; Peek et al., 2013; Reinke and Asher, 42 2019; Sinturel et al., 2020). In mice, the disruption of clock genes leads to multiple metabolic 43 abnormalities, including impaired insulin release, glucose intolerance, insulin resistance, hepatic 44 steatosis and obesity (Bugge et al., 2012; Jacobi et al., 2015; Perelis et al., 2015); conversely, diet- 45 induced metabolic imbalance impairs circadian coordination in part by reprogramming the circadian 46 transcriptome (Eckel-Mahan et al., 2013; Peek et al., 2013). Experimental and epidemiological data 47 support a similar reciprocal relationship between circadian misalignment or disruption and 48 metabolic disorders in humans (Qian and Scheer, 2016; Stenvers et al., 2019). 49 The liver is a major metabolic organ in which circadian gene expression is regulated directly 50 by core clock and clock-controlled transcription factors among which nuclear hormone receptors 51 including REV-ERBs, PPARs, GR, FXR, SHP, play a prominent role (Mukherji et al., 2019). 52 Krüppel-like factors (KLFs) form another family of transcription regulators, a majority of which has 53 been involved the physiology of metabolic organs including the liver (Hsieh et al., 2019). Many of 54 these KLFs are also clock-controlled genes in mouse liver (Ceglia et al., 2018; Yoshitane et al., 55 2014). This suggests KLFs as additional factors for the circadian regulation of hepatic metabolism. 56 KLF15 was for instance shown to control the circadian regulation of nitrogen metabolism and bile 57 acid production (Han et al., 2015; Jeyaraj et al., 2012). KLF10, first described as a TGF-β induced 58 early gene in human osteoblasts and pro-apoptotic factor in pancreatic cancer (Subramaniam et

59 al., 1995; Tachibana et al., 1997), has been further identified as a clock-controlled and glucose- 60 induced transcription factor that helps suppress hepatic gluconeogenesis (Guillaumond et al., 61 2010; Hirota et al., 2002). In Drosophila, the KLF10 ortholog Cabut, has a similar role in sugar 62 metabolism and circadian regulation (Bartok et al., 2015), which suggests the function of the 63 transcription factor is evolutionary conserved. 64 A better understanding of the role of KLF10 in hepatic metabolism may have important 65 translational implications in the context of the rising prevalence of nonalcoholic fatty liver disease 66 (NAFLD) and fructose overconsumption (Jensen et al., 2018). In this study, using a hepatocyte- 67 specific Klf10 knockout mouse model, we show that KLF10 is required for the temporal coordination 68 of various biological pathways associated with energy metabolism, and that it has a protective role 69 in shielding mice from adverse effects associated with sugar overload.

70 Results

71 Hepatocyte-specific deletion of KLF10 reprograms the liver circadian transcriptome

72 To understand the role of the circadian transcription factor KLF10 in hepatic metabolism, we 73 generated a conditional Klf10 knockout mouse model by crossing Klf10flox/flox mice (Weng et al., 74 2017) with a mouse line expressing an inducible CreERT2 recombinase under the control of the 75 endogenous serum albumin (SA) promoter (Schuler et al., 2004) (Figure 1A). Upon tamoxifen 76 treatment of Klf10flox/flox, SA+/CreERT2 mice, we obtained a deletion of Klf10 specifically in hepatocytes 77 (Klf10Δhep). Both qPCR and immunoblot analyses confirmed the targeted reduction of the 78 transcription factor in the liver of Klf10Δhep mice, with residual expression in this organ coming from 79 non-parenchymal cells (Figures 1B and 1C). As expected, we detected a robust circadian 80 oscillation of Klf10 mRNA in the liver of Klf10flox/flox mice, with peak expression of the transcript 81 occurring just before the LD transition while in Klf10Δhep mice, rhythmic expression of Klf10 was 82 absent (Figure 1D). Body weight, food intake and food seeking behavior did not differ between 83 genotypes, indicating that post-natal hepatocyte-specific deletion of KLF10 did not grossly affect 84 the physiology and behavior of unchallenged adult Klf10hepmice (Figures S1A and S1B).

86 Figure 1. Genetic disruption of Klf10 in mouse hepatocytes. 87 (A) Schematic illustrating the strategy for generating of Klf10Δhep mice. (B) Relative gene expression of 88 Klf10 across various tissue in Klf10flox/flox and Klf10Δhep mice as measured at ZT9 by RT-qPCR (mean ± 89 SEM, n = 3-5). (C) Immunoblot showing KLF10 protein abundance in liver extracts from Klf10flox/flox and 90 Klf10Δhep mice at ZT9. EF1α was used as loading control. (D) 24h gene expression profiles of hepatic 91 Klf10 mRNA measured every 3 hours in Klf10flox/flox and Klf10Δhep mice (mean ± SEM, n = 3 mice per 92 time point). Statistics: Non parametric Wilcoxon test. * p < 0.05. See also Figure S1. 93 94 95 96 We next performed RNA sequencing (RNA-seq) on livers collected every three hours across 97 the 24h day from Klf10flox/flox and Klf10Δhep mice entrained to a 12h:12h LD cycle (Figure 2A). Using 98 MetaCycle (Wu et al., 2016), we found that ~12% of genes (1,664) displayed circadian expression 99 in Klf10flox/flox mice (Figure 2B; Table S1), a percentage which is consistent with previous reports 100 (Greenwell et al., 2019; Zhang et al., 2014). In Klf10Δhepmice, we detected a similar percentage of 101 genes with circadian expression (~12%; 1,633) (Figure 2B; Table S1). The phase distributions and 102 relative amplitudes of rhythmically expressed genes in Klf10flox/flox and Klf10Δhep mice were similar 103 (Figures 2C and 2D). Furthermore, the 24h expression profiles of the core clock genes in Klf10Δhep 104 mice were unchanged (Figure S2A), suggesting that hepatocyte KLF10 is not required for normal

105 core clock function in the liver. Despite these similarities, less than 40% of genes (602) displayed 106 rhythmic expression in both genotypes (Figures 2E and 2F). Interestingly, we observed rhythmicity 107 of 1,031 transcripts exclusively in Klf10Δhep mice, indicating that the deletion of KLF10, a known 108 transcriptional repressor, results in de novo oscillation of otherwise non-rhythmically expressed 109 transcripts (Figures 2E and 2F). 110 To better understand the changes in the circadian transcriptome associated with the deletion 111 of Klf10 in hepatocytes, we assessed the temporal coordination of biologically related transcripts 112 in Klf10flox/flox and Klf10Δhep mice. After performing Phase Set Enrichment Analysis (PSEA) (Zhang 113 et al., 2016), we found a decrease in the total number of temporally enriched gene sets in Klf10Δhep 114 mice (Figure 2G). Specifically, gene sets associated with processes such as oxidative 115 phosphorylation, fatty acid metabolism, and mTORC1 signaling peaking around the LD transition 116 in Klf10flox/flox mice were absent in Klf10Δhep mice (Figure 2G and Table S2). Despite the global 117 reduction in temporally coordinated transcripts, Klf10Δhep mice acquired de novo temporal 118 enrichment of transcripts associated with glycolysis and cellular stress processes such as 119 apoptosis, UV response and inflammation (Figure 2G, Table S2). The alterations of biological 120 pathways at the transcript level in Klf10Δhep mice prompted us to assess whether any changes 121 associated with these pathways would be visible at the physiological level. Given the scope of our 122 research question, we focused primarily on processes linked to metabolism. We found that Klf10Δhep 123 hepatocytes displayed increased glucose uptake capacity (Figure S2B), and that Klf10Δhep mice 124 had decreased glycogen levels during their resting phase (Figure S2C), suggesting that the loss of 125 Klf10 in hepatocytes leads to subtle physiological changes in carbohydrate processing at both the 126 cellular and tissue level. Together, these findings indicate that dysregulation of the hepatic 127 circadian transcriptome in Klf10Δhep mice results in the rewiring of various pathways related to 128 energy metabolism, which in turn may compromise the ability of these animals to process energy 129 substrates.

130 131 Figure 2. Deletion of hepatocyte KLF10 alters the circadian transcriptome in the liver. 132 (A) Schematic illustrating the workflow used to assess the circadian transcriptomes of livers in 133 Klf10flox/flox and Klf10Δhep mice. (B) Number of rhythmic transcripts detected in the livers from Klf10flox/flox 134 and Klf10Δhep mice. (C) Phase distributions of rhythmic transcripts in the livers of Klf10flox/flox and Klf10Δhep 135 mice. (D) Relative amplitudes of rhythmic transcripts in the livers of Klf10flox/flox and Klf10Δhep mice. (E) 136 Number of unique and overlapping rhythmic transcripts in Klf10flox/flox and Klf10Δhep mice. (F) Heatmap 137 showing the expression profiles of oscillating transcripts in the livers of Klf10flox/flox and Klf10Δhep mice 138 (pools of 3 livers per time point). (G) Magnitudes and phases of enriched biological processes identified 139 by PSEA in Klf10flox/flox mice only (top), in Klf10Δhep mice only (bottom), and in both genotypes (middle). 140 Statistics: MetaCycle, significance threshold, p < 0.05; PSEA; Kuiper test, significance threshold, q < 141 0.01. See also Figure S2 and Tables S1 and S2.

142 Hepatocyte KLF10 minimizes the adverse metabolic effects associated with a high 143 sugar diet

144 Consumption of glucose and fructose in the form of sugar-sweetened beverages is linked to 145 negative health outcomes (Softic et al., 2020). While previous studies have shown that glucose is 146 a potent inducer of Klf10 expression in primary hepatocytes (Guillaumond et al., 2010; Iizuka et al., 147 2011), the effect of fructose is unknown. We now show that hepatocytes treated with 5 mM fructose 148 induces Klf10 to the same extent as high (25 mM) glucose; while the combination of high glucose 149 and fructose has an additive effect (Figure 3A). Given these results in hepatocytes, combined with 150 transcriptional changes occurring in Klf10Δhep mice, we sought to better understand the role of 151 KLF10 in mediating the effect of the two dietary sugars at the organism level. Klf10flox/flox and 152 Klf10Δhep mice were given ad libitum access to a chow diet with water (chow) or a chow diet with 153 sugar-sweetened water (chow + SSW) for 8 weeks (Figure 3B). Similar to the response we 154 observed in hepatocytes, the chow + SSW diet induced hepatic Klf10 expression, when measured 155 during the animals’ feeding phase (ZT15) (Figure 3C). Mice on the chow + SSW diet adapted their 156 feeding behavior, with less calories coming from the chow pellets. As a result, their total calorie 157 intake was unchanged relative to mice fed the chow diet (Figures S3A). After eight weeks on the 158 chow + SSW diet, mice from both genotypes displayed a significant increase in body mass and 159 epididymal fat pad mass (Figures 3D and 3E), while only Klf10Δhep mice displayed an increase in 160 liver mass (Figure 3F). Chow + SSW diet increased liver triglycerides content, and histologic signs 161 of steatosis were evident in both genotypes; however, to a greater extent in Klf10Δhep mice (Figure 162 3G). Despite the greater increase in liver triglycerides in Klf10Δhep mice, no difference in circulating 163 triglycerides was detected between genotypes (Figure S3B). 164 To further profile these mice, we assessed various parameters associated with sugar 165 metabolism during the animals’ early active (feeding) phase (ZT12-15)—a time of day when insulin 166 sensitivity is at its maximum (la Fleur et al., 2001), and when pathways associated with nutrient 167 intake and processing are activated (Greenwell et al., 2019; Vollmers et al., 2009). Compared to 168 Klf10flox/flox mice, Klf10Δhep mice on the chow + SSW diet had a significant increase in glycemia, and 169 displayed hyperinsulinemia (Figure 3H and 3I). Furthermore, Klf10Δhep mice on the chow + SSW 170 diet exhibited greater glucose intolerance compared to Klf10flox/flox mice on the same diet (Figure 171 3J). Next we profiled carbohydrate, amino acid and TCA related metabolite levels in the four 172 experimental groups of mice. We found that hexose-6-phosphate, myo-inositol, UDP-D-hexose and 173 citric acid were increased and that phospho-enol-pyruvate was decreased to a similar extent in 174 both Klf10flox/flox and Klf10Δhep mice on the chow + SSW diet (Figure 3K). In contrast, glutamine was 175 increased only in Klf10flox/flox on the chow +SSW diet, while phenylalanine and fumarate were

176 increased and decreased respectively only in Klf10Δhep mice on the chow + SSW (Figure 3K). These 177 observations reveal subtle changes in amino acid catabolism in mice lacking hepatocyte KLF10.

178 179 180 Figure 3. Loss of hepatocyte KLF10 exacerbates the adverse effects associated with increased 181 sugar consumption. 182 (A) Response of Klf10 in primary mouse hepatocytes challenged with glucose and/or fructose in vitro 183 (mean ±SEM, n = 6). (B) Schematic illustrating the duration and sampling time points of mice

184 undergoing the SSW paradigm. (C) Gene expression of Klf10 in livers of Klf10flox/flox and Klf10Δhep mice 185 given a chow or chow + SSW diet (mean ± SEM, n = 4). (D) Body mass of mice given a chow or chow 186 + SSW diet (mean ± SEM, n = 9-12). (E) Epididymal white adipose tissue mass of mice given a chow 187 or chow + SSW diet (mean ± SEM, n = 9-12). (F) Liver mass of mice given a chow or chow + SSW 188 diet (mean ± SEM, n = 8-12). (G) Liver triglyceride content (mean ± SEM, n = 6-10) (left) and 189 representative images of liver histology (right) in Klf10flox/flox and Klf10Δhep mice given a chow or chow + 190 SSW diet (scale bar = 50 µm). (H) Blood glucose in Klf10flox/flox and Klf10Δhep mice given a chow or 191 chow + SSW diet (n = 9-12). (I) Insulin levels in Klf10flox/flox and Klf10Δhep mice given a chow or chow + 192 SSW diet (mean ± SEM, n = 9-12). (J) Blood glucose levels assessed at regular intervals over a 2 193 hour period in mice undergoing a glucose tolerance test (GTT) performed at ZT 12 (left) and area 194 under the curve of blood glucose levels over the measurement period (right) (mean ± SEM, n = 9-12). 195 (K) Heatmap showing the normalized concentration of metabolites in the liver of Klf10flox/flox and 196 Klf10Δhep mice given a chow or chow + SSW diet (mean, n = 9-12). Significant pairwise comparisons 197 are boxed. Unless otherwise indicated all measurements were performed during the animals’ feeding 198 period, at ZT15. Statistics: Non parametric Kruskal and Wallis test. *p < 0.05, **p < 0.01, *** p<0.005. 199 See also Figure S3. 200 201 We next assessed the effect of the chow + SSW diet at the transcriptional level in Klf10flox/flox 202 and Klf10Δhep mice by comparing the expression of various gene associated with metabolism. We 203 detected an upregulation of Slc2a4, a glucose transporter with known expression in adipose and 204 skeletal muscle (Klip et al., 2019), and normally undetected in the liver (Figure 4A, left). 205 Furthermore, the absence of KLF10 in hepatocytes dramatically increased this upregulation of 206 Slc2a4 in Klf10Δhep mice on chow + SSW compared to Klf10flox/flox mice on the same diet (Figure 207 4A). In contrast, the chow + SSW diet resulted in similar upregulation of glucose (Slc2a2) and 208 fructose (Slc2a5) transporters in both Klf10flox/flox and Klf10Δhep mice (Figure S4). Upon assessing 209 genes involved in the glucose and fructose metabolism pathway, we found that Klf10Δhep mice on 210 the chow + SSW diet displayed higher expression of the gene encoding for the rate-limiting 211 glycolytic enzyme encoding Pklr compared to Klf10flox/flox mice on the same diet (Figure 4A). 212 Furthermore, the chow + SSW diet altered expression of genes encoding for enzymes associated 213 with gluconeogenesis. However, we detected an upregulation of G6Pase in mice from both 214 genotypes on the chow + SSW diet, whereas the expression of Pck1, a direct KLF10 target gene 215 (Guillaumond et al., 2010), was upregulated exclusively in Klf10Δhep mice (Figure 4B). Hepatocyte- 216 specific deletion of KLF10 also altered the transcriptional response of genes involved in de novo 217 lipogenesis, as Klf10Δhep mice fed the chow + SSW diet displayed a significantly greater induction 218 of the genes encoding for Acly, Acc1, Fasn, Elovl6, ME, and Thrsp relative to their Klf10flox/flox 219 counterparts (Figure 4C). 9

220 Taken together, our physiological, metabolic and molecular data suggests that the induction of 221 KLF10 in response to a dietary sugar challenge serves as a protective mechanism that aids in 222 safeguarding the organism from the detrimental effects associated with excess sugar intake. 223

224

225 Figure 4. Altered metabolic gene expression in Klf10Δhep mice challenged with high sugar. 226 (A) Expression profiles of genes associated with glycolysis/fructolysis in the liver of Klf10flox/flox and 227 Klf10Δhep mice given a chow or chow + SSW diet (mean ± SEM, n = 6). (B) Expression profiles of 228 genes associated with gluconeogenesis in the liver of Klf10flox/flox and Klf10Δhep mice given a chow or 229 chow + SSW diet (mean ± SEM, n = 5-6). (C) Expression profiles of genes associated with 230 lipogenesis in the liver of Klf10flox/flox and Klf10Δhep mice given a chow or chow + SSW diet (mean ± 231 SEM, n = 4-6). Statistics: Non parametric Kruskal and Wallis test. *p < 0.05, **p < 0.01, ***, p<0.005. 232 See also Figure S4. 233

234 KLF10 is a ‘transcriptional brake’ that fine-tunes sugar signaling in hepatocytes

235 As the induction of hepatic KLF10 in chow +SSW fed mice helps to minimize the adverse metabolic 236 effects associated with excess dietary sugar intake, we hypothesized that this transcription factor 237 may be involved in the metabolic adaptation of hepatocytes to acute changes in carbohydrate

238 availability. To test this hypothesis in a cell autonomous manner, we treated Klf10flox/flox and Klf10Δhep 239 primary hepatocytes with either 5 mM glucose (low glucose; LG) or 25 mM glucose and 5 mM 240 fructose (high glucose and fructose; HGF) for 12 hours (Figure 5A and 5B) and performed RNA- 241 seq to profile the transcriptomes under each condition. 242 The switch from the LG to the HGF condition lead to a differential expression of 842 and 608 243 genes in Klf10flox/flox and Klf10Δhep hepatocytes, respectively (Figure 5C, Table S3). Notably, the 244 number of downregulated genes was decreased by ~40% in challenged Klf10Δhep hepatocytes 245 compared to Klf10flox/flox hepatocytes, suggesting that the ability of Klf10Δhep hepatocytes to repress 246 gene expression in response to the HGF challenge is compromised (Figure 5C and S5A). Using 247 the significantly altered transcripts detected under the LG and HGF conditions as an input, we 248 performed pathway enrichment analysis and found several differences between Klf10flox/flox and 249 Klf10Δhep hepatocytes (Figure 5C and Table S4). In the LG condition, pathways associated with 250 amino acid catabolism, fatty acid oxidation, and peroxisome metabolism were significantly enriched 251 in Klf10flox/flox but not Klf10Δhep hepatocytes (Table S4). Underlying these changes, Klf10flox/flox 252 hepatocytes displayed an upregulation of genes coding enzymes linked to branched chain amino 253 acids (Mccc2, Ivd, Acat1, Fah, Echs1, Adlh9a1), alanine, lysine and tryptophan (Echs1, Adlh9a1), 254 and, tyrosine (Fah) degradation, as well as genes encoding for enzymes linked to mitochondrial 255 and peroxisomal fatty acid oxidation (Cpt1a, Echs1, Adlh9a1, Decr2) (Figure 5D, 5E and 5F). In 256 the HGF condition, Klf10Δhep hepatocytes displayed a greater number of enriched pathways relative 257 to Klf10flox/flox hepatocytes, with many of these de novo pathways associated with carbohydrate 258 metabolism (Figure 5C, right panel). Similar to our results in mice challenged with the chow + SSW 259 diet, the HGF condition resulted in an upregulation of the glucose transporter Slc2a4 in the 260 Klf10flox/flox and Klf10Δhep hepatocytes, with the absence of KLF10 further augmenting this response 261 (Figure 5E). In a similar manner, Klf10Δhep hepatocytes in the HGF condition displayed greater 262 induction of genes associated with fructose (Khk, Tfkc) and glycogen (Ppp1r3c) metabolism relative 263 to Klf10flox/flox hepatocytes (Figure 5D and 5E). In addition to the enhanced upregulation of genes 264 associated with carbohydrate metabolism, Klf10Δhep hepatocytes had a greater induction of genes 265 associated with de novo lipogenesis (Acly, Acc1, Fasn, Thrsp) and long chain fatty acid elongation 266 (Elovl6) (Figures 5D, 5E, 5F and S5B). 267 Collectively, our in vitro challenge data supports the notion that KLF10 acts as a brake that 268 aids in fine-tuning the hepatocyte’s transcriptional response to glucose and fructose. Furthermore, 269 the dysregulation of various metabolic pathways in hepatocytes lacking KLF10 is consistent with, 270 and may underlie, the abnormal physiological response of Klf10Δhep mice challenged with the chow 271 + SSW diet.

272

273 Figure 5. KLF10 governs the transcriptional response to hexose sugars in hepatocytes. 274 (A) Schematic illustrating the design of the in vitro sugar challenge in Klf10flox/flox and Klf10Δhep 275 hepatocytes treated with low glucose (LG) or high glucose and fructose (HGF). (B) Representative 276 immunoblot showing KLF10 protein abundance in Klf10flox/flox hepatocytes treated with LG or HGF and in 277 Klf10Δhep primary hepatocytes treated with HGF. EF1α was used as loading control. (C) Volcano plots 278 showing changes in gene expression in Klf10flox/flox and Klf10Δhep hepatocytes treated with LG or HGF. 279 Positive fold change values represent genes upregulated under HGF conditions, negative fold change 280 values represent genes upregulated under LG conditions. Genes significantly upregulated (FDR < 0.5) 281 and displaying a fold change (FC) > 2 in HGF (red dots) and in LG (blue dots) conditions. Dashed 282 horizontal lines, FDR = 0.05; dashed vertical lines, FC = 2. Boxed areas highlight the larger number of 283 downregulated genes in Klf10flox/flox compared to Klf10Δhep hepatocytes. (D) KEGG enriched pathways in 284 Klf10flox/flox and Klf10Δhep primary hepatocytes treated with LG or HGF. (E) Expression profiles of genes 285 associated with key metabolic pathways in Klf10flox/flox and Klf10Δhep primary hepatocytes treated with LG 286 or HGF (mean ± SEM, n = 3). (F) A schematic of key metabolic pathways in hepatocytes. Genes 287 significantly upregulated in Klf10Δhep vs. Klf10flox/flox hepatocytes when treated with HGF are highlighted 288 on the schematic. DHAP, dihydoxyacetone phosphate; LCFA, long chain fatty acids; GA, 289 glyceraldehyde; OAA, oxaloacetic acid; PEP, phospho-enol-pyruvate. Statistics: Non parametric Kruskal 290 and Wallis test. *, p < 0.05; **, p < 0.01; ***, p <0.005. See also Figure S5 and Tables S3 and S4. 291

292

293 KLF10 targets an extensive metabolic gene network in the liver

294 To further explore the role of KLF10 in the regulation of the hepatic metabolism at the transcriptome 295 level, we assessed the genome-wide occupancy of KLF10 in mouse liver at the time of maximal 296 expression of the protein (ZT9, Figure S6) using chromatin immunoprecipitation followed by high 297 throughput sequencing (ChIP-seq) (Figure 6A). Upon filtering the list of KLF10-bound fragments 298 for the presence of at least one KLF10 binding motif (Schmitges et al., 2016), we obtained a 299 repertoire of KLF10 35,523 KLF10 binding sites present between in the -50 kb and +1 kb region of 300 associated genes (Figure 6B). The density of KLF10 binding sites was negatively correlated with 301 their distance from the transcription initiation site with 37 % being located in the -10 kb and +1 kb, 302 thus indicating that KLF10 mainly binds proximal promoter regions (Figure 6B). We next integrated 303 expression data for genes differentially expressed in Klf10Δhep vs Klf10flox/flox hepatocytes in the HGF 304 condition (see Figure 5), ChIP-seq data for genes containing at least one binding site in their -10 305 kb to +1 kb promoter region and HumanCyc metabolic pathways (Romero et al., 2005) in a single 306 network visualized in Cytoscape. In this network containing 795 nodes and 10,980 edges, we found

307 that genes regulated or/and bound by KLF10 were present in 23 annotated subnetworks (Figure 308 6C, Table S5). However, half of these genes clustered in three major processes associated with 309 lipid and amino acid metabolism as well as protein phosphorylation (Figure 6C, Cytoscape file S1). 310 The intersect between differentially expressed genes in HGF and KLF10 bound genes identified 311 93 direct targets (Figure 6D). This list was primarily enriched for genes associated with carboxylic 312 acid metabolism, suggesting a role for KLF10 in central carbon metabolism (Figure 6D). In line with 313 this observation, two of the top ranked direct target genes included acyl-CoA synthetase short chain 314 family member 2 (Acss2) and acetyl-CoA carboxylase beta (Acacb), which encode two key 315 enzymes of acetyl-CoA metabolism (Table S5). ACSS2 converts acetate to acetyl-CoA, while 316 ACACB carboxylates acetyl-CoA to malonyl-CoA which is both a negative regulator of fatty acid 317 oxidation through inhibition of CPT1a mediated fatty acid uptake within the mitochondria and the 318 substrate of FASN for palmitate synthesis (Figure 6E). ChIP experiments targeting the KLF10 319 binding sites identified within the Acss2 and Acacb genes using the ChIP-seq profiling experiment 320 confirmed that both genes were bound by KLF10 at ZT15 in the liver of Klf10flox/flox mice on chow + 321 SSW diet (Figure 6E). Consistently these two genes showed a greaer upregulation in the liver of 322 Klf10Δhep mice on chow + SSW diet compared to Klf10flox/flox mice on the same diet, demonstrating 323 that KLF10 is a transcriptional repressor of these two genes upon induction by high sugar (Figure 324 6F). Interestingly, ACSS2 is one of the highly connected nodes in the KLF10 metabolic network, 325 suggesting that KLF10 dependent repression of Acss2 may indirectly impact multiple components 326 of hepatic energy metabolism (Cytoscape file S1 and table S5). We conclude from this data that 327 KLF10 has an extensive repertoire of metabolic targets in the liver that include key regulators of 328 acetyl-CoA metabolism.

329

330 331 Figure 6. KLF10 regulates a large metabolic network in the liver. 332 (A) Schematic illustrating the workflow used to identify KLF10 bound loci in mouse liver at ZT9. (B) 333 Jaspar logo representing the KLF10 response element DNA sequence and distribution of KLF10 binding 334 sites in mouse liver as a function of distance to the transcription start site. (C) Cytoscape metabolic 335 network integrating RNA-seq data from Klf10flox/flox and Klf10Δhep hepatocytes treated with HGF and 336 ChIP-seq data. (D) Venn diagram showing the number of direct KLF10 target (top) and their associated 337 GO biological process annotation (bottom). (E) ChIP-seq tracks near the Acss2 and Acacb promoters

338 (left) and ChIP of the boxed region at ZT15 in the liver of Klf10flox/flox mice fed the chow + SSW diet 339 (right) (mean ± SEM, n = 4-6). (F) Expression profiles of Acss2 and Acacb at ZT15 in the liver of 340 Klf10flox/flox and Klf10Δhep mice given a chow or chow + SSW diet (mean ± SEM, n = 6). Statistics: Non 341 parametric Kruskal and Wallis test. *, p < 0.05; **, p < 0.01; ***, p <0.005. See also Figure S6, Table S5 342 and Cytoscape file S6.

343 Discussion

344 There is newfound appreciation of KLFs as transcriptional regulators of energy metabolism (Hsieh 345 et al., 2019). This role is underscored by the identification and KLF variants involved in 346 cardiometabolic disorders (Oishi and Manabe, 2018). In the liver, several Klf genes are direct 347 targets of the CLOCK transcription factor (Yoshitane et al., 2014), suggesting that timed expression 348 of these transcription factors may help facilitate the connection between the circadian clock and 349 metabolism (Jeyaraj et al., 2012; Panda, 2016; Reinke and Asher, 2019; Sinturel et al., 2020). 350 In this study, we show that KLF10 is required for the circadian coordination of biological 351 processes associated with energy metabolism in the liver while dispensable for an intact hepatic 352 clock function. This confirms our initial hypothesis that KLF10 acts primarily a clock output 353 (Guillaumond et al., 2010). Alteration of the rhythmic hepatic transcriptome in Klf10Δhep mice also 354 led to de novo oscillation for a large number of genes, reminiscent of the transcriptional changes 355 occurring in KLF15 deficient cardiomyocytes (Zhang et al., 2015, p. 15). In addition to the genetic 356 disruption of clock-controlled KLFs, a variety of other systemic perturbations including high fat and 357 ketogenic diet challenges, arrhythmic feeding, calorie restriction, alcohol consumption, lung cancer 358 and rheumatoid arthritis also generate de novo circadian oscillations in the liver (Eckel-Mahan et 359 al., 2013; Gaucher et al., 2019; Greenwell et al., 2019; Makwana et al., 2019; Masri et al., 2016; 360 Poolman et al., 2019; Tognini et al., 2017). Unchallenged Klf10Δhep mice appear healthy but are 361 prone to develop hepatic steatosis when fed with high sugar while Klf10-/- mice develop greater 362 liver injury in a non-alcoholic steatohepatitis (NASH) model (Leclère et al., 2020). We therefore 363 interpret the de novo oscillation seen in Klf10Δhep mice as the signature of their hepatic vulnerability. 364 In line with this assumption, we identified apoptosis and inflammation related processes as gene 365 sets gaining circadian coordination in Klf10Δhep mice. It is presently unknown whether the extensive 366 reprograming of the hepatic circadian transcriptome seen in many genetic or disease mouse 367 models also occurs in humans. Should this occur, it may have important biomedical implications if 368 diagnostic biomarkers or/and regulators of drug pharmacology that are expressed at low levels and

369 arrhythmic in healthy subjects would become circadian upon disease initiation and progression, 370 advocating for the integration of circadian timing in translational research (Cederroth et al., 2019). 371 Hepatocyte KLF10 furthermore operates on several levels to protect the liver from the negative 372 effects of excess dietary sugars. We found that it suppresses glucose uptake and we obtained 373 correlative evidence that this could be achieved by preventing misexpression of the adipose and 374 muscle specific Glut4 glucose transporter in hepatocytes. Previous studies showed that Klf10 is a 375 glucose-responsive gene (Guillaumond et al., 2010; Hirota et al., 2002), a finding which together 376 with our recent observation strongly suggests that KLF10 directs a negative feedback loop limiting 377 excessive glucose intake. Consistent with this role, glycolytic gene expression was upregulated in 378 Klf10Δhep mice. In addition to glucose, we show that Klf10 is induced by fructose, and the 379 combination of these two hexose sugars amplifies this response. This upregulation is crucial in 380 blunting the negative effects associated with sugar intake as Klf10Δhep mice challenged with a 381 sugar-sweetened beverage containing glucose and fructose leads to greater signs of insulin 382 resistance and hepatic steatosis. Our functional genomics data indicates that KLF10 represses the 383 lipogenic gene network in liver. During the metabolic adaptation of the liver to high sugar, the 384 glucose inducible factor carbohydrate-responsive element-binding protein (ChREBP) plays a 385 critical role by inducing lipogenic genes (Ma et al., 2006; Ortega-Prieto and Postic, 2019). 386 Interestingly, the induction of Klf10 is also mediated by ChREBP (Iizuka et al., 2011). We therefore 387 hypothesize a scenario in which sugar induced steatosis is attenuated by KLF10 via a double action 388 taking place upstream and downstream of ChREBP through repression of glucose intake and de 389 novo lipogenesis respectively. Further, fructose is also converted by the gut microbiota to acetate 390 which fuels ACLY independent lipogenesis upon conversion to acetyl-coA by liver ACSS2 (Zhao 391 et al., 2020, 2016). Our finding that KLF10 is a transcriptional repressor of Acss2 suggests an 392 additional mechanism used by KLF10 to reduce the impact of dietary fructose. Our findings are 393 directly relevant to human health as the increased consumption of fructose in the form of sweetened 394 beverages and processed food unarguably contributes to the currently uncontrolled pandemic of 395 obesity and NAFLD (Hannou et al., n.d.; Jensen et al., 2018; Mortera et al., 2019). Links between 396 the circadian clock and NALFD have been suggested yet they are mostly based on preclinical 397 models with disrupted core clock genes, a situation that is unlikely in humans (Mazzoccoli et al., 398 2018; Mukherji et al., 2019). More presumably, the extensive rewiring of circadian gene expression 399 associated with many pre-disease or disease states and, leaving an intact clock while altering the 400 expression of clock-controlled regulators such as KLF10 may disconnect circadian timing from 401 metabolism and thereby contribute to the development of chronic liver disease including NAFLD.

402 In summary, our work defines KLF10 as a dual transcriptional regulator with a circadian arm 403 participating in the circadian coordination of liver metabolism and a homeostatic arm serving as a 404 negative feedback control of sugar induced lipogenesis. This duality is likely to be coordinated in 405 healthy hepatocytes so that KLF10 peaks at the fasting/feeding transition when sugar production 406 or intake increases. Desynchrony between circadian timing and feeding as observed in metabolic 407 disorders may therefore challenge KLF10 function with associated negative health outcome.

408

409 Acknowledgements

410 This study was supported by French Agence Nationale pour la Recherche (ANR): #ANR-15-CE14- 411 0016-01, #ANR-18-CE14-0019-02 and by the "Investments for the Future" LABEX SIGNALIFE 412 (#ANR-11-LABX-0028-01), the UCAJEDI Investments in the Future project (#ANR-15-IDEX-01) and 413 ATER grant from Université Côte d’Azur (JSR). We thank the Canceropôle Provence-Alpes-Côte 414 d’Azur, and the Provence-Alpes-Côte d’Azur Region for the financial support provided to the 415 MetaboCell project. We thank Pierre Chambon for the gift of SA-CreERT2 mice. We thank Aurélie 416 Biancardini, Angelo Barnabeo, Pierre Chérel, Antoine Landouar and the staff of the mouse facility 417 for assistance with the animal care. We thank Sama Rekami for the help with the histopathology 418 and the PFTC facility for technical support.

419 Authors contributions

420 Conceptualization, FD and MT; Project administration, FD; Supervision, FD and MT; Methodology, 421 LM, FD and MT; Investigation, AAR, AGC, SG, LM, JSR, MM, FD and MT; Writing-original draft, 422 AAR, FD and MT; Writing-review and editing, AAR, MS, FD and MT; Resources, MS, SG, LM, FD 423 and MT; Funding acquisition, FD, MS and MT.

424 Declaration of interest

425 The authors declare no competing interests.

426

427 Materials and methods

428 Key resources table Reagent type Designation Source or Identifiers Additional (species) or reference information resource strain, strain Klf10flox/flox Weng et al, N/A Used males for background (Mus 2017 experiments musculus) strain, strain SA-CreERT2 Schuler et N/A background (Mus al, 2004 musculus) strain, strain Klf10Δhep This study N/A Used males for background (Mus experiments musculus) cells (Mus musculus) Klf10flox/flox hepatocytes This study N/A Used 150,000- 250,000 cells per plate as indicated

cells (Mus musculus) Klf10Δhep hepatocytes This study N/A Used 150,000- 250,000 cells per plate as indicated

biological sample Liver, heart, kidney, This study N/A (Mus musculus) spleen, lung, white adipose tissue biological sample Blood This study N/A (Mus musculus) antibody Human anti-KLF10 mAb CDI Cat. #m14-355 WB (1:800) Laboratories antibody Mouse anti-EGF1a mAb Upstate Cat. #05-235 WB (1:1000) signaling solutions antibody Goat anti-mouse IgG Sigma Cat. #A4416 WB (1:40,000) commercial assay or Glucose Hexokinase Sigma Cat. #GAHK20- kit Assay Kit 1KT commercial assay or PowerUp SYBR green Applied Cat. #A25779 kit Master Mix Biosystems commercial assay or QIAquick PCR Qiagen Cat. #2810 kit Purification Kit commercial assay or Triglycerides Cayman Cat. #10010303 kit colorimetric assay kit commercial assay or NucleoSpin RNA Macherey- Cat. kit Nagel #740955.250 commercial assay or Rneasy mini kit QIAGEN Cat. #74104 kit commercial assay or NEBNext Ultra RNA New Cat. #E7530S kit Library Prep Kit for England Illumina Biolabs commercial assay or QubitTM dsDNA HS Invitrogen Cat. #Q32854 kit Assay Kit

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.22.423999; this version posted December 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

commercial assay or Glucose Uptake-Glo Promega Cat. #J1341 kit Assay commercial assay or Mouse Insulin Elisa Mercodia Cat. #10-1247-01 kit chemical compound, DMEM, low glucose, Gibco Cat. #21885025 drug GlutaMAX™ Supplement, pyruvate

chemical compound, DMEM, high glucose, Gibco Cat. #10569010 drug GlutaMAX™ Supplement, pyruvate

chemical compound, DMEM, glucose-free Gibco Cat. #11966025 drug chemical compound, William's E medium Gibco Cat. #22551022 drug chemical compound, Dexamethasone Sigma Cat. #D0700000 Used 10 nM final drug concentration chemical compound, Insulin-selenium- Gibco Cat. #41400045 Used 1 X final drug transferrin mix concentration chemical compound, Insulin (human, Thermo Cat. #12585014 Used 860 nM fina drug recombinant zinc) Fisher lconcentration Scientific chemical compound, Percoll GE Cat. #17089101 drug Healthcare chemical compound, Protein G dynabeads Thermo Cat. #10003D drug Fisher Scientific chemical compound, Complete protease Roche Cat. drug inhibitor cocktail #11836153001 chemical compound, Collagenase CLS3 Worthington Cat. Used 500 U/mL final drug #WOLS04180 concentration chemical compound, Collagenase CLS4 Worthington Cat. Used 500 U/mL final drug #WOLS04186 concentration chemical compound, Amyloglucosidase Sigma Cat. Used 2 mg/mL final drug #11202332001 concentration chemical compound, SuperScript II Reverse Invitrogen Cat. #18064014 Used 200 U per drug Transcriptase reaction chemical compound, Micrococcal nuclease Thermo Cat. #88216 Used 80 U per drug Fisher reaction Scientific chemical compound, Protein G dynabeads Thermo Cat. #10003D drug Fisher Scientific software, algorithm R Bioconducto https://www.bioc r onductor.org/ software, algorithm Trimmomatic Bolger et al., http://www.usade 2014 llab.org/cms/

software, algorithm Fastqc Babraham https://www.bioin Institute formatics.babrah am.ac.uk/project s/fastqc/ software, algorithm Fastq Groomer Galaxy https://sourceforg project e.net/projects/fas tqgroomer/ software, algorithm Bowtie2 Langmead https://github.co and m/BenLangmead Salzberg, /bowtie2 2012 software, algorithm BEDtools Quinlan and https://github.co Hall, 2010 m/arq5x/bedtools 2 software, algorithm Rsubread Liao et al., https://git.biocon (featureCounts) 2014) ductor.org/packa ges/Rsubread software, algorithm MetaCycle Wu et al., https://github.co 2016 m/gangwug/Meta CycleApp software, algorithm Phase Set Enrichment Zhang et al., https://github.co Analysis 2016 m/ranafi/PSEA software, algorithm MACS2 Feng et al., https://pypi.org/pr 2012 oject/MACS2/ software, algorithm GREAT McLean et http://great.stanf al., 2010 ord.edu/public/ht ml/splash.php software, algorithm FIMO Bailey et al., http://meme- 2015 suite.org/tools/fi mo software, algorithm FastHeinz Beisser et https://www.bioc al., 2010 onductor.org/pac kages/release/bi oc/html/BioNet.ht ml software, algorithm clusterMaker2 Morris et al. http://www.rbvi.u 2011 csf.edu/cytoscap e/clusterMaker2/ software, algorithm Bingo Maere et al , https://github.co 2005 m/cytoscape/BiN GO software, algorithm Cytoscape Shannon et https://cytoscape al, 2003 .org/ other Qubit fluorometer Thermo Cat. #Q33238 Fisher Scientific other FreeStyle InsuLinx Abbott N/A glucometer other Laboratory rodent chow Safe diets Cat. #R03-25 diet other Apistar syrup (organic) ICKO Cat. #HC406

429

430 Mouse models, housing, and diets

431 To generate a hepatocyte-specific conditional Klf10 null allele, mice bearing a Klf10 exon1 flanked 432 by LoxP sites (Klf10flox/flox) described previously (Weng et al., 2017) were crossed with a serum 433 albumin (SA) driven Cre-ERT2 mouse line (Schuler et al., 2004). To obtain mice with a loss of KLF10 434 in hepatocytes (Klf10Δhep), 5 week old male Klf10flox/flox; SA+/CreERT2 mice were gavaged with 435 tamoxifen (10 mg/kg body weight in 200 µl sunflower oil). For all experiments, aged-matched, 436 tamoxifen treated, Klf10flox/flox male mice were used as controls. Genotyping was performed by PCR 437 using the primers listed in Supplementary Table S6. 438 Mice were housed in a temperature- and humidity-controlled room (21±2 °C / 70%), maintained 439 on a 12h:12h LD cycle and fed a standard laboratory chow diet (SafeDiets A03) ad libitum. For the 440 in vivo sugar challenge, mice from each genotype were randomized between the experimental 441 groups and had ad libitum access to standard laboratory chow diet and either water or sugar 442 beverage (water + 30 % Apistar solution [35% fructose, 33% glucose, 32% sucrose]) for 8 weeks. 443 One animal that did not gain weight after eight weeks was excluded. 444 All animal studies were approved by the local committee for animal ethics Comité Institutionnel 445 d'Éthique Pour l'Animal de Laboratoire (CIEPAL-Azur; Authorized protocols: PEA 244 and 557) 446 and conducted in accordance with the CNRS and INSERM institutional guidelines.

447 Primary hepatocyte isolation and culture conditions

448 Primary hepatocytes were isolated using a retrograde in situ perfusion procedure as described 449 previously (Mederacke et al., 2015) with a few modifications. Briefly, mice were anesthetized 450 between ZT3-ZT6, and upon cannulation via the inferior vena cava, the liver was perfused at a flow 451 rate of 6 mL/min with a warmed (37°C), EGTA solution. After 3 minutes, the buffer was exchanged

452 for a glucose- and EGTA-free buffer containing 45 mM CaCl2, collagenase types CLS3 (500 U/mL) 453 and CLS4 (500 U/mL) (Worthington Biochemical Corporation) and the liver was perfused in this 454 buffer for an additional 5 min at a flow rate of 6 mL/min. After perfusion, the liver was excised and 455 placed in Leibovitz L-15 medium supplemented with 10 % fetal calf serum (FCS), 1% penicillin and 456 streptomycin (P/S). Hepatocytes were released by mechanical disruption and filtered through a 70 457 µm cell strainer and resuspended in 50 mL of Leibovitz L-15, 10% FCS, 1% P/S medium. To seed 458 hepatocytes, the solution was centrifuged at 400 rpm for 2 min and hepatocytes were resuspended 459 in 8 mL William’s E medium supplemented with 860 nM insulin, 1% P/S, 10% FCS, and 1% 460 glutamate. Dead cells were removed by Percoll density-gradient centrifugation. Unless otherwise 461 indicated, hepatocytes were seeded in collagen-coated 6-well plates at a density of 250,000 462 cells/plate. After hepatocyte attachment, medium was substituted for low glucose DMEM,

463 supplemented with GlutaMAX, pyruvate, 10 nM dexamethasone (Dex), and 1X insulin-selenium- 464 transferrin (ITS). All hepatocyte manipulations took place the following day, ~12-15 h after cell 465 attachment.

466 Food seeking behavioural analysis

467 Mice were individually housed with ad libitum access to food and water. An infrared detection 468 system (Actimetrics) was used to assess food seeking behavior. Mice were weighed weekly and 469 their daily food intake was estimated by measuring the difference between the quantity of food 470 provided and food remaining after 1 week. Actogram plots were generated using the ImageJ plugin 471 “ActogramJ” (Schmid et al., 2011).

472 Blood analysis

473 Unless stated otherwise, all measurements were performed at ZT 15. Glycemia levels were 474 measured using a FreeStyle Papillon glucometer (Abbott). Plasma triglycerides were measured 475 using a colorimetric assay kit (Cayman) according to the manufacturer’s protocol. Plasma insulin 476 was measured using a Mouse Insulin Elisa kit (Mercodia) in accordance with the manufacturer’s 477 protocol.

478 Glycogen assay

479 Hepatic glycogen was measured as described previously (Feillet et al., 2016). Hepatic glycogen 480 was extracted from samples collected at ZT3 and ZT15. Upon precipitation, the glycogen pellet 481 was either resuspended in 2 mg/mL amyloglucosidase (Sigma-Aldrich) or in 0.2 M sodium acetate 482 (pH 4.9) to measure total glucose and free glucose, respectively. Total and free glucose was 483 assayed using a glucose-hexokinase assay kit (Sigma-Aldrich) according to the manufacturer’s 484 protocol.

485 Glucose tolerance test

486 Mice were fasted at ZT6 for 6h and then injected intraperitoneally with glucose (0.5g/kg body 487 weight) in 1X PBS. Blood glucose was measured from tail-blood microsamples (<0.5 μL) just before 488 injection and at regular intervals over 2 h using a FreeStyle Papillon glucometer (Abbott).

489 Liver triglycerides

490 Total lipids were extracted from liver samples collected at ZT15 using a modified Folch method 491 (Folch et al., 1957). Briefly, 50-100 mg of liver were homogenized in a methanol/chloroform (2:1,

492 v/v) solution. After addition of 1 volume of chloroform and 0.9 volume of water followed by mixing, 493 the organic phase was separated, evaporated and resuspended in NP40. Triglycerides content 494 was measured using a colorimetric assay kit (Cayman) according to the manufacturer’s protocol.

495 Histology

496 Liver tissue was excised, fixed with 4 % PFA solution overnight, processed for paraffin embedding, 497 sectioned at 5 μm thickness, and then stained for hematoxylin and eosin (H&E). Liver sections 498 were then imaged using the Vectra Polaris digital slide scanner (CLS143455, Akoya Biosciences 499 Inc) and subjected to blind scoring.

500 Glucose uptake

501 Primary hepatocytes from Klf10flox/flox and Klf10Δhep were seeded in collagen-coated 24-well plates 502 at a density of 150,000 cells per well and incubated overnight in DMEM containing 25 mM glucose, 503 17 nM ITS and 10 nM Dex. The following day, cells were washed twice with 1x PBS before 504 stimulation with a glucose-free DMEM containing 17 nM ITS and 1 µM 2-deoxyglucose (2-DG). 505 Hepatocytes were left in this medium for 10 min before the reaction was stopped. Lysates were 506 then assayed for glucose uptake using the Glucose Uptake-Glo Assay kit (Promega) as per the 507 manufacturer’s protocol. Relative light unit (RLU) values were normalized to the total protein 508 concentration of each corresponding sample.

509 Liver extract metabolite identification

510 Liver metabolites were extracted following a protocol adapted from Hui et al., 2017. Briefly, 10 511 µL/mg liver of -20°C methanol:acetonitrile:water (40:40:20, v/v/v) solution was added to 512 approximately 50 mg of liver. The mixture was crushed using a pellet pestle for 30 s, followed by 513 sonication for 5 min and vortexing for 15 s. Samples were incubated at 4°C for 10 min to precipitate 514 proteins before centrifugation at 4°C and 15,000 g for 10 min. The supernatant was transferred to 515 LC-MS vials for analyses. 516 Liver crude organic extracts were analyzed using a Vanquish UHPLC coupled with a Thermo 517 Q-Exactive (Thermo Fisher Scientific GmbH, Bremen, Germany) mass spectrometer (MS) and an 518 ESI source operated with Xcalibur (version 2.2, ThermoFisher Scientific) software package. 519 Metabolites separation was achieved on a Waters Acquity BEH C8 (1 x 150 mm, 1.7 µm) column 520 with an injection volume of 5 µL and a flow rate of 0.1 mLmin−1. The mobile phase was composed

521 of octylammonium acetate 4 mM in H2O adjusted to pH = 4.6 with acetic acid (phase A) and 522 methanol (phase B) and the following gradient was used: 1% B for 5 min, 1 to 95 % B in 10 min,

523 95 to 99% B in 5 min, 99% B held for 3 min, then 1% B for 7 min for a total run time of 30 minutes. 524 Data acquisition was realized under full scan switch (positive and negative) mode ionization from 525 m/z 80 to 900 at 70,000 resolution followed by inclusion list dependent MS/MS. Samples were 526 randomized and analyzed in triplicates. A quality control (QC) sample was added by pooling equal 527 volumes of each sample to monitor suitability, repeatability and stability of the system and injected 528 every 10 samples. An extraction blank sample was also injected every 10 runs in order to monitor 529 sample cross-contamination. Pure commercial standards (Fumaric acid, succinic acid, oxaloacetic 530 acid, alpha-ketoglutaric acid, glutamic acid, phosphoenolpyruvic acid, dihydroxy acetone 531 phosphate, citric acid, NADHP and pyruvic acid, Merk) were analyzed to monitor their retention 532 time, m/z of the parent ion and fragment ions. Feature extraction, compound identification and 533 retention time correction were performed using Compound Discoverer 2.1 software (Thermo Fisher 534 Scientific). Processing of the filtered feature table was realized using Metaboanalyst 535 (www.metaboanalyst.ca) (Chong et al., 2019). Samples were normalized using QCs, data were log 536 transformed then Pareto-scaled (mean-centered and divided by the square root of the standard 537 deviation of each variable). The QC were then removed from the normalized data table to perform 538 statistical analyses.

539 In vitro stimulation of hepatocytes with glucose and fructose

540 After seeding of Klf10flox/flox and Klf10Δhep hepatocytes, medium was replaced with DMEM 541 supplemented with GlutaMAX, pyruvate, 10 nM Dex, 1X ITS and containing either (i) 5 mM glucose; 542 (ii) 25 mM glucose; (iii) 5mM glucose and 5mM fructose; or (iv) 25 mM glucose and 5 mM fructose. 543 After a 12 h incubation period, media was aspirated and hepatocytes were lysed with Qiagen RLT 544 buffer. Lysed material containing RNA was collected, flash frozen and stored at -80ºC until RNA 545 extraction.

546 RNA isolation and quantitative real time PCR

547 Total RNA was purified using RNeasy extraction mini kit (Qiagen) as per the manufacturer’s 548 instructions. To generate cDNAs, 1 µg of total RNA was reverse transcribed using random hexamer 549 primers and Superscript II reverse transcriptase (Invitrogen). Diluted cDNAs (1/10) were added to 550 a PowerUp SYBR green master mix (Applied Biosystems) and amplified using a StepOnePlus 551 Real-Time PCR system (Applied Biosystems) as per the manufacturer’s instructions. The relative 552 mRNA abundance was calculated using the ΔCt method and normalized to the value of the 553 expression of the housekeeping gene 36B4. The sequences of primers used are listed in 554 supplementary Table S6.

555 Western blotting

556 Whole cell extracts from liver tissue or isolated hepatocytes were processed using the Active Motif 557 Nuclear Extraction kit as per manufacturer’s instructions. Lysates were separated by SDS–PAGE 558 and transferred to PVDF membranes (Bio-Rad). Immunoreactive protein was detected using ECF 559 reagent (Millipore) and Fusion FX7 (Vilber). The primary antibodies used were human anti-KLF10 560 mAb (1:800), and mouse anti-EGF1a mAb (1:1000). The secondary antibody used was goat anti- 561 mouse IgG (1:40000).

562 RNA-sequencing

563 For the circadian time course experiment, equal amounts of total RNA extracted from 3 mouse 564 livers were pooled at each of the 8 time points assessed over a 24h period. Preparation of libraries 565 and sequencing was performed at UCAGenomiX (Sophia-Antipolis, France). Here, libraries were 566 prepared using the Truseq Stranded Total RNA Preparation Kit with Ribo-Zero Gold (Illumina) 567 following the manufacturer’s recommendations. Normalized libraries were multiplexed, loaded on 568 a flow cell (Illumina) and sequenced on a NextSeq 500 platform (Illumina) using a 2x75bp paired- 569 end (PE) configuration. Image analysis and base calling were conducted by the HiSeq Control 570 Software (HCS). Raw sequencing data (.bcl files) generated from Illumina NextSeq 500 platform 571 was converted into fastq files and de-multiplexed using Illumina's bcl2fastq 2.17 software. 572 For the in vitro sugar challenge experiment, RNA was extracted from Klf10flox/flox and Klf10Δhep 573 hepatocytes treated with either 5 mM glucose or 25 mM glucose and 5 mM fructose for 12 hours 574 (3 biological replicates per experimental condition). Preparation of libraries and sequencing was 575 performed at GENEWIZ (Leipzig, Germany). Here, libraries were prepared using NEBNext Ultra 576 RNA Library Preparation Kit for Illumina (New England Biosystems) following the manufacturer’s 577 recommendations. Normalized libraries were multiplexed, loaded on a flow cell (Illumina) and 578 sequenced on a HiSeq 4000 platform (Illumina) using a 2x150 PE configuration. Image analysis 579 and base calling were conducted by the HiSeq Control Software. Raw sequencing data (.bcl files) 580 generated from Illumina HiSeq 4000 platform was converted into fastq files and de-multiplexed 581 using Illumina's bcl2fastq 2.17 software.

582 Chromatin immunoprecipitation

583 Snap frozen liver (~300 mg) was cross-linked with 1.1% formaldehyde in PBS for 20 min at room 584 temperature and cross-linking was quenched with 125 mM glycine for 5 min. After Dounce 585 homogenization, the crushed tissue was washed twice in cold PBS, resuspended in cold cell lysis 586 buffer (50 mM HEPES [pH 8], 200 mM NaCl, 20 mM EDTA, 12% glycerol, 1% NP40, 0.8% Triton-

587 X100, 1 mM PMSF, protease inhibitor cocktail), and sonicated 3x (15 sec on/30 sec off cycle, low 588 power mode) in a Bioruptor (Diagenode). Nuclei were purified by sedimentation at 1,000 rpm for 5 589 min, the supernatant was removed and the pellet was resuspended in nuclei preparation buffer 590 (0.34 M sucrose, 15 mM Tris [pH 8], 15 mM KCl, 0.2 mM EDTA, 0.2mM EGTA, protease inhibitor 591 cocktail). To digest chromatin, the suspension was incubated with 80 Units micrococcal nuclease 592 (MNase) for 10 min at 37°C. Next, the suspension was spun at 6,600 rpm for 5 min, the supernatant 593 was discarded, and nuclei were resuspended in nuclei lysis buffer (50 mM Tris [pH8], 150 mM 594 NaCl, 10 mM EDTA, 0.5% NP40, 1% Triton-X100, 0.4% SDS and protease inhibitor cocktail). The 595 sample was sonicated for 12x (30 sec on/30 sec off, high power mode) in a bioruptor to release 596 digested chromatin. The suspension containing the fragmented chromatin was centrifuged at 597 10,000 rpm for 8 min at 4°C to remove debris and the supernatant was diluted in ChIP dilution 598 buffer (50 mM Tris pH8, 150 mM NaCl, 1% Triton-X100, protease inhibitor cocktail). Protein-DNA 599 complexes were immunoprecipitated using 3-4 µg of human anti-KLF10 mAb (CDI laboratories) 600 incubated at 4°C overnight and subsequently isolated with protein G Dynabeads for 2 h at room 601 temperature. Beads were washed 4 times in low salt buffer (50 mM Tris [pH7.4], 5 mM EDTA, 150 602 mM NaCl, 1% Triton X-100, 0.5% NP40), 4 times in high salt wash buffer (100 mM Tris [pH7.4], 603 250 mM NaCl, 1% NP40, 1% sodium deoxycholate), once in TE buffer (50 mM Tris [pH7.4], 10 mM

604 EDTA) and then eluted with 1% SDS-0.1 M NaHCO3. Reverse crosslinking of protein-DNA 605 complexes was performed overnight by incubation at 65 °C followed by proteinase K digestion for 606 4 h at 42 °C. DNA fragments were purified using QIAquick PCR purification columns (Qiagen) 607 quantified using a Qubit fluorometer (Invitrogen). Fragmented DNA was flash frozen and stored at 608 -80ºC until further processing for high throughput sequencing or qRT-PCR.

609 ChIP-sequencing

610 Preparation of libraries and sequencing was performed at GENEWIZ (Leipzig, Germany). Here, 611 libraries were prepared using NEBNext Ultra DNA Library Preparation Kit for Illumina (New England 612 Biosystems) following the manufacturer’s recommendations. Briefly, the ChIP DNA was end 613 repaired and adapters were ligated after adenylation of the 3’ends. Adapter-ligated DNA was size 614 selected, followed by clean up, PCR enrichment. ChIP libraries were validated using an Agilent 615 TapeStation, and quantified using Qubit 2.0 Fluorometer as well as qRT-PCR (Applied Biosystems, 616 Carlsbad, CA, USA). Normalized libraries were multiplexed, clustered, and loaded on one lane of 617 a flow cell (Illumina). Libraries were sequenced on a HiSeq 4000 platform (Illumina) using a 2x150 618 PE configuration. Image analysis and base calling were conducted by the HiSeq Control Software 619 (HCS). Raw sequence data (.bcl files) generated from Illumina HiSeq was converted into fastq files

620 and de-multiplexed using Illumina's bcl2fastq 2.17 software. One mis-match was allowed for index 621 sequence identification.

622 Network analysis

623 A Network integrating RNA-seq data from the HGF condition of the in vitro sugar challenge 624 experiment and ChIP-seq data was generated using FastHeinz from the R BioNet package 625 (Beisser et al., 2010) with the HumanCyc metabolic pathways (http://humancyc.org/) downloaded 626 from http://www.ndexbio.org (Pillich et al., 2017). Next, a weight matrix was constructed with all 627 expressed genes in the Klf10flox/flox and Klf10Δhep primary hepatocytes taking the lowest p value 628 from the ChIP-seq RNA-seq datasets. Network visualization was done using Cytoscape and 629 AutoAnnotate app (Kucera et al., 2016; Shannon et al., 2003). Clustering was performed using the 630 MCL partitioning algorithm from clusterMaker2 (Morris et al., 2011) and gene ontology analysis 631 was performed using BiNGO (Maere et al., 2005) at http://impala.molgen.mpg.de/ (Kamburov et 632 al., 2011) and at https://www.uniprot.org/.

633 Quantification and statistical analyses

634 For the circadian time course experiment, sequences were preprocessed using FASTQ groomer 635 and checked using Fastqc (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads 636 were mapped against the Mus musculus genome (mm10) downloaded from Ensembl.org using 637 STAR_2.4.0a (Dobin et al., 2013) with the Encode RNA-seq best practices options indicated (“– 638 outFilterIntronMotifsRemoveNoncanonicalUnannotated –alignMatesGapMax 1000000 – 639 outReadsUnmapped Fastx –alignIntronMin 20 –alignIntronMax 1000000 –alignSJoverhangMin 8 640 –alignSJDBoverhangMin 1 –outFilterMultimapNmax 20”). After alignment to the genome reads 641 were counted with featureCounts (Liao et al., 2014) with “–primary –g gene_name –p –s 1 –M –C” 642 options indicated. Gene stable IDs (GRCm38.p6) were downloaded from Ensembl Genes 94 643 (Ensembl) and used to filter for mouse protein coding genes. Matrix of raw counts from both 644 genotypes across time was generated in R (R Core Team, 2018). Transcripts with less than 8 645 counts across all samples were excluded from downstream analyses. All transcript counts were 646 normalized using edgeR (Robinson et al., 2010) and expressed in counts per million (cpm). 647 For the in vitro sugar challenge experiment, sequences were preprocessed using FASTQ 648 groomer and checked using Fastqc (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). 649 Reads were mapped against the Mus musculus genome (mm10) downloaded from Ensembl.org 650 using HISAT2with default parameters (Kim et al., 2015). After alignment to the genome reads were 651 counted with featureCounts (Liao et al., 2014) with with default parameters. Matrices containing

652 the counts were loaded into R. For downstream analyses, we analyzed protein coding genes with 653 a sum >20 counts across all samples.

654 Circadian expression of mRNA was determined using MetaCycle (Wu et al., 2016) with the 655 following options: minper = 24, maxper = 24, cycMethod = c (“ARS,” “JTK,” “LS”), analysisStrategy 656 = “auto,” outputFile = TRUE, outIntegration = “both,” adjustPhase = “predictedPer,” combinePvalue 657 = “fisher,” weightedPerPha = TRUE, ARSmle = “auto,” and ARSdefaultPer = 24. Transcripts with 658 an integrated p-value (meta2d_pvalue) < 0.05 were considered rhythmically expressed. 659 Circadian pathways were determined by Phase Set Enrichment Analysis (PSEA) (Zhang et al., 660 2016) based on the sets of circadian transcripts with a relative amplitude value (rAMP; the ratio 661 between amplitude and baseline expression) of ≥ 0.1. Gene sets were downloaded from the 662 Molecular Signatures database (MSigDB) H (HallmarkGene Sets) (Subramanian et al., 2005). Sets 663 containing fewer than 5 circadian transcripts were excluded from the analysis. The Kuiper test was 664 used to identify circadian gene sets by comparing the acrophases of all circadian transcripts 665 belonging to each gene set to a uniform background distribution and by testing for non-uniformity 666 (q < 0.01). 667 For the in vitro sugar challenge experiment, differential expression analysis between conditions 668 was performed using the TestDE function in the R package “edgeR” (Robinson et al., 2010). Genes 669 with an FDR value < 0.05 when comparing LG vs HGF conditions in each respective genotype 670 were considered differentially expressed and used to assess the pathways enriched under the LG 671 and HGF conditions. Gene lists were uploaded to enrichR web server (Chen et al., 2013; Kuleshov 672 et al., 2016) and pathway enrichment was assessed by comparing the overlap of genes in the query 673 to the Kyoto Encyclopedia of Genes and Genomes (KEGG) 2019 mouse gene sets (Kanehisa and 674 Goto, 2000). For a schematic of the workflow used, refer to Figure S5C. 675 For ChIP-seq read mapping, peak calling and quantification, sequences were preprocessed 676 using FASTQ groomer and checked using Fastqc 677 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). PE reads were trimmed to 50 bases 678 with Trimmomatic (Bolger et al., 2014), and reads were aligned to mouse genome 679 (GRCm38/mm10) using Bowtie2 (Langmead and Salzberg, 2012) with the default parameters. 680 Next, peak calls were made in MACS2 (Feng et al., 2012) with default parameters, except allowing 681 for duplicates. Filters were applied to keep peaks with a log p-value < -2 and a Mfold > 2. Results 682 from the independent ChIP-seq runs were concatenated and intervals were merged using the 683 MergeBED function from BEDtools (Quinlan and Hall, 2010). Identified peaks within the -10 kb to 684 +1 kb region relative to the TSS were annotated using GREAT (McLean et al., 2010) and further

685 filtered for the presence of a KLF10 motif (matrix ID: MA1511.1, http://jaspar.genereg.net/) using 686 FIMO (Bailey et al., 2015). 687 Numerical values are presented as mean ± SEM. Replicates (n) are indicated in the figure 688 legends. Pairwise comparisons were tested using the non-parametric Wilcoxon test. Differences 689 between more than 2 experimental conditions were tested using the Kruskal and Wallis rank sum 690 test for multiple comparison followed by a pairwise post-hoc test using the Benjamini-Hochberg 691 adjustment. P values less than 0.05 were considered significant.

692 Resource and data availability

693 Further information and requests for resources and reagents should be directed to and will be 694 fulfilled by the Lead Contact, Michèle Teboul ([email protected]).

695 The datasets generated during this study are available at the European Nucleotide Archive (ENA) 696 (https://www.ebi.ac.uk/ena/browser/home) under ENA project numbers PRJEB39035, 697 PRJEB39036, PRJEB40195.

698 Supplemental files

699 Table S1. Hepatic circadian transcriptome of Klf10flox/flox and Klf10Δhep mice 700 Related to Figure 1. Metacycle analysis of the hepatic transcriptome of Klf10flox/flox and 701 Klf10Δhep mice fed ad libitum and entrained in a 12:12 LD cycle. (.xlsx 9.65 MB) 702 703 Table S2. Temporal coordination of the hepatic transcriptome of Klf10flox/flox and Klf10Δhep 704 mice 705 Related to Figure 1. PSEA analysis of the hepatic transcriptome of Klf10flox/flox and 706 Klf10Δhep mice fed ad libitum and entrained in a 12:12 LD cycle. (.xlsx 0.02 MB) 707 708 Table S3. Transcriptome of Klf10flox/flox and Klf10Δhep hepatocytes challenged with high 709 sugar 710 Related to Figure 5. Lists of differentially expressed genes in Klf10flox/flox and Klf10Δhep 711 hepatocytes challenged with high sugar compared to unchallenged cells. (.xlsx 0.13 MB) 712 713 Table S4. Enriched pathways in Klf10flox/flox and Klf10Δhep hepatocytes challenged with 714 high sugar

715 Related to Figure 5. Lists of KEGG enriched pathway in Klf10flox/flox and Klf10Δhep 716 hepatocytes challenged with high sugar compared to unchallenged cells. (.xlsx 0.08 MB) 717 718 Table S5. KLF10 regulated network in mouse liver. 719 Related to Figure 6. List of genes bound by KLF10 in their -10/+1kb region or 720 differentially expressed in Klf10flox/flox vs Klf10Δhep hepatocytes challenged with high 721 sugar. (.xlsx 0.16 MB) 722 723 Table S6. List of Primers used in the study (.xlsx 0.07 MB) 724 725 Cytoscape file S1. KLF10 regulated metabolic network 726 Ralated to Figure 6. File used to generate, visualize and analyze the KLF10 regulated 727 metabolic network using Cytoscape (.cys 0.47 MB) 728

729 References

730 Bailey, T.L., Johnson, J., Grant, C.E., Noble, W.S., 2015. The MEME Suite. Nucleic Acids Res. 43, W39-49. 731 https://doi.org/10.1093/nar/gkv416 732 Bartok, O., Teesalu, M., Ashwall-Fluss, R., Pandey, V., Hanan, M., Rovenko, B.M., Poukkula, M., Havula, E., 733 Moussaieff, A., Vodala, S., Nahmias, Y., Kadener, S., Hietakangas, V., 2015. The transcription factor Cabut 734 coordinates energy metabolism and the circadian clock in response to sugar sensing. EMBO J 34, 1538–53. 735 https://doi.org/10.15252/embj.201591385 736 Bass, J., Lazar, M.A., 2016. Circadian time signatures of fitness and disease. Science 354, 994–999. 737 https://doi.org/10.1126/science.aah4965 738 Beisser, D., Klau, G.W., Dandekar, T., Müller, T., Dittrich, M.T., 2010. BioNet: an R-Package for the 739 functional analysis of biological networks. Bioinformatics 26, 1129–1130. 740 https://doi.org/10.1093/bioinformatics/btq089 741 Bolger, A.M., Lohse, M., Usadel, B., 2014. Trimmomatic: a flexible trimmer for Illumina sequence data. 742 Bioinformatics 30, 2114–2120. https://doi.org/10.1093/bioinformatics/btu170 743 Bugge, A., Feng, D., Everett, L.J., Briggs, E.R., Mullican, S.E., Wang, F., Jager, J., Lazar, M.A., 2012. Rev-erbα 744 and Rev-erbβ coordinately protect the circadian clock and normal metabolic function. Genes Dev 26, 657– 745 67. https://doi.org/10.1101/gad.186858.112 746 Cederroth, C.R., Albrecht, U., Bass, J., Brown, S.A., Dyhrfjeld-Johnsen, J., Gachon, F., Green, C.B., Hastings, 747 M.H., Helfrich-Förster, C., Hogenesch, J.B., Lévi, F., Loudon, A., Lundkvist, G.B., Meijer, J.H., Rosbash, M., 748 Takahashi, J.S., Young, M., Canlon, B., 2019. Medicine in the Fourth Dimension. Cell Metab 30, 238–250. 749 https://doi.org/10.1016/j.cmet.2019.06.019 750 Ceglia, N., Liu, Y., Chen, S., Agostinelli, F., Eckel-Mahan, K., Sassone-Corsi, P., Baldi, P., 2018. CircadiOmics: 751 circadian omic web portal. Nucleic Acids Res. 46, W157–W162. https://doi.org/10.1093/nar/gky441

752 Chen, E.Y., Tan, C.M., Kou, Y., Duan, Q., Wang, Z., Meirelles, G.V., Clark, N.R., Ma’ayan, A., 2013. Enrichr: 753 interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics 14, 128. 754 https://doi.org/10.1186/1471-2105-14-128 755 Chong, J., Wishart, D.S., Xia, J., 2019. Using MetaboAnalyst 4.0 for Comprehensive and Integrative 756 Metabolomics Data Analysis. Curr Protoc Bioinformatics 68, e86. https://doi.org/10.1002/cpbi.86 757 Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P., Chaisson, M., Gingeras, 758 T.R., 2013. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21. 759 https://doi.org/10.1093/bioinformatics/bts635 760 Dyar, K.A., Lutter, D., Artati, A., Ceglia, N.J., Liu, Y., Armenta, D., Jastroch, M., Schneider, S., de Mateo, S., 761 Cervantes, M., Abbondante, S., Tognini, P., Orozco-Solis, R., Kinouchi, K., Wang, C., Swerdloff, R., Nadeef, 762 S., Masri, S., Magistretti, P., Orlando, V., Borrelli, E., Uhlenhaut, N.H., Baldi, P., Adamski, J., Tschöp, M.H., 763 Eckel-Mahan, K., Sassone-Corsi, P., 2018. Atlas of Circadian Metabolism Reveals System-wide 764 Coordination and Communication between Clocks. Cell 174, 1571-1585.e11. 765 https://doi.org/10.1016/j.cell.2018.08.042 766 Eckel-Mahan, K.L., Patel, V.R., de Mateo, S., Orozco-Solis, R., Ceglia, N.J., Sahar, S., Dilag-Penilla, S.A., Dyar, 767 K.A., Baldi, P., Sassone-Corsi, P., 2013. Reprogramming of the circadian clock by nutritional challenge. Cell 768 155, 1464–78. https://doi.org/10.1016/j.cell.2013.11.034 769 Feillet, C., Guérin, S., Lonchampt, M., Dacquet, C., Gustafsson, J., Delaunay, F., Teboul, M., 2016. Sexual 770 Dimorphism in Circadian Physiology Is Altered in LXRα Deficient Mice. PLoS One 11, e0150665. 771 https://doi.org/10.1371/journal.pone.0150665 772 Feng, J., Liu, T., Qin, B., Zhang, Y., Liu, X.S., 2012. Identifying ChIP-seq enrichment using MACS. Nat Protoc 773 7, 1728–1740. https://doi.org/10.1038/nprot.2012.101 774 Folch, J., Lees, M., Sloane Stanley, G.H., 1957. A simple method for the isolation and purification of total 775 lipides from animal tissues. J Biol Chem 226, 497–509. 776 Gaucher, J., Kinouchi, K., Ceglia, N., Montellier, E., Peleg, S., Greco, C.M., Schmidt, A., Forne, I., Masri, S., 777 Baldi, P., Imhof, A., Sassone-Corsi, P., 2019. Distinct metabolic adaptation of liver circadian pathways to 778 acute and chronic patterns of alcohol intake. Proc Natl Acad Sci U S A 116, 25250–25259. 779 https://doi.org/10.1073/pnas.1911189116 780 Greenwell, B.J., Trott, A.J., Beytebiere, J.R., Pao, S., Bosley, A., Beach, E., Finegan, P., Hernandez, C., 781 Menet, J.S., 2019. Rhythmic Food Intake Drives Rhythmic Gene Expression More Potently than the 782 Hepatic Circadian Clock in Mice. Cell Rep 27, 649-657.e5. https://doi.org/10.1016/j.celrep.2019.03.064 783 Guillaumond, F., Gréchez-Cassiau, A., Subramaniam, M., Brangolo, S., Peteri-Brünback, B., Staels, B., 784 Fiévet, C., Spelsberg, T.C., Delaunay, F., Teboul, M., 2010. Krüppel-Like Factor KLF10 Is a Link between the 785 Circadian Clock and Metabolism in Liver. Molecular and cellular biology 30, 3059–3070. 786 Han, S., Han, S.S., Zhang, R., Jain, R., Shi, H., Zhang, L., Zhou, G., Sangwung, P., Tugal, D., Atkins, G.B., 787 Prosdocimo, D.A., Lu, Y., Han, X., Tso, P., Liao, X., Epstein, J.A., Jain, M.K., 2015. Circadian control of bile 788 acid synthesis by a KLF15-Fgf15 axis. Nat Commun 6, 7231. https://doi.org/10.1038/ncomms8231 789 Hannou, S.A., Haslam, D.E., McKeown, N.M., Herman, M.A., n.d. Fructose metabolism and metabolic 790 disease. J Clin Invest 128, 545–555. https://doi.org/10.1172/JCI96702 791 Hirota, T., Okano, T., Kokame, K., Shirotani-Ikejima, H., Miyata, T., Fukada, Y., 2002. Glucose down- 792 regulates Per1 and Per2 mRNA levels and induces circadian gene expression in cultured Rat-1 fibroblasts. J 793 Biol Chem 277, 44244–51. https://doi.org/10.1074/jbc.M206233200

794 Hsieh, P.N., Fan, L., Sweet, D.R., Jain, M.K., 2019. The Krüppel-Like Factors and Control of Energy 795 Homeostasis. Endocr Rev 40, 137–152. https://doi.org/10.1210/er.2018-00151 796 Hui, S., Ghergurovich, J.M., Morscher, R.J., Jang, C., Teng, X., Lu, W., Esparza, L.A., Reya, T., Le Zhan, null, 797 Yanxiang Guo, J., White, E., Rabinowitz, J.D., 2017. Glucose feeds the TCA cycle via circulating lactate. 798 Nature 551, 115–118. https://doi.org/10.1038/nature24057 799 Iizuka, K., Takeda, J., Horikawa, Y., 2011. Krüppel-like factor-10 is directly regulated by carbohydrate 800 response element-binding protein in rat primary hepatocytes. Biochem Biophys Res Commun 412, 638– 801 43. https://doi.org/10.1016/j.bbrc.2011.08.016 802 Jacobi, D., Liu, S., Burkewitz, K., Kory, N., Knudsen, N.H., Alexander, R.K., Unluturk, U., Li, X., Kong, X., 803 Hyde, A.L., Gangl, M.R., Mair, W.B., Lee, C.H., 2015. Hepatic Bmal1 Regulates Rhythmic Mitochondrial 804 Dynamics and Promotes Metabolic Fitness. Cell Metab 22, 709–20. 805 https://doi.org/10.1016/j.cmet.2015.08.006 806 Jensen, T., Abdelmalek, M.F., Sullivan, S., Nadeau, K.J., Green, M., Roncal, C., Nakagawa, T., Kuwabara, M., 807 Sato, Y., Kang, D.H., Tolan, D.R., Sanchez-Lozada, L.G., Rosen, H.R., Lanaspa, M.A., Diehl, A.M., Johnson, 808 R.J., 2018. Fructose and sugar: A major mediator of non-alcoholic fatty liver disease. J Hepatol 68, 1063– 809 1075. https://doi.org/10.1016/j.jhep.2018.01.019 810 Jeyaraj, D., Scheer, F.A., Ripperger, J.A., Haldar, S.M., Lu, Y., Prosdocimo, D.A., Eapen, S.J., Eapen, B.L., Cui, 811 Y., Mahabeleshwar, G.H., Lee, H.G., Smith, M.A., Casadesus, G., Mintz, E.M., Sun, H., Wang, Y., Ramsey, 812 K.M., Bass, J., Shea, S.A., Albrecht, U., Jain, M.K., 2012. Klf15 orchestrates circadian nitrogen homeostasis. 813 Cell Metab 15, 311–23. https://doi.org/10.1016/j.cmet.2012.01.020 814 Kamburov, A., Pentchev, K., Galicka, H., Wierling, C., Lehrach, H., Herwig, R., 2011. ConsensusPathDB: 815 toward a more complete picture of cell biology. Nucleic Acids Res. 39, D712-717. 816 https://doi.org/10.1093/nar/gkq1156 817 Kanehisa, M., Goto, S., 2000. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 28, 27– 818 30. https://doi.org/10.1093/nar/28.1.27 819 Kim, D., Langmead, B., Salzberg, S.L., 2015. HISAT: a fast spliced aligner with low memory requirements. 820 Nat Methods 12, 357–60. https://doi.org/10.1038/nmeth.3317 821 Klip, A., McGraw, T.E., James, D.E., 2019. Thirty sweet years of GLUT4. J. Biol. Chem. 294, 11369–11381. 822 https://doi.org/10.1074/jbc.REV119.008351 823 Kucera, M., Isserlin, R., Arkhangorodsky, A., Bader, G.D., 2016. AutoAnnotate: A Cytoscape app for 824 summarizing networks with semantic annotations. F1000Res 5, 1717. 825 https://doi.org/10.12688/f1000research.9090.1 826 Kuleshov, M.V., Jones, M.R., Rouillard, A.D., Fernandez, N.F., Duan, Q., Wang, Z., Koplev, S., Jenkins, S.L., 827 Jagodnik, K.M., Lachmann, A., McDermott, M.G., Monteiro, C.D., Gundersen, G.W., Ma’ayan, A., 2016. 828 Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 44, 829 W90-97. https://doi.org/10.1093/nar/gkw377 830 la Fleur, S.E., Kalsbeek, A., Wortel, J., Fekkes, M.L., Buijs, R.M., 2001. A daily rhythm in glucose tolerance: a 831 role for the suprachiasmatic nucleus. Diabetes 50, 1237–43. 832 Langmead, B., Salzberg, S.L., 2012. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359. 833 https://doi.org/10.1038/nmeth.1923 834 Leclère, P.S., Rousseau, D., Patouraux, S., Guérin, S., Bonnafous, S., Gréchez-Cassiau, A., Ruberto, A.A., 835 Luci, C., Subramaniam, M., Tran, A., Delaunay, F., Gual, P., Teboul, M., 2020. MCD diet-induced

836 steatohepatitis generates a diurnal rhythm of associated biomarkers and worsens liver injury in Klf10 837 deficient mice. Sci Rep 10, 12139. https://doi.org/10.1038/s41598-020-69085-w 838 Liao, Y., Smyth, G.K., Shi, W., 2014. featureCounts: an efficient general purpose program for assigning 839 sequence reads to genomic features. Bioinformatics 30, 923–930. 840 https://doi.org/10.1093/bioinformatics/btt656 841 Ma, L., Robinson, L.N., Towle, H.C., 2006. ChREBP*Mlx is the principal mediator of glucose-induced gene 842 expression in the liver. J Biol Chem 281, 28721–28730. https://doi.org/10.1074/jbc.M601576200 843 Maere, S., Heymans, K., Kuiper, M., 2005. BiNGO: a Cytoscape plugin to assess overrepresentation of gene 844 ontology categories in biological networks. Bioinformatics 21, 3448–3449. 845 https://doi.org/10.1093/bioinformatics/bti551 846 Makwana, K., Gosai, N., Poe, A., Kondratov, R.V., 2019. Calorie restriction reprograms diurnal rhythms in 847 protein translation to regulate metabolism. FASEB J 33, 4473–4489. 848 https://doi.org/10.1096/fj.201802167R 849 Masri, S., Papagiannakopoulos, T., Kinouchi, K., Liu, Y., Cervantes, M., Baldi, P., Jacks, T., Sassone-Corsi, P., 850 2016. Lung Adenocarcinoma Distally Rewires Hepatic Circadian Homeostasis. Cell 165, 896–909. 851 https://doi.org/10.1016/j.cell.2016.04.039 852 Mauvoisin, D., Gachon, F., 2019. Proteomics in Circadian Biology. J Mol Biol. 853 https://doi.org/10.1016/j.jmb.2019.12.004 854 Mazzoccoli, G., De Cosmo, S., Mazza, T., 2018. The Biological Clock: A Pivotal Hub in Non-alcoholic Fatty 855 Liver Disease Pathogenesis. Front Physiol 9, 193. https://doi.org/10.3389/fphys.2018.00193 856 McLean, C.Y., Bristor, D., Hiller, M., Clarke, S.L., Schaar, B.T., Lowe, C.B., Wenger, A.M., Bejerano, G., 2010. 857 GREAT improves functional interpretation of cis-regulatory regions. Nat. Biotechnol. 28, 495–501. 858 https://doi.org/10.1038/nbt.1630 859 Mederacke, I., Dapito, D.H., Affò, S., Uchinami, H., Schwabe, R.F., 2015. High-yield and high-purity 860 isolation of hepatic stellate cells from normal and fibrotic mouse livers. Nat Protoc 10, 305–15. 861 https://doi.org/10.1038/nprot.2015.017 862 Morris, J.H., Apeltsin, L., Newman, A.M., Baumbach, J., Wittkop, T., Su, G., Bader, G.D., Ferrin, T.E., 2011. 863 clusterMaker: a multi-algorithm clustering plugin for Cytoscape. BMC Bioinformatics 12, 436. 864 https://doi.org/10.1186/1471-2105-12-436 865 Mortera, R.R., Bains, Y., Gugliucci, A., 2019. Fructose at the crossroads of the metabolic syndrome and 866 obesity epidemics. Front Biosci (Landmark Ed) 24, 186–211. 867 Mukherji, A., Bailey, S.M., Staels, B., Baumert, T.F., 2019. The circadian clock and liver function in health 868 and disease. J Hepatol 71, 200–211. https://doi.org/10.1016/j.jhep.2019.03.020 869 Oishi, Y., Manabe, I., 2018. Krüppel-Like Factors in Metabolic Homeostasis and Cardiometabolic Disease. 870 Front Cardiovasc Med 5, 69. https://doi.org/10.3389/fcvm.2018.00069 871 Ortega-Prieto, P., Postic, C., 2019. Carbohydrate Sensing Through the Transcription Factor ChREBP. Front 872 Genet 10. https://doi.org/10.3389/fgene.2019.00472 873 Panda, S., 2016. Circadian physiology of metabolism. Science 354, 1008–1015. 874 https://doi.org/10.1126/science.aah4967 875 Peek, C.B., Affinati, A.H., Ramsey, K.M., Kuo, H.Y., Yu, W., Sena, L.A., Ilkayeva, O., Marcheva, B., Kobayashi, 876 Y., Omura, C., Levine, D.C., Bacsik, D.J., Gius, D., Newgard, C.B., Goetzman, E., Chandel, N.S., Denu, J.M.,

877 Mrksich, M., Bass, J., 2013. Circadian clock NAD+ cycle drives mitochondrial oxidative metabolism in mice. 878 Science 342, 1243417. https://doi.org/10.1126/science.1243417 879 Perelis, M., Marcheva, B., Ramsey, K.M., Schipma, M.J., Hutchison, A.L., Taguchi, A., Peek, C.B., Hong, H., 880 Huang, W., Omura, C., Allred, A.L., Bradfield, C.A., Dinner, A.R., Barish, G.D., Bass, J., 2015. Pancreatic β 881 cell enhancers regulate rhythmic transcription of genes controlling insulin secretion. Science 350, 882 aac4250. https://doi.org/10.1126/science.aac4250 883 Pillich, R.T., Chen, J., Rynkov, V., Welker, D., Pratt, D., 2017. NDEx: A Community Resource for Sharing and 884 Publishing of Biological Networks. Methods Mol Biol 1558, 271–301. https://doi.org/10.1007/978-1-4939- 885 6783-4_13 886 Poolman, T.M., Gibbs, J., Walker, A.L., Dickson, S., Farrell, L., Hensman, J., Kendall, A.C., Maidstone, R., 887 Warwood, S., Loudon, A., Rattray, M., Bruce, I.N., Nicolaou, A., Ray, D.W., 2019. Rheumatoid arthritis 888 reprograms circadian output pathways. Arthritis Res Ther 21, 47. https://doi.org/10.1186/s13075-019- 889 1825-y 890 Qian, J., Scheer, F.A.J.L., 2016. Circadian System and Glucose Metabolism: Implications for Physiology and 891 Disease. Trends Endocrinol Metab 27, 282–293. https://doi.org/10.1016/j.tem.2016.03.005 892 Quinlan, A.R., Hall, I.M., 2010. BEDTools: a flexible suite of utilities for comparing genomic features. 893 Bioinformatics 26, 841–842. https://doi.org/10.1093/bioinformatics/btq033 894 Reinke, H., Asher, G., 2019. Crosstalk between metabolism and circadian clocks. Nat Rev Mol Cell Biol 20, 895 227–241. https://doi.org/10.1038/s41580-018-0096-9 896 Rijo-Ferreira, F., Takahashi, J.S., 2014. Genomics of circadian rhythms in health and disease. Genome Med 897 11, 82. https://doi.org/10.1186/s13073-019-0704-0 898 Robinson, M.D., McCarthy, D.J., Smyth, G.K., 2010. edgeR: a Bioconductor package for differential 899 expression analysis of digital gene expression data. Bioinformatics 26, 139–140. 900 https://doi.org/10.1093/bioinformatics/btp616 901 Robles, M.S., Cox, J., Mann, M., 2014. In-vivo quantitative proteomics reveals a key contribution of post- 902 transcriptional mechanisms to the circadian regulation of liver metabolism. PLoS Genet 10, e1004047. 903 https://doi.org/10.1371/journal.pgen.1004047 904 Robles, M.S., Humphrey, S.J., Mann, M., 2017. Phosphorylation Is a Central Mechanism for Circadian 905 Control of Metabolism and Physiology. Cell Metab 25, 118–127. 906 https://doi.org/10.1016/j.cmet.2016.10.004 907 Romero, P., Wagg, J., Green, M.L., Kaiser, D., Krummenacker, M., Karp, P.D., 2005. Computational 908 prediction of human metabolic pathways from the complete human genome. Genome Biol. 6, R2. 909 https://doi.org/10.1186/gb-2004-6-1-r2 910 Saini, C., Suter, D.M., Liani, A., Gos, P., Schibler, U., 2011. The Mammalian Circadian Timing System: 911 Synchronization of Peripheral Clocks. Cold Spring Harb Symp Quant Biol 76, 39–47. 912 https://doi.org/10.1101/sqb.2011.76.010918 913 Schmid, B., Helfrich-Förster, C., Yoshii, T., 2011. A new ImageJ plug-in “ActogramJ” for chronobiological 914 analyses. J. Biol. Rhythms 26, 464–467. https://doi.org/10.1177/0748730411414264 915 Schmitges, F.W., Radovani, E., Najafabadi, H.S., Barazandeh, M., Campitelli, L.F., Yin, Y., Jolma, A., Zhong, 916 G., Guo, H., Kanagalingam, T., Dai, W.F., Taipale, J., Emili, A., Greenblatt, J.F., Hughes, T.R., 2016. 917 Multiparameter functional diversity of human C2H2 zinc finger proteins. Genome Res. 26, 1742–1752. 918 https://doi.org/10.1101/gr.209643.116

919 Schuler, M., Dierich, A., Chambon, P., Metzger, D., 2004. Efficient temporally controlled targeted somatic 920 mutagenesis in hepatocytes of the mouse. Genesis 39, 167–72. https://doi.org/10.1002/gene.20039 921 Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., Amin, N., Schwikowski, B., Ideker, 922 T., 2003. Cytoscape: a software environment for integrated models of biomolecular interaction networks. 923 Genome Res 13, 2498–504. https://doi.org/10.1101/gr.1239303 924 Sinturel, F., Petrenko, V., Dibner, C., 2020. Circadian Clocks Make Metabolism Run. J Mol Biol. 925 https://doi.org/10.1016/j.jmb.2020.01.018 926 Softic, S., Stanhope, K.L., Boucher, J., Divanovic, S., Lanaspa, M.A., Johnson, R.J., Kahn, C.R., 2020. Fructose 927 and hepatic insulin resistance. Crit Rev Clin Lab Sci 1–15. 928 https://doi.org/10.1080/10408363.2019.1711360 929 Stenvers, D.J., Scheer, F.A.J.L., Schrauwen, P., la Fleur, S.E., Kalsbeek, A., 2019. Circadian clocks and insulin 930 resistance. Nat Rev Endocrinol 15, 75–89. https://doi.org/10.1038/s41574-018-0122-1 931 Subramaniam, M., Harris, S.A., Oursler, M.J., Rasmussen, K., Riggs, B.L., Spelsberg, T.C., 1995. 932 Identification of a novel TGF-beta-regulated gene encoding a putative zinc finger protein in human 933 osteoblasts. Nucleic Acids Res 23, 4907–4912. https://doi.org/10.1093/nar/23.23.4907 934 Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., 935 Pomeroy, S.L., Golub, T.R., Lander, E.S., Mesirov, J.P., 2005. Gene set enrichment analysis: a knowledge- 936 based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U.S.A. 102, 937 15545–15550. https://doi.org/10.1073/pnas.0506580102 938 Tachibana, I., Imoto, M., Adjei, P.N., Gores, G.J., Subramaniam, M., Spelsberg, T.C., Urrutia, R., 1997. 939 Overexpression of the TGFbeta-regulated zinc finger encoding gene, TIEG, induces apoptosis in pancreatic 940 epithelial cells. J Clin Invest 99, 2365–2374. https://doi.org/10.1172/JCI119418 941 Tognini, P., Murakami, M., Liu, Y., Eckel-Mahan, K.L., Newman, J.C., Verdin, E., Baldi, P., Sassone-Corsi, P., 942 2017. Distinct Circadian Signatures in Liver and Gut Clocks Revealed by Ketogenic Diet. Cell Metab 26, 943 523-538.e5. https://doi.org/10.1016/j.cmet.2017.08.015 944 Vollmers, C., Gill, S., DiTacchio, L., Pulivarthy, S.R., Le, H.D., Panda, S., 2009. Time of feeding and the 945 intrinsic circadian clock drive rhythms in hepatic gene expression. Proc Natl Acad Sci U S A 106, 21453–8. 946 https://doi.org/10.1073/pnas.0909591106 947 Weidemann, B.J., Ramsey, K.M., Bass, J., 2018. A day in the life of chromatin: how enhancer-promoter 948 loops shape daily behavior. Genes Dev 32, 321–323. https://doi.org/10.1101/gad.314187.118 949 Weng, C.-C., Hawse, J.R., Subramaniam, M., Chang, V.H.S., Yu, W.C.Y., Hung, W.-C., Chen, L.-T., Cheng, K.- 950 H., 2017. KLF10 loss in the pancreas provokes activation of SDF-1 and induces distant metastases of 951 pancreatic ductal adenocarcinoma in the KrasG12D p53flox/flox model. Oncogene 36, 5532–5543. 952 https://doi.org/10.1038/onc.2017.155 953 Wu, G., Anafi, R.C., Hughes, M.E., Kornacker, K., Hogenesch, J.B., 2016. MetaCycle: an integrated R 954 package to evaluate periodicity in large scale data. Bioinformatics 32, 3351–3353. 955 https://doi.org/10.1093/bioinformatics/btw405 956 Yoshitane, H., Ozaki, H., Terajima, H., Du, N.-H., Suzuki, Y., Fujimori, T., Kosaka, N., Shimba, S., Sugano, S., 957 Takagi, T., Iwasaki, W., Fukada, Y., 2014. CLOCK-controlled polyphonic regulation of circadian rhythms 958 through canonical and noncanonical E-boxes. Mol. Cell. Biol. 34, 1776–1787. 959 https://doi.org/10.1128/MCB.01465-13

960 Zhang, L., Prosdocimo, D.A., Bai, X., Fu, C., Zhang, R., Campbell, F., Liao, X., Coller, J., Jain, M.K., 2015. 961 KLF15 Establishes the Landscape of Diurnal Expression in the Heart. Cell Rep 13, 2368–2375. 962 https://doi.org/10.1016/j.celrep.2015.11.038 963 Zhang, R., Lahens, N.F., Ballance, H.I., Hughes, M.E., Hogenesch, J.B., 2014. A circadian gene expression 964 atlas in mammals: implications for biology and medicine. Proc Natl Acad Sci U S A 111, 16219–24. 965 https://doi.org/10.1073/pnas.1408886111 966 Zhang, R., Podtelezhnikov, A.A., Hogenesch, J.B., Anafi, R.C., 2016. Discovering Biology in Periodic Data 967 through Phase Set Enrichment Analysis (PSEA). J Biol Rhythms 31, 244–57. 968 https://doi.org/10.1177/0748730416631895 969 Zhao, S., Jang, C., Liu, J., Uehara, K., Gilbert, M., Izzo, L., Zeng, X., Trefely, S., Fernandez, S., Carrer, A., 970 Miller, K.D., Schug, Z.T., Snyder, N.W., Gade, T.P., Titchenell, P.M., Rabinowitz, J.D., Wellen, K.E., 2020. 971 Dietary fructose feeds hepatic lipogenesis via microbiota-derived acetate. Nature 579, 586–591. 972 https://doi.org/10.1038/s41586-020-2101-7 973 Zhao, S., Torres, A., Henry, R.A., Trefely, S., Wallace, M., Lee, J.V., Carrer, A., Sengupta, A., Campbell, S.L., 974 Kuo, Y.-M., Frey, A.J., Meurs, N., Viola, J.M., Blair, I.A., Weljie, A.M., Metallo, C.M., Snyder, N.W., Andrews, 975 A.J., Wellen, K.E., 2016. ATP-Citrate Lyase Controls a Glucose-to-Acetate Metabolic Switch. Cell Rep 17, 976 1037–1052. https://doi.org/10.1016/j.celrep.2016.09.069 977

Supplemental Figures

KLF10 integrates circadian timing and sugar signaling to coordinate hepatic metabolism

Anthony A. Ruberto, Aline Gréchez-Cassiau, Sophie Guérin, Luc Martin, Johana S. Revel, Mohamed Mehiri, Malayannan Subramaniam, Franck Delaunay, Michèle Teboul

A B Klf10flox/flox Klf10Δhep 40 5 flox/flox Klf10 ZT (h) 24/0 12 24/0 24/0 12 24/0 Klf10Dhep 30 4 Day 1 3 Mouse #1 Mouse #1 20 Day 7 2 Day 1 10 Mouse #2 Mouse #2

Body mass (g) (g) mass Body 1 Day 7 Food intake (g/day) intake Food 0 0 Day 1 Mouse #3 Mouse #3 Day 7

Figure S1 (related to Fig. 1). Genetic disruption of Klf10 in mouse hepatocytes (A) Body weights of Klf10flox/flox and Klf10Δhep mice (mean ± SEM, n = 6). (B) Daily food intake of of Klf10flox/flox and Klf10Δhep mice (mean ± SEM, n = 6). (C) Representative actograms of double-plotted food-seeking activity records of Klf10flox/flox and Klf10Δhep mice. Mice were placed in individual cages containing a Minimetter infrared detection system. Vertical bars represent time points that animals were in the area of the cage containing food. bioRxiv preprint doi: https://doi.org/10.1101/2020.12.22.423999; this version posted December 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

A Klf10flox/flox Klf10Dhep Bmal1 Clock Cry1 Cry2 Per1 Per2

Per3 Reverbα Reverbβ Rorc Npas2 Chrono Relative expression Relative

Zeitgeber time (h)

B C *** 250 1200 flox/flox * Klf10flox/flox * Klf10 /g) Dhep Dhep 1000 Klf10 200 Klf10 *** mol 800 150

/µg protein /µg protein 600 3 100 400 50

glycogen (µ glycogen 200

RLU x10 RLU 0 0 ZT3 ZT15

Figure S2 (related to Fig. 2). Deletion of hepatocyte KLF10 alters the circadian transcriptome in the liver. (A) Gene expression profiles of clock genes in the livers of Klf10flox/flox and Klf10Δhep mice (n = 3 pools of liver per time point). (B) Glucose uptake in Klf10flox/flox and Klf10Δhep mice (mean ± SEM, n = 6). (C) Liver glycogen content in Klf10flox/flox and Klf10Δhep mice measured at ZT3 and ZT15 (mean ± SEM, n = 5-6). Statistics: Non parametric Wilcoxon (B) or Kruskal and Wallis © test. * p < 0.05; ***, p <0.005. bioRxiv preprint doi: https://doi.org/10.1101/2020.12.22.423999; this version posted December 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

A B

0.8 200 Pellet Drink

0.6 ) 150 mg/dl

0.4 ( 100 TG TG

mass/day 0.2 50 Calorie/g body body Calorie/g 0.0 0 - SSW - SSW - SSW - SSW Klf10flox/flox Klf10Dhep Klf10flox/flox Klf10Dhep

Figure S3 (related to Fig. 3). Loss of hepatocyte KLF10 exacerbates the adverse effects associated with increased sugar consumption (A) Daily caloric intake in Klf10flox/flox and Klf10Δhep mice under the chow and chow + SSW dietary conditions (mean ± SEM, n = 8-12). (B) Blood triglyceride content at ZT15 in Klf10flox/flox and Klf10Δhep mice under the chow and chow + SSW dietary conditions (mean ± SEM, n = 7-9).

Slc2a2 Slc2a5 2.0 8 * *** 1.5 *** *** 6 1.0 4

0.5 2 Relative expression Relative Relative expression Relative 0.0 0 - SSW - SSW - SSW - SSW Klf10flox/flox Klf10Dhep Klf10flox/flox Klf10Dhep

Figure S4 (related to Fig.4). Altered metabolic gene expression in Klf10Δhep mice challenged with high sugar. Expression profiles of the genes encoding for facilitated glucose transporter Slc2a2 and the facilitated glucose/fructose transporter Slc2a5 at ZT15 in Klf10flox/flox and Klf10Δhep mice given a chow or chow + SSW diet (mean ± SEM, n = 6). Statistics: Non parametric Kruskal and Wallis test. *, p < 0.05; ***, p <0.005. bioRxiv preprint doi: https://doi.org/10.1101/2020.12.22.423999; this version posted December 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

A B Elovl6 Klf10flox/flox Klf10Δhep *** Klf10flox/flox + LG 15 **

Klf10flox/flox + HGF 3 12 Klf10Dhep + LG Up 504 ** Dhep

400 10 x 9 Klf10 + HGF

DEGs 6

208 CPM 3

341 Down 0

C Klf10flox/flox Klf10Δhep 1. edgeR FDR < 0.05 HGF vs LG HGF vs LG KEGG mouse 2019 gene sets DEGs DEGs

341 504 400 208 down- up- down- up- regulated regulated regulated regulated

2. enrichR + .gmt + .gmt + .gmt + .gmt FDR < 0.05 pathway enrichment pathway enrichment .gmt

LG HGF LG HGF

9 2 0 14

Figure S5 (related to Fig. 5). KLF10 governs the transcriptional response to hexose sugars in hepatocytes. (A) Total number of DEGs in Klf10flox/flox and Klf10Δhep hepatocytes treated with LG or HGF for 12 hours. (B) Expression profiles of Elovl6 in Klf10flox/flox and Klf10Δhep primary hepatocytes treated with LG or HGF (mean ± SEM, n = 3). (C) Schematic of the workflow used for the assessment of enriched gene sets in Klf10flox/flox and Klf10Δhep hepatocytes treated with LG or HGF. A data table containing the enrichment results (input set of genes overlapping with ‘KEGG mouse 2019 gene sets’) was downloaded from the EnrichR server. Gene sets with a q value (FDR) of < 0.05 were used for the dot plot visualization of enriched pathways shown in Figure 3C. Statistics: Non parametric Kruskal and Wallis test. **, p < 0.01; ***, p <0.005. bioRxiv preprint doi: https://doi.org/10.1101/2020.12.22.423999; this version posted December 23, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

ZT0 ZT3 ZT6 ZT9 ZT12 ZT15 ZT18 ZT21

KLF10

EF1a

Figure S6 (related to Fig. 6). KLF10 regulates a large metabolic network in liver Immunoblot showing the oscillation of KLF10 protein abundance in the liver of mice entrained in a LD12:12 light/dark cycle.