Intestinal lacteals are essential lymphatic channels for absorption and transport of dietary lipids and drive the pathogenesis of debilitating metabolic diseases. However, organ-specific mechanisms linking lymphatic dysfunction to disease etiology remain largely unknown. In this study, we uncover an intestinal lymphatic program that is linked to the left-right (LR) asymmetric transcription factor Pitx2. We show that deletion of the asymmetric Pitx2 enhancer ASE alters normal lacteal development through the lacteal-associated contractile smooth muscle lineage. ASE deletion leads to abnormal muscle morphogenesis induced by oxidative stress, resulting in impaired lacteal extension and defective lymphatic system-dependent lipid transport. Surprisingly, activation of lymphatic system-independent trafficking directs dietary lipids from the gut directly to the liver, causing diet-induced fatty liver disease. Our study reveals the molecular mechanism linking gut lymphatic function to the earliest symmetry-breaking Pitx2 and highlights the important relationship between intestinal lymphangiogenesis and the gut-liver axis.
Intestinal lacteals are essential lymphatic channels for absorption and transport of dietary lipids and drive the pathogenesis of debilitating metabolic diseases. However, organ-specific mechanisms linking lymphatic dysfunction to disease etiology remain largely unknown. In this study, we uncover an intestinal lymphatic program that is linked to the left-right (LR) asymmetric transcription factor Pitx2. We show that deletion of the asymmetric Pitx2 enhancer ASE alters normal lacteal development through the lacteal-associated contractile smooth muscle lineage. ASE deletion leads to abnormal muscle morphogenesis induced by oxidative stress, resulting in impaired lacteal extension and defective lymphatic system-dependent lipid transport. Surprisingly, activation of lymphatic system-independent trafficking directs dietary lipids from the gut directly to the liver, causing diet-induced fatty liver disease. Our study reveals the molecular mechanism linking gut lymphatic function to the earliest symmetry-breaking Pitx2 and highlights the important relationship between intestinal lymphangiogenesis and the gut-liver axis.
Lymphatic vessels are a crucial component of the cardiovascular system (Alexander et al., 2010a, 2010b; Alitalo, 2011; Alitalo et al., 2005; Bernier-Latmani and Petrova, 2017; Betterman and Harvey, 2016; Cifarelli and Eichmann, 2019; Cueni and Detmar, 2008; Zhang et al., 2018). In the gut, they are essential for absorption and transport of fats and fat-soluble nutrients, a function that distinguishes them from all other lymphatics. However, the developmental programs governing their specialized function remain unknown. Gut lymphangiogenesis depends on prior establishment of arteries within the left side of the dorsal mesentery, the major conduit for blood and lymphatic vessels in the gut (Mahadevan et al., 2014). This process is regulated by the home-odomain transcription factor Pitx2, the major player during the evolutionarily conserved left-right (LR) organ asymmetry (Campione et al., 1999; Logan et al., 1998; Ryan et al., 1998; Yoshioka et al., 1998). Pitx2−/− mouse embryos have gut arterial patterning defects and fail to initiate gut but not dermal lymphangiogenesis, suggesting an organ-specific lymphatic role for Pitx2 (Mahadevan et al., 2014).The Pitx2 gene includes three distinct isoforms: Pitx2a, Pitx2b, and Pitx2c (Figures 1A-1C). The Pitx2ab splice variants share the same promoter and are expressed bilaterally, whereas Pitx2c is transcribed from a separate promoter and is expressed asymmetrically (Liu et al., 2001; Schweickert et al., 2000; Shiratori et al., 2001, 2006). Left-specific Pitx2c expression is conferred by the conserved enhancer ASE (Shiratori et al., 2001, 2006). Pitx2 mice fail to manifest most of the left-sided Pitx2c expression and exhibit laterality defects similar to those observed in mice lacking Pitx2c (Liu et al., 2002). Unlike Pitx2−/− embryos lacking all isoforms (Pitx2HD/HD) (Lu et al., 1999), most Pitx2 embryos survive to birth, and a few can reach weaning, allowing analyses of gut lymphatic function in the absence of ASE (Figures 1B and 1C).
Figure 1.
Pitx2 mice fail to properly gain weight
(A) Pitx2 gene structure. E, exon; ATG, start codon; black arrow, transcriptional start site; ASE, asymmetric enhancer.
(B) Pitx2-deficient mouse models. Pitx2 has homeobox deletion from exon 5 to exon 6. Pitx2Δ has Pitx2c-specific exon 4 deletion. Pitx2 contains ASE deletion in intron 5.
(C) Survival rates of Pitx2 mice. Data displayed on the y axis are represented as incidence of live Pitx2 mice on collection (n = 28 at E10.5, 125 at P0, 500 at P1.5, 533 at P9, and 169 at P21 [weaning age]). Orange marks, survival rate of Pitx2 mice based on the number of Pitx2 live mice/expected number of mice per Mendelian ratio (25%); black marks, live birth rate of Pitx2 mice based on the number of Pitx2 live mice/number of total live mice.
(D) Pitx2 mice show abnormal weight gain. Left: body weights measured at E18.5 and birth are counted as P0. Center: a pair of littermates showing significant growth retardation of Pitx2 versus the WT at P9. Right: stomach at collection of P1.5 WT fed, WT unfed, and Pitx2 fed mates. Data displayed on the y axis in the left graph are represented as mean ± SEM. P0 WT = 1.388 ± 0.03382 g, n = 21; P0 Pitx2 = 1.295 ± 0.03531 g, n = 22. Difference between WT and Pitx2 = 0.09264 ± 0.04897 g, p = 0.0656. P1.5 WT = 1.791 ± 0.2251 g, n = 41; P1.5 Pitx2 = 1.495 ± 0.1435 g, n = 10. Difference between WT and Pitx2 = −0.2960 ± 0.07494 g, p = 0.0003***. P9WT = 5.963 ± 0.8161 g, n = 16; P9 Pitx2 = 4.25 ± 1.211 g, n = 7. Difference between WT and Pitx2 = −1.713 ± 0.4286 g, ***p = 0.0007. Scale bars, 500 μm.
See also Figure S1.
We first focused on lacteals, the absorptive lymphatic capillaries responsible for dietary fat uptake (Cifarelli and Eichmann, 2019; Miller et al., 2010). Fatty nutrients use lacteals as their major transport route, where chylomicron-packed lipids drain into larger mesenteric collectors and, ultimately, to the systemic blood circulation (Bernier-Latmani and Petrova, 2017; Dixon, 2010; Ko et al., 2020; Wang et al., 2013). In contrast, non-lipid nutrients enter the portal (venous) circulation for delivery to the liver (Bernier-Latmani and Petrova, 2017; Dixon, 2010; Ko et al., 2020; Wang et al., 2013). Lacteals form around embryonic day 17.5 (E17.5) and are functionally ready to absorb lipids from milk at birth (Kim et al., 2007). Signaling via lymphangiogenic vascular endothelial growth factor C (VEGF-C), its receptor VEGFR-3, and its co-receptor neuropilin 2 (NRP2) plays a key role in driving postnatal lacteal sprouting and growth (Tammela et al., 2008; Xu et al., 2010). Although most adult lymphatics are quiescent, lacteals undergo filopodium-mediated regeneration (Bernier-Latmani et al., 2015). Although most lymphatic capillaries lack smooth muscle coverage, lacteals are surrounded by contractile axial smooth muscle cells (SMCs) (Hosoyamada and Sakai, 2007), which secrete VEGF-C (Nurmi et al., 2015) and facilitate lipid transport by squeezing the lacteal (Choe et al., 2015).Here we show that Pitx2 is required for development and function of lacteals through a non-cell-autonomous pathway that involves lacteal-associated axial SMCs. ASE deletion results in abnormal axial SMC morphogenesis caused by oxidative stress, leading to impaired lacteal extension and defective lymphatic system-dependent lipid transport. Strikingly, ASE deletion leads to excessive portal entry of dietary lipids and severe hepatic lipidosis in pups. Our study reveals a molecular mechanism linking gut lymphatic function to Pitx2 and highlights the important relationship between the intestinal lymphatic vasculature and gut-liver axis. We show that Pitx2 continues to orchestrate postnatal intestinal function beyond its well-established roles in early LR asymmetry.
RESULTS
Loss of the LR Pitx2 asymmetric enhancer (ASE) results in growth retardation and early postnatal lethality
Although most Pitx2 mice die within hours after birth (C57BL/6 × 129 mixed background), some ASE newborns on the FVB background remain viable for ~10 days (21.6%; Figure 1C; Figure S1) or longer (4.7% at weaning; Figure 1C), allowing analyses of gut lymphatic development and function in the absence of ASE. To evaluate intestinal lymphatic function in dietary lipid transport, we first compared post-feeding weight gain between wild-type (WT), Pitx2, and Pitx2 neonates. Despite their comparable body weights at birth, Pitx2 had severely retarded growth at 24–48 h and at post-natal day 9 (P9) (Figure 1D; Figure S1). This was independent of variable food ingestion; ample milk was seen in all neonatal stomachs analyzed, regardless of genotype (Figure 1D). We hence hypothesize that the failure of proper weight gain seen in Pitx2 mice was due to malabsorption caused by compromised lymphatic transport function.
ASE is required for lacteal extension and formation of lacteal filopodia
Intestinal lipid transport is driven by lacteals (Figure S2A); hence, we characterized lacteal development in Pitx2 mice, in which they are detectable as early as E17.5, using the lymphatic endothelial cell marker Lyve-1 (Kim et al., 2007; Figure S2A). We compared embryonic and postnatal lacteal development between WT and Pitx2 mice using IF on gut slices (Figure S2B). First we observed a significant reduction in the number of Lyve-1+ lacteals in Pitx2 embryos at E18.5 (Figure S2C) and again at P9 (Figure S2C; Figures 3B and 3D, asterisks) despite unaffected villus length (Figure S2C). Lymphatic endothelial cells (LECs) at the tip of the lacteals extend long filopodia in WT neonatal intestines, consistent with active sprouting and migration (P1.5; Figure 2A, arrows; Adams and Alitalo, 2007). In the WT, ~73% of lacteals harbor at least one filopodium at P1.5 (Figure 2A; 72.90% ± 2.760%, n = 16), in agreement with prior studies in adulthood (Bernier-Latmani et al., 2015). In contrast, most Pitx2 lacteals are missing filopodia at P1.5 (Figure 2A, arrowheads; 41.28% ± 3.545%, n = 10, p = 0.0001). Similar results were obtained following isoform-specific deletion of Pitx2c (Figure 2B; WT: 81.38% ± 4.660%, n = 3; Pitx2c−/−, 41.13% ± 5.604%, n = 4; p = 0.0034). Filopodium formation remained suppressed during later stages of development (P9), a phenotype that was exacerbated over time (Figure 2C; WT, 49.66% ± 4.200%, n = 8; Pitx2, 8.543% ± 2.610%, n = 3; p = 0.0003).
Figure 3.
Malformation of the muscular-lacteal complex in Pitx2 mice
(A) Lineage tracing with Pitx2 - ROSA26 and ASE - ROSA26 mice; whole-mount villus from Pitx2;ROSA26 mice at P9 (A’ and A”) and 6 months (A”’). Note the colocalization of αSMA (green) and tdTomato (tdT, red) that marks axial SMCs of Pitx2 descent (white arrowhead) and axial SMC heterogeneity (A”). White arrows indicate additional cell types marked by Pitx2 daughters (A’). (A””) Whole mount villus from ASE - ROSA26 mice at P9. Note the colocalization of αSMA (red) and YFP (green) that marks axial SMCs of ASE descent (white arrowheads). White arrows indicate additional cell types marked by ASE daughters, distinct from BECs (red). Scale bar, 20 μm.
(B) Whole-mount muscular-lacteal complexes of WT and ASE at P9. Green is lacteal (Lyve-1); red is αSMA (smooth muscle). From left to right in the Pitx2 panel: aberrant muscular arrangement with no lacteal (white asterisk), loss of contact between axial SMCs and lacteal, and open axial SMCs with shorter and dilated lacteal. Scale bar, 40 μm.
(C) Percent lacteals surrounded by “open” axial SMCs based on data represented in (B). We defined the muscles as “open” when they were divergently aligned around the lacteal tip. Data are represented as mean ± SEM. P9 WT = 33.51% ± 3.950%, n = 6; P9 Pitx2 = 76.12% ± 7.775%, n = 3. Difference between WT and Pitx2 = 42.61% ± 7.703%, p = 0.0009***. White characters mark the mean in each group.
(D) Muscular-lacteal complexes at P9 in transverse sections. Green is lacteal (Lyve-1); red is αSMA (smooth muscle); blue is DAPI. Note the close (white arrows) versus distanced (white arrowheads) proximity of red and green signals. White stars mark villi with missing lacteals. Scale bar, 20 μm.
(E) Cartoon demonstrating muscular-lacteal arrangement of WT and Pitx2 mice. Green is lacteal; red is muscle. See also Figure S3.
Figure 2.
Shortened lacteals with missing filopodia in Pitx2 mice
(A) Whole-mount lacteals (Lyve-1, green) of WT and Pitx2 villi at P1.5. Top right: percent lacteals with filopodia. Bottom right: average lacteal length. Data are represented as mean ± SEM. P1.5 WT lacteals with filopodia = 72.90% ± 2.760%, n = 16; P1.5 Pitx2 lacteals with filopodia = 41.28% ± 3.545%, n = 10. Difference between WT and Pitx2 = −31.62% ± 4.476%. ****p < 0.0001. P1.5 WT lacteal length = 176.1 ± 6.401 μm, n = 17; P1.5 Pitx2 lacteal length = 143.7 ± 6.392 μm, n = 10. Difference between WT and Pitx2 = −32.41 ± 9.704 μm. **p = 0.0026. White characters mark the mean in each group. Scale bars, 30 μm.
(B) Whole-mount lacteals (Lyve-1, green) of WT and Pitx2 villi at P1.5. Right: percent lacteals with filopodia. Data are represented as mean ± SEM. WT = 81.38% ± 4.660%, n = 3; Pitx2 = 41.13% ± 5.604%, n = 4; difference between WT and Pitx2 = −40.25% ± 7.692%; **p = 0.0034. White characters mark the mean in each group. Scale bar, 20 μm.
(C) Whole mount lacteals (Lyve-1, green) of WT and Pitx2 villi at P9. Left: percent lacteals with filopodia. Right: average lacteal length. Note the extended filopodium in WT lacteal tips (white arrows) versus the blunt-ended lacteals (white arrowhead) in ASE. Data are represented as mean ± SEM. P9 WT lacteals with filopodia = 49.66% ± 4.200%, n = 8; P9 Pitx2 lacteals with filopodia = 8.543% ± 2.610%, n = 3. Difference between WT and Pitx2 = −41.12% ± 7.237%. ***p = 0.0003. P9 WT lacteal length = 222.7 ± 8.252 μm, n = 8; P9 Pitx2 lacteal length = 144.6 ± 9.518 μm, n = 3. Difference between WT and Pitx2 = −78.12 ± 14.90 μm. ***p = 0.0005. Black characters mark the mean in each group. Scale bars, 30 μm.
See also Figure S2.
Consistent with filopodium defects, Pitx2 lacteals were significantly shorter than those of WT littermates at P1.5 (Figure 2A; WT: 176.1 ± 6.401 μm, n = 17; Pitx2: 143.7 ± 6.392 μm, n = 10; p = 0.0026) and again at P9 (Figure 2C; WT: 222.7 ± 8.252 μm, n = 8; Pitx2: 144.6 ± 9.518 μm, n = 3; p = 0.0005). Although WT lacteals elongated by ~50 μm from P1.5 to P9 (Figures 2A and 2C; P1.5: 176.1 ± 6.401 μm, n = 17; P9: 222.7 ± 8.252 μm, n = 8), such lacteal extension was halted in Pitx2 mice despite normal villus length (Figures 2A and 2C; P1.5: 143.7 ± 6.392 μm, n = 10; P9: 144.6 ± 9.518 μm, n = 3). ASE is required for proper lacteal extension and lacteal filopodium formation.
To determine whether Pitx2 acts autonomously in LECs or in a non-cell-autonomous manner, we performed Pitx2 (knockin) lineage tracing to fate map Pitx2 descendants (Figures S3A-S3A”). Pitx2 is a null Pitx2 allele that drives Cre activity in cells that have expressed any Pitx2 isoform throughout embryonic development (Liu et al., 2002; Figure S3A). To follow Pitx2 daughter cells, we crossed Pitx2 heterozygotes with Rosa26 (R26R-tdTomato) reporter mice (Figure S3A).TdTomato-marked Pitx2 daughters were detected in the intestine, consistent with the described role of Pitx2 during midgut development (Figure S3A’). Pitx2 daughters did not populate CD31-positive blood endothelial cells (BECs) or Lyve1+ LECs at any time point analyzed (E14.5, E16.5, E18.5, P1.5, P9, and 6 months; n = 19) (Figure S3A”). Instead, we found Pitx2 daughter cells among the α-smooth muscle actin (αSMA)-positive SMCs of the intestine, including villus-axial SMCs that are closely associated with lacteals (Bernier-Latmani and Petrova, 2016; Hosoyamada and Sakai, 2007; Figures 3A′-3A‴, white arrowheads, and 3B, WT). We confirmed this observation using ASE-specific lineage tracing with Rosa26 (R26R-EYFP) reporter mice, where the activity of Cre is driven by an ~18-kb genomic fragment containing ASE and the Pitx2c promoter (Shiratori et al., 2014; Figure 3A″″, white arrowheads).In addition to axial SMCs, we found a discrete fraction of TdTomato-marked Pitx2 (Figure 3A′, white arrows) and YFP-marked ASE (Figure 3A″″, white arrows) daughter cells morphologically different from axial SMCs and near, but distinct from, the CD31+ blood capillary plexus of the villous lamina propria (Figure 3A″, white arrows). Based on their location and morphology, these cells may lie upstream in the lineage compared with the more differentiated axial SMCs and may be involved in muscular regeneration or expansion at later stages (Figure 3A‴, white arrows, 6 months). Pitx2 regulates lacteal development through a non-cell-autonomous pathway and may contribute to SMC patterning of the intestinal lamina propria.
Loss of ASE impairs lacteal-associated axial smooth muscle development
Contraction of axial SMCs drives lipid transport in adjacent lacteals (Choe et al., 2015), whereas a subset of axial SMCs secretes VEGF-C (Nurmi et al., 2015). Loss of VEGF-C-driven VEGFR-3 activation leads to arrested lacteal growth (Nurmi et al., 2015). We hence hypothesized that Pitx2 is indirectly required for lacteal morphogenesis via SMC patterning. Under histological examination of P26 weanlings, robust villus axial SMCs were seen in the WT, but limited SMCs were observed in Pitx2 mice (Figure S3B, red arrowheads). To characterize the muscular-lacteal complex at high resolution, we examined WT and Pitx2 gut slices using confocal microscopy with αSMA (smooth muscle) and Lyve-1 (lacteal) IF. In WT pups at P9, αSMA+ cells were closely intertwined with Lyve-1+ lacteals. This striking pattern of “closed” muscle bundle morphology suggests active muscular contractions squeezing the lacteal (Figures 3B and 3E). In contrast, age-matched Pitx2 mice displayed a wide spectrum of muscle-lacteal abnormalities, including misaligned axial SMCs with absent lacteals, reduced axial SMC numbers, loss of direct SMC-LEC contact, and an “open” bundle conformation leaving spaces between individual muscle bundles and lacteals (Figures 3B and 3E). We quantified a significant increase in open axial smooth muscle bundles within Pitx2 villi compared with control mice (Figures 3B and 3C; WT: 33.51% ± 3.950%, n = 6; Pitx2: 76.12% ± 7.775%, n = 3; p = 0.0009). Using villus sections, we confirmed that axial SMCs and lacteals fail to align in Pitx2 mice (Figures 3D and 3E). This supports the hypothesis that ASE regulates villus axial SMC morphogenesis.
Lacteal-associated axial SMCs arise by ASE-dependent remodeling
Villus-axial SMCs were first identified in 1909 (Trautman, 1909) and were considered vital to provide structural tension against interstitial pressure while driving villous rhythmical contraction in close relationship with the adjacent fibroblasts (Güldner et al., 1972; Hosoyamada and Sakai, 2007). More recent in vivo imaging revealed that villus SMCs promote lateral villus contractions dependent on the autonomic nervous system and facilitate lipid transport by squeezing the lacteal (Choe et al., 2015). Despite this crucial interplay between axial SMCs and lacteal function, the origins and patterning mechanisms of axial SMCs remain unexplored.We investigated muscle formation alongside lacteals in WT mice from E16.5–P9 (Figure 4A; Figure S4A). Cells with low levels of αSMA (αSMA+) were first detected in the villus at E16.5, before emergence of lacteals (Figure S4A). αSMA+ cells within the villus appeared star-shaped with multiple protrusions (Figure 4A, inset, “star cells”) and aligned closely with the CD31+ blood vascular plexus (Figure S4A). Interestingly, αSMA+ star cells were reminiscent of TdTomato-marked Pitx2 and YFP-marked ASE daughters (Figures 3A’ and 3A″″, white arrows). Coincident with the appearance of lacteals at E18.5, αSMA+ cells underwent a robust transformation to form elongated, spindle-like cell types with strong immunoreactivity for αSMA (αSMA++) that closely associate with lacteals (Figure 4A, graph). At P1.5, the number of spindle-like αSMA++ cells became predominant at the expense of αSMA+ star cells (Figure 4A, graph). At P9, all αSMA++ spindle cells were located adjacent to the lacteal, extending radially from the submucosa to the filopodium at the tip of the lacteal, whereas very few αSMA+ star cells were detected at this time (Figure 4A; Figure S4D).
Figure 4.
Lacteal-associated axial SMCs arise by ASE-dependent remodeling
(A) Whole-mount WT villous muscular-lacteal development from E16.5–P9. Green is lacteal (Lyve-1); red is αSMA (smooth muscle). Insets: note the shape difference of cells at E16.5 (star) versus at P9 (spindle). Scale bar, 40 μm. Graph: percent villi with αSMA+ star cells (black squares) or αSMA++ spindle cells (gray dots). Note the negative correlation between αSMA+ star cells (dominant in prenatal stages) and αSMA++ spindle cells (dominant postnatally). Data displayed on the y axis are represented as mean ± SEM. E16.5 villi with star cells = 90.7% ± 4.6%, n = 6; E16.5 villi with bundles = 0% ± 0%, n = 6; E18.5 villi with star cells = 48.02% ± 5.9%, n = 6; E18.5 villi with muscle bundles = 62.98% ± 7.5%, n = 6; P1.5 villi with star cells = 12.25% ± 4.9%, n = 6; P1.5 villi with muscle bundles = 83.95% ± 4.0%, n = 6; P9 villi with star cells = 9.454% ± 4.5%, n = 8; P9 villi with muscle bundles = 93.51% ± 2.5%, n = 8.
(B) Star but not spindle cells are proliferative (E18.5). Whole-mount WT villus; green is αSMA (green), red is phospho-histone 3 (pH3), and blue is DAPI. Note that the pH3 signal colocalized only with star- but not spindle-like SMCs. Scale bar, 10 μm.
(C) Whole-mount muscular-lacteal complexes from WT and ASE siblings at P9. Green is lacteal (Lyve-1); red is αSMA (smooth muscle). Note the retention of SMA+ star cells in Pitx2. Scale bar, 20 μm.
(D) Percent villi with star or spindle cells in WT and ASE siblings. Left: percent villi with spindle cells in E18.5 embryos. Right: percent villi with star cells in P1.5 pups. Data displayed on y axes are represented as mean ± SEM. E18.5 WT villi with spindles = 62.98% ± 7.52%, n = 6; E18.5 Pitx2 villi with spindles = 36.93% ± 4.41%, n = 6. Difference between WT and Pitx2 = −26.05% ± 8.72%, *p = 0.0136. P1.5 WT villi with stars = 12.25% ± 4.917%, n = 6; P1.5 Pitx2 villi with stars = 42.42% ± 3.510%, n = 5. Difference between WT and Pitx2 = 30.18% ± 6.292%, ***p = 0.0010.
(E) Cartoon depicting star-to-spindle smooth muscle transition to form the axial SMC-lacteal complex.
See also Figure S4.
αSMA+ star cells were proliferative (Figure S4B; Figure 4B ~10%) compared with αSMA++ spindle-like axial SMCs (Figure S4B; Figure 4B). This agrees with previous observations that only round muscular progenitors (but not mature spindle muscles) are proliferative (Owens, 1995; Owens et al., 2004). αSMA++ spindle cells expressed high levels of myosin heavy chain 11 (Myh11), resembling more differentiated muscle cells (Gomez et al., 2015; P9, Figure S4D). In contrast, αSMA+ star cells showed very weak Myh11 immunoreactivity (P9; Figure S4D, white arrows). These observations suggest that villus-axial smooth muscle development is commensurate with formation of lacteals and proceeds by a robust spatiotemporal remodeling program where αSMA+ precursor star cells proliferate and assemble to create and shape the mature axial SMCs network of the lacteal (Figure 4E). Axial SMCs were largely juxta-vascular, and this was especially evident for SMCs at the lacteal filopodia (Figure S4C), highlighting an important reciprocal interaction with the adjacent lymphatic endothelium.Next we examined villus-axial SMCs dynamics in the absence of ASE. As in WT villi, we first detected αSMA+ star cells in the Pitx2 at E16.5, with no differences in the proportion of villi containing αSMA+ star cells among all genotypes tested (data not shown). However, 2 days later, at E18.5, we found fewer αSMA++ spindle-like axial SMCs in Pitx2 mice compared with WT littermates (Figure 4D; WT: 62.98% ± 7.52%, n = 6; Pitx2: 36.93% ± 4.41%, n = 6; p = 0.0136). In contrast to WT neonates, a significant number of αSMA+ star cells remained in the lamina propria of Pitx2 at P1.5 (Figures 4C and 4D; WT: 12.25% ± 4.917%, n = 6; Pitx2: 42.42% ± 3.510%, n = 5; p = 0.0010). These star cell-retained Pitx2 villi (Figure 4C, white arrows) had less prominent muscular bundles surrounding the lacteals (Figure 4C, white arrowheads).TUNEL staining of WT and Pitx2 at P9 revealed no difference in cell death of villous SMCs (Figure S4E). Under histopathological analyses, apoptotic or necrotic cells were not seen in E18.5, P1.5, and P9 Pitx2 villi, whereas villous SMC proliferation was similarly unaffected in Pitx2 mice at E18.5 (data not shown). These observations suggest that neither cell death nor cell cycle arrest at the stages examined contribute to the ASE-dependent axial SMC phenotypes. Instead, we propose that ASE functions in villous-axial SMC morphogenesis during their fetal to neonatal transition (Figure 4E).
ASE deletion leads to redox imbalance in intestinal smooth muscle
To explore the cause of muscular-lacteal impairment in Pitx2 intestines, whole-intestine transcriptomics (RNA sequencing [RNA-seq]) of WT and Pitx2 tissue at P1.5 was performed. Notably, ~20% of genes differentially expressed between WT and Pitx2 intestines corresponded to reactive oxygen species (ROS) production, antioxidant scavenging genes, and redox-sensitive signaling pathways (Figure 5A). Unsupervised gene set enrichment analysis (GSEA) demonstrated an enriched ROS pathway in Pitx2 intestine (normalized enrichment score = 1.43, nominal p = 0.03, false discovery rate [FDR] q = 0.03) (Subramanian et al., 2005; Figure S5A). Among potential targets were genes that protect the cell from oxidative stress, such as the glutamate-cysteine ligase (Gclc), the first rate-limiting enzyme of glutathione synthesis (Fedorova et al., 2014; Lu, 2013; Wang et al., 2014). Interestingly, half of the ROS-related differentially expressed genes are bound by Pitx2 in vivo (Wang et al., 2014), including the noncoding region of Gclc (Figure S5B). These data raised the possibility that Pitx2c may directly regulate the redox balance in intestinal smooth muscle.
Figure 5.
ASE deletion leads to redox imbalance and actin carbonylation in intestinal smooth muscle
(A) RNA-seq. Left: heatmap of differentially expressed genes at P1.5. The relative minimum expression is depicted in white and maximum in red. Red dots show genes bound by Pitx2 based on prior chromatin immunoprecipitation sequencing (ChIP-seq); green dots show genes with known ROS function. Right: Venn diagrams of differentially expressed genes bound by Pitx2 and/or with reported function in redox balance.
(B) 2,4-dinitrophenylhydrazine (DNPH, or HZ) reacts with protein carbonyl groups, forming stable 2,4-dinitrophenylhydrazone (hydrazone), quantified spectrophotometrically (B’). Protein carbonyl concentration measured from WT (blue), Pitx2 (orange), and Pitx2 (black) intestines. Data are represented as mean ± SEM. Protein carbonyl in the WT/Pitx2 group = 2.76 ± 0.3194 nmol/mg, n = 7; protein carbonyl in Pitx2 = 6.833 ± 0.4567 nmol/mg, n = 3. Difference between WT/Het and Pitx2 = 4.073 ± 0.5741 nmol/mg, ***p = 0.0001.
(C–E) P1.5 gut protein lysates labeled with biotin-HZ, separated by SDS-PAGE, then transferred to a nitrocellulose membrane.
(C) Total protein carbonyl detected by streptavidin-HRP (SA-HRP) (top blot) and total αSMA expression by αSMA antibody on the same blot (bottom blot).
(D) Left: αSMA detected by an αSMA antibody. Right: total protein carbonyl detected by SA-HRP on the same blot.
(E) P1.5 gut protein extracts were labeled with biotin-HZ following αSMA immunoprecipitation (SMA-IP), separated by SDS-PAGE, then transferred to a nitrocellulose membrane. Total protein carbonyl was detected by SA-HRP (top blot). SMA-IP efficiency was confirmed by total αSMA expression, detected by αSMA antibody on the same blot.
See also Figure S5.
To investigate whether ASE deletion leads to a redox imbalance in the intestine, we spectrophotometrically detected and quantified protein carbonylation in WT, Pitx2, and Pitx2 neonatal intestines at P1.5. Carbonyl groups (Figure 5B) are irreversible products of ROS-mediated protein oxidation and a general biomarker for oxidative injury (Chevion et al., 2000; Dalle-Donne et al., 2003b; Fedorova et al., 2014). We observed that Pitx2 intestines harbor 2.5 times more protein carbonyl content compared with their WT or Pitx2 littermates (Figure 5B’; WT/Het: 2.760 ± 0.3194 nmol/mg, n = 7; Pitx2: 6.833 ± 0.4567 nmol/mg, n = 3; p = 0.0001), demonstrating that ASE loss causes oxidative damage. We also assessed protein carbonyl by electrophoresis, after labeling carbonyl groups in protein extracts from WT, Pitx2, and Pitx2 intestines at P1.5 with biotin-hydrazine (biotin-HZ). Proteins were SDS-PAGE-separated and probed with streptavidin-conjugated horseradish peroxidase (SA-HRP). Consistent with the spectroscopic measurements, a higher total level of carbonylated proteins (overall darker lane signal) was observed in the Pitx2 intestine lysate relative to controls (Figure 5C).Interestingly, a well-defined carbonylated protein band of ~42 kDa was enriched in Pitx2 intestinal extracts versus controls (Figure 5C). This band is consistent in size with monomeric actin. Probing the same lysates for αSMA showed that actin levels were similar among all genotypes (Figure 5C), suggesting the possibility that levels of oxidized actin were enriched in Pitx2 intestines.The dynamic actin cytoskeleton is susceptible to oxidation, which regulates cell behavior and contractility pathways (Xu et al., 2017). Oxidized actin is linked to decreased actin monomer (G-actin) assembly and actin filament (F-actin) stability (Wilson and González-Billault, 2015). Oxidation of select cysteine and methionine residues in actin causes structural changes, monomer aggregation, decreased actin polymerization and stability, and defective cell contractility (Dalle-Donne et al., 2002, 2003a; Hung et al., 2011, 2013). Actin carbonylation is associated with even higher oxidant levels than those associated with modification of critical cysteine and methionine residues, and an accumulation of carbonylated actin has been proposed to indicate severe oxidative stress and functional impairment (Dalle-Donne et al., 2001). Interestingly, we noted discontinuous patterns of αSMA arrangement in the axial SMCs of the more severely affected Pitx2 mice, reminiscent of the severed F-actin filaments seen under oxidative stress (Figure S5C). These data led us to hypothesize that ASE deletion impaired villus-axial smooth muscle morphogenesis because of abnormal accumulation of ROS in the differentiating smooth muscle.To test this, we directly assessed the oxidation status of SMA in WT and Pitx2 neonatal intestines (Figures 5D and 5E). Streptavidin enrichment of biotin-labeled carbonylated proteins confirmed that actin was carbonylated and that Pitx2 yielded the highest amount of carbonylated αSMA (Figure 5D). Immunoisolation of αSMA from biotin-HZ-labeled lysates similarly confirmed an increased level of carbonylated αSMA in Pitx2 intestine in comparison with WT or Pitx2 littermates (Figure 5E). These results demonstrate that loss of ASE induced irreversible oxidative damage in αSMA+ cells of the neonatal intestine. Our data suggest that elevated levels of oxidative stress caused muscular-lacteal impairment in Pitx2 intestines, pointing to a crucial role of Pitx2 in the control of ROS during early smooth muscle development.
ASE is required for the normal route of dietary lipid transport
The lymphatic vasculature is the major route through which long-chain fatty acids are absorbed into the body (Zhou et al., 2020). In contrast, non-fatty nutrients and shorter-chain fatty acids are able to directly enter the hepatic portal circulation for processing within the liver (Zhou et al., 2020). This lymphatic system-independent portal transport is insignificant compared with the lymphatic route, except under pathological conditions leading to hepatic lipid accumulation (Anwar et al., 2007; Cabré et al., 2005; Courtney and Warner, 2017; Ee et al., 2000; Fujimoto et al., 1991; Hussain, 2014; Kurtel et al., 1991; McDonald et al., 1980; Onufer et al., 2019; Young et al., 1995). To assess whether ASE deletion impaired dietary lipid transport, we orally administered fluorescent BODIPY C16, a fluorescent long-chain fatty acid tracer, to WT and Pitx2 neonates to follow lipid trafficking (Figure 6A; Dixon et al., 2009). As expected, BODIPY C16 entered the lumen of the intestine and mesenteric collecting vessels after oral administration (Figure 6A), with very little staining of the hepatic portal circulation in WT pups (Figure 6B). Strikingly, long-chain BODIPY C16 accumulation was found in the hepatic portal vein and liver parenchyma of Pitx2 pups (Figure 6B; liver, liver section, and hepatic portal vein), confirming abnormal lipid transport in the absence of ASE. Notably, BODIPY C16 remained detectable in the mesenteric collectors of Pitx2 pups, implying a partial rather than complete shift in lipid transport in the absence of ASE (Figure S6). Thus, ASE deletion impairs normal long-chain fatty acid lymphatic trafficking, resulting in direct lipid entry into the hepatic portal circulation (Figure 6C).
Figure 6.
ASE is required for the normal route of dietary lipid transport
(A) Pups were fed with BODIPY FL C16 and then left with the dam for 5 h before analyses. Note entry of the tracer in mesenteric collectors in the WT validating feeding assay. Scale bar, 200 μm.
(B) Hepatic analysis of tracer accumulation in Pitx2 at P1. Left to right: analysis of whole livers, frozen liver sections with DAPI counterstain (blue), and hepatic portal vein (HPV; white arrows) in WT (top) and Pitx2 (bottom) neonates. Note the enriched green signal in HPVs and livers of Pitx2 neonates in comparison with the WT. Scale bars: 1 mm (left, liver, whole organ), 20 μm (center, liver section), and 300 μm (right, HPV).
(C) Lymphatic system-dependent transport (green) of dietary fats (yellow dots) is dominant in WT mice, whereas additional activation of lymphatic system-independent fat transport via the HPV (light blue) is evident in ASE mice. Red marks star and axial SMCs.
Aberrant dietary lipid transport by the hepatic portal vein can accumulate lipid within hepatocytes, resulting in hepatic lipidosis. We compared liver lipid content in prenatal and postnatal WT and Pitx2 mice using oil red O, a fat-soluble dye that stains neutral triglycerides and lipids (Mehlem et al., 2013; Figure 7A). Consistent with our BODIPY C16 data, increased oil red O staining in postnatal but not prenatal (unfed) Pitx2 liver suggested an etiology related to post-partum nursing. Excessive hepatic lipid accumulation was observed in 100% of Pitx2 pups at P1.5 (Figure 7A). This was confirmed by ultrastructural examination (Figure S7A) and quantitative triglyceride analysis (Figure S7C).
Figure 7.
Pitx2 mice develop diet-induced fatty liver
(A) Hematoxylin (blue) and oil red O (red) staining of liver sections from E18.5, P1.5, and P11 WT (top) and Pitx2 mice (bottom). Scale bar, 50 μm.
(B) Histology of WT (top) and Pitx2 (bottom) livers isolated from suckling pups (left) and weanlings (right). Note lipid vacuoles and loss of eosinophilic parenchymal stain in Pitx2 pups, resolved in Pitx2 weanlings. Images shown on the right are high magnifications of the left adjacent image. Scale bar, 50 μm.
(C) Structural muscular-lacteal defects remain in post-weaning Pitx2 cubs. Left: whole-mount muscular-lacteal complexes from WT and ASE cubs at P26. Note shorter lacteals (Lyve-1, green) and a more open muscular arrangement (αSMA, red) in ASE but not WT weanlings. Right: quantification of percent lacteals with filopodia and open muscle bundles. Note that the following mice were pooled: n = 4 WT at P21 with n = 1 Pitx2 at P26; the Pitx2 group consisted of n = 3 at P21 and n = 1 at P26. Data displayed on y axes are represented as mean ± SEM. Lacteals with filopodia in WT weanlings = 28.81% ± 7.915%, n = 5; lacteals with filopodia in Pitx2 weanlings = 5.833% ± 3.967%, n = 4. Difference between WT and Pitx2 = −22.98% ± 4.375%; **p = 0.0012. Lacteals with open muscle bundles in WT weanlings = 13.08% ± 2.637%, n = 5; lacteals with open bundles in Pitx2 weanlings = 73.85% ± 9.599%, n = 4. Difference between WT and Pitx2 = 60.77% ± 8.945%; ***p = 0.0003. Black characters represent the mean of each group. Scale bar, 50 μm.
See also Figure S7.
At P9–P11, oil red O staining was even more robust (Figure 7A). Histopathology analysis at P9 revealed a 92% incident of hepatic steatosis in Pitx2 mice (Figure S7B) with variable parenchymal extinction and multifocal subcapsular mineralization (Figure 7B). Importantly, no significant changes in hepatic structure, mitochondrial morphology, or extramedullary hematopoiesis were observed between Pitx2 and WT littermates (data not shown), ruling out primary liver impairment.Although the lipid content of murine maternal milk is ~17%–30% (Görs et al., 2009), the lipid content in standard mouse chow diet is 6- to 10-fold lower (~3%). Hepatic steatosis in Pitx2 suckling pups gradually resolved when weaned (P21) and transitioned to a dry pellet diet (Figure 7B). Severe hepatic lipidosis in Pitx2 mice was seen in only 50% of pups at P21 and was reduced further to 25% at P26 and 0% after P30 (Figure S7B). Importantly, the intestinal lesions caused by ASE deletion did not resolve with age or weaning/change of diet. These lesions included lacteal filopodium defects (Figure 7C; WT: 28.81% ± 7.915%, n = 5; Pitx2: 5.833% ± 3.967%, n = 4; p = 0.0012), open axial smooth muscle (Figure 7C; WT: 13.08% ± 2.637%, n = 5; Pitx2: 73.85% ± 9.599%, n = 4; p = 0.0003), and severe axial smooth muscle hypoplasia (Figure S3B). This is a strong indication that mature mice remain susceptible to lipid accumulation, depending on their diet. These data suggest that a high-fat diet (in this case maternal milk) drives fatty liver in Pitx2 mice, whereas Pitx2c protects against such pathogenesis in the WT (Figure S9).
DISCUSSION
Pitx2 in intestinal smooth muscle development
LR asymmetry is fundamental to the vertebrate body plan, errors in which cause important human birth defects. The intestine begins as a symmetrical midline tube but later rotates and loops in a conserved pattern to pack into the body cavity. The first critical leftward midgut rotation is driven by molecular and morphologic asymmetries within the left and right sides of the dorsal mesentery. These events are orchestrated by Pitx2 expressed in all cells of the left dorsal mesentery (Davis et al., 2008). Although Pitx2 mice exhibit duodenal (foregut) positioning defects (Shiratori et al., 2006), no midgut rotation, strangulation, or vascular obstruction are apparent in these mice (Shiratori et al., 2006; Figures S6 and S8A). Further analysis of blood vascular organization (Figure S8B), histopathology (Figure S8C), and staining for the hypoxia marker HIF-1α (data not shown) support that the lacteal-muscle defects caused by loss of ASE are not caused indirectly by confounding physiological factors or intestinal topology.The dorsal mesentery is also the major source of gut smooth muscle progenitors (Wilm et al., 2005; Winters et al., 2014). Pitx2 descendants populate multiple intestinal smooth muscle layers, including the lacteal-associated axial SMCs that drive lacteal contraction and postnatal lipid transport (Choe et al., 2015). Interestingly, not all axial SMCs are descendants of Pitx2 (Figures 3A” and A”’), suggesting heterogeneity among axial SMCs and hinting at an additional progenitor source. Although Pitx2 is the primary determinant in the left dorsal mesentery, the transcription factor Tbx18 is enriched exclusively on the right side (Davis et al., 2008; Welsh et al., 2013). In the kidney, Tbx18-derived stromal progenitors give rise to vascular SMCs, where stromal cell differentiation requires Tbx18 to form the renal vasculature (Airik et al., 2006; Xu et al., 2014a). The right-sided origin Tbx18 lineage may provide an additional source of axial SMCs that is functionally distinct from the Pitx2 lineage. Similar heterogeneity has been demonstrated recently among villous fibroblasts, highlighting the lamina propria milieu in the overall integrity and function of intestinal lymphatics (Hong et al., 2020).Pitx2 daughters were also found near the villous blood capillary plexus, suggesting that they are blood capillary-associated mesenchymal precursors to axial SMCs (Figures 3A’ and A””). A Pitx2 daughter subpopulation was also present in the circular and longitudinal smooth muscle of the muscularis externa (data not shown), suggesting that Pitx2 may regulate gut motility and lymph propulsion through peristalsis. Indeed, Pitx2 modulates the expression of contractile proteins in skeletal muscle, including myosin heavy chains (MyHCs) and troponin I and T (Zhou et al., 2011). During myogenesis, Pitx2 is downstream of Pax3/7 and upstream of Myf5, Myf6, and MyoD (L’honoré et al., 2010; Lagha et al., 2010). Mutations of Pitx1, Pitx2, and Pitx3 result in human developmental defects involving muscle disorders, such as facioscapulohumeral muscular dystrophy (Dixit et al., 2007), Axenfeld-Rieger syndrome (Semina et al., 1996), and anterior segment mesenchymal dysgenesis (Semina et al., 1998), respectively. Current literature on Pitx2 in control of myogenesis concerns skeletal muscle (from paraxial mesoderm). Here we provide insight into Pitx2 function in visceral smooth muscle (splanchnic mesoderm). We speculate that Pitx transcription factors regulate muscle biogenesis and function at their distinct spatiotemporal sites of expression.
Pitx2-dependent formation of the lacteal-associated axial SMC complex
Recent studies have revealed the detailed developmental program of the muscularis externa (Huycke et al., 2019), but the origins of lacteal-associated axial SMC remain unexplored. We demonstrate that axial SMCs do not appear to be continuous with the submucosal lymphatic vasculature, which is devoid of muscle coverage (Miller et al., 2010). Instead, we suggest that axial SMCs arise from villous star-like progenitors associated with blood capillaries, regulated by ASE. In adult vessels, differentiated muscles are not proliferative and express unique contractile proteins (Gomez et al., 2015; Wang et al., 2015) reminiscent to axial SMCs. In contrast, the proliferative αSMA+ star cells show weak immunoreactivity for αSMA and Myh11, a progenitor-like phenotype. A few αSMA+ star cells remain in WT villi (Figure S4D). It is also conceivable that star-like progenitors remain in the villus but lose αSMA expression, akin to SMC phenotypic switching (Owens, 1995; Owens et al., 2004), and become undetectable by smooth muscle markers. We speculate that these cells may contribute to postnatal development or regeneration.The transition from αSMA+ star to elongated spindle-like αSMA++ cells coincided with the appearance of lacteals (E18.5), implying lacteal-axial SMC crosstalk. Postnatal lacteals in WT and Pitx2 intestines always appeared with adjacent axial SMCs. Conversely, mispatterned axial SMCs with aberrant interconnecting branches were occasionally observed in Pitx2 villi without lacteals (Figure 3B). This suggests that axial SMCs are necessary for lacteal formation, whereas lacteals are required for SMC organization. Here we reveal the interplay between lacteal and axial SMC development during the pre- to postnatal transition that is ASE dependent.
Pitx2-mediated changes in ROS levels in the intestinal muscle
Oxidative stress is known to inhibit mesenteric lymphatic contractility and reduce lymph flow by affecting lymphatic smooth muscles (Thangaswamy et al., 2012; Zawieja et al., 1991). Our studies suggest that increased oxidative stress causes muscular-lacteal impairment in Pitx2 intestines, pointing to a crucial role of Pitx2 in control of embryonic redox balance. During myogenesis, the post-mitotic transition of progenitors to mature contractile muscle involves a glycolytic-to-oxidative metabolic switch that increases ROS output (Blanchet et al., 2011; Remels et al., 2010). A network of antioxidant genes is critical to mitigate excessive ROS (Mittler, 2017; Schieber and Chandel, 2014). Recent work described an essential role of Pitx2 and Pitx3 in redox balance among differentiating myoblasts and satellite stem cells (L’honoré et al., 2014, 2018). Similarly, Pitx2 maintains cardiac cellular composition during heart regeneration by preserving mitochondrial function (Li et al., 2018). In a neonatal cardiac injury model, Pitx2c has been shown to promote heart repair by activating the antioxidant response and ROS scavenger genes (Tao et al., 2016). We propose that Pitx2 manages redox homeostasis in the developing gut.Proteins are vulnerable to oxidation, depending on their structure and amino acid composition, affording specificity in post-translational ROS regulation. Although carbonylated proteins are cleared by the proteasome, oxidative stress can impede this process and prolong the half-life of carbonylated proteins (Grune et al., 2004; Stadtman and Levine, 2000), causing sublethal functional impairment and lesions such as actin aggregation (Castro et al., 2012; Dalle-Donne et al., 2001). We observed fragmented αSMA patterns in Pitx2 axial SMCs at P21 (Figure S5C) and profound villus smooth muscle hypoplasia at P26 (Figure S3B). Our data suggest that Pitx2-mediated control of ROS is critical to maintain dynamic organization of the actin cytoskeleton during organogenesis and in differentiated muscles. These data implicate Pitx2 as a rheostat of redox status during muscle development, inviting future studies that should define the roles of Pitx2c in smooth muscle regeneration after intestinal damage.
Pitx2 regulates lacteal development and function in a non-cell autonomous manner
Lymphatic function depends on the microenvironment (Bernier-Latmani et al., 2015; Davis et al., 2017; Hong et al., 2020; Nurmi et al., 2015; Petrova and Koh, 2018). Although ROS can act on nearby cells with (Niethammer et al., 2009; Yoo et al., 2011) or without (Lim et al., 2014) diffusion, ROS-generating axial SMCs can influence adjacent lacteal filopodium behavior. Filopodia are formed by linear polymerization of G-actin-ATP at their barbed ends, whereas filopodia actin bundle dynamics are regulated by ROS (Mattila and Lappalainen, 2008; Munnamalai and Suter, 2009; Shneyer et al., 2017). Lacteal filopodia ensure lacteal function and are crucial for LEC sprouting, extension, and regeneration (Bernier-Latmani et al., 2015; Xu et al., 2010). During adult lacteal regeneration, filopodium-mediated lacteal migration is regulated by the Notch ligand delta-like 4 (DLL4), downstream of VEGF-C (Bernier-Latmani et al., 2015). Deletion of DLL4 in LECs (Bernier-Latmani et al., 2015) or postnatal deletion of VEGF-C (Nurmi et al., 2015) leads to lacteal regression and impaired fat uptake, where VEGFR-3 interaction with NRP2 mediates lacteal sprouting in response to VEGF-C (Xu et al., 2010). Interestingly, we identified three highly conserved Pitx2 binding regions upstream of the VegfC transcriptional start site and found reduced VegfC expression in Pitx2 embryos by E10.75 (data not shown). Our data suggest that Pitx2 provides the context of how the lymphangiogenic VEGF-C/R3 signaling is integrated with the body and organ plan. Collectively, elevated levels of ROS and reduced VEGF-C/R3 signaling in the absence of ASE may compromise filopodium-mediated lacteal extension, causing shorter lacteals and impaired dietary fat uptake and/or trafficking (Figure S9).The portal lipid transport is insignificant compared with the lymphatic pathway but increases during disease where lymphatic clearance is impaired (Cabré et al., 2005; Courtney and Warner, 2017; Fujimoto et al., 1991; Kurtel et al., 1991; Onufer et al., 2019). Lymphatic transport distributes dietary lipids to peripheral tissues for direct energy consumption, whereas the lymphatic system-independent portal route delivers lipids to the liver and reduces accessibility to peripheral tissues.In our model, lymphatic lipid transport dominates in WT neonates, whereas muscular-lacteal defects in the absence of ASE cause lymphatic insufficiency. Compensatory activation of lymphatic system-independent (portal) lipid trafficking overwhelms hepatocytes and causes fatty liver in neonatal Pitx2 mice. Interestingly, these lesions resolved when Pitx2 weanlings transitioned to a low-fat diet. This suggests that non-alcoholic fatty liver disease (NAFLD) may be reversible with early diet adjustments, whereas a prolonged high-fat diet may lead to non-alcoholic steatohepatitis (NASH). These questions require further study using targeted mouse models.The mechanism of portal lipid circulation in Pitx2 intestine is a key outstanding question. The liver is primarily perfused via the portal vein, and portal venous cargo strongly affects hepatic function (as in first-pass drug metabolism). Fatty acids and other nutrients, including carbohydrates and amino acids, are absorbed by enterocytes, a function that is broadly zonated along the villus axis, where genes associated with chylomicron packaging and lipid transport are most highly expressed in the tip-most enterocytes (Moor et al., 2018). This produces a chylomicron concentration gradient along the villus, affording chylomicrons priority access to taller lacteals (Takahara et al., 2013). Although chylomicrons are assumed to be too large to enter blood capillaries, smaller chylomicrons have been detected in hepatic portal blood (Takahara et al., 2013), perhaps via the fenestrated venous vasculature of the villi that may admit smaller lipoprotein particles (Furuya and Furuya, 2007). It is also well established that smaller lipoprotein particles (very-low-density lipoproteins [VLDLs], low-density lipoproteins [LDLs], and high-density lipoproteins [HDLs]) that routinely transit the lymphatic system-independent portal route (Hussain, 2014; Kindel et al., 2010) also participate in dietary lipid uptake. This becomes more evident in disease conditions of mice, rats, and humans (Anwar et al., 2007; Cabré et al., 2005; Courtney and Warner, 2017; Ee et al., 2000; Fujimoto et al., 1991; Hussain, 2014; Kurtel et al., 1991; McDonald et al., 1980; Onufer et al., 2019; Young et al., 1995) and may similarly contribute to lipid transport in ASE mice. Finally, fenestration pore sizes of blood capillaries may be altered in ASE mice, favoring chylomicron uptake, but that possibility has not been addressed in this work.Fatty liver disease remains the single most common chronic liver disease worldwide (Tesfay et al., 2018). Our study provides one possible etiology–defective gut lymphatic development and intestinal fat transport–pointing to a role of Pitx2-driven organ laterality in regulating the gut-liver axis. This understanding may drive future discoveries of clinical targets involving alternative fat absorption pathways. Our findings unravel a role of the LR gene Pitx2 during muscular-lacteal morphogenesis and reinforce the importance of the reciprocal interplay between intestinal muscle development and lacteal function in dietary lipid transport.
Limitations of the study
Loss of Pitx2 results in early embryonic lethality and hampers postnatal studies. To circumvent this, we have leveraged the hypomorphic ASE allele and uncovered that Pitx2 protects against fatty liver disease. The low postnatal survival rate and global loss of ASE have limited our interpretation of detailed molecular pathogenesis. In this regard, an inducible lineage-specific Pitx2 knockout model is needed to precisely parse out the mechanisms downstream of Pitx2. The major challenge in building this model is the lack of appropriate genetic tools that can dissect the roles of lymphatic and gut biology and the remaining questions about the cell of origin and developmental program of lacteal-associated SMCs. Our future work centers on generating gene expression maps of all intestinal cell types at single-cell resolution and across key stages to develop tools that will delete Pitx2 with spatiotemporal precision.Although the role of Pitx2 in muscular redox homeostasis is well documented (L’honoré et al., 2014, 2018; Tao et al., 2016), the precise source of ROS causing oxidative damage in our model remains unclear. Moreover, the unstable nature of ROS escalates the challenge to capture real-time ROS production as well as the source of origin in vivo. We measured protein carbonylation instead as a functional readout of irreversible oxidative damage. However, protein carbonyl is a non-specific marker for protein oxidative damage, which provides limited information regarding the origin of ROS and the nature of oxidized protein. Biochemical analyses of sorted cell populations are needed to identify ROS origins and specific targets to better scrutinize the link between Pitx2 and intestinal lymphatic function.
STAR★METHODS
RESOURCE AVAILABILITY
Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Natasza A. Kurpios (natasza.kurpios@cornell.edu).
Materials availability
This study did not generate new unique reagents.
Data and code availability
The RNaseq database generated during this study is available at GEO: GSE160677This paper does not report original codeAny additional information required to reanalyze the data reported in this paper is available from the lead contact upon request
EXPERIMENTAL MODEL AND SUBJECT DETAILS
All experiments adhered to guidelines of the Institutional Animal Care and Use Committee of Cornell University, under the Animal Welfare Assurance on file with the Office of Laboratory Animal Welfare. Pitx2ASE/ASE, Pitx2, Pitx2cre, ASEcre, ROSA26, and ROSA26-EYFP mice were previously described (Liu et al., 2001,2002; Madisen et al., 2010; Shiratori et al., 2006; Srinivas et al., 2001). All strains of mice in these studies were maintained in a specified pathogen-free barrier facility under a 12-hour light cycle with free access to standard chow pellets and water unless specified. All mice used for mating were between 2-6 months old. For timed pregnancies, dams and studs were housed separately until the scheduled mating. Time mated breeding pairs were separated upon the morning when vaginal plugs were found and staged as E0.5. Postnatal stages are defined upon birth as postnatal day 0 (P0). Each mouse used was numbered by toe clip and genotyped by PCR amplification at the age between 7-10 days. Embryos and pups under 9 days were genotyped after isolation. For full list of genotyping primers, see Supplemental Table. Each group of comparison contained at least three mice from representative genotypes at the specified developmental stages. Mice of different genotypes and gender were randomly chosen and sorted into different procedures and measurements. When possible, littermates of different genotypes were sorted into the same procedure and measurements as pairs.
METHOD DETAILS
Mouse tissue collection, fixation, processing
For sectioning, tissues were dissected in ice-cold 1X PBS, immediately followed by PBS washes and fixed overnight at 4°C in 4% paraformaldehyde (PFA)/PBS, ice-cold 100% MeOH, or PAXgene® fixative. For tissues fixed in PFA, tissues were thoroughly washed with PBS, then preserved in 4°C until use. Tissues fixed in MeOH were stored in −20°C until further use. Tissues fixed with PAXgene® fixative were processed following manufacturer’s instructions. For intestinal tissue that proceeded to frozen sections, tissues were dehydrated in gradient sucrose/PBS solution, then embedded in Tissue-Tek* OCT. Compound and stored in −80°C until sectioning. Hepatic tissue for frozen sections following oil red O stain was embedded in 100% OCT immediately after dissection. Tissues intended for histology analyses were processed with standard paraffin embedding protocol in the Animal Health Diagnosis Center, Cornell University College of Veterinary Medicine. All frozen sections were 15 μm thick; paraffin sections were 5 μm thick.
Section and whole mount immunofluorescence
Embryos were processed for immunofluorescence (IF) on sections or whole mount (tissue slice). For section IF, frozen sections were permeabilized with 0.1% Triton X-100/PBS for 45 min, blocked in 3% BSA in PBS for 3 hours at room temperature, and incubated with 100X diluted primary antibodies/blocking overnight at room temperature. 500X diluted secondary antibodies were co-incubated with 1000x diluted DAPI for nuclear labeling at room temperature for 1 hour. Samples were mounted in ProLong™ Gold Antifade Mountant (P36930, Thermo Fisher) and preserved in room temperature until imaging. For whole mount IF staining, the protocol was adapted from (Bernier-Latmani and Petrova, 2016). In brief, tissue was blocked in BSA, serum, and 0.3% Triton X-100 in PBS in 4°C for 3 hours, then incubated with primary antibodies overnight at 4°C. Tissues were thoroughly washed in 0.3% Triton X-100/PBS for 5 hours with hourly buffer changes, then incubated with secondary antibody overnight at 4°C. Tissue was thoroughly washed in 0.3% Triton X-100/PBS for 5 hours with buffer changes every half hour, then fixed in 4% PFA/PBS for two overnights at 4°C. For confocal imaging, tissue was washed with PBS and properly sliced to 100-200 μm with spring scissors before mounted in ProLong™ Gold Antifade Mountant. Primary antibodies are listed below: GFP (ab290, Abcam), CD31 (PECAM-1, 553371, BD Bioscience), Lyve-1 (Ab14917, Abcam), αSMA (F3777, C6198, A2547, Sigma-Aldrich), and Myh11 (BT-563, Alfa Aesar). Secondary antibodies used are listed below: AlexaFluor anti-goat, anti-mouse or anti-rabbit 568, and anti-rabbit 488 (source).
Liver oil red O stain
Oil red O stain was performed to detect liver neutral lipids as previously described (Mehlem et al., 2013). In brief, tissue sections were processed as described in the previous section. Slides with paired tissues from WT and Pitx2 littermates collected on the same slide were equivalate to room temperature from −80°C when stored, following oil red O and hematoxylin incubation. Slides were rinsed in water before mounted in MOWIOL® 4-88 Reagent (EM475904, EMD Millipore).
In vivo lipid tracer feeding assay
BODIPY™FL C16 (Thermo Fisher D3821) was added to a warmed Intralipid (20% emulsion, Sigma I141) solubilizing agent to make final concentration at 0.4 μg/μl. 50 μL of 37°C reconstituted Bodipy™/Intralipid solution was fed to each pup through 24-gauge reusable feeding needles (Fine Science Tools 18061-24). To maintain proper body temperature and hydration, the pups were left with the dam for 5 hours (milk ad lib) before imaging and tissue collection. Of note, 2 hours was not sufficient for the ASE mixed strain background (C57BL/6 x C3H, mutant or WT) to see BODIPY. For both WT and ASE mice, 4-5 hours was the shortest time needed to detect BODIPY signal reaching ileum and the ileal mesenteric lymphatics. Images were taken under a Zeiss dissection microscope immediately after sacrifice by decapitation. Images were taken from liver tissue flash frozen in liquid nitrogen, then cryo-sectioned at 20 μm and mounted in ProLong™ Gold Antifade Mountant.
Tissue carbonyl assay detection
Protein Carbonyl Colorimetric Assay Kit (Cayman 10005020) was adapted for intestinal protein carbonyl measurement. In brief, small intestines were collected from animals euthanized by hypothermia followed by PBS perfusion. Intestine contents were removed by gentle swirling in ice-cold PBS after exposing the interior. Liquids were removed as much as possible before flash freezing the tissue in liquid nitrogen. 150 mg of tissue was homogenized in 50 mM phosphate buffer pH 6.7 then centrifuged at 10,000 x g for 15 min at 4°C. Supernatants were collected and incubated with 1% streptomycin sfate in 50mM potassium phosphate pH 7.2 at room temperature following centrifugation at 6,000 x g for 10 min at 4°C to remove excessive nucleic acids. Supernatants were collected and stored in −80°C until further use. The tissue carbonyl assay was then performed and analyzed following manual instruction.
Carbonylated total protein and smooth muscle actin-alpha measurement
EZ-link hydrazine-biotin was used to label protein carbonyl as previously described (Grimsrud et al., 2007; Xu et al., 2014b). Small intestines were collected in ice-cold PBS following removal of intestinal contents. Tissues were stored in −80°C until further processed. In brief, tissues were homogenized in ice-cold homogeneous buffer (pH5.5-100mM sodium acetate, 20mM NaCl, 0.1mM EDTA with protease and phosphatase inhibitor cocktail). Supernatant was incubated with 2% SDS in homogeneous buffer in 65°C for 5 min after centrifugation at 1200 rpm for 10 min in 4°C. Samples were further centrifuged at 13,000 g for 1 hour in 4°C. Supernatant was collected for protein concentration measurement and 100-150 μg of protein samples were then incubated with EZ-link hydrazine-biotin at 125uM for 2 hours at room temperature with constant mixing. Hydrazine-labeled samples were then dialyzed into PBS for streptavidin pulldown or αSMA IP. Specific bands of hydrazine-labeled proteins and αSMA were detected by western blot: samples were separated on SDS-PAGE then transferred to nitrocellulose membrane following 5% BSA blocking at room temperature for 1 hour before primary antibody or HRP-streptavidin incubation at 4°C overnight. For αSMA detection, HRP-conjugated secondary antibody was incubated at room temperature for 1 hour before signal detection.
Liver Lipid Analysis
20 mg of fresh liver samples were flash frozen and homogenized in 1.8 mL of Chloroform: Methanol (2:1 v/v). The homogenates were centrifuged at 4000 rpm for 10 min. the Supernatants were collected into a clean tube and vortexed with 250 μl of 50mM NaCl. After 30 min of phases separation, 100 μL of lower phase extract was collected and mixed with 20 μL of Triton X-100/Chloroform mixture. The samples were then air-dried on 37°C heat block for 1 hour before resuspension in 80 μL of ddH2O. 10 μl extract were used in the Serum Triglyceride Determination Kit from Sigma-Aldrich (TR0100) to determine hepatic triglyceride level. Hepatic neutral lipid accumulation was evaluated by Oil Red O stain as previously described (Mehlem et al., 2013). 10 μm frozen section tissues were stained with Oil Red O working solution prepared as described for 5 min, following Hematoxylin Solution, Mayer’s (MHS1, Sigma-Aldrich) stain for 15 s. Tissue slides were rinsed in running tap water for 15-30 min until they were ready to mount in Gelvatol mounting medium. The Oil Red O stained slides were then imaged the following day.
Transmission electron microscopy
Tissues were isolated in ice-cold PBS and gently agitated to remove blood. Liver tissues were cut into approximately 1mm size cubes and fixed in 2% Glutaraldehyde, 2% Paraformaldehyde in 0.1M Sorensen’s Phosphate Buffer overnight in 4°C then washed in ice-cold PBS 3 times. Tissues were then post fixed in 1% Osmium Tetroxide aqueous for 1 hour before they were dehydrated in graded ethanol series following Propylene Oxide washes and immersed in Araldite/Propylene Oxide 1:1 solution overnight. The samples were then soaked in fresh Araldite for 4 hours before final embedding and heated to polymerize in silicon rubber molds. Thin sections (60-90nm) were cut on a Leica EM UC6 Ultramicrotome. Sections were collected onto 200 mesh copper grids and stained with 2% Uranyl Acetate and Reynold’s Lead Citrate before examined on the FEI Tecnai G2 Spirit Transmission Electron Microscope operated at 80 Kv.
RNA-sequencing
4 WT and 3 Pitx2 P1.5 small intestines were included in the analysis. Total RNA was isolated with Trizol, with an extra chloroform extraction to remove residual phenol and addition of glyco-blue as a carrier to promote RNA precipitation. RNA concentration was determined on a Nanodrop. TruSeq-barcoded RNA libraries were generated with the NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs) with polyA+ enrichment using 1 μg total RNA as input. The libraries were sequenced on a NextSeq500 with single-end 85nt reads. Illumina pipeline software v1.8 was used for base calling. Trim_galore was used to trim and filter reads (–nextseq 20–length 50). STAR was used to map reads to the Mus musculus mm10 reference using Ensembl gene annotations and output raw counts (–quantMode GeneCounts). Differential gene expression was defined by false discovery rate (FDR, adjusted p value for multiple tests) ≤ 0.05 from raw counts between WT and Pitx2.
Gene Set Enrichment Analysis
All RNA-seq data from 4 WT and 3 Pitx2 P1.5 small intestines were included in the Hallmark gene sets v7.4 analysis from Molecular Signatures Database (MSigDB). The following parameters were used to generate Figure S5A: Number of permutations = 1000, Permutation type = phenotype, Enrichment statistic = weighted, Metric for ranking genes = Signal2Noise.
QUANTIFICATION AND STATISTICAL ANALYSIS
Imaging and quantifications
Microscopic images were acquired using a Zeiss Observer.Z1/Apotome. Confocal images were taken on Zeiss LSM880 confocal microscopes. To capture the entire lacteal and villus in whole mount immunofluorescent stain, multiple adjacent fields of images were taken under a 40X lens then stitched on Zen before exporting for further analyses. Imaris 9.5 was used for image processing, filtering, background subtraction, confocal image stacking, and quantifications. Measurements (lacteal length, filopodia counts etc.) were done on Imaris 9.5 or Fiji. Z stack images were all analyzed on Imaris 9.5.
Statistics
Statistical analyses were performed in GraphPad Prism 8 (La Jolla, CA). Comparison between different genotypes were tested by unpaired, two tailed Student’s t test. All data were first analyzed as mean values of measurements (lacteal length, filopodia number, etc.) from multiple villi in multiple fields across the proximal small intestine for each animal. Lacteal filopodium was defined as cell protrusion longer than or equal to 6 μm. The mean values from all animals were then plotted in groups based on genotype and developmental stages. Almost all data were expressed as mean of the means from each mouse ± standard error of mean (SEM).
Authors: H Shiratori; R Sakuma; M Watanabe; H Hashiguchi; K Mochida; Y Sakai; J Nishino; Y Saijoh; M Whitman; H Hamada Journal: Mol Cell Date: 2001-01 Impact factor: 17.970
Authors: Manjusha Dixit; Eugénie Ansseau; Alexandra Tassin; Sara Winokur; Rongye Shi; Hong Qian; Sébastien Sauvage; Christel Mattéotti; Anne M van Acker; Oberdan Leo; Denise Figlewicz; Marietta Barro; Dalila Laoudj-Chenivesse; Alexandra Belayew; Frédérique Coppée; Yi-Wen Chen Journal: Proc Natl Acad Sci U S A Date: 2007-11-05 Impact factor: 11.205
Figure 1.
Figure 3.
Figure 2.
Figure 4.
Figure 5.
Figure 6.
Figure 7.