ABSTRACT
There are fundamental differences in how neonatal and adult intestines absorb nutrients. In adults, macromolecules are broken down into simpler molecular components in the lumen of the small intestine, then absorbed. In contrast, neonates are thought to rely on internalization of whole macromolecules and subsequent degradation in the lysosome. Here, we identify the Maf family transcription factors MAFB and c-MAF as markers of terminally differentiated intestinal enterocytes throughout life. The expression of these factors is regulated by HNF4α and HNF4γ, master regulators of enterocyte cell fate. Loss of Maf factors results in a neonatal-specific failure to thrive and loss of macromolecular nutrient uptake. RNA-Seq and CUT&RUN analyses defined an endo-lysosomal program as being downstream of these transcription factors. We demonstrate major transcriptional changes in metabolic pathways, including fatty acid oxidation and increases in peroxisome number, in response to loss of Maf proteins. Finally, we show that loss of BLIMP1, a repressor of adult enterocyte genes, shows highly overlapping changes in gene expression and similar defects in macromolecular uptake. This work defines transcriptional regulators that are necessary for nutrient uptake in neonatal enterocytes.
INTRODUCTION
The small intestine plays an essential role in nutrient uptake. Central to this function are enterocytes, the primary cell type responsible for absorption in the gut. Enterocytes are highly polarized columnar cells that line the villi of the small intestine. They are derived from the stem/progenitor cells that lie within the intestinal crypts and are the most abundant cell type in the intestinal epithelium. Enterocyte turnover is rapid and it is estimated that billions are generated every day in adult humans (Sender and Milo, 2021).
The enterocyte cell fate is developmentally linked to the morphogenesis of the intestine. Prior to embryonic day (E) 14.5 in the mouse, the intestine is a flat, pseudostratified epithelium (Wang et al., 2018). At E15.5, villi begin to form and enterocytes are first specified. Whereas adult enterocytes are terminally differentiated, it has been proposed that embryonic enterocytes remain competent to contribute to the stem cell pool through fission of villi (Guiu et al., 2019).
Recent work has begun to elucidate the transcriptional pathways that control enterocyte fate. A number of transcriptional regulators, including CDX1/2 (caudal type homeobox 1 and caudal type homeobox 2), HNF4α/γ (hepatocyte nuclear factor 4 alpha and hepatocyte nuclear factor 4 gamma), GATA4 (gata-binding protein 4) and ELF3 (E74-like ETS transcription factor 4) play important roles in enterocyte specification (Beuling et al., 2011; Chen et al., 2019a,b; Gao and Kaestner, 2010; Verzi et al., 2011). Of these, HNF4α and HNF4γ have emerged as essential regulators of the enterocyte program. They are required for brush border assembly as well as expression of many canonical enterocyte genes (Chen et al., 2021b, 2019a,b). Although HNF4α and HNF4γ are essential for enterocyte gene expression, they are not sufficient to drive this cell fate. Expression of HNF4α/γ throughout the crypt/villus axis, not exclusively in enterocytes, indicates that additional factors are required to regulate enterocyte gene expression.
Our lab previously identified MAFB (v-maf avian musculoaponeurotic fibrosarcoma oncogene homolog B) as an enterocyte-specific transcription factor in the intestinal epithelium (Sumigray et al., 2018). We initially found it to be enriched in villar cells by RNA-sequencing (RNA-Seq) analysis and subsequently demonstrated specific expression in enterocytes and not goblet or enteroendocrine cells. MAFB and c-MAF (MAF; avian musculoaponeurotic fibrosarcoma oncogene homolog) are both members of the Maf family of proteins and are characterized as large Maf proteins, containing both DNA-binding and transactivating domains. MAFB and c-MAF act redundantly in both the epidermis, by regulating genes associated with differentiated keratinocytes, and in macrophages, by controlling exit from the cell cycle (Aziz et al., 2009; Lopez-Pajares et al., 2015). Recent work has shown that deletion of c-MAF from the intestinal epithelium leads to a slight decrease in adult body weight, with disruption of zonation (the specific expression of distinct transcripts/proteins along the villar axis) and decreased uptake of amino acids (Cosovanu et al., 2022; González-Loyola et al., 2022). Given their redundant functions in other tissues, we sought to characterize the functions of both MAFB and c-MAF in the intestinal epithelium and determine their collective role in enterocytes.
Enterocytes are a heterogeneous group of cells in which gene expression is regulated both spatially and temporally. There are distinct enterocyte signatures and functions along the proximal-distal axis of the intestine, and in zones along the villus axis (Haber et al., 2017; Moor et al., 2018; Park et al., 2019). In addition, there are significant changes in enterocytes between neonatal/suckling and post-suckling/adult stages (Wilson et al., 1991). During the neonatal stage, there is both rapid growth and a relative immaturity of the intestine and other organs that support digestion, such as the liver and pancreas (Greengard, 1977; Robberecht et al., 1971). The major changes in expression of digestive enzymes that occur as animals approach weaning indicate a shift in expression program to support the changes in the diet from the milk-based diet of neonates to solid food post-weaning. The differences between neonatal and adult enterocytes are reflected in fundamentally different morphologies of the apical domain of the cell (Muncan et al., 2011; Skrzypek et al., 2007), and in mechanisms of nutrient uptake. In adults, most macromolecules are broken down in the lumen of the intestine and then absorbed. In contrast, neonates are thought to rely heavily upon uptake of whole macromolecules and subsequent degradation within lysosomes (Gonnella and Neutra, 1984; Wilson et al., 1991). In support of this, we previously demonstrated that the endocytic adaptor protein DAB2 is required for protein uptake in suckling mice (Park et al., 2019). Similarly, knockout of Mamdc4 (also known as endotubin), results in decreased weight in neonates and endocytic defects in Caco-2 cells (Cox et al., 2018). These studies have begun to identify the cellular mechanisms of nutrient uptake; however, many questions remain about the machinery required for this process, differences in uptake in neonatal and adult enterocyte, and the transcriptional regulation of these pathways.
Little is known about the transcriptional regulation of neonatal enterocyte function, with one interesting exception. The transcriptional repressor BLIMP1 (PRDM1) is expressed in neonatal mice and its expression is largely lost around weaning. Ablation of Blimp1 in the intestinal epithelium results in premature expression of an adult enterocyte gene signature (Harper et al., 2011; Muncan et al., 2011). However, it is not known whether BLIMP1 is required for bulk protein uptake in the neonatal intestine in addition to its role in repressing adult genes.
Here, we demonstrate that the transcription factors MAFB and c-MAF are expressed in enterocytes beginning at the earliest stage of their embryonic specification and through adulthood. Their expression is controlled by the enterocyte master regulators HNF4α and HNF4γ. Furthermore, we show that loss of Maf factors results in a failure to thrive and an inability of neonatal enterocytes to take up macromolecules from the intestinal lumen. In addition, we find that loss of BLIMP1 results in very similar changes in both gene expression and in an inability to take up nutrients. Together, these data begin to define the transcriptional regulation of nutrient uptake in the neonatal gut.
RESULTS
Developmental expression of Maf proteins
Enterocytes are first specified around E15.5 of mouse development, concomitant with the folding of villi. Prior to this, the intestinal epithelium is a flat, pseudostratified epithelium (Wang et al., 2018). Using two different antibodies, one that recognizes both MAFB and c-MAF (referred to herein as Mafs), as well as one that uniquely recognizes c-MAF (Fig. S1), we did not detect protein expression of Maf factors in the E14.5 intestinal epithelium (Fig. 1A). A day later, at E15.5, Maf factors were found in the nuclei of cells at the tops of the emerging villi (Fig. 1A). At both stages, there were also occasional mesenchymal cells that were Maf positive. To determine the relationship between villi folding and Maf expression, we quantified villi height and the emergence of Maf expression. There was a clear demarcation in villus height and Maf expression as emerging villi shorter than 50 µm were largely Maf negative, whereas almost all villi taller than 50 µm contained Maf-expressing cells (Fig. 1B). Villi are short when first formed and get taller as morphogenesis proceeds; thus, these data are consistent with MAFB expression turning on shortly after morphogenesis begins.
Maf proteins are expressed from the onset of enterocyte differentiation. (A) Expression of MAFB (red) in the developing intestine at E14.5 before villi emerge (left) and at E15.5, once villi have begun to form (right). Basement membranes are outlined by dashed white lines; the arrow indicates the epithelium. Schematics above depict intestinal morphology at the indicated developmental stages. (B) Quantification of Maf expression by villus height at E15.5. Villus height was measured from the base of the villus to the PDGFRα cluster at the top of the villus. Villi were grouped into two categories: those expressing Maf proteins (n=33), or those negative for Maf proteins (n=25). Horizontal lines indicate the group mean and circles represent individual villi. (C-E) Quantification of Maf protein (red) and Ki67 (green) colocalization at E15.5 (C), E17.5 (D) and P0 (E). The positions of individual nuclei were measured by their height along the villus axis. Each cell was classified as Ki67+ only (n=166 cells at E15.5; n=82 at E17.5; n=135 at P0), Maf+ only (n=102 at E15.5; n=189 at E17.5; n=323 at P0) or Ki67/Maf double positive (n=102 at E15.5; n=14 at E17.5; n=11 at P0). Only nuclei within 50 µm of the villus base were analyzed at P0. Above this region, there were never any Ki67+ cells. n=3 mice at each developmental time point. Dashed and dotted lines in violin plots represent the median and quartiles, respectively. Scale bars: 50 µm. (F,G) Lineage trace of Mafb+ cells. Mafb-Cre driving the expression of GFP. (F) By E17.5, most embryonic enterocytes are labeled with GFP as visualized in tissue sections. The image on the right shows a villus at higher magnification. (G) Crypts are uniformly GFP negative in P7 intestines, as shown in this wholemount image. Scale bars: 100 µm (F, main image); 50 µm (F, inset; G); dashed lines indicate the basement membrane.
Maf proteins are expressed from the onset of enterocyte differentiation. (A) Expression of MAFB (red) in the developing intestine at E14.5 before villi emerge (left) and at E15.5, once villi have begun to form (right). Basement membranes are outlined by dashed white lines; the arrow indicates the epithelium. Schematics above depict intestinal morphology at the indicated developmental stages. (B) Quantification of Maf expression by villus height at E15.5. Villus height was measured from the base of the villus to the PDGFRα cluster at the top of the villus. Villi were grouped into two categories: those expressing Maf proteins (n=33), or those negative for Maf proteins (n=25). Horizontal lines indicate the group mean and circles represent individual villi. (C-E) Quantification of Maf protein (red) and Ki67 (green) colocalization at E15.5 (C), E17.5 (D) and P0 (E). The positions of individual nuclei were measured by their height along the villus axis. Each cell was classified as Ki67+ only (n=166 cells at E15.5; n=82 at E17.5; n=135 at P0), Maf+ only (n=102 at E15.5; n=189 at E17.5; n=323 at P0) or Ki67/Maf double positive (n=102 at E15.5; n=14 at E17.5; n=11 at P0). Only nuclei within 50 µm of the villus base were analyzed at P0. Above this region, there were never any Ki67+ cells. n=3 mice at each developmental time point. Dashed and dotted lines in violin plots represent the median and quartiles, respectively. Scale bars: 50 µm. (F,G) Lineage trace of Mafb+ cells. Mafb-Cre driving the expression of GFP. (F) By E17.5, most embryonic enterocytes are labeled with GFP as visualized in tissue sections. The image on the right shows a villus at higher magnification. (G) Crypts are uniformly GFP negative in P7 intestines, as shown in this wholemount image. Scale bars: 100 µm (F, main image); 50 µm (F, inset; G); dashed lines indicate the basement membrane.
Because differentiation is typically associated with loss of proliferation, we sought to determine whether there is a correlation between the expression of Maf proteins and loss of proliferation. At E15.5, the Maf-negative cells in inter-villar spaces and in cells less than 50 µm from the base of villi tended to express the proliferation marker Ki67 (MKI67). Conversely, cells over 50 µm from the base of villi generally expressed Maf proteins and many cells at this stage were positive for both Ki67 and Mafs (Fig. 1C). At E17.5, most of the Ki67+ cells were located less than 35 µm from the base of villi and fewer Maf and Ki67 double-positive cells were detected (Fig. 1D). At postnatal day (P) 0, the day the mice are born, the Ki67+ region was restricted to inter-villar spaces and the 10 µm at the base of villi. Epithelial cells over 10 µm from the base of the villi were negative for Ki67 and most expressed Maf proteins (Fig. 1E). Together, these data demonstrate that Maf factors are expressed at the earliest stages of enterocyte specification and that cells become post-mitotic shortly after Maf expression.
To verify these results, we took advantage of an existing Mafb-Cre knock-in mouse line and a fluorescent reporter to visualize and lineage trace Mafb-expressing cells (Muzumdar et al., 2007; Wu et al., 2016). We observed Mafb lineage-labeled cells in the embryonic intestine, with a pattern similar to endogenous Maf-expressing cells, but shifted slightly later in developmental time, likely reflecting the delay in recombination and subsequent expression of the GFP reporter. At E16.5, 24% of epithelial cells were fluorescently labeled as determined by fluorescence-activated cell sorting (FACS) (n=7 mice, >1000 cells counted/mouse) and by E17.5 most cells on villi were lineage labeled (Fig. 1F). In postnatal mice, villi were uniformly labeled whereas crypts were negative (Fig. 1G). We found that fewer than 1% of crypts were labeled at P7 (n=3 mice, >100 crypts/mouse). Although these data are consistent with our antibody staining, they are inconsistent with a proposed model that embryonic villi undergo fission with villar cells giving rise to adult stem cells (Guiu et al., 2019). Our findings demonstrate that Mafb-positive embryonic enterocytes do not substantially contribute to the adult stem cell lineage. These data are incongruent with the idea that all embryonic intestinal epithelial cells have equal abilities to contribute to adult stem cells (Guiu et al., 2019). Importantly, the modeling studies of Krt19-CreER lineage-traced cells, upon which the villar fission model is based, assumed that lineage labeling occurred at equal frequencies in villar and inter-villar cells (Guiu et al., 2019). However, when we analyzed recombination 24 h after Krt19-CreER induction, we found that there was a preference for recombination to occur within the inter-villar regions (Fig. S2). This may reflect differences in expression of Krt19 in different cell types, or it may reflect increased recombination activity in proliferative cells, which has been previously noted (Mascré et al., 2012). In either case, our data demonstrate that not all embryonic intestinal cells give rise to adult stem cells and the Mafb-positive population is a differentiated cell type that does not contribute to adult stem/crypt cells.
Maf factor expression is regulated by HNF4α/γ
As Maf factors mark differentiated enterocytes, we wanted to address the upstream transcriptional regulators of their expression. HNF4α/γ transcription factors are crucial for enterocyte specification and intestinal homeostasis (Chen et al., 2021b, 2019a,b). We therefore analyzed mice in which HNF4α and HNF4γ were deleted throughout the intestinal epithelium, using Villin-CreER (el Marjou et al., 2004). Adult mice were collected 3 days after deletion of HNF4α/γ was initiated in the intestinal epithelium (HNF4 DKO). Multi-omic analysis of these mice has previously been performed, and we mined those datasets to determine the effects of HNF4 DKO on Maf factor expression (Fig. 2A,D,E) (Chen et al., 2019a,b). RNA-Seq analysis demonstrated a substantial reduction in both Mafb and Maf mRNA in HNF4 DKO intestinal epithelia (Fig. 2A). This result was validated at the protein level, with a decrease in both the intensity of staining and the percentage of MAFB-positive cells in the mutant tissue (Fig. 2B,C). In addition, we found that HNF4α/γ binding was enriched on/around the Mafb and Maf genes by chromatin immunoprecipitation with sequencing (ChIP-Seq) analysis (Fig. 2D,E, top panels, in green), suggesting that this regulation may be direct. Furthermore, the content of H3K27ac was clearly reduced at Mafb and Maf loci in HNF4 DKO mice, consistent with a conversion to less active transcription (Fig. 2D,E, compare gray and red peaks). Prior work demonstrated that HNF4α and HNF4γ control chromatin looping at many genes that they regulate (Chen et al., 2021a). We therefore examined our published HiChIP (in situ Hi-C library preparation followed by chromatin immunoprecipitation) dataset to determine effects at the Mafb and Maf loci. The interactions between chromosomal domains were clearly decreased at both loci in HNF4 DKO intestine (Fig. 2D,E, bottom panels). Together, these data indicate that Maf factors are direct transcriptional targets of HNF4α/γ.
Maf proteins are downstream of HNF4 factors in the intestinal epithelium. (A) Analysis of Mafb and Maf mRNA levels by RNA-Seq of HNF4 DKO adult mice 3 days after tamoxifen administration. (B) Immunostaining for Maf proteins (red) in control and HNF4 DKO tissue. Insets show magnified views of the boxed areas. (C) Quantification of the number of Maf-positive nuclei on villi of control and HFN4 DKO intestines (n=3 control mice, 2835 cells counted; n=4 HNF4 DKO mice, 4036 cells counted). Error bars in A and C represent s.d. ***P<0.001 (unpaired t-test). (D,E) Mafb (D) and Maf (E) genetic loci are depicted. Shown are ChIP-Seq peaks for HNF4α/γ binding on Mafb and Maf genes (green), and H3K27ac peaks in wild type (WT; gray) and HNF4 DKO (red). Arrow indicates transcriptional start site. Graphs below show DNA looping in control (blue) and HNF4α/γ DKO (red) intestines at Mafb and Maf gene loci.
Maf proteins are downstream of HNF4 factors in the intestinal epithelium. (A) Analysis of Mafb and Maf mRNA levels by RNA-Seq of HNF4 DKO adult mice 3 days after tamoxifen administration. (B) Immunostaining for Maf proteins (red) in control and HNF4 DKO tissue. Insets show magnified views of the boxed areas. (C) Quantification of the number of Maf-positive nuclei on villi of control and HFN4 DKO intestines (n=3 control mice, 2835 cells counted; n=4 HNF4 DKO mice, 4036 cells counted). Error bars in A and C represent s.d. ***P<0.001 (unpaired t-test). (D,E) Mafb (D) and Maf (E) genetic loci are depicted. Shown are ChIP-Seq peaks for HNF4α/γ binding on Mafb and Maf genes (green), and H3K27ac peaks in wild type (WT; gray) and HNF4 DKO (red). Arrow indicates transcriptional start site. Graphs below show DNA looping in control (blue) and HNF4α/γ DKO (red) intestines at Mafb and Maf gene loci.
Loss of Maf factors results in a failure to thrive
To examine the function of Maf proteins in the intestinal epithelium, we obtained floxed alleles for Mafb and Maf and used Villin-Cre and Villin-CreER to drive their conditional deletion in the intestinal epithelium (Maf DKO and Maf iDKO, respectively) (Wende et al., 2012; Yu et al., 2013). Villin-Cre is active in the intestinal epithelium starting at E14.5; therefore, in this model, the intestinal epithelium never expresses Maf factors (Fig. 3A). Immunofluorescence analysis confirmed the loss of Maf proteins in Maf DKO mice, demonstrating efficient deletion, yet there were no major changes in the morphology of the intestine (Fig. 3B, Fig. S3). Maf DKO mice were born at Mendelian ratios; however, mutant mice were clearly identifiable as being smaller than littermate controls by P3. From this time onward, Maf DKO mice were about 70% the weight of control mice, a phenotype consistent with deficits in intestinal function (Fig. 3C,D). Although some animals needed to be euthanized at weaning because of their small size, most survived into adulthood and were able to reproduce, although their smaller size persisted.
Developmental loss of Maf proteins leads to decreased body weight of neonates. (A-D) Embryonic ablation of Mafb and Maf, driven by Villin-Cre. (A) Schematic showing intestinal epithelium developing without Maf proteins. (B) Control and Maf DKO villus at P7. Basement membrane is outlined in white. (C) Maf DKO (left) and control (right) mice at P7. (D) Weight of Maf DKO mice as a percentage of the average weight of control littermates. *P<0.05; ***P<0.001; ****P<0.0001 (unpaired t-test). (E) Schematic showing normal Maf expression (pink) during development and deletion of Mafb and Maf in adult mice after addition of tamoxifen. (F) Control villus and Maf DKO villus in adult mice. Basement membrane is outlined in white. (G) Weights of control and Maf iDKO mice 30 days after the first of two 2.5 mg intraperitoneal tamoxifen injections given at day 0 and day 2. n=3 control, n=3 Maf DKO. ns, not significant: P=0.6475 (unpaired t-test). Error bars in D and G represent s.d. Scale bars: 10 µm (B); 25 µm (F).
Developmental loss of Maf proteins leads to decreased body weight of neonates. (A-D) Embryonic ablation of Mafb and Maf, driven by Villin-Cre. (A) Schematic showing intestinal epithelium developing without Maf proteins. (B) Control and Maf DKO villus at P7. Basement membrane is outlined in white. (C) Maf DKO (left) and control (right) mice at P7. (D) Weight of Maf DKO mice as a percentage of the average weight of control littermates. *P<0.05; ***P<0.001; ****P<0.0001 (unpaired t-test). (E) Schematic showing normal Maf expression (pink) during development and deletion of Mafb and Maf in adult mice after addition of tamoxifen. (F) Control villus and Maf DKO villus in adult mice. Basement membrane is outlined in white. (G) Weights of control and Maf iDKO mice 30 days after the first of two 2.5 mg intraperitoneal tamoxifen injections given at day 0 and day 2. n=3 control, n=3 Maf DKO. ns, not significant: P=0.6475 (unpaired t-test). Error bars in D and G represent s.d. Scale bars: 10 µm (B); 25 µm (F).
To determine potential functions of Maf proteins in adult mice, we used the Villin-CreER driver and tamoxifen to delete Mafb and Maf after weaning (Maf iDKO) (Fig. 3E). Similar to the developmental deletion of MAF proteins, no major changes in morphology were detected in Maf iDKO mice (Fig. 3F). Weights of Maf iDKO mice and controls were measured weekly for 1 month following tamoxifen injection. The weights of Maf iDKO mice did not differ from those of controls (Fig. 3G). Therefore, Maf factors play a clear role in intestinal function at neonatal stages, but are largely dispensable for intestinal morphology and function in adulthood. Recent reports on Maf knockout in adults suggests that there are strain-dependent and subtle effects for c-MAF loss on animal weight that emerged after several months (Cosovanu et al., 2022; González-Loyola et al., 2022).
Consistent with the normal epithelial architecture, we did not detect changes in proliferation following deletion of Mafb and Maf whether the deletion occurred during development or postnatally (Fig. S4A,B,E,F,I,J). We also investigated whether loss of enterocyte specific factors leads to a misregulation of lineage specification and results in increased numbers of secretory cells. We quantified the percentage of secretory goblet cells using Muc2 as a marker and observed no change in Maf DKO neonates or adults or Maf iDKO adults (Fig. S4C,D,G,H,K,L). Additionally, by RNA-Seq analysis (discussed in more detail below), we saw no change in markers for enteroendocrine or Paneth cells (data deposited in GEO under accession number GSE208763). Therefore, although the decreased body weight of Maf DKO mice suggests functional consequences of Maf loss during development, no major changes were detected in gross morphology, proliferation, and secretory lineage specification.
Maf factors are required for expression of phagocytic and endolysosomal genes
In order to assess the transcriptional changes resulting from Maf DKO, we turned to bulk RNA-Seq. We isolated epithelial cells from proximal, medial and distal intestine of three control and three Maf DKO mice at P7 and pooled all three regions into a single tube per mouse. Bulk RNA-Seq indicated that over 3400 genes were differentially expressed in Maf DKO mice (log2 fold change>0.25; P<0.05), with similar numbers of up- and downregulated genes (Fig. 4A,B). Maf proteins are thought to function largely as transcriptional activators; thus, we initially focused on genes that were downregulated in the mutant. KEGG pathway analysis of this gene set revealed a striking downregulation of genes involved in lysosomes, endocytosis and phagocytosis (Fig. 4C). These are all genes/pathways implicated in the macromolecule uptake and degradation of nutrients, including Dab2 and Mamdc4, which are required in neonates for endocytic uptake (Cox et al., 2018; Park et al., 2019). In adults, enterocytes express a diverse array of digestive enzymes and nutrient transporters that enable initial nutrient breakdown to occur extracellularly, and digested nutrients are transported into the enterocytes. In neonates, these pathways are not yet mature and enterocytes are thought to rely on bulk intake of nutrients (i.e. through endocytic and phagocytic pathways, as distinguished from transport of subunits) and subsequent digestion in lysosomes (Gonnella and Neutra, 1984; Wilson et al., 1991). However, much of the machinery regulating this intake, as well as the transcriptional regulation of neonatal nutrient uptake, has not been identified.
Maf proteins control expression of endosomal, phagocytic and lysosomal genes in the neonatal intestine. (A,B) Epithelial cells were isolated from intestines of P7 mice (n=3 control and n=3 Maf DKO mice). Cells were collected from proximal, medial and distal regions and pooled into a single sample for each mouse. Bulk RNA-Seq identified 3425 differentially regulated genes, 1731 downregulated (red) and 1694 upregulated (blue) (P<0.05). (C) KEGG analysis of downregulated genes. The top 12 pathways are shown. (D) Quantification of western blots of protein isolated from distal intestinal epithelium of P7 mice (n=3 control and n=3 Maf DKO) showing decreased protein expression of GULP1 (P=0.0143), legumain (P= 0.0147) and cathepsin (P=0.0073). Unpaired t-test. Band intensities were normalized to tubulin loading control. (E) Immunohistochemistry stain for LAMP1 in P7 intestinal epithelium of control and Maf DKO. Dashed lines delineate the basement membrane. Scale bar: 50 µm. (F) Western blot analysis of LAMP1 levels. Protein was isolated from distal intestinal epithelium of P7 mice (n=3 control and n=3 Maf DKO). ns, not significant. P=0.2894 (unpaired t-test). Error bars in D and F represent s.d.
Maf proteins control expression of endosomal, phagocytic and lysosomal genes in the neonatal intestine. (A,B) Epithelial cells were isolated from intestines of P7 mice (n=3 control and n=3 Maf DKO mice). Cells were collected from proximal, medial and distal regions and pooled into a single sample for each mouse. Bulk RNA-Seq identified 3425 differentially regulated genes, 1731 downregulated (red) and 1694 upregulated (blue) (P<0.05). (C) KEGG analysis of downregulated genes. The top 12 pathways are shown. (D) Quantification of western blots of protein isolated from distal intestinal epithelium of P7 mice (n=3 control and n=3 Maf DKO) showing decreased protein expression of GULP1 (P=0.0143), legumain (P= 0.0147) and cathepsin (P=0.0073). Unpaired t-test. Band intensities were normalized to tubulin loading control. (E) Immunohistochemistry stain for LAMP1 in P7 intestinal epithelium of control and Maf DKO. Dashed lines delineate the basement membrane. Scale bar: 50 µm. (F) Western blot analysis of LAMP1 levels. Protein was isolated from distal intestinal epithelium of P7 mice (n=3 control and n=3 Maf DKO). ns, not significant. P=0.2894 (unpaired t-test). Error bars in D and F represent s.d.
To validate the transcriptomic data, we examined the levels of proteins predicted to be downregulated in the Maf DKO. The phagocytic receptor GULP1, as well as legumain and cathepsins (both lysosomal proteases), were all detected at lower levels in intestinal epithelial lysates collected from the Maf DKO consistent with the RNA-Seq results (Fig. 4D, Fig. S5). Given the decrease in expression of many lysosomal genes and proteins, we assessed lysosomal architecture in the Maf DKO enterocytes. Although many of the mRNAs for lysosomal enzymes were dramatically decreased, those encoding structural components were only modestly decreased (Lamp1 down by 1.72-fold and Lamp2 by 1.66-fold). Consistent with this, we did not see a significant decrease in lysosomes as assayed by LAMP1 staining or western blotting (Fig. 4E,F, Fig. S5). This suggests that the function but not the biogenesis of lysosomes is affected in Maf DKO intestine.
We next sought to identify chromosomal loci bound by MAFB using CUT&RUN analysis. Two independent samples were generated using cells isolated from P7 villi. We identified ∼4000 bound loci that were present in both samples, roughly equivalent to the number of loci bound by MAFB in keratinocytes (Lopez-Pajares et al., 2015). Importantly, comparison with the genes identified as both downregulated in Maf DKO and loci bound by MAFB revealed KEGG terms for endocytosis, phagocytosis and lysosomes, supporting the idea that Maf factors may directly regulate these targets (Fig. S6). Analysis of genes both downregulated in Maf DKO intestines and bound by MAFB revealed a statistically significant enrichment of shared genes (P<4.0×10−41), supporting regulatory functions of MAFB at these loci. When analyzing genes that are both bound and downregulated by Maf factors, ‘actin regulators’ was the top-enriched term. Of the 13 actin regulators on this list, 11 have been demonstrated to play roles in endocytosis or phagocytosis. Furthermore, Gulp1 (a phagocytic receptor), Dab2 (an endocytic adapter), Lrp2 (a scavenger receptor) and the endocytic recycling proteins Rab11a and Rab11fip2 were identified as bound by MAFB (Fig. S6). Together, these data support a model in which Maf proteins directly regulate the expression of genes necessary for intake of macromolecular nutrients.
Neonatal Maf DKO enterocytes fail to take up protein and dextran
The decreased expression of genes/proteins important for endocytosis and lysosome function suggests that these mice might have defects in nutrient uptake and degradation. In suckling neonates, enterocytes internalize milk, which contains components that are autofluorescent, especially in the green wavelengths. The intake of autofluorescent compounds from the milk is especially evident in the distal intestine. However, this autofluorescence was not observed in the intestines of Maf DKO neonates (Fig. 5A). The absence of milk fluorescence in Maf DKO enterocytes was not secondary to lack of feeding as both control and Maf DKO pups had milk in their stomachs and throughout the lumen of the intestine. These observations demonstrate that the failure to take in nutrients is at the level of the enterocytes.
Defective bulk intake of nutrients in Maf DKO enterocytes. (A) Autofluorescence from milk (green) taken in by control enterocytes is not detected in Maf DKO mice. Images taken from the distal intestine of P7 mice. MAFB is in magenta. Boxed areas are shown at higher magnification in the central images. Scale bar: 50 µm. (B) Top: Schematic of the experimental set up. P7 control and Maf DKO mice were gavaged with 40 µg fluorescently labeled protein (Lys-RRX, red) and 40 µg fluorescently labeled carbohydrate (Dext-488, green). After 3 h, mice were sacrificed and intestines were collected. Bottom: Maximum intensity projections of intestinal epithelium collected from the medial intestine of control and Maf DKO mice. Scale bar: 50 µm. (C) Quantification of uptake in individual cells isolated from the medial small intestine by FACS (n=3 control, and n=5 Maf DKO; 10,000 cells/sample were counted). ****P<0.0001 (unpaired t-test). Error bars represent s.d. (D) Cross-sections of villi from medial intestine of mice gavaged with mCherry protein (red) 3 h prior to collection. Phalloidin was used to stain actin (green). mCherry protein is found in many vesical structures in enterotypes of control mice, but is absent or decreased in Maf DKO enterocytes. Scale bar: 10 mm.
Defective bulk intake of nutrients in Maf DKO enterocytes. (A) Autofluorescence from milk (green) taken in by control enterocytes is not detected in Maf DKO mice. Images taken from the distal intestine of P7 mice. MAFB is in magenta. Boxed areas are shown at higher magnification in the central images. Scale bar: 50 µm. (B) Top: Schematic of the experimental set up. P7 control and Maf DKO mice were gavaged with 40 µg fluorescently labeled protein (Lys-RRX, red) and 40 µg fluorescently labeled carbohydrate (Dext-488, green). After 3 h, mice were sacrificed and intestines were collected. Bottom: Maximum intensity projections of intestinal epithelium collected from the medial intestine of control and Maf DKO mice. Scale bar: 50 µm. (C) Quantification of uptake in individual cells isolated from the medial small intestine by FACS (n=3 control, and n=5 Maf DKO; 10,000 cells/sample were counted). ****P<0.0001 (unpaired t-test). Error bars represent s.d. (D) Cross-sections of villi from medial intestine of mice gavaged with mCherry protein (red) 3 h prior to collection. Phalloidin was used to stain actin (green). mCherry protein is found in many vesical structures in enterotypes of control mice, but is absent or decreased in Maf DKO enterocytes. Scale bar: 10 mm.
To probe further the ability of Maf DKO enterocytes to take in nutrients in a more controlled manner and to determine whether the uptake of specific classes of nutrients is affected, we turned to a gavage assay. In this assay, we used oral gavage to deliver fluorescently labeled nutrients directly to the stomachs of P7 Maf DKO and control mice. To assess protein uptake, we used lysozyme tagged with rhodamine red (Lys-RRX). In addition, we used a dextran conjugated to Alexa Fluor 488 (Dex-488) as a carbohydrate macromolecule that can be taken up via endocytic/phagocytic pathways. Three hours after oral gavage, we collected the intestines (Fig. 5B). First, we examined the fluorescence content in intact intestines and observed green and red signal throughout the lumen of the small intestine and colon in both control and Maf DKO mice. Therefore, the gavaged solution had progressed throughout the digestive tract. Next, we used fluorescence microscopy to determine whether uptake of the Lys-RRX and Dex-488 could be detected in enterocytes. Epithelial wholemounts of control mice showed efficient uptake of both lysozyme and dextran; however, very little uptake of labeled nutrients was detected in enterocytes of Maf DKO mice (Fig. 5B). Intestinal epithelial cells were dissociated into single cells and the percentage of cells positive for Lys-RRX and Dextran-488 was determined via FACS. The percentage of cells that took up the fluorescently labeled nutrients was significantly decreased in Maf DKO enterocytes (Fig. 5C). Closer examination of control enterocytes in cross-section showed several discrete compartments of varying sizes filled with labeled protein. Smaller vesicles were located near the apical region of the cells, and larger structures, presumably lysosomes, were seen in some cells closer to the nucleus (Fig. 5D). Taken together, the decreased expression of phagocytic, endocytic and lysosomal genes and the resulting inability to take in protein and dextran demonstrate that Maf proteins are crucial for nutrient intake in neonatal enterocytes.
Metabolic changes in Maf DKO enterocytes
Although there was a dramatic loss of uptake of proteins and carbohydrates in Maf DKO mice, they were still able to grow, albeit at a reduced rate. To determine whether there were other transcriptional changes that may promote growth, we examined the genes that were upregulated following loss of Maf proteins. KEGG analysis indicated major rewiring of metabolic pathways. Interestingly, genes required for fatty acid degradation and peroxisomes, organelles specialized for lipid breakdown (Lazarow and De Duve, 1976), were upregulated (Fig. 6A). In addition, genes for fatty acid transfer proteins (Slc27a4/a2), microsomal triglyceride transfer proteins (Mttp) and even apolipoproteins and their regulators (Apoc1/2/3, Apobec1, Apoa1, Apoe, Apoa4) were increased. This upregulation suggests increased lipid metabolism in Maf DKO mice. To interrogate this possibility, we examined lipid uptake in Maf DKO neonates. We gavaged P7 mice with corn oil and collected the intestines 3 h later. We used Oil Red O to stain the lipids and observed that both control and Maf DKO mice efficiently take in lipids (Fig. 6B,C). The ability of Maf DKO enterocytes to take in lipids is consistent with the fact that fatty acids and cholesterol do not rely on the same intake pathways as proteins (Johnston and Borgstroem, 1964; Shiau, 1981). Next, we stained for the peroxisomal marker PMP70 (ABCD3) to determine whether the increased levels of peroxisomal mRNAs reflected changes in the levels of this organelle. We found a clear increase in PMP70-stained peroxisomes (Fig. 6D,E). These data, in addition to the upregulation of fatty acid β-oxidation enzymes (Fig. 6A), point to potential compensation for the lack of protein and carbohydrate intake through increased metabolism of lipids. Furthermore, the expression of enzymes for the biosynthesis of amino acids and co-factors was also upregulated (Fig. 6A). This could indicate use of metabolic building blocks from lipid breakdown to synthesize other metabolites that are limited in Maf DKO mice.
Increased expression of metabolic genes in Maf DKO neonates. (A) KEGG analysis of genes upregulated in Maf DKO villi. The top 14 pathways with disease terms excluded from list are shown. (B) Oil Red O staining of proximal intestines of mice gavaged with 10 µl corn oil and collected 3 h later. (C) Quantification of Oil Red O signal in Maf DKO and control mice (n=3 control and 3 Maf DKO mice; five measurements from four different images per mouse were quantified). ns, not significant. P=0.18 (unpaired t-test). (D) Staining for the peroxisomal marker PMP70 (red) in P7 medial intestine. Boxes indicate regions shown at higher magnification below. (E) Quantification of PMP70 signal in control and Maf DKO mice (n=3 control and 3 Maf DKO mice; five measurements from four different images per mouse were quantified). **P=0.0031 (unpaired t-test). Error bars in C and E represent s.d. Dashed lines in B and D indicate the basement membrane. Scale bars: 50 µm.
Increased expression of metabolic genes in Maf DKO neonates. (A) KEGG analysis of genes upregulated in Maf DKO villi. The top 14 pathways with disease terms excluded from list are shown. (B) Oil Red O staining of proximal intestines of mice gavaged with 10 µl corn oil and collected 3 h later. (C) Quantification of Oil Red O signal in Maf DKO and control mice (n=3 control and 3 Maf DKO mice; five measurements from four different images per mouse were quantified). ns, not significant. P=0.18 (unpaired t-test). (D) Staining for the peroxisomal marker PMP70 (red) in P7 medial intestine. Boxes indicate regions shown at higher magnification below. (E) Quantification of PMP70 signal in control and Maf DKO mice (n=3 control and 3 Maf DKO mice; five measurements from four different images per mouse were quantified). **P=0.0031 (unpaired t-test). Error bars in C and E represent s.d. Dashed lines in B and D indicate the basement membrane. Scale bars: 50 µm.
Premature expression of adult enterocyte genes in Maf DKO neonates
In addition to metabolic genes for fatty acid degradation, mRNAs for several brush border enzymes, including sucrase isomaltase (Sis) and trehalase (Treh), are increased in Maf DKO neonates (Fig. 7D). We validated these changes at the protein level as well. Immunohistochemistry of sucrase isomaltase revealed increased brush border staining in the Maf DKO intestine, compared with control littermates (Fig. 7A), consistent with its localization in adult enterocytes. Trehalase levels were also increased in the mutant mice as determined by western blot analysis (Fig. 7B, Fig. S7). Finally, Arg2, a gene that is important for urea metabolism, was also upregulated at the protein level (Fig. 7C, Fig. S7).
Blimp1 KO neonates display a similar phenotype to Maf DKO neonates. (A) Immunohistochemistry analysis of sucrase isomaltase in P7 intestine of control and Maf DKO mice. (B,C) Quantification of western blots of protein isolated from distal intestine of P7 mice (n=3 control and n=3 Maf DKO) showing increased protein expression of Arg2 (*P=0.0174) and trehalase (*P=0.0177) (unpaired t-test). Band intensity was normalized to tubulin loading control. (D,E) Comparisons of Maf DKO sequencing hits and published array data following Blimp1 KO, on a select set of genes (D) and on the entire gene list (E). (F) Protein uptake assay on Blimp1 KO mice, consisting of oral gavage of 40 mg Lys-RRX (red) into the stomachs of P7 mice, with intestines collected 3 h later. Epithelial wholemounts of medial intestine are shown. (G) Quantification Lys-RRX brightness in control and Maf DKO (n=3 control and 3 Prdm1 KO; five measurements from four different images per mouse were quantified). **P=0.0030 (unpaired t-test). Error bars in B and G represent s.d. Scale bars: 25 µm.
Blimp1 KO neonates display a similar phenotype to Maf DKO neonates. (A) Immunohistochemistry analysis of sucrase isomaltase in P7 intestine of control and Maf DKO mice. (B,C) Quantification of western blots of protein isolated from distal intestine of P7 mice (n=3 control and n=3 Maf DKO) showing increased protein expression of Arg2 (*P=0.0174) and trehalase (*P=0.0177) (unpaired t-test). Band intensity was normalized to tubulin loading control. (D,E) Comparisons of Maf DKO sequencing hits and published array data following Blimp1 KO, on a select set of genes (D) and on the entire gene list (E). (F) Protein uptake assay on Blimp1 KO mice, consisting of oral gavage of 40 mg Lys-RRX (red) into the stomachs of P7 mice, with intestines collected 3 h later. Epithelial wholemounts of medial intestine are shown. (G) Quantification Lys-RRX brightness in control and Maf DKO (n=3 control and 3 Prdm1 KO; five measurements from four different images per mouse were quantified). **P=0.0030 (unpaired t-test). Error bars in B and G represent s.d. Scale bars: 25 µm.
Strikingly, many of these genes identified in Maf DKO RNA-Seq were also differentially regulated following deletion of Blimp1 in the intestinal epithelium (Harper et al., 2011; Muncan et al., 2011). BLIMP1 is a transcriptional repressor that is expressed in the intestine from late embryonic development until weaning and is proposed to repress the expression of mature enterocyte genes. Loss of BLIMP1 leads to a failure to thrive in neonatal mice, and premature maturation of neonatal enterocytes (Harper et al., 2011; Muncan et al., 2011). Sucrase isomaltase, trehalase and Arg2, which are expressed in adult enterocytes (Henning, 1981, 1985; Hurwitz and Kretchmer, 1986), are inappropriately expressed in neonatal intestines of Maf DKO and Blimp1 knockout (KO) mice (Fig. 7D). Furthermore, there was a similar decrease in lysosomal protease gene expression in Blimp1 KO and Maf DKO intestines (Fig. 7D). These are just a few examples highlighting an overall shift in the expression of Maf DKO enterocytes towards an adult enterocyte signature. Thus, we sought to compare all the differentially regulated genes identified by microarray in P7 Blimp1 KO intestines (Muncan et al., 2011) with those identified as differentially regulated in Maf DKO intestines. We determined that 133 out of 208 genes identified as being significantly changed in Blimp1 KO intestine by microarray analysis were also differentially regulated in Maf DKO mice (Fig. 7E). The hypergeometric test indicated that the enrichment is 4.5-fold over random (P=4.6×10−65).
Given these similarities, we next investigated whether there were similar deficits in nutrient uptake. Using the same gavage scheme described above, we found that Blimp1 KO enterocytes fail to take in fluorescently labeled protein (Fig. 7F,G). These data suggest that these factors may either act together or regulate each other to control nutrient absorption. Mafb and Maf mRNA levels were not significantly altered in the Blimp1 KO microarrays (Harper et al., 2011; Muncan et al., 2011). In contrast, we noted a small decrease (0.58-fold) in Blimp1 mRNA levels in Maf DKO intestines (Fig. S8). However, antibody staining detected similar levels of BLIMP1-positive nuclei in Maf DKO intestines (Fig. S8). Together, these data support these proteins acting in concert rather than regulating each other's expression.
DISCUSSION
In this work, we demonstrated that Maf family transcription factors, MAFB and c-MAF, are required for nutrient absorption in the neonatal intestine. These factors, expression of which is controlled by master enterocyte regulators HNF4α and HNF4γ, are found specifically in enterocytes in the intestinal epithelium from their earliest emergence through adulthood. MAFB and c-MAF are required for internalization of nutrients in neonates, and their loss results in decreased expression of lysosomal, phagocytic and endosomal genes. Notably, there is also an upregulation of many metabolic pathways, including fatty acid degradation pathways in the peroxisome. Similar functional defects and gene expression changes are also found in Blimp1 mutant intestines. Together, this work defines important transcriptional regulators of neonatal enterocyte function, and identifies both putative machinery for nutrient uptake, and potential compensatory pre-maturation of neonatal enterocytes when these intake pathways are perturbed.
Although there is clear heterogeneity in enterocyte function both spatially and temporally, the underlying transcriptional regulation of this has remained unclear. HNF4α/γ factors are required for a general enterocyte program, and regulate pathways for brush border assembly and metabolic processes (Chen et al., 2019a,b). However, we have limited knowledge of the downstream regulators that control the specialized functions of different types of enterocytes. As an example, Gata4 is expressed in the duodenum and jejunum, but not the ileum, and loss of Gata4 leads to enterocyte expression that is typically associated with the ileal region of the intestine (Bosse et al., 2006). Our data demonstrate that Maf factors lie downstream of HNF4α/γ and strongly suggest that they are direct targets of these transcriptional regulators, based on RNA-Seq, ChIP-Seq and chromatin conformation analysis. However, whereas HNF4α and HNF4γ are expressed throughout the crypt/villus axis (Chen et al., 2019b), Maf factors are only present in enterocytes, suggesting additional levels of regulation to control Maf expression. Recent work suggests that BMP signaling promotes expression of Maf, consistent with the enrichment of BMP signals from the villar tip (Cosovanu et al., 2022; González-Loyola et al., 2022).
Maf expression is first evident as villi begin to form. Although initially these cells are also proliferative, they appear to quickly become post-mitotic. By the time the mice are born, there is a clear delineation between the postmitotic Maf-positive cells of the villi and the proliferative inter-villar cells. Indeed, lineage tracing of Mafb-expressing cells demonstrated that their progeny do not significantly contribute to crypts, the stem cell niche. This was surprising given a prior study that suggested that all embryonic intestinal epithelial cells are equally capable of contributing to the stem cell pool. Mechanistically, this model suggested that fission of embryonic villi repositions enterocytes into inter-villar regions where they continue to proliferate (Guiu et al., 2019). Our data demonstrate that at least some embryonic epithelial cells (Mafb-positive cells) are committed to their differentiated fate and do not produce stem cell progeny, but do not rule out that Mafb-negative cells of the emerging villi can contribute to stem cell pools. However, it is important to note that the data supporting this model are based on Krt19-CreER-based lineage tracing and the assumption that cells throughout the intestinal epithelium are evenly labeled. We found that under the conditions used, recombination preferentially occurs in inter-villar regions, not in an unbiased way throughout the entire epithelium, thus lending support to the model that most cells on embryonic villi do not contribute to the stem cell population.
Although Maf factors are expressed in enterocytes throughout life, our data suggest they are most important for intestinal function in neonates. Maf DKO mice exhibited failure to thrive, with weights only about 70% of their wild-type littermates. The Maf mutant intestine, although largely normal histologically, had a profound defect in the uptake of large macromolecules from the lumen. This mode of nutrient absorption is specific to the neonate, in which endocytic and phagocytic uptake followed by degradation in the lysosome is thought to play a major role (Gonnella and Neutra, 1984; Wilson et al., 1991).
The machinery for nutrient uptake in neonatal enterocytes remains largely unstudied. Work in zebrafish, which are thought to use similar uptake pathways throughout their lives, has identified important roles for scavenger receptors, including cubilin and endocytic adaptors, such as DAB2 (Park et al., 2019). The function of DAB2 is conserved in neonatal mice, as we demonstrated lack of efficient protein uptake in Dab2 mutant neonatal intestines (Park et al., 2019). Additionally, Mamdc4 is required for the apical endocytic phenotype of neonatal enterocytes (Cox et al., 2018). Interestingly, CDX2, which regulates the expression of HNF4α/γ in early development and later binds similar genetic loci as HNF4α/γ (Chen et al., 2019a), is required for lysosome maturation in enterocytes (Gao and Kaestner, 2010). The downregulated genes in Maf knockout mice provide a potential parts list for genes involved in uptake of nutrients. Notably, we find that many endocytic, phagocytic and lysosomal genes, including Dab2, are downregulated. Owing to the defect in internalization, we were unable to assess possible defects in subsequent vesicle trafficking. That said, at the transcript level, many regulators of vesicle trafficking, including Rab11 and Rab11fip2 were downregulated. In addition to these pathways, many actin regulators were identified in the genes that are both bound by MAFB (by CUT&RUN) and downregulated in the Maf mutant intestine. Of the 13 genes in this category, 11 are actin regulators that have been demonstrated to play important roles in endocytosis/phagocytosis. These endocytic and cytoskeleton genes are excellent candidates for future function validation of roles in nutrient uptake.
In addition to the downregulated genes in the Maf DKO, there was also a substantial number of genes that were upregulated. These include many ‘adult’ enterocyte genes and suggest that either Mafs inhibit their expression directly, or their upregulation is secondary to disruption of other genes and pathways following loss of Mafs. Loss of endocytic internalization may lead to increased expression of adult genes to compensate for the lack of nutrient uptake. A very similar set of differentially regulated genes was identified in the Blimp1 mutant intestine. BLIMP1 is a transcriptional repressor proposed to inhibit adult enterocyte gene expression in neonates (Harper et al., 2011; Muncan et al., 2011). Remarkably, we find that Blimp1 is also required for nutrient uptake in neonates, consistent with the previously published gene expression data. The nature of the interaction of Maf factors and Blimp1 remains unclear in the intestine, although there is precedent for these factors working cooperatively to induce the expression of IL10 in CD4+ T-cell subsets (Neumann et al., 2014).
Finally, the intriguing finding that many metabolic pathways are upregulated in both Maf DKO and Blimp1 knockout intestine demonstrates a close integration of metabolism and physiology in these cells. Notably, we found that the mice, although deficient in uptake via phagocytosis and endocytosis, were able to take in lipids similar to control littermates. In Maf DKOs, there is upregulation of almost all enzymes of the fatty acid β-oxidation pathway as well as peroxisomal genes and number of peroxisomes. These data suggest that these mice may become more reliant on energy and/or building blocks from fatty acid catabolism. It is not clear whether many of these changes are directly controlled by these transcription factors and/or whether they are controlled by changes in the nutritional status of the mice as a result of loss of macromolecular uptake. Previous work demonstrated that enterocytes change their expression of digestive enzymes in response to the nutrients available. Upon prolonged exposure to a lactose-rich diet, enterocytes maintain the expression and activity of lactase (Peuhkuri et al., 1997). Together, these data underscore the dynamic modes of nutrient uptake, the capacity of enterocytes to utilize the macromolecules available, and the transcriptional network that governs these changes.
MATERIALS AND METHODS
Mice
Animal work was performed in accordance with Duke University's Institutional Animal Care and Use Committee guidelines and approval. Mouse lines used in this study were: CD1 (Charles River), Mafb-Cre (Wu et al., 2016), mTmG (Muzumdar et al., 2007), Villin-Cre (Madison et al., 2002), Villin-CreER (gift from Sylvie Robine, Institut Curie, France) (el Marjou et al., 2004), Mafbfl/fl (Yu et al., 2013), Maffl/fl (Wende et al., 2012) and Prdm1fl/fl (Shapiro-Shelef et al., 2003). Both male and female mice were used, and genotyping was performed by PCR. Mice were maintained in a facility with 12 h light/dark cycles. See Table S1 for details of experimental strains used.
Tissue preparation
Isolated intestinal tissue was either embedded immediately in Optical Cutting Temperature (Sakura) or prefixed in 4% paraformaldehyde (PFA) in PBS-T (PBS+0.2% Triton X-100) overnight before embedding. Proximal, medial and distal regions, corresponding to duodenum, jejunum and ileum, were collected. Blocks of frozen tissue were cut at 7 µm thickness using a cryostat. Blocks and tissue sections were stored at −80°C. For each experiment, a minimum of three age-matched mice from each condition were collected and analyzed.
Intestinal epithelium isolation
Mice were sacrificed and intestines were dissected. Several ∼1 cm pieces of small intestine were isolated, cut open longitudinally and washed in PBS. Any luminal contents were removed by gentle shaking. Up to five ∼1 cm sections were incubated in 5 ml of 30 mM EDTA in PBS. Intestines from P7 mice were incubated at 37°C for 10 min without rotation and intestines from adult mice were incubated for 30 min at 4°C with rotation. The intestines were transferred to a Petri dish with PBS, grasped with forceps and then shaken to release the epithelium. The epithelial sheets were transferred to a 15 ml conical tube and allowed to gravity pellet. The tissue was further processed according to the downstream application.
Isolation of single cells from intestinal epithelium
Isolated intestinal epithelium was incubated in PBS+0.8 mg/ml dispase for 20 min at 37°C with intermittent shaking. Cells were then passed through a 70 µm strainer, washed with PBS+10% fetal bovine serum (FBS) and cells were pelleted for 5 min at 2400 RPM (500 g). Supernatant was aspirated and cells were washed again with PBS+10% FBS. Cells were either fixed in 4% PFA in PBS for 15 min at room temperature (RT), washed twice with PBS, and stored at 4°C or used fresh.
Immunofluorescence
After thawing the tissue sections, they were fixed with 4% PFA in PBS-T for 8 min. When staining for Muc2, antigen retrieval was performed by boiling samples for 10 min in 10 mM sodium citrate buffer, pH 6.0. Tissue sections were washed with PBS-T, then blocked with 3% bovine serum albumin (BSA) (Cytiva), 5% normal goat serum (NGS) (Gibco), 5% normal donkey serum (NDS) (Sigma-Aldrich) in PBS-T for 15 min. Primary antibodies were diluted in blocking buffer and added to the sections for 15 min to 1 h at RT. After three washes with PBS-T, secondary antibody and any stains, such as Hoechst 34580 or Phalloidin (see Table S1), were added to the tissue sections for 15 min at RT. After three washes with PBS-T, the slides were mounted with 90% glycerol in PBS plus 2.5 mg/ml p-phenylenediamine (Sigma-Aldrich) and sealed with clear nail polish. The sections were imaged using Zeiss AxioImager Z1 microscope with Apotome.2 attachment, Plan-APOCHROMAT 20×/0.8 objective or Plan-NEOFLUAR 40×/1.3 oil objective, Plan-APOCHROMAT 63×/1.4 oil objective, Axiocam 506 mono camera, and Zen software (Zeiss). See Table S1 for details of antibodies used.
Immunohistochemistry
After thawing the tissue sections, they were fixed with 4% PFA in PBS-T for 8 min. When staining for Lamp1 and sucrase isomaltase, antigen retrieval was performed by boiling samples for 10 min in 10 mM sodium citrate buffer, pH 6.0. Tissue was blocked first with 0.3% hydrogen peroxide in PBS for 10 min, then in 3% BSA, 5% NGS, 5% NDS in PBS-T for 15 min at RT. The tissues were incubated with primary antibody diluted 1:100 in blocking solution for 1 h at RT then washed three times with PBS-T. Next, the tissues were incubated with horseradish peroxidase-conjugated secondary antibodies at 1:100 dilution in blocking solution for 15 min then washed three times with PBS-T. DAB reagent (Vector Laboratories, SK-4100) was prepared fresh according to the directions in the kit then added to the tissue sections for 30 s to 10 min. Tissue sections were washed, then allowed to dry completely. Slides were mounted with Permount (Fisher Scientific) and imaged as described above. See Table S1 for details of antibodies used.
Immunostaining intestinal epithelial wholemounts
After isolating intestinal epithelium as described above, the tissue was fixed overnight in 4% PFA in PBS at 4°C. For immunostaining, the tissue was blocked in 3% BSA, 5% NGS, 5% NDS in PBS-T for 45 min with rocking at RT, then incubated with primary antibody diluted in blocking solution for 1.5 h with rocking at RT. Tissue was allowed to gravity pellet, and washed three times for 5 min each wash, with rocking. Secondary antibody along with nuclear stain (Hoechst 3450) and actin stain (Phalloidin; see Table S1) were diluted in blocking solution and incubated for 45 min at RT with rocking and protected from light. In cases in which only stains or nuclear dye were used, the blocking and primary steps were not performed. Epithelial wholemounts were mounted on slides using melted VALAP (vasaline, lanolin and paraffin in a 1:1:1 ratio) to create a border within which the tissue was placed. After the VALAP had solidified, the tissue was pipetted inside the border, Antifade (p-Phenylenediamine, pH 9, in 90% glycerol and 10% water) was added to cover the tissue and a coverslip was placed on top and sealed with melted VALAP. Slides were stored at 4°C and imaged within 1 week of preparation on Zeiss 780 upright confocal with a 20×/0.8 Plan-Apochromat objective or 63×/1.4 oil immersion Plan-Apochromat objective.
Western blotting
After isolating intestinal epithelium as described above, four times the volume of sample buffer, with 15% BME, was added to the tissue, which was then incubated at 95°C for 10 min with intermittent trituration. Samples were aliquoted, flash-frozen, and stored at −80°C. Samples were run on 10% polyacrylamide gels, and transferred to nitrocellulose membranes. Membranes were blocked in 5% BSA in PBS+1% Tween (PBS-Tween) for 45 min at RT. Primary antibodies were diluted in 5% BSA in PBS-Tween and added to the membrane overnight at 4°C with rocking. Tubulin is the exception to this and was added to the membranes for 30 min at RT with rocking. Membranes were washed three times for 5 min each wash with PBS-Tween. LI-COR secondary antibodies were diluted in 5% BSA in PBS-Tween and added to the membrane for 45 min at RT with rocking. After washing, membranes were imaged using a LI-COR Odyssey imaging system and band intensity was quantified using LI-COR software. Protein amounts were standardized using Revert total protein stain (LI-COR). Additionally, protein levels were normalized to tubulin for quantification. See Table S1 for details of antibodies used.
Oil Red O staining
P7 mice were gavaged with 15 µl corn oil. Three hours later, intestines were collected as described above. This Oil Red O stain protocol is based on Lillie and Ashburn (1943). A stock solution of 0.5 g Oil Red O (Sigma-Aldrich) in 100 ml isopropanol was prepared. A working solution was prepared fresh by adding 15 ml of stock solution to 10 ml water, incubating at RT for 10 min then gravity filtering through Whatman paper. At the same time, slides with 7-μm-thick small intestine sections were dried completely then fixed with 4% PFA in PBS for 1 h at RT. Slides were washed three times with diH2O then incubated in the filtered Oil Red O working solutions for 30 min at RT. Slides were washed three times again and run under tap water for 5 min. Slides were dried completely then mounted with Permount.
In vivo uptake assays
P7 mice were gavaged with 40 µg Lys-RRX (Nanocs, LS1-RB-1) and 40 µg Dex-488 (Invitrogen, D22910) in a total volume of 10 µl or, for lipid experiments, with 15 µl corn oil using 22-gauge plastic feeding tubes. Mice were euthanized and intestines were collected 3 h post-gavage. Pieces of whole intestine, epithelial isolates, and single cells were collected as described above depending on the experiment.
FACS analysis
Fixed intestinal epithelial cells were isolated as described above and stored at 4°C until the time of analysis. Immediately before FACS analysis, the cells were passed through 100 µm Celltrix filters. Samples were processed on a Fortessa Analyzer with assistance from the Flow Cytometry core facility at Duke University.
For FACS to quantify Mafb-Cre lineage tracing, early labeling of intestinal epithelial cells with GFP by Mafb-Cre was assessed at E16.5. Gating was performed on a Mafb-Cre-negative embryo, and the percentage of GFP+ lineage labeled cells was determined.
For FACS to quantify uptake of fluorescently labeled nutrients, samples were normalized to a ‘no gavage’ control. The percentage of Alexa Fluor 488- and RRX-positive cells in neonatal mice gavaged with Lys-RRX and Dex-488 was determined. For each sample, 10,000 cells were counted (n=3 control mice and n=5 Maf DKO mice).
RNA-Seq
Epithelial cells were isolated from proximal, medial and distal P7 intestine of three control and three Maf DKO mice as described above and pooled into a single tube per mouse. RNA was isolated using the QIAGEN RNeasy kit (74104). The RNA concentration for each sample was determined using a NanoDrop spectrophotometer, then samples were aliquoted and flash-frozen. Samples were sent to Novogene for sequencing and analysis. Genes with changes greater than 0.2 log2 fold change or −0.25 log fold change and had a P-value of less than 0.05 were considered differentially regulated.
CUT&RUN
Cells were isolated from the distal intestine of a P7 CD1 mouse. Intestinal epithelial cells were isolated and further processed into single cells as described above. The CUT&RUN Assay Kit from Cell Signaling Technologies was used to perform the CUT&RUN. The MinElute PRC Purification Kit (QIAGEN, 28004) was used to isolate the DNA. Libraries were prepared as previously described (Meers et al., 2019; Skene and Henikoff, 2017). The concentration of the libraries was determined using a Qubit 4 fluorometer (Invitrogen) and equal amounts of sample were pooled and sent to Novogene for sequencing. In brief, the results were analyzed by standard pipeline: trimgalore→bowtie2 alignment→calling peaks by MACS2.
Statistical analysis
Figure legends contain details about statistics for each figure. Generally, unpaired t-tests were performed using PRISM software and P-values were reported as follows: *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.
Acknowledgements
We thank members of the Lechler Lab for comments on the manuscript, and Carmen Birchmeier (Max Delbruek-Centrum, Berlin, Germany), Lisa Goodrich (Harvard University, Cambridge, MA, USA), and Michel Bagnat (Duke University, Durham, NC, USA) for reagents. In addition, we thank Bin Li from the Duke Flow Cytometry Shared Resource for assistance with cell sorting, Yasheng Gao from the Duke Light Microscopy Core facility for imaging assistance, and Jianhong Ou for CUT&RUN analysis.
Footnotes
Author contributions
Conceptualization: T.L.; Methodology: A.M.B., L.C., C.M., R.S.M., K.S., T.S., T.L.; Formal analysis: A.M.B., L.C., C.M., J.U., R.S.M.; Investigation: A.M.B.; Resources: Y.D., M.V., T.L.; Data curation: A.M.B., L.C., C.M., J.U., R.S.M., K.S.; Writing - original draft: A.M.B.; Writing - review & editing: A.M.B., J.U., R.S.M., K.S., Y.D., M.V., T.L.; Visualization: A.M.B.; Supervision: M.V., T.L.; Project administration: T.L.; Funding acquisition: Y.D., M.V., T.L.
Funding
This work was supported by the National Institutes of Health (R01DK121915 and R01DK126446 to M.Z.; R01DK11798 and R01AR067203 to T.L.; and R35HG011328 and U01HL156064 to Y.D.). Deposited in PMC for release after 12 months.
Peer review history
The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201251.reviewer-comments.pdf
References
Competing interests
The authors declare no competing or financial interests.