ACLY ubiquitination by CUL3-KLHL25 induces the reprogramming of fatty acid metabolism to facilitate iTreg differentiation

Inducible regulatory T (iTreg) cells play a central role in immune suppression. As iTreg cells are differentiated from activated T (Th0) cells, cell metabolism undergoes dramatic changes, including a shift from fatty acid synthesis (FAS) to fatty acid oxidation (FAO). Although the reprogramming in fatty acid metabolism is critical, the mechanism regulating this process during iTreg differentiation is still unclear. Here we have revealed that the enzymatic activity of ATP-citrate lyase (ACLY) declined significantly during iTreg differentiation upon transforming growth factor β1 (TGFβ1) stimulation. This reduction was due to CUL3-KLHL25-mediated ACLY ubiquitination and degradation. As a consequence, malonyl-CoA, a metabolic intermediate in FAS that is capable of inhibiting the rate-limiting enzyme in FAO, carnitine palmitoyltransferase 1 (CPT1), was decreased. Therefore, ACLY ubiquitination and degradation facilitate FAO and thereby iTreg differentiation. Together, we suggest TGFβ1-CUL3-KLHL25-ACLY axis as an important means regulating iTreg differentiation and bring insights into the maintenance of immune homeostasis for the prevention of immune diseases.


Introduction
As a critical subset of immunosuppressive cells, regulatory T (Treg) cells play a central role in the suppression of aberrant or excessive immune responses and are critical for the maintenance of immune homeostasis (Dominguez-Villar and Hafler, 2018). Defects in Treg frequency or function are often associated with distinct autoimmune disorders, such as system lupus erythematosus (SLE), rheumatoid arthritis (RA), and inflammatory bowel disease (IBD) (Sakaguchi et al., 2008;Tao et al., 2017). Treg cells are composed of thymus-derived natural Treg (nTreg) and peripheral inducible Treg (iTreg) cells, which are originated and matured at different locations in the body (Lee et al., 2009;Noack and Miossec, 2014). In general, nTreg cells are dominant in the periphery under unperturbed conditions, whereas iTreg cells are often boosted upon special immune challenges to balance inflammatory responses (Murai et al., 2010). Furthermore, enhancing iTreg cells by stimulating iTreg differentiation or employing adoptive transfer appeared to be effective in the therapeutic treatment for several immune diseases (Dall'Era et al., 2019;Esensten et al., 2018;Gö schl et al., 2019;Ryba-Stanisławowska et al., 2019).
Accumulating evidence has revealed that different types of T cells are usually associated with distinct metabolic characteristics (Chen et al., 2015;Kempkes et al., 2019). For instance, Th0, Th1, Th2, and Th17 prefer de novo fatty acid synthesis (FAS) that substantially supports anabolism to meet the need for biological macromolecules during rapid proliferation (Lochner et al., 2015;Ma et al., 2017;Maciolek et al., 2014;Wang et al., 2011). In contrast, iTreg relies on fatty acid oxidation (FAO) (Chen et al., 2015;Gerriets et al., 2015;Ma et al., 2017;Michalek et al., 2011). Because carnitine palmitoyltransferase 1 (CPT1) is the rate-limiting enzyme in FAO, upregulating CPT1 by its substrate palmitate and downregulating CPT1 by the specific inhibitor etomoxir, respectively, could result in enhancement and impairment in FAO and iTreg differentiation (Gualdoni et al., 2016;Michalek et al., 2011). Obviously, FAS and FAO are reciprocal pathways, only one of which can be dominant in a specific type of T cells (Foster, 2012;Wolfgang and Lane, 2006). During iTreg differentiation, cell metabolism undergoes a series of dramatic changes, including a shift from FAS to FAO. Although the downregulation of FAS appears to be important (Berod et al., 2014), the mechanism regulating FAS in the process of iTreg differentiation is still unclear.
FAS is directly controlled by a series of metabolic enzymes in cytoplasm, including three rate-limiting enzymes ATP-citrate lyase (ACLY), acetyl-CoA carboxylase (ACC), and fatty acid synthase (FASN) (Lochner et al., 2015). In brief, ACLY converts mitochondrial-derived citrate into acetyl-CoA and oxaloacetic acid (OAA), and provides the main acetyl-CoA source for de novo FAS. ACC catalyzes acetyl-CoA into malonyl-CoA to provide an active two-carbon-unit donor for carbon chain extension. FASN uses both malonyl-CoA and acetyl-CoA as substrates to continuously synthesize long-chain fatty acids. Although regulation of the three enzymes at the transcriptional level was reported (Kidani et al., 2013), direct modulation of their activities by posttranslational modifications (PTMs), particularly ubiquitination that is tightly coupled with protein degradation, has set a new stage for the exploration of FAS control. For instance, nutrient deficiency induced ACLY ubiquitination at K540/K546/K554 in human lung cancer cells. This change resulted in the degradation of ACLY proteins and suppressed de novo lipid synthesis, eventually causing impairment in cell proliferation (Lin et al., 2013;Zhang et al., 2016). ACC stability as well as de novo lipid synthesis in mouse adipocytes was found to be regulated by ubiquitination (Qi et al., 2006). Additionally, FASN ubiquitination in HEK293T cells was also linked to its degradation, which in turn compromised de novo lipid synthesis (Lin et al., 2016). However, it still remains poorly understood whether and how ubiquitination is involved in the control of FAS during iTreg differentiation.
In this study, we have revealed that ACLY activity declined significantly as iTreg cells were differentiated. Upregulation in ACLY expression hindered the switch from FAS to FAO and thereby impaired iTreg differentiation. ACLY-dependent malonyl-CoA, an intermediate in FAS, could directly finetune CPT1 activity and therefore FAO. In response to TGFb1 stimulation, CUL3-KLHL25 mediated ACLY ubiquitination to facilitate its degradation. When we removed CUL3 to block ACLY degradation, iTreg differentiation and colitis alleviation by adoptive iTreg cells in mice were compromised. Simultaneous depletion of ACLY partially rescued CUL3 deficiency-induced defects in iTreg differentiation and colitis alleviation. Overall, we demonstrated the regulation of iTreg differentiation via TGFb1-CUL3-KLHL25-ACLY axis and bring insights into the maintenance of immune homeostasis for the prevention of colitis.

ACLY is critical for the efficient differentiation of iTreg cells
To understand the alteration in FAS during iTreg differentiation, we examined the three rate-limiting enzymes ACLY, ACC, and FASN in FAS pathway. It appeared that the enzymatic activity of ACLY was reduced markedly, while ACC and FASN activities remained largely unchanged ( Figure 1A, Figure 1-figure supplement 1A). Meanwhile, acetyl-CoA and OAA, the two products of ACLY-mediated reaction, were also reduced upon TGFb1 treatment, confirming the decreased ACLY activity during iTreg differentiation ( Figure 1B, Figure 1-figure supplement 1B). Importantly, preventing the decline in ACLY activity by supplementing cells with exogenous ACLY drastically impaired iTreg differentiation ( Figure 1C, Figure 1-figure supplement 1C). In contrast, knocking down Acly by small interfering RNAs (siRNAs) or blocking ACLY activity by the specific inhibitor SB204990 led to elevated iTreg differentiation ( Figure 1D-E, Figure 1-figure supplement 1D-E). Together, these data suggest that ACLY regulates iTreg differentiation.
ACLY is required for the reprogramming of fatty acid metabolism during iTreg differentiation ACLY converts citrate into acetyl-CoA and OAA, and acetyl-CoA is the main source supporting de novo FAS, mevalonate-cholesterol synthesis, and histone acetylation (Zaidi et al., 2012). understand ACLY-dependent regulation of metabolic reprogramming during iTreg differentiation, we performed [U-13 C] glucose-tracing experiment. Upon ACLY inhibition, 13 C-labeled fatty acids were reduced as expected ( Figure 2A). However, key intermediates in mevalonate-cholesterol synthesis, including 3-hydroxy-3-methyl glutaryl coenzyme A (HMG-CoA), mevalonate, mevalonate 5pyrophosphate, and cholesterol remained rather stable irrespective of the presence of ACLY inhibitor ( Figure 2A). Considering that palmitate, the terminal product of FAS, may not be the limiting factor coordinating ACLY-dependent impact and iTreg differentiation, other intermediate products in FAS can be potentially at play (Figure 2-figure supplement 4B). Next, we targeted key enzymes in FAS separately with different inhibitors to gain insights into the pivotal metabolite(s). While FASN inhibition abolished ACLY-inhibition-induced iTreg differentiation, the inhibition of ACC barely had any effect (Figure 2-figure supplement 4C). This strongly indicates that malonyl-CoA, the substrate of FASN, might be important for ACLY-inhibition-induced iTreg differentiation. Upon ACLY inhibition, we did observe a clear reduction in malony-CoA (Figure 2-figure supplement 4D). More importantly, add-back of malonyl-CoA completely abolished ACLY-inhibition-induced iTreg differentiation ( Figure 2C). These data suggest that ACLY-dependent regulation of iTreg differentiation is probably through malonyl-CoA.
Because malonyl-CoA is a well-characterized metabolic intermediate capable of inhibiting CPT1, the key enzyme in FAO, we asked whether ACLY inhibition could regulate CPT1 activity and thereby FAO to promote iTreg differentiation (Figure 2-figure supplement 4B). When ACLY was inhibited by SB204990, both CPT1 activity ( Figure 2D) and FAO ( Figure 2E) were enhanced. Under the same condition, upregulating the cellular level of malonyl-CoA by exogenous supplementation of malonyl-CoA or FASN inhibition abrogated ACLY-inhibition-induced upregulation in CPT1 activity ( Figure 2D

TGFb1 induces ACLY ubiquitination and degradation during iTreg differentiation
To further understand the decline in ACLY activity during iTreg differentiation, we first isolated ACLY proteins from Th0 and iTreg cells and normalized them to the same level ( Figure 3-figure supplement 1A-B). Interestingly, both ACLY activity and Ser455 phosphorylation, which is a marker directly reflecting the enzymatic activity of ACLY (Das et al., 2015;Sivanand et al., 2017), remained unchanged ( Figure 3-figure supplement 1A-B). However, a dramatic decrease in ACLY protein, but not in its mRNA, was observed in the process of iTreg differentiation ( Figure 3A, Figure 3-figure supplement 1C), strongly arguing a mechanism regulating ACLY at its protein level.
Given that protein levels are controlled by both synthesis and degradation, we next examined ACLY protein level upon the treatment of cycloheximide (CHX), a well-documented ribosome inhibitor, often used for blocking protein synthesis (Schneider-Poetsch et al., 2010). Our time-course analysis showed that TGFb1 stimulation caused a reduction in ACLY protein even in the presence of CHX ( Figure 3B), indicating that protein degradation played an important role in TGFb1-mediated ACLY downregulation. Indeed, when we blocked protein degradation with MG132, which inhibits ubiquitination-proteasome degradation , TGFb1-induced downregulation in ACLY was abolished ( Figure 3C). In contrast, the addition of leupeptin, which blocks autophagy-lysosome degradation , did not affect ACLY protein level ( Figure 3C). These results  Figure 1E for 24 hr in the presence of [U-13 C] glucose (11 mM). Cells were collected and subjected to metabolic flux analysis for FAS by ultra-high performance liquid chromatography-high resolution mass spectrometry (UHPLC-HRMS) analysis. n = 4, Student's t-test, suggested that, during iTreg differentiation, TGFb1 induced ACLY downregulation via ubiquitination-proteasome degradation.
It is known that CUL3-KLHL25 complex belongs to Cullin-RING ubiquitin ligase family, the largest class of ubiquitin ligases (Chen and Chen, 2016;Zheng and Shabek, 2017). In brief, CUL3 is the core scaffolding protein holding the entire complex together, while KLHL25 serves as an adaptor for substrate recognition . When we knocked down Cul3 or Klhl25 with targeting siRNAs, ACLY ubiquitination was dramatically reduced ( Figure 4B Figure 4D-E). These data demonstrated that TGFb1-mediated ACLY ubiquitination and degradation rely on CUL3-KLHL25.
In addition, we generated CD4 + T-cell-specific CUL3-deficient mice (Cd4 Cre Cul3 fl/fl ) by crossing Cd4 Cre mice (Dong et al., 2008) with Cul3  mean ± SD. *p<0.05, **p<0.01, ****p<0.0001; ns, nonsignificant. (B, C) Functional evaluation of metabolic intermediates from FAS on iTreg differentiation. Naive CD4 + T cells were treated with SB204990 (100 mM) and cultured as in Figure 1E in the presence of different doses of palmitate (B) or malonyl-CoA (C). CD4 + CD25 + Foxp3 + iTreg cells were assayed by flow cytometry (FCM) (left) and quantified (right). (D, E) ACLY inhibition increases carnitine palmitoyltransferase 1 (CPT1) activity and fatty acid oxidation (FAO). In the presence of malonyl-CoA (50 mM) or cerulenin (4 mM), naive CD4 + T cells treated with SB204990 were cultured as in Figure 1E for 24 hr. Cell lysates were used for analyzing CPT1 activity (D). For oxygen consumption rate (OCR) detection, cells were transferred to XF Base Medium containing palmitate and carnitine. Diagram (left) illustrating the OCR at various conditions and associated quantifications (right) are shown (E). (F) Impact of Cpt1 knockdown on iTreg differentiation. Naive CD4 + T cells transfected with small interfering RNA (siRNA) against Cpt1 were cultured as in Figure 1E. CD4 + CD25 + Foxp3 + iTreg cells were assayed by FCM (left) and quantified (right). (B-F) Data represent mean ± SD of three (D-F) or four (B, C) independent experiments, with significance determined by one-way analysis of variance (ANOVA) test. **p<0.01, ***p<0.001, and ****p<0.0001; ns, nonsignificant. The online version of this article includes the following figure supplement(s) for figure 2:      Figure 1A were subjected to western blot (WB) with antibodies against ACLY (left). Quantification of ACLY levels (right). (B) Impact of transforming growth factor b1 (TGFb1) on ACLY degradation. Naive CD4 + T cells were cultured as in Figure 1A for 24 hr and treated with cycloheximide (CHX) (10 mg/ml) as indicated prior to WB analysis (left). Quantification of ACLY levels (right). (C) Pathway analysis for ACLY degradation. Th0 and iTreg cells polarized as in Figure 1A for 24 hr were treated with MG132 (left) or leupeptin (right) and analyzed with WB for ACLY protein. (D, E) Impact of TGFb1 on ACLY ubiquitination. Cells polarized as in Figure 1A for 24 hr were treated with MG132 (10 mM) and cell extracts were immunoprecipitated with antibodies against ACLY (D) or ubiquitin (Ub) (E). IgG serves as a negative control. (F-H) Functional comparison between ACLY WT and ACLY 3KR . Naive CD4 + T cells transfected with GFP-ACLY WT or -ACLY 3KR were polarized as in Figure 1A for 24 hr. (F) Extracts from cells pre-treated with MG132 were assayed with immunoprecipitation (IP) for ACLY ubiquitination. IgG serves as a negative control. (G) WB analysis for ACLY protein. (H) Cell lysates were prepared for the analysis of carnitine palmitoyltransferase 1 (CPT1) activity. (I) ACLY ubiquitination impacts on iTreg differentiation. Cells transfected with GFP-ACLY WT or -ACLY 3KR were polarized as in Figure 1A and assessed with flow cytometry (FCM) (left). Quantification of GFP + CD25 + Foxp3 + iTreg cells (right). (A-C, G-I) Quantification shows mean ± SD based on three independent experiments. Student's t-test (A), two-way analysis of variance (ANOVA) test (B), or one-way ANOVA test (C, G-I) was used. *p<0.05, . Notably, when naive CD4 + T cells from these CUL3-deficient mice were induced into iTreg cells in vitro, both ACLY ubiquitination and degradation induced by TGFb1 were blocked ( Figure 4F-H). These observations fully recapitulated what we previously found in siRNA-mediated CUL3-depleted cells. More importantly, alterations induced by CUL3 deficiency, including increase in malonyl-CoA level and decrease in CPT1 activity and iTreg differentiation, were dampened by ACLY depletion (Figure 4-figure supplement 3D, Figure 4I-J). Together, TGFb1 induces CUL3-KLHL25-mediated ACLY ubiquitination and degradation for iTreg differentiation.

Activated T (Th0) and inducible regulatory T (iTreg) cells prepared as in
It is worth mentioning that the suppressive function of iTreg cells did not seem to be affected by ACLY ubiquitination and degradation. When we knocked down Cul3 and/or Acly with targeting siRNAs, levels of distinct surrogate molecules typically associated with the suppressive activity of iTreg cells, including CTLA4, ICOS, TIGIT, and IL-10, remained rather stable (Figure 4-figure supplement 4A). In agreement with this observation, we took the advantage of the available Foxp3 YFP-Cre mice (Priyadharshini et al., 2018) and induced YFP-iTreg cell differentiation under various conditions in vitro. We found that YFP-iTreg cells lacking CUL3 alone or in combination with ACLY could still retain the potent suppressive function (Figure 4-figure supplement 4B). Hence, instead of the suppressive function of iTreg, ACLY ubiquitination and degradation mainly affect iTreg differentiation.

ACLY inhibition alleviates mouse colitis by promoting iTreg cell generation
As a common digestive system disease, IBD poses a serious threat to public health, and the number of patients with IBD keeps growing (Bernstein et al., 2010). Manipulation of iTreg functions holds promise for the therapeutic treatment of IBD (Clough et al., 2020;Izcue et al., 2006;van Herk and Te Velde, 2016). Next, we performed a functional assay for CUL3-KLHL25-mediated ACLY ubiquitination in a classic mouse IBD model . Same amount of naive CD4 + T cells was isolated from wild-type (WT) or Cd4 Cre Cul3 fl/fl (Cul3 -/-) mice and transfected with siRNAs prior to polarization to obtain Th0 (WT), iTreg (WT), iTreg (Cul3 -/-), or iTreg (Cul3 -/-+siAcly) cells ( Figure 6A). Compared to control, the lack of CUL3 impaired iTreg differentiation, while ACLY depletion compromised CUL3-deficiency-induced changes (Figure 6-figure supplement 1, Figure 6B). Upon the adoptive transfer of these cells along with WT-mice-derived naive CD4 + T cells used to induce colitis into Rag1 -/recipient mice ( Figure 6A), Th0 (WT) group suffered serious colitis lying in body weight loss, disease activity index (DAI) score increase, colon length shortening,  TGFβ1 -  Figure 1A for 24 hr were treated with MG132 and assayed by immunoprecipitation (IP) with ACLY antibody. IgG serves as a negative control. (B-E) Functional assessment of CUL3-KLHL25 depletion in inducible regulatory T (iTreg) differentiation. Naive CD4 + T cells transfected with desired small interfering RNAs (siRNAs) were polarized as in Figure 1A for 24 hr (B-D) or 72 hr (E). They were pre-treated with MG132 before IP for Figure 4 continued on next page inflammatory cell infiltration, mucosal edema and injury as well as crypt damage increase ( Figure 6C-E). As expected, iTreg (WT) mice showed significantly alleviated pathological alterations associated with colitis ( Figure 6C-E). When we injected recipient mice with iTreg (Cul3 -/-) cells, less alleviation in colitis was observed as compared to that in iTreg (WT) group ( Figure 6C-E). Importantly, knocking down Acly in iTreg (Cul3 -/-+siAcly) cells prior to the adoptive transfer rescued CUL3deficiency-induced failures in colitis alleviation ( Figure 6C-E). Using the same T-cell adoptive transfer strategy, we found that direct inhibition of ACLY by SB204990, which led to elevation in iTreg population from both mesenteric lymphatic node (MLN) and colonic lamina propria (cLP) and alleviation of dextran sodium sulfate (DSS)-induced colitis ( Figure 6-figure supplement 2A-F), also effectively relieved T-cell-transfer-induced colitis in mice ( Figure 6-figure supplement 3A-E). Together, these in vivo data confirm the important role of CUL3-KLHL25-mediated ACLY ubiquitination in colitis alleviation.

Discussion
iTreg cells are essential for the regulation of immune homeostasis, and their establishment relies on TGFb1 induction (Chen et al., 2003;Dominguez-Villar and Hafler, 2018). Recent findings suggest that a shift from FAS to FAO is required for efficient iTreg differentiation (Berod et al., 2014;Michalek et al., 2011), but the control of this process is still unclear. Here we unveiled a TGFb1-CUL3-KLHL25-ACLY axis that is important for the metabolic reprogramming in fatty acid metabolism during iTreg differentiation.
As one of the most well-studied signaling pathways, TGFb1 governs a number of signaling cascades, including canonical Smad signals and non-canonical MAPK (ERK/JNK/P38), PI3K/Akt, and Rho-like GTPase signals, mainly by regulating different kinases to phosphorylate their downstream factors (Vander Ark et al., 2018;Zhang, 2017). In the process of iTreg differentiation, TGFb1 activates TGFRI/II, which are transmembrane protein serine/threonine kinases and directly phosphorylate Smad2/3, to improve Foxp3 expression (Schlenner et al., 2012). Interestingly, our work here demonstrated that, through ubiquitination of a key metabolic enzyme, TGFb1 launched a reprogramming in fatty acid metabolism and promoted iTreg differentiation. These findings indicate Figure 4 continued the detection of ACLY ubiquitination (B), analyzed by western blotting (WB) with indicated antibodies (C), evaluated for carnitine palmitoyltransferase 1 (CPT1) activity (D), and stained for CD4/CD25/Foxp3 to identify iTreg cells prior to flow cytometry (FCM) analysis (E). (F-H) Functional assays for ACLY ubiquitination in cells derived from Cul3 knockout mice. Naive CD4 + T cells isolated from wild-type (WT) or Cd4 Cre Cul3 fl/fl mice were polarized as in Figure 1A for 24 hr. Cells were pre-treated with MG132 before IP with antibodies against ACLY (F) or ubiquitin (Ub) (G), or analyzed for ACLY expression with WB (H). (I, J) Impact of CUL3 deficiency on fatty acid oxidation (FAO) and iTreg differentiation. Naive CD4 + T cells transfected with siRNAs were induced as in Figure 1A for 24 hr (I) or 72 hr (J). Cells were lysed for CPT1 activity detection (I) or stained for CD4/CD25/Foxp3 to identify iTreg cells prior to FCM analysis (J). (C-E, H-J) Quantification shows mean ± SD based on three independent experiments, with significance determined by one-way analysis of variance (ANOVA) test. *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001; ns, nonsignificant. For WB in (A-D, F-I), one representative experiment out of three is represented. Associated scores indicate mean intensities based on three biological replicas. The online version of this article includes the following figure supplement(s) for figure 4:    T-cell proliferation was evaluated using flow cytometry (FCM) (left). Quantification analysis for CD4 + T-cell proliferation (right) based on three independent experiments. One-way analysis of variance (ANOVA) test, mean ± SD. **p<0.01, ***p<0.001, ****p<0.0001. (C-E) Systematic evaluation of Figure 6 continued on next page responses, and cell apoptosis (Davidge et al., 2019;Genschik et al., 2013). The lack of CUL3 has been associated with defects in embryogenesis, hypertension, diminished angiogenesis, progressive interstitial inflammation, exacerbated colonic inflammation, and swelled spleens and lymph nodes Li et al., 2017;Maekawa et al., 2017;Mathew et al., 2012;Saritas et al., 2019;Singer et al., 1999). Interestingly, some of these defects, for example, progressive interstitial inflammation, exacerbated colonic inflammation, and swelled spleens and lymph nodes, are clearly relevant to immune system imbalance. In this study, the functional characterization of CUL3-KLHL25 in iTreg differentiation implies that disturbed iTreg cell homeostasis may contribute to CUL3-deficiency-associated defects, in particular, inflammatory-related phenomena. In addition, this may provide a new angle for re-evaluating and understanding severe pathological changes induced by the loss of CUL3.
It is also worth mentioning that the loss of CUL3 did not affect the differentiation of Th1, Th2, or Th17 in vitro, but enhanced follicular helper T (Tfh) cell number and responses in the germinal center (GC). These observations and our findings suggest that CUL3 may play distinct roles in different types of T cells, such as Tfh and Treg cells. In GC, Tfh cells can be suppressed by follicular regulatory T (Tfr) cells, which are also crucial for the maintenance of immune homeostasis (Essig et al., 2017;Huang et al., 2019). This specific type of T cells is actually derived from Treg cells (Essig et al., 2017;Huang et al., 2019). Whether CUL3 could finetune the balance between Tfh and Tfr for the control of GC function remains to be addressed in future.
Growing evidence shows that the regulation of cell metabolism is vital for T-cell proliferation, differentiation, and functions Guo, 2017;Maciolek et al., 2014;Wang and Solt, 2016). Different types of T cells are usually associated with distinct metabolic status and features. Unlike effector T cells, including Th1, Th2, and Th17, which 'prefer' FAS, iTreg cells 'favor' FAO (Maciolek et al., 2014;Michalek et al., 2011;Wang et al., 2011). During iTreg differentiation, suppressing FAS could probably be advantageous to keep the differentiation process on the right track and prevent differentiating cells from rushing into unwanted pathways. Also, we found that downregulation of FAS reduced malonyl-CoA, the potent physiological inhibitor of CPT1, and this could facilitate the unleashing of CPT1 activity for FAO and thereby iTreg differentiation. Therefore, reprogramming in fatty acid metabolism, in particular, the switch from FAS to FAO, is vital for the efficient differentiation of iTreg cells.
Although the essential role of FAO and CPT1 in iTreg differentiation has been documented (MacIver et al., 2013;Michalek et al., 2011;Patel and Powell, 2017;Zhang et al., 2017), a more recent investigation challenged this conclusion and proposed that CPT1 is dispensable for iTreg differentiation (Raud et al., 2018). Raud et al., 2018 found that naive CD4 + T cells isolated from mice with CPT1 deletion in T cells (TCPT1 mice) did not seem to have any defect in iTreg differentiation and proposed that the phenotypes induced by high-dose CPT1 inhibitor, etomoxir (>100 mM), were due to the off-target effect. However, we found that siRNA-mediated Cpt1 knockdown or low-dose etomoxir (5 mM), which inhibits CPT1 without introducing off-target effects, impaired iTreg differentiation in vitro (Hao et al., 2021;O'Connor et al., 2018; Figure 2F, Figure 2-figure supplement  5B). Of note, TCPT1 mice reported in Raud et al., 2018 survived from the lack of CPT1. One thus can envision that various metabolic genes/regulators have to be re-adjusted to compensate for the loss of CPT1, and metabolic homeostasis needs to be re-established. As a consequence, the defect (s) associated with the loss of CPT1 would be gradually compensated or masked during mouse Figure 6 continued colitis in mice adoptively transferred with different cells. (C) Body weight (left) and disease activity index (DAI) score (right) based on body weight loss, stool consistency, and blood in the stool were recorded weekly. 7 weeks later, entire colons from mice treated in different ways were removed for length assessment (D) and hematoxylin and eosin (H and E) staining, which is used for revealing histopathological changes (E). Scale bar, 200 mm. One representative experiment out of four biological replicas is represented. (C, D) Data represent mean ± SD (n = 4), with significance determined by twoway ANOVA test (C) or one-way ANOVA test (D). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. The online version of this article includes the following figure supplement(s) for figure 6:    development. In contrast, blocking CPT1 function by a specific inhibitor or siRNA that mimics an acute disruption in FAO might be useful for minimizing long-term metabolic adaption and directly evaluating CPT1 function in iTreg differentiation. In addition, our in vitro functional assays showed that alterations in CPT1 activity induced by ACLY and/or CUL3 depletion failed to change the suppressive function of iTreg cells ( Figure 4D,I, Figure 4-figure supplement 4A-B). In a recent study, Saravia et al., 2020 also showed that the deletion of CPT1 in Foxp3 + Treg cells did not influence cell proliferation or function (Saravia et al., 2020). Together, these findings argue that CPT1 may be dispensable for the regulation and maintenance of iTreg function.
FAS is controlled by a series of enzymes, three of which are rate-limiting enzymes, namely ACLY, ACC, and FASN. During iTreg differentiation, only ACLY changed significantly. These observations suggest that, rather than ACC and FASN, ACLY is apt to function as a sensor for FAS to respond to TGFb1 signal. ACLY is responsible for the conversion of citrate into acetyl-CoA and OAA. Its enzymatic activity and expression are tightly coupled with the level of citrate, which could block glycolysis by inhibiting phosphofructokinase (PFK) (Iacobazzi and Infantino, 2014;Zaidi et al., 2012). Notably, inhibition of glycolysis has a positive impact on iTreg differentiation . Therefore, TGFb1-induced downregulation of ACLY, which is probably accompanied by citrate accumulation, might also contribute to the repression of glycolysis for iTreg differentiation.
Because of the great potential as a target for pharmaceutical intervention against hyperlipidemia, hypercholesterolemia, obesity, and cancer, ACLY has attracted considerable amount of attention over the recent decade (Feng et al., 2020;Migita et al., 2008). A growing list of ACLY inhibitors, including bempedoic acid (BemA), SB204990, curcumin, cinnamon polyphenol, and hydroxycitric acid, is under extensive investigation. It is worth mentioning that BemA, a small molecule inhibiting liver-specific ACLY, has been tested in a phase-three clinical trial for the control and treatment of hypercholesterolemia (Feng et al., 2020;Pinkosky et al., 2016). Results in this work have shown that ACLY inhibition by SB204990 effectively alleviated colitis in mice, suggesting that this small-molecule inhibitor (SB204990) may have a promising role in the clinical treatment for IBD. Thus, future investigations systematically evaluating ACLY as a drugable target for IBD and other immune diseases are in need.
Functional characterization of CUL3-KLHL25-mediated ACLY ubiquitination pinpoints ACLY as a regulatory node integrating TGFb1 signaling with iTreg differentiation and therefore brings insights into the modulation and maintenance of immune homeostasis. Moreover, ACLY inhibitors with pharmaceutical potentials are apparently of interest for immunotherapy of autoimmunity and unwanted inflammation and certainly require further investigations. Mice were bred and cohoused, four to six mice per cage, in a specific pathogen-free facility with a standard 12-hr alternate light/dark cycle at an ambient temperature of 22 ± 2˚C and 30-70% humidity at the Animal Research Center of Northeast Normal University (Changchun, China). Health status of mice was determined via daily observation by technicians supported by veterinary care. All mouse experiments were conducted in accordance with the protocols for animal use, treatment, and euthanasia approved by the local ethics committee (reference number: AP2019085, the institutional animal care and use committee (IACUC) of the Northeast Normal University).

Induction of human iTreg in vitro
Human peripheral blood samples were obtained from Jilin Blood Center (Changchun, China). All work with human blood samples was approved by the local ethics committee (reference number: AP2019085, the ethical committee of the Northeast Normal University), and informed consent was obtained from all subjects. After Ficoll (GENVIEW) gradient, naive CD4 + T cells were isolated with MojoSort human CD4 + Naïve T Cell Isolation Kit (Biolegend) according to the manufacturer's instructions (Ghosh et al., 2016). For Th0 induction, 2 to 3 Â 10 5 naive CD4 + T cells were cultured in RPMI 1640 medium (Sigma) supplemented with 10% FBS (Biological Industries), penicillin-streptomycin (500 U; PAN biotech), and b-mercaptoethanol (50 mM; Sigma-Aldrich) and stimulated with Dynabeads human T-Activator CD3/CD28 (Gibco) at a cell:bead ratio of 1:1 in the presence of rhIL-2 (200 U/ml; Gibco). To obtain iTreg cells, rhTGFb1 (2 or 0.5 ng/ml; Peprotech) was simultaneously added along with the reagents used for inducing Th0 cells.

Flow cytometry analysis
Cells were collected in phosphate-buffered saline (PBS) containing 1% FBS (v/v) and fixed/permeabilized using the Transcription Factor Buffer Set (BD Biosciences) according to the manufacturer's instructions. For mouse iTreg analysis, selected protein markers were stained with fluorescein isothiocyanate (FITC) rat anti-mouse CD4, phycoerythrin (PE) rat anti-mouse CD25, and Alexa Fluor 647 rat anti-mouse Foxp3 monoclonal antibodies (BD Biosciences). For human iTreg analysis, protein markers were stained with FITC rat anti-human CD4, PE mouse anti-human CD25, and Alexa Fluor 647 mouse anti-human Foxp3 monoclonal antibodies (Biolegend).

Quantitative real-time PCR
Total RNA was isolated using the TRIzol reagent (Invitrogen). Complementary DNA (cDNA) was synthesized with the PrimeScript RT reagent Kit plus gDNA Eraser (Takara) according to the manufacturer's instructions. Real-time PCR was performed on the QuantStudio 3 Real-Time PCR Instrument (Applied Biosystems) with a TB Green Premix Ex Taq (Tli RNaseH Plus) reagent (Takara).
The mRNA expression of genes was normalized to the expression of b-actin gene. Data were analyzed using the comparative cycling threshold method. Quantitative PCR primers are shown in Supplementary file 1.

siRNA transfection
All siRNA transfections were performed with Mouse T-cell nucleofection kit (Lonza) according to the instructions using X-001 progress as described previously (Mantei et al., 2008). Cells were transiently transfected with siRNAs (siRNAs against Acly, Cul3, and Klhl25 were commercially synthesized; Shanghai GenePharma) at a final concentration of 20 nM and rested in RPMI-1640 (with 10% FBS) for 4-5 hr before incubation under different conditions (Th0 or iTreg) for a desired period of time. siRNA sequences are shown in Supplementary file 1.

Lentivirus transduction of CD4 + T cells
HEK293T cells (without mycoplasma contamination) were grown in 10-cm cell-culture dishes to 70% confluence and were transfected with a three-plasmid system (pWPXLd-ACLY, PAX, PMD). The supernatant was harvested at 48, 72, and 96 hr after transfection and was concentrated 100-fold in an ultracentrifuge. Aliquots were stored at À80˚C. Naive CD4 + T cells were cultured under Th0inducing condition as described above for 16-24 hr and were spin infected for two rounds with lentivirus supernatants (ACLY WT , ACLY 3KR , or Empty) in the presence of 8 mg/ml polybrene (Solarbio) at 1000 Âg, 90 min, at 32˚C as described previously (Lin et al., 2019;Sun et al., 2015). After 24 hr of infection, supernatants were removed and cells were selectively cultured under either Th0-or iTreginducing condition for a desired period of time.

Assessment of metabolic molecules
For acetyl-CoA and OAA detection, cytosolic fractions from Th0 or iTreg cells were separated using Cell Mitochondria Isolation Kit (Beyotime). In brief, cells were washed with cold PBS, followed by homogenation for 20 s, and then centrifuged twice (first round, 600 Âg for 5 min; second round, 11,000 Âg for 5 min). Supernatants that represent cytosolic fractions without nuclei and mitochondria were confirmed by western blotting (WB) and deproteinized with a 10-kD spin filter by centrifuge for effluent preparation. Nuclear fraction was separated using Cell Nuclear/Cytoplasm Protein Isolation Kit (Beyotime) and also confirmed by WB. Cytosolic/nuclear acetyl-CoA and cytosolic OAA were assessed using PicoProbe acetyl-CoA fluorometric assay kit (Biovision) and OAA assay kit (Sigma), respectively, according to the manufacturer's protocol. For the analysis of malonyl-CoA, cells were cultured under iTreg-polarization condition for 24 hr and lysed with ice-cold 5% sulfosalicylic acid (Sigma-Aldrich) containing 50 mM dithioerythritol (DTE) (Sigma-Aldrich). After 600 Âg centrifugation for 10 min at 4˚C, the supernatants were collected for the detection of malonyl-CoA by high-performance liquid chromatography (HPLC). For the detection of HMG-CoA, mevalonate, and mevalonate-5-pyrophosphate, cells were homogenized in PBS and the supernatants were collected for detection of HMG-CoA, mevalonate, and mevalonate-5-pyrophosphate using HMG-CoA enzyme-linked immunosorbent assay (ELISA) kit (Shanghai Ruifan Biological Technology), Mevalonate ELISA kit (Sino Best Biological Technology), and Mevalonate-5-pyrophosphate ELISA kit (Shanghai Ruifan Biological Technology) according to the manufacturers' instructions. Equal amount of cells was used for different groups in the same assay.

Extracellular flux analysis
For the FAO-associated oxygen consumption rate (OCR) assay by Seahorse XFp Analyzer (Agilent), Palmitate-BSA reagent (Agilent) and Mito Stress Test Kits (Agilent) were used according to the manufacturers' instructions with minor modifications. In brief, cells were collected and adhered to poly-D-lysine-coated XFp plates (4 to 6 Â 10 5 cells/well) via centrifugation in serum-free XF Base Medium (Agilent). For evaluation of palmitate oxidation, cells were cultured with XF Base Medium (Agilent) in a non-CO 2 incubator for 15-20 min at 37˚C. Subsequently, the XF Palmitate-BSA FAO Substrates (Agilent) palmitate-BSA (167 mM palmitate conjugated with 28 mM BSA) and 0.5 mM L-carnitine (Sigma Aldrich) were added to the medium for the assessment of OCR. Immediately, XF Cell Mito Stress Test was performed in a Seahorse XFp Analyzer (Agilent) by sequentially adding several modulators of mitochondrial function, including 1 mM oligomycin (Oligo) (Agilent), 1.5 mM fluorocarbonyl cyanide phenylhydrazone (FCCP) (Agilent), and 100 nM rotenone plus 1 mM antimycin A (Rot/ Ant A) (Agilent). Data were analyzed with Wave software version 2.3.0 (Agilent) and proper statistical methods using GraphPad Prism version 6 (GraphPad Software).

C-tracing assessment
Cell samples cultured in a medium containing [U-13 C] glucose (11 mM) were mixed with 10 ml formate and 800 ml chloroform, and processed by four cycles of 1 min ultra-sonication and 1 min interval in ice-water bath followed by standing for 30 min at room temperature prior to centrifugation at 3000 Âg for 15 min. All chloroform was transferred to a new glass vial and mixed with 500 ml 75% ethanol (containing 0.5 M KOH) prior to incubation at 80˚C for 60 min. Then, the mixture was added with 600 ml of hexane and vortexed for 1 min and made to stand for 30 min. After centrifugation at 3000 Âg for 15 min, all hexane layers were evaporated to dryness. The dry residue deposit was mixed with 10 ml 1-hydroxybenzotriazole (HoBt) (in dimethyl sulfoxide [DMSO]), 20 ml cholamine (in DMSO with 200 mM triethylamine [TEA]), and 10 ml O-(7-azabenzotriazol-1yl)-N,N,N',N'-tetramethyluronium hexafluorophosphate (HATU) (in DMSO) followed by incubating for 5 min at room temperature. 60 ml acetonitrile was added, followed by centrifuging at 3000 Âg for 15 min at 4˚C prior to UHPLC-HRMS analysis.
The ultra-high performance liquid chromatography-tandem mass-spectrometry (UHPLC-MS/MS) analysis was performed on an Agilent 1290 Infinity II UHPLC system coupled to a 6470A Triple Quadrupole mass spectrometre (Santa Clara, CA, United States). Samples were injected onto a Waters UPLC BEH C18 column (100 mm Â 2.1 mm, 1.7 mm) at a flow rate of 0.3 ml/min. The mobile phase consisted of water (phase A) and acetonitrile (phase B), both with 0.1% formate. The chromatographic separation was conducted by a gradient elution program as follows: 0 min, 10% B; 4 min, 40% B; 8 min, 45% B; 11 min, 50% B; 14 min, 70% B; 15 min, 90% B; 15.5 min, 100% B; 18 min, 100% B; 18.1 min, 10% B; 20 min, 10% B. The column temperature was 40˚C. The raw data were processed by Agilent MassHunter Workstation Software (version B.08.00) by using the default parameters and assisting manual inspection to ensure the qualitative and quantitative accuracies of each compound. The peak areas of target compounds were integrated and the output was run for IsoCorrection analysis.

Assessment of metabolic enzyme activities
Cells were lysed by ultrasonic treatment in extracting buffer from the kits. Supernatants were collected and used for the analyses of enzymatic activities. ACC Activity Assay Kit (Suzhou Comin), ACLY Activity Assay Kit (Suzhou Comin), FASN Activity Assay Kit (Suzhou Comin), and CPT1 Activity Assay Kit (Suzhou Comin) were used according to the manufacturers' instructions.

Immunoprecipitation and western blotting
Cells were lysed in cold western blot-immunoprecipitation (WB-IP) lysis buffer (Beyotime) for 30 min and centrifuged (at 4˚C, 5 min, at 14,000 Âg) to remove cell debris. The supernatant was collected as a whole-cell lysate. ACLY, ubiquitin (Ub), and CUL3 were immunoprecipitated from the whole-cell lysates using anti-ACLY (Abcam), anti-ubiquitin (Abcam), and anti-CUL3 (Santa Cruz Biotechnology) antibodies coupling to Protein A/G PLUS Agarose Beads (Santa Cruz Biotechnology) overnight at 4C . Nonspecific rabbit or mouse IgG antibody was used as a negative control. Immunoprecipitated proteins were mixed with 1Â loading buffer and boiled at 100˚C for 10 min followed by WB. For WB, protein samples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels and then proteins were transferred to polyvinylidene fluoride membranes (Millipore). The membranes were blocked with 5% nonfat milk for 1 hr at room temperature before incubation with primary antibodies overnight at 4˚C. Mouse-or rabbit-conjugated horseradish peroxidase (HRP) secondary antibodies were added to the membranes for 50 min at room temperature, followed by washing three times with 1Â Tris-buffered saline with Tween 20 (TBST), and visualized using Chemiluminescent HRP Substrate (Millipore) on Chemiluminescence Image System (Tanon Science and Technology Co, Ltd). The antibodies for WB are shown in Key Resources Table. In vitro iTreg suppression assay Naive CD4 + T cells from WT or Cd4 Cre Cul3 fl/fl mice were transfected with Acly-specific or scramble siRNA and cultured under Th0-or iTreg-polarization condition for 72 hr to obtain Th0 (WT), iTreg (WT), iTreg (Cul3 -/-), and iTreg (Cul3 -/-+siAcly) cells. Meanwhile, 2 Â 10 5 freshly isolated naive CD4 + T cells from WT mice pulse-labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) (Biolegend) at 0.5 mM for 20 min were co-cultured with different groups of iTreg cells at a ratio of 1:1 in the presence of Dynabeads Mouse T-Activator CD3/CD28 for 72 hr as described previously . The flow cytometry was used for the assessment of CFSE dilution, which indicates the suppressive function of iTreg cells in vitro.
Naive CD4 + T cells from Foxp3 YFP-Cre mice were transfected with siRNAs against Acly and (or) Cul3 and cultured under Th0-or iTreg-polarization condition for 72 hr followed by sorting yellow fluorescent protein (YFP)-positive cells to obtain Th0, YFP-iTreg, YFP-iTreg (siCul3), and YFP-iTreg (siCul3+siAcly) cells. Meanwhile, 2 Â 10 5 freshly isolated naive CD4 + T cells from WT mice pulselabeled with CellTrace Violet (CTV) (Thermo Fisher) at 1 mM for 20 min were co-cultured with different groups of iTreg cells at a ratio of 1:1 in the presence of Dynabeads Mouse T-Activator CD3/ CD28 for 72 hr as described previously. The flow cytometry was used for the assessment of CTV dilution (Johnson et al., 2018), which indicates the suppressive function of iTreg cells in vitro.

Mouse model of colitis
T-cell transfer colitis based on adoptive transfer of naive CD4 + T cells into Rag1 -/mice is widely used for studying chronic intestinal inflammation (Furusawa et al., 2013). As described previously, 5 Â 10 5 freshly isolated naive CD4 + T cells were administrated into Rag1 -/mice by intraperitoneal injection to induce colitis. Concomitant injection of 5 Â 10 5 Th0 (WT), iTreg (WT), iTreg (Cul3 -/-), or iTreg (Cul3 -/-+siAcly) cells was conducted for the evaluation of CUL3 and ACLY function. Meanwhile, Rag1 -/mice of blank control group were administrated with PBS. Moreover, the function of ACLY in acute colitis was also assessed in DSS-induced colitis model. DSS (MP Biomedicals) dissolved in drinking water (2%, w/v) was given ad libitum to 6-to 8-week-old female C57BL/6J mice for 7 days followed by water for 3 days (Wirtz et al., 2017;. Mice supplied with regular drinking water were included as a negative control. ACLY inhibitor, SB204990 (37.5 mg/kg/d) (Tocris Bioscience), was administrated orally every 2 days before the assessment of colitis.

Evaluation of colitis in mice
To assess the severity of colitis, DAI was weekly (T-cell transfer colitis) or daily (DSS-induced colitis) recorded by scoring the body weight loss, stool consistency, and blood in the stool as described in the literature (Furusawa et al., 2013;Wirtz et al., 2017;. 7 weeks later (T-cell transfer colitis) or 10 days later (DSS-induced colitis), mice were euthanized for the examination of histopathological changes in colon tissues. Entire colons were removed and colon length was determined. Following cleaning with saline and fixation with formalin, colon samples were embedded in paraffin. Sections (3 mm) were stained with hematoxylin and eosin (H and E) and slides were blindly analyzed for intestinal inflammation according to the previous description (Wirtz et al., 2017;. In addition, for DSS-induced colitis, the mononuclear cells from MLNs and the cLP were prepared as described previously (Atarashi et al., 2011) for the analysis of Treg cells by flow cytometry.

Induction of allergic diarrhea
Mice were sensitized with 50 mg of OVA (Sigma Aldrich) and 1 mg of alum (Thermo) in sterile saline by interperitoneal (ip) injection. From day 14 on, mice were challenged by oral gavage with 50 mg of OVA in saline three times per week for 3 weeks as described. Mice showing profuse liquid stool within 60 min after oral OVA challenge were recorded as diarrhea-positive. Th0 and iTreg cells under various conditions were adoptively transferred into mice, respectively, by tail vein injection (iv) after the first and forth challenge. 2 hr after the last oral administration, mice were sacrificed and blood samples and intestinal tissues were collected for detection of OVA-specific IgE and MCPT1 in serum by ELISA and eosnophil infiltration by H and E staining.

Data analysis and statistics
All experiments were done at least three times independently. Statistical analysis was performed using GraphPad Prism version 6 (GraphPad Software). Data are displayed as mean ± standard (SD). The p-values were calculated from Student's unpaired t-test when comparing within two groups. One-way or two-way analysis of variance (ANOVA) was performed in the indicated figures when more than two groups were compared. p-values <0.05 were considered significant. p-values were indicated on graphs as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001; ns, nonsignificant.