Delivery of MMP-12 siRNA with a Nanocarrier Improves the Homeostasis of the Small Intestine and Metabolic Dysfunction in High-Fat Diet Feeding-Induced Obese Mice

The changes of small intestinal homeostasis have been recognized to contribute essentially to the obese development. However, the core small intestinal regulator which mediates overnutrient impacts on the homeostasis of the small intestines remains elusive. Here, we identify the MMP-12 as such a responsive factor in mouse small intestines. Taking advantages of the nano delivery system, we demonstrate that small intestine-specic MMP-12 knockdown alleviates high-fat diet feeding-induced metabolic disorders and improves intestinal homeostasis in mice, including a signicant decrease in lipid transportation, bile acid reabsorption, and inammation. In parallel, the small intestinal integrity is recovered and the gut microbiota composition is reversed towards that under normal diet feeding. Mechanistically, MMP-12, differing from its traditional elastolytic function, acts as a transcriptional factor to activate Fabp4 transcription through epigenetic modication. In translational medicine, clinical applications of our nanosystem and therapeutic interventions targeting MMP-12 will benet patients with obesity and associated diseases. an we blocked MMP-12 enzymatic the MMP-specic inhibitor GM6001 We unexpectedly found this inhibitor shows modest effects on HFD feeding-induced metabolic physiology, indicating that the benecial role of MMP12 knockdown did not rely on its enzymatic function. Unprecedentedly, both HFD feeding and FFA stimulation triggered a marked translocation of MMP-12 into the cell nucleus, as evidenced by IHC and immunocytochemistry (ICC) analyses. These results are in accordance with a previous study, which suggested that MMP-12 also exerts DNA-binding abilities 30 and thus functions as a transcriptional factor in response to virus infection. Similarly, in mouse small intestines, MMP-12 activates FABP4 transcription by directly binding to the poly(A) tract presented on its proximal promoter. Histones respond sensitively to external signals, such as food and light, and thus undergo multiple posttranslational modications, including acetylation and methylation 49,50 . However, the MMP-12-triggered epigenetic regulation of histones remains elusive. Our ndings showed that overexpression of MMP-12 transformed the local chromatin environment of the poly(A) tract on the FABP4 promoter into an active state, which was conrmed by increased AcH3 (an active biomarker) accumulation. However, we found that the levels of H3K9-me2 were not altered by MMP-12 overexpression. These results indicated that MMP-12 may function as an enhancer to facilitate gene transcription by altering histone acetylation rather than methylation. Hence, our ndings extend the current recognition of the epigenetic regulation of lipid transportation and intestinal homeostasis via the MMP-12/FABP4 axis. Additionally, since the FABP4 was demonstrated as the MMP-12


Introduction
With the development of the global social economy in the 21 st century, obesity has become a global pandemic disease 1,2 . Over the last 40 years, the prevalence of obesity has almost tripled worldwide 3 . The development of obesity is closely associated with multiple diseases, such as type 2 diabetes, hepatic steatosis, cardiovascular diseases, and even cancer [4][5][6][7][8][9] . Hence, obesity is a great threat to public health and causes an economic burden for society. Although effective strategies, including lifestyle intervention, conventional pharmacology and weight-loss surgery, are well established to counteract obesity, their undesirable inconstancy and side effects are not exempt in all cases 7,10 . Therefore, an urgent medical need exists for the large and heterogeneous population of obese patients.
Although obesity is caused by various pathological factors, changes in the small intestinal homeostasis have been recognized to contribute essentially to obesity progression 11 . As the rst organ exposed to nutrients, the small intestine has to change its capacity to adapt external nutrient signals, such as a HFD 12 . When challenged with dietary fat, our stomach functions as a bioreactor for dietary lipid peroxidation. These lipids are further engulfed by enterocytes within the duodenum and are re-esteri ed into complex lipid molecules. 4 Then, they bind with lipoproteins (e.g., ApoB48 and ApoB100) and are assembled into primarily chylomicrons in the jejunum 13 . These chylomicrons enter the lymphatics through lacteals and further drain into the venous circulation 14,15 . Subsequently, the ileum plays a nonredundant role in handling lipid over ow by controlling bile acid (BA) reabsorption, cholesterol absorption and metabolism [16][17][18] . In addition, disruptions of lipid homeostasis can lead to unexpected intestinal in ammation and permeability destruction, as well as alterations in the composition of the intestinal microbiota [19][20][21] . Collectively, the coordination of such processes in the whole intestine guarantees metabolic homeostasis within our bodies. However, diets with high fat have been associated with various metabolic syndromes, including obesity and diabetes mellitus 4, 10, 22 . These HFD feeding-induced detrimental effects are ascribed to their profound impacts on the overall intestinal homeostasis as stated above. Thus, correction of the small intestinal homeostasis has become a promising approach to treat obesity.
To screen out the direct mediator that transfers HFD signals to intestinal homeostasis, we clustered highthroughput RNA-seq results from the jejunum and ileum (two major organs for lipid absorption) collected from mice subjected to 11 weeks of HFD feeding. We found that the mRNA expression levels of MMP-12 were markedly elevated in both small intestinal tissues. MMP-12 was rst demonstrated as an in ammatory macrophage-secreted elastolytic MMP 23 . In addition, MMP-12 is able to degrade a broad spectrum of extracellular matrix (ECM) components, such as collagen IV, bronectin and laminin. Hence, in ammation-induced MMP-12 degrades the basement membrane, allowing macrophages to penetrate into injured tissue 24 . Accordingly, the expression and enzymatic activity of MMP-12 are increased in multiple in ammatory diseases, including atherosclerosis and chronic obstructive pulmonary disease 25 .
Importantly, MMP-12 is involved in obesity development. For example, MMP-12 is highly expressed in the adipose tissue of HFD-fed mice. MMP-12 de ciency alleviates obesity-induced in ammation and improves metabolic dysfunction. Such a bene cial effect may be dependent on the organization and composition of the ECM, raising the possibility that MMP-12-orchestrated ECM deposition and degradation are involved in maintaining metabolic homeostasis. Notably, intestinal ECM homeostasis is abolished in HFD-fed mice. Hence, identi cation of the role of MMP-12 in the regulation of small intestinal function is of great urgency in the treatment of obesity.
In the present study, to effectively and speci cally knock down MMP-12 in mouse small intestines, we established a new nanoparticle (NP) system consisting of chitosan (CS), PLGA (poly(lactic-coglycolic acid)) and polyethylene glycol (PEG) (CS@PLGA NPs, CPA NPs) to speci cally deliver MMP-12 siRNAs into mouse small intestines. By making the best of this NP system, we found that it improved the small intestinal homeostasis of HFD feeding-induced obese mice. As a consequence, these mice exhibited improved metabolic physiology and reduced systemic in ammation. Mechanistically, MMP-12 exerted DNA-binding abilities, rather than its elastolytic activity, to induce fatty acid-binding protein 4 (Fabp4) transcription. Taken together, we demonstrated a mediatory role of MMP-12 in transferring HFD signals to small intestinal homeostasis and systemic metabolic homeostasis.

Results
MMP-12 responses to HFD signals in mouse small intestines. To address the aforementioned question, we performed high-throughput RNA sequencing by using small intestinal samples (including jejunum and ileum) dissected from 11-week HFD-fed mice. Bioinformatic analysis revealed a cluster of 8 genes that were robustly regulated by HFD signals in both the mouse jejunum and the ileum (Fig. 1a, b and  Supplementary Table 1), which was con rmed by independent RT-qPCR and semiquantitative analyses (Fig. 1c, Supplementary Fig. 1a and Supplementary Table 2). Moreover, in vitro analysis revealed that 0.4 mM free fatty acids (FFAs, an equimolar mixture of oleic acid and palmitic acid) increased the mRNA expression levels of MMP-12, LEP and CYP1A1 in Caco-2 cells, a colorectal adenocarcinoma-derived cell line. However, other genes, including GPR17 and KCNJ14, were modestly altered by such stimulation (Fig.  1d). Given that MMP-12 was the most responsive gene corresponding to the hyperlipidemic signals in vivo and in vitro, we ultimately focused our research on this gene. Consistently, the protein expression levels of MMP-12 were similarly increased ( Fig. 1e and Supplementary Fig. 1b, c).
Of note, MMP-12 is also a macrophage metal elastase that possesses biological and pathophysiological functions based on its enzymatic activity. Hence, we evaluated the enzymatic activities of MMP-12 and found that they were consistently increased in the jejunum and ileum of HFD-fed mice, as well as in FFAtreated Caco-2 cells (Fig. 1f and Supplementary Fig. 1d, e). Moreover, MMP-12 is known as a secretory protease that degrades the ECM during the destruction of in ammatory tissues. As shown in Fig. 1g, the serum and fecal levels of MMP-12 were correspondingly increased in HFD-fed mice. Meanwhile, FFAs also increased the MMP-12 concentrations in the supernatant from Caco-2 cells (Fig. 1h).
Taking these data together, we concluded that MMP-12 may serve as an important factor responding to HFD signals in the small intestine.
Preparation and Characterization of CPA NPs. To speci cally and effectively knock down MMP-12 in the small intestines of HFD-fed mice, we designed an oral NP system using PLGA-PEG and CS based on the pH-sensitive and adhesive effects of CS. In brief, PLGA-PEG was rst synthesized by conjugating COOH-PEG-NH2 to PLGA-N-hydroxysuccinimide (PLGA-NHS) through amide bond formation (Fig. 2a). In the 1 H nuclear magnetic resonance (NMR) spectra of PLGA-PEG, the peaks at 5.2, 4.8, 3.7, and 1.6 ppm were ascribed to (3, -OCH-), (2, -OCH 2 -), (4, -CH 2 -CH 2 O-), and (5, -OCH(CH 3 )-), respectively (Fig. 2b). The degradable CPA NPs were prepared using PLGA-PEG polymer (as a core) coated with a CS shell (as a protective shelter to avoid gastric acid) by double-emulsion and ionic gelation strategies (Fig. 2c). The CPA NPs were roughly spherical in shape and had a particle diameter of ~227 nm, as analyzed by transmission electron microscopy (TEM) and light scattering spectrometry (Fig. 2d). The polymer disparity index (PDI) and zeta potential were 0.23 ± 0.01 and 57.8 ± 4.98 mV, respectively (Fig. 2e).
Essential to the NP system, the protonation of the CS shell stabilized the PLGA-PEG NPs under acidic conditions (Fig. 2f). Next, cationic poly-L-lysine (PLL) was used for siRNA complexation (PLL/siRNA weight ratio of 0.5, Fig. 2g). In contrast, this shell collapsed in the simulated intestinal uid (SIF), an in vitro system that mimics the basic environment in the small intestines (Supplementary Table 3) 26 .
Endocytosis has been well recognized as an important process through which extracellular cargos are internalized by cells 27 . To identify the mechanisms through which the CPA NPs entered the intestinal cells, we treated human Caco-2 cells with CPA NPs carrying FAM siRNA (with green uorescence) for 4 h in the presence of different inhibitors for the endocytic process. As shown in Fig. 3a, CPA NP transfection was partially blocked by treatment with either EIPA (a macropinocytic inhibitor) or chlorpromazine (a clathrin-mediated endocytic inhibitor). Notably, cellular uptake was dramatically diminished in combination with lipin (a caveolae-mediated endocytic inhibitor). Taken together, these results suggested that caveolae-mediated endocytosis is the predominant pathway for the cellular uptake of CPA NPs. In addition, the CPA NP-coated siRNA was more prone to absorption than the PLGA NPs (Fig. 3b), which could be explained by the stronger positive electricity of the CPA NPs.
To evaluate the biodistribution of CPA NPs, Cy5 was grafted onto PLGA to form uorescent NPs for in vivo tracking (Fig. 3c, d). Next, the mice were gavaged with Cy5-labeled CPA NPs (40 μg/kg) and equivalent Cy5-labeled PLGA NPs for 48 h. We found that the uorescent signals of Cy5-labeled CPA NPs accumulated in mouse small intestines and exhibited retentive ability compared with that of the Cy5labeled PLGA NPs ( Fig. 3e-g). In addition, we did not observe any uorescent signals in other metabolic organs, including the liver, epididymal white adipose tissue (eWAT), kidney, or spleen (Fig. 3h).
Biosafety analysis of CPA NPs. To investigate the potential toxic effects of CPA NPs, we treated Caco-2 cells with different components or doses of CPA NPs. We found that none of them affected the cell viability of Caco-2 cells, indicating that CPA NPs were safe for intestinal cells ( Supplementary Fig. 2a, b). More importantly, CPA NPs modestly affected the mouse body weight and food intake when compared to the PBS-treated group ( Supplementary Fig. 3c, d). In addition, we did not observe any changes in serological hepatic injury parameters, including alanine aminotransferase (ALT) and aspartate transaminase (AST) (Supplementary Fig. 3e). Similar results were observed for renal injury markers, such as serum creatinine and blood urea nitrogen (BUN) ( Supplementary Fig. 2f, g). Histologically, we found that no signi cant differences existed in any examined organs between groups ( Supplementary Fig. 2h).
MMP-12-siRNA-CPA NPs improve HFD feeding-induced obesity and insulin resistance. To further identify the speci c role of MMP-12 in mouse small intestines, we used a state-of-art CPA NP system carrying an MMP-12 siRNA mixture to construct mice with small intestine-speci c MMP-12 knockdown (with 93.46 ± 0.54% encapsulation e ciency, Supplementary Fig. 3a). The CPA NP-mediated knockdown e ciency and speci city are shown in supplementary Fig. 3b-e. Note that enzyme-linked immunosorbent assay (ELISA) analysis revealed that knockdown of MMP-12 in mouse small intestines signi cantly reduced the serum and fecal levels of MMP-12, indicating that the small intestine was a dominant organ for MMP-12 secretion ( Supplementary Fig. 3f). As shown in Fig. 4a, b, HFD feeding-induced body weight gain was attenuated by 19.86% in mice with oral administration of MMP-12-siRNA-CPA NPs, while food intake was modestly altered by MMP-12-siRNA-CPA NPs in HFD-fed mice (Fig. 4c). Since the small intestine is an important organ for maintaining lipid homeostasis and contributes essentially to obese development, we determined the impact of MMP-12 knockdown on whole-body energy metabolism using metabolic cage analysis. We found that administration of MMP-12-siRNA-CPA NPs signi cantly increased O 2 consumption and CO 2 production in HFD-fed mice ( Fig. 4d-f). Consistently, HFD feeding-induced hypertriglycemia, hypercholesterolemia, fasting hyperglycemia, hyperinsulinemia, the increased homeostasis model assessment (HOMA) of insulin resistance (HOMA-IR), and hyperlipidemia were alleviated in these mice ( Fig. 4g and Supplementary Fig. 3g, h). In contrast, the fecal levels of total cholesterol (TC) and nonesteri ed fatty acids (NEFAs) were correspondingly increased ( Supplementary   Fig. 3i). In parallel, glucose intolerance and insulin sensitivity were correspondingly improved by MMP-12 knockdown (Fig. 4h, i and Supplementary Fig. 3j). These results suggested that the overall metabolic disorders were attenuated by MMP-12 manipulation.
MMP-12-siRNA-CPA NPs improve lipid transportation and BA reabsorption in the small intestines of HFDfed mice. To identify the potential mechanism underlying the protective effects of MMP-12 knockdown in the mouse small intestines on diet-induced obesity, we next performed transcriptional pro ling on small intestine samples (a mixture of jejunum and ileum) to globally screen out the downstream events of MMP-12. As shown in Fig. 6a, MMP-12 knockdown increased the expression levels of 630 genes. In contrast, the expression levels of 454 genes were decreased. Gene ontology (GO) analysis revealed a cluster of genes involved in lipid homeostasis (e.g., lipid uptake/transportation and BA reabsorption) and the immune response (Fig. 6b). Hence, we examined whether MMP-12 was essential for the transport of dietary lipid absorption into the circulation since the small intestine is an important organ for such a process. As shown in Fig. 6c, postprandial TG response assays indicated that administration of MMP-12-siRNA-CPA NPs markedly decreased serum TG levels in HFD-fed mice. Histological and Nile red staining analyses indicated that refeeding-induced lipid accumulation in the enterocytes of the mouse small intestines was almost absent in these mice (Fig. 6d).
Of note, chylomicron contributes importantly to mediating the transportation of dietary lipids into the circulation system. 14 Transmission electronic microscopy (TEM) scanning analysis indicated that smaller chylomicron particles existed in the serum from MMP-12 knockdown mice after 2-h HFD refeeding compared to the control group ( Fig. 6e and Supplementary Fig. 5a, b). Coincident with these ndings, the serum collected from 2 h HFD-refed mice administered MMP-12-siRNA-CPA NPs was transparent, whereas the control serum was milky (Fig. 6e). At the molecular level, ApoB-48 contents in the chylomicron fraction and serum were correspondingly reduced, while its expression was consistently reduced in mouse small intestines (Fig. 6f, Supplementary Fig. 5c).
BAs are involved in the absorption and metabolism of dietary lipids in the small intestines, as well as gut permeability and bacterial composition. 21 As shown in Fig. 6g, we found that serum levels of total BAs were decreased by approximately 45.15% in HFD-fed mice treated with MMP-12-siRNA-CPA NPs compared to HFD-fed mice. Furthermore, we examined the mRNA expression levels of BA reabsorption biomarkers in the mouse jejunum and ileum, including Asbt, Ibabp, Osta and Ostb, and found that they were downregulated in response to MMP-12 knockdown ( Supplementary Fig. 5d), which is consistent with the reduced serum levels of BAs in these mice. Furthermore, fecal levels of MMP-12 and serum lipids were positively correlated in clinical samples (Fig. 6h).
MMP-12-siRNA-CPA NPs attenuate in ammation and maintain the permeability of mouse small intestines. We next explored the in ammatory status in the small intestines of HFD-fed mice treated with MMP-12-siRNA-CPA NPs. As shown in Fig. 7a, HFD feeding signi cantly decreased the small intestine length, which was reversed by MMP-12 knockdown. Histological analysis indicated that MMP-12-siRNA-CPA NPs signi cantly attenuated HFD feeding-induced villous atrophy and crypt hyperplasia (Fig. 7b). Since intestinal goblet cells contribute importantly to intestinal mucin excretion, we therefore examined the numbers of these cells using alcian blue and periodic acid Schiff (AB-PAS) staining. As shown in Fig.  7b, HFD feeding decreased the numbers of goblet cells in the mouse small intestines compared to the normal diet (ND) feeding groups. MMP-12-siRNA-CPA NP treatment restored this reduction. To con rm the severity of intestinal in ammation, we evaluated macrophage (F4/80), neutrophil (Gr-1), and T lymphocyte (CD4) in ltration. Compared with modest changes observed in the ND group, HFD-fed mice displayed increased immune cell in ltration in the small intestines. In contrast, administration of MMP-12-siRNA-CPA NPs dramatically blocked these in ltrations (Fig. 7c). Accordingly, the mRNA expression levels of pro-in ammatory cytokines, including Il-1β and Tnf-α, were markedly reduced in the small intestines of these mice (Supplementary Fig. 6a). These results indicated that MMP-12 was a critical mediator in HFD feeding-induced small intestinal in ammation.
Small intestinal permeability is affected by in ammation under metabolic stress 29 . We found that HFD feeding induced marked tight junction (TJ) impairments in the mouse small intestinal epithelium, as evidenced by increased serum levels of D-lactitol (D-Lac), impaired TJ structure and reduced protein expression levels of TJ biomarkers, including Occludin and ZO-1. In contrast, MMP-12-siRNA-CPA NP intervention signi cantly attenuated these impairments by restoring them close to the levels of the ND-fed group (Fig. 7d, Supplementary Fig. 6b-d). Of note, the bene cial effects of MMP-12-siRNA-CPA NPs were further con rmed by in vitro studies with increased levels of TJ-associated protein expression and transepithelial electrical resistance (TEER) (Supplementary Fig. 7a-c). Therefore, we speculate that lipids may in uence intestinal permeability through the regulation of MMP-12. These results suggested that MMP-12 was a critical factor in mediating the HFD feeding-induced permeability impairments of mouse small intestines.
MMP-12-siRNA-CPA NPs alter the composition of gut microbiota and SCFAs. Gut dysbiosis has been shown to facilitate the development of obesity, whereas obesity further disrupts the composition of gut microbiota, thus forming a vicious circle. 18 By using bacterial 16S rRNA sequencing (V4 region), a total of 1,648,417 raw reads were obtained after quali cation and were annotated to 699 OTUs. Rarefaction analysis showed that the current sequencing depth covered almost all the gut microbiota within each sample ( Supplementary Fig. 8a, b). Unsupervised principal coordinate analysis (PCoA) was used to analyze the whole structural alterations in the gut microbiota among groups. Along the PCo2 axis, MMP-12-siRNA-CPA NP treatment reversed HFD feeding-induced dysbiosis, whereas no signi cant changes were observed in the structure along PCo1 (Supplementary Fig. 8c). Unweighted pair-group method with arithmetic mean (UPGMA) analysis indicated signi cant separations in the microbiota between ND-and HFD-fed mice and between HFD-fed groups in presence with or without MMP-12-siRNA-CPA NPs, suggesting that MMP-12 knockdown could alter the microbiota composition and shift the microbial community pro le from a HFD feeding-induced dysbiotic state towards homeostasis in ND-fed mice ( Supplementary Fig. 8d). In addition, taxonomic pro ling revealed that MMP-12-siRNA-CPA NP administration markedly reduced the contents of the Proteobacteria phylum in the feces from HFD-fed obese mice ( Supplementary Fig. 8e-h). Given that the Proteobacteria phylum is a well-known infective pathogen that contributes to the development of metabolic endotoxemia, MMP-12-siRNA-CPA NPs consistently reduced serum levels of lipopolysaccharide (LPS) in HFD-fed mice ( Supplementary Fig. 8i).
Such a reduction in LPS could be a critical explanation for the alleviated in ammation in these mice.
Next, we used redundancy analysis (RDA) to clarify the speci c bacterial phylotypes changed by HFD signals and MMP-12-siRNA-CPA NP treatment (Supplementary Fig. 8j). Compared to the ND-fed group, HFD signals dramatically altered the 228 OTUs, among which 167 were upregulated and 61 were downregulated. In HFD-fed mice, administration of MMP-12-siRNA-CPA NPs altered 15 OTUs (9 upregulated and 6 downregulated). Interestingly, 6 OTUs were altered by both HFD feeding and MMP-12-siRNA-CPA NP administration, and their changes exhibited opposite trends. Among these, OTU_26 and OTU_144 were classi ed in the detrimental Lachnospiraceae family. This family belongs to the Firmicutes phylum, which is known to aggravate LPS transfer into the circulation system. In addition, 3 OTUs (OTU_30, OTU_1513 and OTU_129) were annotated to the bene cial Eggerthellaceae family, which is a bene cial bacteria responsible for metabolizing secondary plant polyphenols to ameliorate gut in ammation (Fig. 7e, Supplementary Table 4).
Among gut microbiota-generated metabolites, SCFAs, which are metabolized by intestinal microbiota through indigestible carbohydrates, have been determined to downregulate in ammatory cytokine production in macrophages. We thus evaluated the contents of 6 dominant SCFAs in mouse feces. As shown in Fig. 7f (Supplementary Table 5), HFD signals signi cantly reduced the contents of SCFAs compared to ND signals. In contrast, such a reduction was reversed by MMP-12-siRNA-CPA NP treatment.
Taken together, these data suggested that MMP-12-siRNA-CPA NPs may exert their anti-in ammatory effects and improve gut health by reshaping the gut microbiota and increasing SCFA production. GM6001-CPA NPs modestly affect HFD feeding-induced obesity. To further dissect whether these bene cial roles of MMP-12 knockdown in mouse small intestines were dependent on its reduced enzymatic activity or expression, we used the CPA NP system carrying the speci c MMP-12 inhibitor GM6001 to neutralize its enzymatic activity (91.73±1.74% encapsulation e ciency, Supplementary Fig.  9a). As shown in supplementary Fig. 9b, GM6001 signi cantly decreased the enzymatic activity of MMP-12 in the small intestines of mice fed a HFD. However, GM6001-CPA NPs modestly affected the HFD feeding-induced increase in body weight gain as well as the reduction in food intake (supplementary Fig.  9c, d). In addition, HFD feeding increased blood glucose levels and impaired glucose tolerance, while GM6001-CPA NPs did not alter these parameters. Morphologically, administration of GM6001-CPA NPs modestly affected HFD feeding-induced fasting hyperglycemia, hyperlipidemia, hypercholesterolemia, and glucose intolerance ( Supplementary Fig. 9e-g). Similarly, HFD feeding-induced hepatic and epididymal fat accumulation was unaltered in GM6001-CPA NP-treated obese mice ( Supplementary Fig.  9h-k). In addition, HFD signals, including increased serum AST and ALT, as well as TG and TC, were similarly unchanged ( Supplementary Fig. 9l, m). Taken together, these data revealed that the bene cial role of MMP-12 knockdown was not dependent on its enzymatic function.
FABP4 serves as an effector of MMP-12 in the small intestines of HFD-fed mice. It should be noted that MMP-12 exerts DNA-binding abilities to control gene expression in the nucleus 30 . As shown in Fig. 8a, we found that HFD feeding signi cantly increased the nuclear and cytosolic contents of MMP-12 proteins in mouse small intestines (jejunum and ileum). Similar results were observed in the in vitro analysis (Fig.  8b). Considering the modest effect of the MMP-12 inhibitor, we speculate that MMP-12 indeed functions as a transcriptional factor involved in the regulation of small intestinal homeostasis. Hence, we performed Spearman correlation analysis to evaluate potential associations among intestinal gene expression, gut microbiota composition and SCFA contents. Venn analysis clustered 2 genes, including Fabp4 and Nnmt, which were positive, in line with the above bene cial effects of MMP-12 knockdown (Fig. 8c and Supplementary Table 8-11). We further evaluated the mRNA expression levels of these two genes in response to MMP-12 knockdown in vivo and in vitro and found that Fabp4 mRNA expression was most robustly decreased in the small intestines of HFD-fed mice treated with MMP-12 siRNA NPs and in FFA-treated Caco-2 cells with MMP-12 knockdown (Fig. 8c). The protein expression of FABP4 exhibited a similar tendency (Fig. 8d and Supplementary Fig. 10a-d).
Bioinformatics analysis indicated that a classic MMP-12-binding motif featuring a poly(A) tract of 6 bases was present on the proximal promoter of FABP4 (-148 bp --142 bp, Fig. 8e). Luciferase activity analysis indicated that overexpression of MMP-12 signi cantly increased the transcriptional activity of the FABP4 promoter. However, such an activation effect was abolished when this poly(A) tract was mutated (Fig. 8e). Consistently, the mRNA and protein expression levels of FABP4 were correspondingly increased in MMP-12-overexpressing Caco-2 cells (Supplementary Fig. 10e, f). At the epigenetic level, acetyl-histone H3 (AcH3) and K9-dimethylated histone H3 (H3K9-me2) are two hallmarks associated with chromatin activity and gene transcription 31 . In our study, overexpression of MMP-12 led to robust AcH3 (activation) accumulation, accompanied by a modest impact on the H3K9-me2 levels around the FABP4 proximal promoter (Fig. 8f), suggesting that MMP-12 is a potential enhancer of FABP4 transcription by increasing the accumulation of histone acetylation.
Finally, we wanted to explore whether FABP4 linked MMP-12 signals to the homeostasis of the small intestinal homeostasis. We knocked down FABP4 expression in MMP-12-overexpressing Caco-2 cells. The knockdown e ciency of FABP4 siRNA and the overexpression e ciency of MMP-12 are presented in supplementary Fig. 10g. We found that knockdown of FABP4 partially antagonized MMP-12-facilitated lipid accumulation, pro-in ammatory cytokine expression and secretion, and TJ impairment (Fig. 8g-i and Supplementary Fig. 10g).

Discussion
MMP-12 is a macrophage metal elastase that cleaves elastin, which is a major component in the media of the small intestines. Macrophages are the major source of MMP-12, the expression of which is markedly increased in various in ammatory diseases 25,32 . Given the causative relationship between in ammation and obesity, MMP-12 expression is increased in mouse WAT in response to HFD-feeding signals. Although MMP-12 is also expressed in the small intestines, whether it contributes to homeostasis and whole-body metabolism remains elusive. In the present study, our ndings revealed that HFD signals induced marked MMP-12 expression in mouse small intestines and feces. To speci cally and effectively knock down MMP-12 in the small intestines of HFD-fed mice, we successfully constructed an NP system to effectively and orally deliver MMP-12 siRNA. Making the best of this NP system, we found that small intestine-speci c MMP-12 knockdown improved HFD feeding-induced metabolic disorders, including hyperglycemia, hyperinsulinemia, glucose intolerance, impaired insulin sensitivity and systemic in ammation. In addition, the HFD-feeding-induced disruptions of lipid uptake/transportation, BA reabsorption, permeability and in ammation in mouse small intestines were recuperated by MMP-12 knockdown. We also revealed that small intestinal MMP-12 exerts DNA-binding abilities in the cellular nucleus, rather than its elastolytic activity, when corresponding to HFD-feeding signals, while FABP4 served as its key transcriptional downstream effector. Our ndings shed some light on the pathophysiological and transcription-regulatory function of MMP-12 and developed a state-of-the-art siRNA delivery system for gene-speci c intervention in mouse small intestines (Fig. 9).
To tissue-speci cally regulate MMP-12 expression, we rst focused on small intestines and designed a CPA NP system. Given the extensive applications of nanotechnology in delivering various RNAs into tumors [33][34][35] , herein, we developed a novel NP system for the systemic delivery of siRNA into the small intestines to treat metabolic diseases. Our ndings broadened the current applications of the NP system for gene therapies for metabolic disorders. In our present study, CPA NPs exert the following bene cial advantages. For basal materials, CS, isolated from lobsters, crabs and other marine invertebrates, is safe and biocompatible, accompanied by pH-sensitive and adhesive properties 36 . Herein, we established a CS to serve as a protective shelter to avoid gastric acid and to retain the functional siRNA in the small intestines. In addition, as another core component of our NPs, PLGA is a widely used polymer with biodegradable and biocompatible abilities that has been approved in the clinic to deliver antitumor drugs for multiple cancer treatments, including colorectal cancer and ovarian cancer 27,37 . We used this material as a core to coat the PLL-modi ed siRNAs 38 . By taking advantage of double-emulsion and ionic gelation strategies, we constructed an NP system and effectively applied it for the in vivo knockdown of MMP-12 in mouse small intestines based on the protective effects of CS. Our CPA NP system did not affect MMP-12 expression in other organs, including the liver, kidney and colon, demonstrating its excellent speci city in the small intestines. Therefore, the CPA NP system possesses great potential in delivering either RNAs or drugs into the small intestines. Notably, adeno-associated viruses (AAVs) have also been developed to manipulate gene expression in small intestines 39 . However, technical limitations, such as di culties in the achievement of high-titer AAVs, still restrict the clinical applications of this method 40 . In contrast, our CPA NP system was host-friendly due to its transient impacts on gene expression, since they were prone to be excreted by intestinal peristalsis within the feces. In addition, although MMP inhibition holds signi cant therapeutic potentials for the treatment of in ammatory diseases, this approach unfortunately has been hindered by signi cant side effects of MMP inhibitors due to the important physiologic role of MMPs in tissue remodeling 41 . In the present study, we took advantages of the nanosystem to speci cally knock down MMP-12 expression in the mouse small intestine and improve the HFD feeding-induced metabolic disorders. This strategy may partially avoid the global side effects caused by MMP inhibitors. Lastly, it does not escape our attention that the small intestines consist of various types of cells, including intestinal villous epithelial cells, columnar cells and goblet cells 41 . Notably, when the mice were fed a HFD, neutrophils and macrophages also in ltrated into the small intestines to trigger in ammatory responses, which facilitated metabolic disorders 42,43 . In our study, we believed that increased MMP-12 expression in mouse small intestines was due to net increased expression levels within all of the above cells. Hence, deep single-cell RNA-seq analyses and cell type-speci c NP systems based on targeted peptides should be considered for the precision treatment of MMP-12-based metabolic diseases [44][45][46] .
Considering that MMP-12 is classically identi ed as an elastase, we blocked MMP-12 enzymatic activity by using the MMP-speci c inhibitor GM6001 47,48 . We unexpectedly found that this inhibitor shows modest effects on HFD feeding-induced metabolic physiology, indicating that the bene cial role of MMP-12 knockdown did not rely on its enzymatic function. Unprecedentedly, both HFD feeding and FFA stimulation triggered a marked translocation of MMP-12 into the cell nucleus, as evidenced by IHC and immunocytochemistry (ICC) analyses. These results are in accordance with a previous study, which suggested that MMP-12 also exerts DNA-binding abilities 30 and thus functions as a transcriptional factor in response to virus infection. Similarly, in mouse small intestines, MMP-12 activates FABP4 transcription by directly binding to the poly(A) tract presented on its proximal promoter. Histones respond sensitively to external signals, such as food and light, and thus undergo multiple posttranslational modi cations, including acetylation and methylation 49,50 . However, the MMP-12-triggered epigenetic regulation of histones remains elusive. Our ndings showed that overexpression of MMP-12 transformed the local chromatin environment of the poly(A) tract on the FABP4 promoter into an active state, which was con rmed by increased AcH3 (an active biomarker) accumulation. However, we found that the levels of H3K9-me2 were not altered by MMP-12 overexpression. These results indicated that MMP-12 may function as an enhancer to facilitate gene transcription by altering histone acetylation rather than methylation. Hence, our ndings extend the current recognition of the epigenetic regulation of lipid transportation and intestinal homeostasis via the MMP-12/FABP4 axis. Additionally, since the FABP4 was demonstrated as the MMP-12 downstream effector, it might function as a more speci c and safer therapeutic target instead of MMP-12 and minimize the MMP-12 inactivity-induced side effects in future studies.
In conclusion, the present study highlights a nonclassic transcriptional regulatory function of MMP-12 in mouse small intestines that activates Fabp4 transcription to regulate the homeostasis of small intestinal homeostasis and whole-body metabolism. In addition, we established a small intestine-speci c CPA NP system, thus providing proof-of-concept siRNA-based metabolic gene manipulations. Translational research that advances the clinical applications of the CPA NP system and therapeutic interventions targeting MMP-12 will bene t patients with obesity and its associated diseases.

Methods
Animals. Male C57BL/6J mice were purchased from the Model Animal Research Center of Nanjing University (Nanjing, Jiangsu, China). Male C57BL/6J mice were maintained in a 12 h LD cycle and in a temperature-and humidity-controlled environment. Note that we housed these mice at a thermoneutral temperature (26 ± 1℃) to avoid the mild cold (22 ± 1℃)-induced activation impacts on the energy expenditure of mice. All experiments were approved and conducted according to the guidance of the Laboratory Animal Care & Use Committee at China Pharmaceutical University (Permit number SYXK-2016-0011). To establish the HFD-fed obese model, 10-week-old mice were fed with HFD (60% kcal from fat, Research Diets, New Brunswick, NJ, USA) for 11 weeks. To dissect the role of MMP-12 in the regulation of intestinal homeostasis, MMP-12 expression was knocked down in the mouse small intestine by using CPA NPs carrying either a MMP-12 siRNA mixture (at a dose of 40 μg/kg body weight once every 3 days) or its corresponding control for a total of 27 days. The following day after nal gavage, mice were analyzed and later sacri ced to collect sera, tissues and feces for further analyses.
Statistical analysis. Statistical analysis was performed by using the GraphPad Prism 7 programme (California, USA). Groups of data were expressed as the mean ± standard deviation (SD). One-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test. was performed to analyze the data. Spearman correlation analysis was performed between short chain fatty acids (SCFAs), gene expression, and operational taxonomic units (OTUs) using psych package of R, and the correlation heatmap was drawn through the heatmap package. A p value < 0.05 was considered statistically signi cant. Control group. All data are presented as the mean ± SD. One-way ANOVA followed by Tukey's multiple comparisons test. Jej: jejunum; Ile: ileum.   levels. n=6 for each group. *p <0.05, **p <0.01 vs. ND scra siRNA/NP group, #p <0.05, ##p <0.01 vs. HFD MMP-12 siRNA/NP group. All data are presented as the mean ± SD. One-way ANOVA followed by Tukey's multiple comparisons test.