Acyl-CoA-binding protein (ACBP): a phylogenetically conserved appetite stimulator

Recently, we reported that, in mice, hunger causes the autophagy-dependent release of a protein called “acyl-CoA-binding protein” or “diazepam binding inhibitor” (ACBP/DBI) from cells, resulting in an increase in plasma ACBP concentrations. Administration of extra ACBP is orexigenic and obesogenic, while its neutralization is anorexigenic in mice, suggesting that ACBP is a major stimulator of appetite and lipo-anabolism. Accordingly, obese persons have higher circulating ACBP levels than lean individuals, and anorexia nervosa is associated with subnormal ACBP plasma concentrations. Here, we investigated whether ACBP might play a phylogenetically conserved role in appetite stimulation. We found that extracellular ACBP favors sporulation in Saccharomyces cerevisiae, knowing that sporulation is a strategy for yeast to seek new food sources. Moreover, in the nematode Caenorhabditis elegans, ACBP increased the ingestion of bacteria as well as the frequency pharyngeal pumping. These observations indicate that ACBP has a phylogenetically ancient role as a ‘hunger factor’ that favors food intake.


Introduction
Approximately 40% of the adult population of the United States is obese, and other countries follow the same trend or even attain higher proportions 1,2 . Logically, an entire industry proposes strategies for the behavioral, nutritional, and pharmacological treatment of obesity, a condition that nowadays is considered as a disease 3 , even before it leads to metabolic syndrome and entails comorbidities including diabetes, non-alcoholic fatty liver, atherosclerosis, and cancer 4 . Nowadays, obesity can be considered as the epidemiologically most important risk factor for premature aging, causing a marked reduction in both health span and lifespan [5][6][7][8][9] .
In spite of an ever-expanding scientific literature, current knowledge on the pathogenesis of excessive appetite is limited, and research performed in rodents has not been translated to human obesity. Thus, the first appetite-controlling hormone that has been characterized in mice, leptin 10 , is usually overabundant in obese persons, leading to the proposal of a "leptin resistance" that would account for the obesity-related hyperphagy 11,12 . Only a handful of obese patients bear genetic aberrations in leptin and its receptors that are equivalent to those encountered in Ob/Ob and Db/Db mice, which lack leptin or its receptor, respectively [13][14][15] . Similarly, a major appetite-stimulating hormone, ghrelin 16 , is paradoxically low in obese individuals 17,18 .
Recently, we identified acyl-CoA-binding protein (ACBP), also known as diazepam binding inhibitor (DBI), as a novel appetite stimulating factor 19 . Indeed, plasma concentrations of ACBP are elevated in obese patients, as well as in mice that were rendered obese by a high-fat diet or that became obese on a normal diet due to the Ob/Ob mutation. Neutralization of ACBP by suitable antibodies reduced multiple obesity-related aberrations including increased nutrient uptake, stimulated lipo-catabolism (lipolysis, triglyceride breakdown, fatty acid oxidation, and conversion of glycerol into glucose) and suppressed lipo-anabolism, thus reducing body weight, adiposity, diabetes, and steatosis. These findings could be recapitulated by inducible knockout of the Dbi gene. Thus, in contrast to the leptin and ghrelin systems, ACBP appears to play a convergent (rather than divergent) role in the obesity-associated hyperphagy of humans and rodents 19 .
ACBP is a small (13 kDa), phylogenetically conserved protein (Supplemental Fig. 1) that can be found in some eubacteria as well as all three eukaryotic kingdoms (plants, fungi and animals), meaning that it is more ancestral than leptin and ghrelin 20,21 . ACBP has the peculiarity to be secreted as a leaderless protein through a non-conventional (Golgi-dependent) pathway that depends on autophagy [22][23][24] . In human and mouse cells, ACBP also regulates autophagy. Both the depletion of intracellular ACBP and its addition to the extracellular milieu inhibit autophagy, suggesting that the autophagy-related translocation of ACBP from the intracellular to the extracellular compartment acts as a feedback control system to limit autophagy 19 .
Here, we investigated the possibility that ACBP would act as phylogenetically conserved regulator of autophagy and appetite in two model systems; namely, in the yeast Saccharomyces cerevisiae (that undergoes sporulation to seek new food sources) and the nematode Caenorhabditis elegans (which can actively search for food and accelerate pharyngeal pumping). We show that ACBP plays an evolutionarily ancient role in appetite control.

Results
Opposed autophagy-regulatory effects of ACBP in unicellular and multicellular organisms Knockout of S. cerevisiae ACB1 (the yeast of ACBP) inhibited autophagy during chronological aging (Fig. 1a-e), although this knockout did not affect maximum autophagy stimulated by rapamycin (Fig. 1f) or nitrogen starvation (Fig. 1g), as determined by assessing the proteolysis of green fluorescent protein (GFP) fused to autophagyrelated gene 8 protein (GFP-Atg8) to free GFP detectable Fig. 1 Autophagy regulation by ACBP in Saccharomyces cerevisiae. a, b Immunoblotting analysis of protein extracts from wild type (WT) and Δacb1 cells expressing a GFP-Atg8 fusion protein. Blots were probed with antibodies against GFP to detect GFP-Atg8 and free GFP, which is indicative of autophagic flux, and against GAPDH as loading control. Representative results (a) and densitometric quantification (b) at 1 and 2 days are shown. (n = 4). c Relative alkaline phosphatase (ALP) activity at 1, 4, and 6 days of chronological aging of WT and Δacb1 cells expressing Pho8pΔN60 (n = 3). d, e Fluorescence microscopy of WT and Δacb1 cells expressing a GFP-Atg8 chimera at day 2 of chronological aging. Propidium iodide (PI) counterstaining served to visualize dead cells. Scale bar = 5 μm. Autophagic cells were defined as cells with clear vacuolar GFP fluorescence and depicted as percentage of viable (PI − ) cells. Per strain and replicate, 500-650 cells were manually counted. (n = 5). f, g ALP activity of WT and Δacb1 cells expressing Pho8pΔN60 at the indicated times of chronological aging with or without 40 nM rapamycin (Rapa) (f) or upon nitrogen starvation (−N) for 4 h and 24 h (g) (n = 3-5). Quantitative results are reported as means ± SEM. Symbols indicate statistical (Student's t-test) comparisons with controls (n.s, not significant; *p < 0.05, ***p < 0.001).
by immunoblot (Fig. 1a, b), the enzymatic activity of alkaline phosphatase (ALP) Pho8 (Fig. 1c, f, g), or the redistribution of a GFP-Atg8 to the yeast vacuole detectable by fluorescence microscopy (Fig. 1d, e). Thus, in yeast, Acb1 acts as a facilitator of autophagy.
In sharp contrast, knockout of C. elegans acbp-1 (the nematode orthologous of ACBP), alone or together with several homologs acbp-3, acbp-4 and/or acbp-6 (which exist in this species but not in yeast nor in mammals) 25 , stimulated autophagy, as indicated by the subcellular redistribution of a GFP::LGG-1 fusion protein (LGG-1 is the nematode orthologous of yeast Atg8 and mammalian LC3) to cytoplasmic puncta (Fig. 2a, b) and the concomitant decrease of SQST-1/p62::GFP (the nematode orthologous of mammalian SQSTM1 fused to GFP) puncta (Fig. 2c, d). Knockdown of daf-2 (the insulin/ insulin growth factor 1 receptor) which induces autophagy 26 also decreased SQST-1/p62::GFP puncta, while knockdown of bec-1 (the nematode orthologous of mammalian BECN1) robustly increased them, proving that this reporter can be reliably utilized for measuring autophagic flux (Fig. 2c, d). Twelve hours of starvation led to a similar decrease of SQST-1::GFP particles in control animals and acbp-1;3 and acbp-4;6 knockout worms (Fig. 2e). Of note the increase in autophagy induced by deletion of acbp genes was partially reduced by mutation of aak-2 (an orthologous of human PRKAA1 and PRKAA2, which encode subunits of AMP activated kinase, AMPK) which is implicated in autophagy induction via ULK-1 phosphorylation 27 . However, knockdown of acbp genes was unable to induce a further increase in autophagy in daf-2 mutants (which lack a functional insulin/insulin growth factor 1 receptor) (Supplemental Fig. 2). Thus, in nematodes, acbp genes act as endogenous inhibitors of autophagy.

Convergent effects of ACBP depletion on feeding behavior in yeast and nematodes
In the next step, we determined whether the effect of ACBP on feeding behavior is phylogenetically conserved.
Yeast is devoid of ameboid movement, and the only possibility for this organism to seek new sources of nutrients consists in sporulation, which occurs in response to prolonged exhaustion of external resources 28 . The knockout of S. cerevisiae ACB1 (Δacb1) led to a defect in sporulation and this sporulation defect of Δacb1 cells was blunted by adding recombinant yeast Acb1 (yAcb1) protein to the cultures (Fig. 3a, b), confirming that Acb1 stimulates sporulation. To investigate possible non-cell-autonomous effects of ACBP on sporulation, we co-cultured WT Hho1-mCherry tagged cells (which emits a red fluorescence) with Δacb1 Hho1-GFP cells (expressing green fluorescent protein (GFP)). As controls these fluorescent protein tagged strains were co-cultured with respective non-fluorescent variants. This procedure revealed that Δacb1 yeast co-cultured with Δacb1 yeast cells exhibit a sporulation defect that is attenuated in the presence of WT cells (Fig. 3c). Thus, the sporulation defect of Δacb1 cells could not only be partially rescued by adding recombinant yAcb1 protein to the cultures but was also blunted by co-culturing the cells with Acb1expressing yeast cells (Fig. 3a-c), confirming that Acb1 stimulates sporulation. Finally, the rescue of the sporulation defect by Acb1 protein depended on Ste3 (but not Ste2), which is one of the two G protein-coupled receptors encoded by the yeast genome (Fig. 3d). These results indicate that extracellular Acb1 protein can act on Ste3 receptors to stimulate sporulation in yeast.
Next, we turned to the nematode model. In C. elegans, knockout of one or several genes coding for ACBP orthologous (acbp-1, acbp-3, acbp-4, and/or acbp-6) reduced the uptake of bacteria expressing red fluorescent protein (RFP) both in ad libitum feeding conditions (Fig. 4a, b) and after 12 h of starvation and refeeding (Fig. 4c, d). This result was confirmed by the analysis of pharyngeal pumping, revealing that removal of acbp-1, Fig. 3 Role of Acb1 in the sporulation of S. cerevisiae. a, b Representative pictures and quantification of sporulation frequencies of wild type (WT) and Δacb1 cells with or without the addition of recombinant yeast Acb1 (yAcb1). Addition of yAcb1 to Δacb1 cultures partially reversed the sporulation frequency defect induced by the ACB1 deletion. c Representative pictures and quantification of sporulation frequencies of yeast cells with the indicated genotypes that were co-cultured are shown. Note that co-culturing of the Δacb1 strain (Hho1-GFP tagged) with a WT strain (Hho1-mCherry tagged) reduced the sporulation defect of this mutant. d Additional deletion of STE3 (coding for membrane receptor, that couples factor a pheromone binding to a MAP kinase cascade) renders mutants resistant to external yAcb1 regarding sporulation frequency, whereas deletion of STE2 (coding for alpha factor membrane receptor) did not affect the response to extracellular yAcb1. Results are expressed as means ± SEM (n = 3-6). n.s, not significant; *p < 0.05, ***p < 0.0001 (Student's t-test), as compared to untreated controls (Ctrl) or $ p < 0.05 (Student's t-test), as compared within acb1 deficient conditions. acbp-3, acbp-4, and/or acbp-6 reduced food intake (Supplemental Fig. 3) in C. elegans. The generation of acbp-1(sv62);daf-2(e1370) double mutants revealed that the daf-2 mutation is epistatic to the acbp-1 mutation, since the acbp-1;daf-2 double mutants behave quite similarly to single daf-2 mutants in respect to feeding (Fig. 4e, f). These observations indicate that the worm orthologous of ACBP stimulate feeding behavior, as additionally corroborated by comparisons with tax-4 (p678) mutants (Fig. S4A, B) or animals treated with clozapine ( Fig. S4C, D). TAX-4 is a nucleotide-gated channel which is broadly expressed in the C. elegans nervous system and in ASI neurons which mediate satiety quiescence 29,30 , while clozapine is a second generation antipsychotic drug, which inhibits pharyngeal pumping of nematodes 31 . Of note, acbp-1(sv62) mutants seem more sensitive to clozapine administration compared to their wild-type counterparts (Fig. S4C, D).

Discussion
ACBP is an evolutionarily ancient protein, based on sequence alignments and structural similarities suggesting that the physicochemical properties of this protein have been conserved throughout the eukaryotic radiation 20,21,32 . The present data suggest that the function of ACBP as a regulator of appetite is phylogenetically conserved as well.
At a first level, knockout of the gene coding for the (single) ACBP orthologous from yeast reduces sporulation, while its addition in the form of a recombinant protein restores sporulation, presumably through an action on Ste3, which is one of the two signaltransducing receptors present in this species. Ste3, a seven transmembrane G protein-coupled receptor, is best known to for the peptide pheromone alpha1factor 33 , the mating factor of yeast, pointing to an intriguing cross-talk between the signal transduction pathways involved in mating and in the control of food intake. At a second level, knockout of the genes coding for the (several) ACBP orthologous from C. elegans reduces the uptake of bacteria as it reduces the frequency of pharyngeal pumping. Together, these findings favor the contention that the appetite control function of ACBP is conserved throughout evolution. Indeed, these results echo a previous report on a gene named Anorexia (Anox) that codes for an acyl-CoA-binding protein with an ankyrin repeat domain and that, if mutated, reduces feeding activity and mouth hook movement (the fly equivalent of mastication) in Drosophila melanogaster 34 . That said, Anox has been involved in central appetite control because it is mostly expressed in the central nervous system and in ganglions, differing from mouse ACBP that has been attributed a predominantly peripheral role in appetite control 19 .
The effects of ACBP on autophagy appear to be distinct in yeast and in nematodes. Thus, removal of the ACBP orthologous from yeast inhibited autophagy during chronological aging, contrasting with the observation that, in C. elegans, deletion of ACBP orthologous resulted in enhanced autophagy. As a possibility, these seemingly contradictory results reflect the distinct cellular organization of these species (monocellular for yeast, multicellular for nematodes), as well as the differential effects of intracellular vs. extracellular ACBP on autophagy. Indeed, when (mostly intracellular) ACBP is depleted from cultured human cells by RNA interference, this results in autophagy inhibition. However, when (mostly extracellular) ACBP is neutralized by antibodies, this results in autophagy induction, both in cultured human cells and in mice 19 . Thus, removal of ACBP from S. cerevisiae might reflect a situation in which physiological effects are secondary to the depletion of intracellular ACBP, while removal of ACBP from C. elegans might reflect the effects of a reduction in extracellular ACBP.
Irrespective of the aforementioned uncertainties, it appears clear that ACBP plays a major appetitestimulatory role throughout eukaryotic evolution, meaning that it triggers a range of different feeding behaviors in yeast (sporulation), nematodes (pharyngeal pumping), insects (mouth hook movement) and mammals (food intake). Since it operates independently from the leptin and ghrelin systems (which only exist in mammals) 35,36 , ACBP may indeed represent the phylogenetically most ancient "hunger factor". In a plausible scenario, starvation causes autophagy, resulting in the release of ACBP from cells, and ACBP then acts on cell surface receptors to stimulate feeding behaviors. In this sense, ACBP would act as a neuroendocrine factor that participates in a primate homeostatic feedback loop or 'hunger reflex' designed to mitigate the effects of nutrient deprivation.
Pharyngeal pumping was measured as previously described 39 . Grinder movements of free-moving animals were measured under the stereomicroscope. Three independent measurements were performed for each individual and the average number of pumps per animal was recorded. Starvation was performed by placing the animals on NGM plates without bacterial lawn for 12 h. For assessing food intake using fluorescent bacteria, we fed synchronized day one (D1) adult animals for 5 min with HT115 bacteria transformed with a IPTG-inducible RFP expressing plasmid (modification of a previous protocol published 40 ). Upon this short feeding period, the animals were immediately immobilized with levamisole and mounted on slides for microscopic observation with a Zeiss AxioImager Z2 epifluorescence microscope. Image J software was used for the quantification of mean RFP intestinal fluorescence. Clozapine (Sigma-Aldrich, product number: C6305) was diluted in 100% ethanol (stock 11 mg/mL) and was added before pouring of NGM plates at a final concentration of 200 μg/mL per plate, as previously described 31 . Plates with solvent alone (1.8% ethanol per plate) were used for comparison. Synchronized animals at day 1 of adulthood were placed overnight (14-16 h) on clozapine or ethanol-containing plates to avoid undesired developmental effects which have been previously described 31 ). Ingestion of RFP+ bacteria at day 2 adult animals was measured as described above.

S. cerevisiae experiments Strains
Hho1 GFP-tagged or mCherry-tagged variants thereof (L5366 mCherry and L5366 Δacb1 GFP) were used. HHO1 codes for the Histone H1 protein. Analyses of autophagy in yeast were carried out in BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) and respective Δacb1 mutant obtained from EUROSCARF. Strains were grown at 28°C on synthetic minimal medium containing 0.17% yeast nitrogen base (Difco), 0.5% (NH 4 ) 2 SO 4 and 30 mg/L of all amino acids (except 80 mg/L histidine, 120 mg/L lysine, and 200 mg/L leucine), 30 mg/L adenine and 320 mg/L uracil with 2% glucose. For chronological aging, cells were inoculated to OD 600nm 0.1 from fresh overnight cultures and grown at 28°C. Where indicated, cultures were supplemented with 40 nM rapamycin (AG Scientific; 1 mg/mL stock in DMSO) at the time point of inoculation. For nitrogen starvation, cells where grown to OD 600nm 1 on synthetic minimal medium and then transferred to medium without amino acids and (NH 4 ) 2 SO 4 .

Autophagy measurement in yeast
peroxidase-conjugated secondary antibodies (Sigma). Densitometric quantification was performed with Image Lab 5.2 Software (Bio-Rad). Quantification and statistical analysis: Micrographs of cells expressing GFP-Atg8 were manually counted, and 500-650 cells were evaluated per strain and per experiment. Thereby, cells displaying clear vacuolar GFP fluorescence were scored as autophagic cells and were depicted as percentage of viable (PI negative) cells. Data represents mean of five independent experiments. Densitometric quantification of immunoblots was performed with Image Lab 5.2 Software (Bio-Rad), and the ratio Free GFP/GAPDH was plotted. Data represent mean ± SEM of four independent experiments. Data showing ALP activity represent mean ± SEM of three independent experiments. Statistical analyses were performed using Students T-test (one-tailed, unpaired), with *p < 0.05, **p < 0.01, and ***p < 0.001.

Microscopy studies
Microscopy was performed as previously described 51 In short, for sporulation experiments with addition of yAcb1, cells were harvested after 3 days and stained with PI (0.1 µg/mL in 1×PBS).
Cells were viewed and documented by fluorescence microscopy with the use of a small-band dsRed filter (Zeiss) on a Zeiss Axioskop microscope. Sample images were taken with a Diagnostic Instruments camera (Model: SPOT 9.0 Monochrome-6), acquired and processed (coloring) using the Metamorph software (version 6.2r4, Universal Imaging Corp.) Subsequently, pictures were quantified by evaluating the amount of cells, which were sporulated, and the cells that were stained red (dead cells), relative to all pictured cells. At least 250-900 cells per strain per independent experiment were manually counted.
For sporulation co-culturing experiments with Hho1tagged mCherry and GFP strains, cells were harvested after 3 days. Cells were viewed and documented by fluorescence microscopy with the use of a small-band eGFP and dsRed filter (Zeiss) on a Zeiss Axioskop microscope. Microscopy pictures were quantified by evaluating the amount of sporulated and un-sporulated cells of wild-type vs. acb1 deletion strains, compared to their respective mono-cultures. 250-900 cells per strain per experiment were manually counted. Data represent results of four independent experiments. Statistical analyses for sporulation experiments were performed using Students t-test (one-tailed, unpaired), with *p < 0.05, **p < 0.01, and ***p < 0.001.

Statistical analysis
Data are reported as the mean ± standard deviation (SD), mean ± standard error of the mean (SEM), or Box and whisker plots (mean, first and third quartiles, and maximum and minimum values) as specified. The number of independent data points (n) is indicated in the figure legends of the corresponding graphs or in the legends. For statistical analyses, p values were calculated by two-way ANOVA, one-way ANOVA with Tukey's multiple comparisons test, two-tailed unpaired Student's t-test, Wilcoxon matched pairs signed rank test, Pearson's coefficients of correlation (R) or false discovery rate (FDR) as indicated (Prism version 7, GraphPad Software). Differences were considered statistically significant when pvalues *(p < 0.05), **(p < 0.01), ***(p < 0.001) and n.s. = not significant (p > 0.1).