Abstract
The lymphatic system plays important roles in draining fluids from interstitial spaces, absorbing lipids from the intestinal tract, and transporting white blood cells, including antigen-presenting cells, to lymphoid organs. Based on these functions, a number of disorders are associated with lymphatic vascular abnormalities including lymphedema, inflammatory disorders, and tumor-associated lymphangiogenesis. Therefore, understanding the mechanisms underlying the development of the lymphatic network can guide the treatment of lymphatic diseases. Activation of the transcription factor p53 has been implicated in several developmental syndromes where p53 is stimulated by cellular stressors like ribosomal imbalance. Once induced, p53 triggers important anti-proliferative programs like cell-cycle arrest and apoptosis and is therefore maintained at very low physiological levels during embryogenesis. Here, we report for the first time, a critical role of p53 overexpression in defects of lymphatic development. We generated two murine models that carry increased p53 activity induced by ribosomal stress concomitant with the loss of its negative regulators. These high-p53 models are embryonic lethal and present with cutaneous hemorrhaging, severe edema, and distended blood-filled lymphatic vessels. Cellular characterization of the lymphatic endothelial vessels at late-gestation show a reduced proliferation of endothelial cells, upregulation of the growth arrest marker p21 and a potential reduction of initial lymphatics that absorb the interstitial fluid. We also demonstrate that the lymphatic phenotypes are p53-dependent as genetic deletion of one copy of p53 or pharmacologic inhibition of p53 abolishes the cutaneous hemorrhaging, drastically reduces the edema, and rescues the embryonic lethality. Importantly, we detected overexpression of p53 in lymphatic endothelium of lymphedema associated disorders, linking our murine findings to the human disease. Taken together, we discovered that wild type p53 plays a central role in lymphedema predominantly through cell cycle arrest. Our findings indicate that targeting the p53 pathway, an unrecognized mechanism thus far in the genesis of lymphatic deficiencies, may offer therapeutic options for treating incurable lymphatic maladies.
Main
The transcription factor p53 is a major sensor of cellular stress, including ribosomal imbalance, DNA damage and oncogene activation1,2. Once induced, p53 triggers multiple important cellular programs such as cell cycle arrest, apoptosis and senescence that are deleterious to healthy normal cells3-5. Therefore, p53 is tightly regulated by its main inhibitors, Mdm2 and Mdm4, and is maintained at undetectable levels in normal fetal and adult cells. Indeed, homozygous deletion of Mdm2 or Mdm4 in mice causes embryonic lethality due to excessive p53 activity resulting in apoptosis or cell cycle arrest respectively6-8. In contrast, Mdm2 or Mdm4 haploinsufficient mice survive to adulthood and reproduce normally, despite an endogenously active p53, unless challenged by cellular stressors like ionizing radiation or oncogene activation. These mice exhibit radiosensitivity, decreased in-vitro transformation potential, and reduced in vivo tumorigenesis9,10. Several studies also revealed a role for p53 overexpression in developmental disorders11,12. For example, the ribosomopathy model with reduced expression of ribosomal protein Rpl27a, displays endogenously elevated p53 and phenocopies mouse models of high p5313. To examine for a potential genetic interaction between Rpl27a, Mdm2 and Mdm4, and to study the effects of ribosomal stress induced p53 during embryonic development, we created compound mice with low Rpl27a and heterozygosity for Mdm2 or Mdm4 (Rpl27a:Mdm2+/- and Rpl27a:Mdm4+/- mice). These mice demonstrated cutaneous hemorrhaging, severe edema and late-gestational lethality. Histopathology examination revealed lymphatic specific defects in both mouse models.
During embryogenesis, the earliest lymphatic endothelial cells arise from the venous branch of the previously established primitive blood vascular network. The lymphatic network in the adult is essential for regulating tissue fluid balance, immune function, and lipid uptake from the gut14. Currently, defects in the lymphatic function have been linked to several pathologies such as obesity, cancer, lymphedema, and inflammation15. Lymphedema results from a failure to collect excess tissue fluid from the interstitial space and return it back to the circulation via the thoracic duct. While lymphedema affects millions of people worldwide, our understanding of lymphatic development and its molecular pathogenesis trails behind that of the blood vascular system. Therefore, current treatments remain largely palliative, including manual lymph drainage and compression garments (Complex Decongestive Therapy) to reduce swelling and drain excessive fluid16 and antibiotics to treat the life-long recurrent infections that lymphedema patients suffer from.
In the last two decades, genetic studies have identified several molecular players in the development of the lymphatic network17. One of these genes, Prospero Homeobox 1 (Prox-1), is important for maintaining lymphatic identity given its role as the master switch of lymphatic differentiation from the veins. When Prox-1 is expressed in a subpopulation of blood endothelial cells (BECs) of the cardinal vein (CV) around E9.5, they give rise to two populations of lymphatic endothelial cells (LECs)18-20. The first LEC population sprouts from the venous endothelium starting from E10.5-E11.5 to form the primary lymph sac. These cells also express the lymphatic endothelial markers Hyaluronan Receptor-1 (Lyve-1) and Vascular Endothelial Growth Factor Receptor-3 (Vegfr-3), prompting their proliferation and development into the peripheral lymphatic vessels by E14.514. These vessels eventually sprout into the skin21-25, after which they undergo remodeling and maturation to make up the lymphatic network of capillaries and collecting vessels26. The second subset of Prox-1+ LECs persists in the CV to form the lymphovenous valves (LVV) that prevent the retrograde flow of blood into the lymphatic circulation27. Given the polarized expression and stage-dependent function of Prox-1 during development, dysregulation of this gene can lead to abnormalities within the entire lymphatic system, including reduced proliferation of the lymph vessels and failed separation of the lymphatics from the venous system25. Accordingly, mice haploinsufficient in Prox-1 exhibit dermal edema by E13.5, smaller lymph sacs, and lack of functional LVVs21,27,28. Essentially, mouse models have been powerful in providing insight into the molecular players of lymphedema and lymphatic-related diseases. Mutations in approximately 12 genes have been found in lymphedema cases29, many of which were identified in animal models. This allowed the establishment of genetic testing (Lymphatic Malformations and Related Disorders Panel by Blueprint Genetics) that supported a more accurate diagnosis and classification of lymphatic defects. Nevertheless, additional research is needed to identify other factors, genes, and mechanisms that may lead to more therapeutically viable options for lymphedema. Recently, anti-inflammatory drugs30 have been effective in managing symptomatic lymphedema that remains with no medicinal option. Our data substantiate that when the master tissue surveyor p53 goes rogue during development, it can elicit nefarious activities. The lymphatic tissue seems to be particularly sensitive to this p53 gene dosage. Therefore, controlling p53 activity at the stage of lymphatic proliferation may greatly benefit patients with lymphatic disease and relieve them from the ordeals of symptomatic lymphedema.
Results
Genetic interaction between Rpl27a, Mdm2, and Mdm4 leads to embryonic lethality due to severe cutaneous edema and hemorrhaging
Mice haploinsufficient for Mdm2, Mdm4, or Rpl27a displayed an endogenously stable p53 resulting in p53-dependent cellular outcomes such as apoptosis and cell cycle arrest10,13. These conditions were subsequently rescued by the deletion of one copy of p53. To test for potential genetic interactions between these genes and observe the impact of an augmented p53 activity, we crossed Rpl27alow/+ mice to Mdm2+/- or Mdm4+/- animals. From hundreds of crosses, we did not observe the expected double heterozygotes Rpl27alow/+:Mdm2+/- (RP27M2) or Rpl27alow/+:Mdm4+/- (RP27M4). Timed mating (E11.5-E18.5) indicated the presence of these genotypes until E16.5 (Tables S1 and S2). MRI imaging (Fig. 1a) confirmed these observations and detected the termination of mutant embryos as indicated by the gradual decrease of perfused volume to 0 mm3 between E15.5-E17.5 (Fig. 1b). As we are unable to genotype in utero, we presumed that the deceased embryos during MRI were the double heterozygotes not seen at birth. All the other genotypes developed normally with an increase in total volume proportional to the gestational age. These findings demonstrate the existence of a genetic interaction between the three genes that ultimately results in fetal lethality.
A closer examination showed that Mdm2+/- and Mdm4+/- embryos were similar to WT with no overt phenotypic abnormalities (SFig. 1). Rpl27alow/+ embryos displayed occasional light hemorrhaging and edema starting at E14.5 that was predominantly localized to the dorsal skin (Fig. 1c). These embryos were born at a lower body weight and with a developmental delay that persisted until around 8 weeks of age. They were able to recover from the delay and reproduce and live normally though maintaining a slightly lower body weight13. On the other hand, 100% of RP27M2 and RP27M4 embryos exhibited hemorrhage and/or edema at late-gestation that resulted in 100% mortality post-E16.5 (Fig. 1c). Typically, other mouse models of edema demonstrate an involvement of the lung, the heart or the liver 31,32. Histopathological examination of Hematoxylin and Eosin (H&E) stained sections of major organs such as the heart, lungs, brain or liver of mutant embryos showed no overt abnormalities (SFig. 2a). When we pulled away the bloody skin of both mutants, we did not observe an internal hemorrhaging (SFig. 2b). This led us to reason that these phenotypes were restricted to the skin. Looking at the H&E of dorsal skin, we observed large fluid-filled gaps in both models and vessels engorged with blood (Fig. 1d). Hemorrhaging and edema severity scoring on a scale of 0 (none) to 3 (severe) (SFig. 3a) revealed that these conditions gradually worsened with gestational age and ended by death at E16.5. RP27M4 phenotypes were significantly more pronounced than those of RP27M2 (SFigs. 3b, 3c). This is surprising given that Mdm2 is a more powerful inhibitor than Mdm4 due to its E3 ligase activity that degrades p53. Therefore, typically mice with conditional loss of Mdm2 in several tissues are invariably much sicker and at an earlier time than those with Mdm4 loss33-35.
RP27M2 and RP27M4 mice display lymphatic defects
RP27M2 and RP27M4 mutants displayed severe hemorrhaging and cutaneous edema. We performed immunofluorescence staining (IF) of blood and lymphatic vessels on embryos using markers such as Platelet Endothelial Cell Adhesion Molecule-1 (PECAM-1 or CD31) and Lyve-1 respectively. We observed small and flat cutaneous lymph vessels (Lyve-1+) in WT mice, while RP27M2 and RP27M4 lymphatics looked distended and filled with blood (Fig. 2a, 2b). The measures of lymphatic vessel size showed a proportional increase in the severity of the phenotypes. As such, the average vessel diameter for RP27M4 measured ∼257 μm, which is approximately two times larger than that of RP27M2 (∼138 μm), and ∼4.3 times bigger than WT lymphatic vessels (∼58 μm). To a much lesser extent, Rpl27alow/+ lymphatic vessels were also enlarged (∼110 μm) compared to WT vessels and often filled with erythrocytes (Fig. 2b). Confocal microscopy on whole-mount embryos stained with lymphatic markers Vegfr-3 and Prox-1 (Fig. 2c) showed free erythrocytes in the interstitial space of the skin and blood-filled (olive color) lymphatics (Vegfr-3+ and Prox-1+ double stained) in the mutants. We also noted sharply reduced density and networking of lymphatics when compared to the other genotypes. Changes in the distribution of mutant blood vessels (PECAM-1+) indicated that they may also be affected. However, the apparent reduction of blood vessel density may be secondary and a consequence of the pronounced edema. The mutants seem to have increased number of filopodia extended by the LECs and the BECs compared to WT, giving the impression that blood and lymphatic vessels are aligned. Given the anti-proliferative role of p53, we double-stained E15.5 skin for Lyve-1 and the proliferation marker Ki-67 and cell cycle arrest marker p21. WT, Mdm2+/-, Mdm4+/- and Rpl27alow/+ embryos demonstrated an active proliferation in lymphatic vessels and an absence of cell cycle arrest (Fig. 2d, SFig. 4c). In contrast, RP27M2 and RP27M4 cutaneous lymphatic vessels showed no detectable Ki-67 and significant upregulation of p21 in mutant LECs (Figs. 2d, 2e). These results indicate a hindrance in proliferation and growth arrest of lymphatic vessels of both models, which may explain the rudimentary network of lymphatics in the mutants. We also double stained for Caspase-3 and Prox-1 in four E13.5-E16.5 tissues per genotype to check for apoptosis, another major cellular process that may be induced by elevated p53. Both RP27M2 and RP27M4 embryos showed no obvious upregulation of apoptosis in lymphatic endothelium compared to the other genotypes (data not shown). While we cannot rule out cell death or the presence of a senescence phenotype, p53 appeared to be acting on the lymphatic network largely through cell cycle arrest.
Ultramicroscopic imaging of the CV and its connecting structures at E11.5 showed that the primordial thoracic duct (pTD) and the CV were physiologically normal in all embryos. However, the primordial valves, which form the contact side between the pTD and the CV, did not develop properly in RP27M2 and RP27M4 mice (Fig. 3a, SFig.4e). Since several primary lymphedema models showed lymphatic defects in the mesenteries, we checked E14.5 mesenteric vessels of RP27M2 and RP27M4. Only a few Prox-1+ cells were present at the hilus, but not around the major blood vessels as seen in WT embryos. We also noted very few LECs and reduced lymphatic branching. The small population of LECs present in Rpl27alow/+ and mutant mesenteries were rather concentrated near the lymphatic sac, the structure that gives rise to the lymphatic vessels. Prox-1 staining (red) was also detected outside of blood and lymphatic vessels. These Prox-1+ cells were not of lymphatic or blood fate given the absence of Vegfr-3 or PECAM-1 staining respectively (Fig. 3b, SFig. 4f). We speculate that these cells may be macrophages that have engulfed Prox-1+ cells but we do not know their exact origin. Staining of E16.5 mesenteries showed that lymphatic vessels were present in Rpl27alow and RP27M2 mice running in parallel along artery-vein pairs that extend from the mesenteric root. In RP27M4, some lymphatic vessels were observed but looked truncated. These observations indicate that lymphatics eventually develop past E14.5 but with a clear delay to WT and single heterozygous mice (Fig. 3c, S4g).
Gene expression in skin cells of select p53 targets and lymphatic regulators at gestational ages E13.5-E15.5 showed the expected p53 increase in RP27M2 (Fig. 4a) and a slight but significant upregulation in Puma (p53 upregulated modulator of apoptosis) at E14.5 (SFig. 6a) that evidently did not translate to increased Caspase-3 positivity in E13.5-16.5 affected tissues (data not shown). Interestingly, p53 overexpression in RP27M4 mice was detected at E13.5 (Fig 4a), which was a day earlier than in RP27M2 mice. The surge of p53 at an earlier stage of development in RP27M4 embryos could partially explain the more severe phenotypic presentations. Puma mRNA was also significantly elevated in RP27M4 at E15.5 (SFig. 6a). Both RP27M2 and RP27M4 skin had dramatically diminished Prox-1 at E15.5 (57% and 73% respectively) in comparison to Rpl27alow/+ or WT littermates (Fig. 4a) reflecting a potential rarefication of the lymphatic vessels in the skin. Another lymphatic marker Lyve-1 remained elevated at E14.5 and E15.5, ostensibly due to myeloid infiltration as Lyve-1 is also expressed on macrophages36. Vegfr-3, a Prox-1 target whose expression correlates with lymphatic branching, was reduced in both models at E15.5 (SFig. 6a). This is consistent with the IF of skin lymphatics that showed fewer lymphatic vessels in mutant mice (Fig. 2b). Stanczuk et al. (2015) showed that c-Kit+ hemogenic endothelial cells in the mesentery gave rise to lymphatic vessels, indicating that some lymphatics can originate from non-venous hematopoietic progenitors37. Since c-Kit is a p53 target that was affected in the mesentery of lymphedema models presented by the Mäkinen group, we checked its levels in our mutants. Amazingly, c-Kit was 80-90% lower in mutant skin compared to WT (Fig. 4a), suggesting that the reduction in c-Kit can contribute to some extent to the lymphatic abnormalities in both models. However, we were still perplexed by the accentuated hemorrhaging and edema in RP27M4 mice compared to RP27M2 mice since Mdm2 is a more powerful inhibitor of p53 than Mdm4. Since Mdm2 also interacts with Mdm4 to regulate p5338, we checked the levels of Mdm2 expression in the skin. Strikingly, Mdm2 was very low in RP27M4, which potentially resulted in the disruption of the Mdm2-Mdm4 interaction and augmented p53 activity induced by Mdm2 and Mdm4 haploinsufficiency (Fig. 4a). Rpl27alow/+ and Mdm4+/- mice also had slightly reduced Mdm2 expression which may have been further exacerbated by the interaction of both genetic conditions as seen in RP27M4 embryos. This could well explain the severity of RP27M4 phenotypes compared to RP27M2. In summary, our findings suggest that p53 upregulation triggered by ribosomal stress leads to low Prox-1 and Vegfr-3 impairing normal lymphatic development.
To further characterize the endothelial cell populations in E12.5-E15.5 skin of edemic mice, we separated CD45- stromal cells by FACS sorting using the established endothelial markers PdPn, CD31, and Lyve-139-43 (Figs. 4b, 4c, and data not shown). We identified four distinct populations: CD31mid:PdPnlow:Lyve-1- (Population I), CD31high:PdPnlow:Lyve-1- (Population II), CD31mid:PdPnhigh:Lyve-1low (LECs, Population IIIA) and CD31mid:PdPnhigh:Lyve-1high (LECs, Population IIIB). Prior to E14.5, no significant differences across these populations were detected in WT, RP27M2, and RP27M4 (data not shown). However, at E15.5, the ratio of Population II to Population III strongly decreased, while the cell frequency of Population III remained constant (Fig. 4b). A closer look at the state of LECs in Population III revealed a Lyve-1low (Population IIIA) and a Lyve-1high (Population IIIB) subpopulation. Population IIIB was reduced proportionally to the increase in Population IIIA (Fig. 4c). If Population IIIB are the initial lymphatics that absorb the interstitial fluid, their decrease would corroborate the IF skin stains, where lymphatics were reduced in number (Fig. 2b) and elucidate the edema in RP27M2 and RP27M4 skin. Thus, it is understandable that the increase in putative collector lymphatics (Population IIIA), did not sufficiently compensate for the hampering of the formation of initials by merely ramping up the capacity to transport excess fluid back to the vein.
Gene deletion of one p53 copy reversed hemorrhaging, edema and embryonic lethality in RP27M2 and RP27M4 mice
To determine if the lymphatic anomalies were driven by p53, we deleted a single allele of p53 in both models. For this, we crossed Rpl27alow/+ animals to mice lacking Mdm2 and p53 (Mdm2-/-: p53-/- mice) or Mdm4 and p53 (Mdm4-/-:p53-/- mice) to obtain 50% Rpl27alow/+:Mdm2+/-:p53+/- (RP27M2:p53+/- mice) and 50% Rpl27alow/+:Mdm4+/-:p53+/- (RP27M4:p53+/-) mice respectively. Both crosses resulted in the expected 1:1 Mendelian ratio of progeny (Fig. 5a) that lived to more than 9 months with no overt pathologies, edema or hemorrhaging (Figs. 5b, 5c). Moreover, Lyve-1 staining of RP27M2:p53+/- and RP27M4:p53+/- skin demonstrated that the lymphatics reverted to normal and were similar in size to those in the WT skin (Figs. 5b, 5c). Molecular examination of RP27M2:p53+/- skin indicated that p53, Puma, and c-Kit normalized to WT levels, as did the lymphatic markers Prox-1, Lyve-1, and PdPn (Fig. 5d, SFig. 6c). This normalization of gene expression was also seen in RP27M4:p53+/- skin. Consistent with these data, the distribution of Population I, II, and III returned to a level equivalent to the WT distribution (Fig. 5f). Taken together, the aberrant lymphatic development in our mutants was p53-dependent, indicating that p53 levels must be kept in check for normal lymphangiogenesis.
Chemical modulation of p53 eliminates cutaneous hemorrhaging, ameliorates edema and rescues embryonic lethality of mutant mice
To confirm the contribution of p53 overexpression to lymphedema, we used a pharmacological approach by testing a known reversible p53 negative modulator, Pifithrin-α (PFT)44,45. For this, we set up time mating of Rpl27alow/+ mice with Mdm2+/- or Mdm4+/- mice. Females carrying both models were injected intraperitoneally (IP) daily from E11.5 to E17.5 and their weight gain was recorded until delivery. Amazingly, pups from both groups were born at a Mendelian ratio with no signs of hemorrhaging (Figs. 6a, 6b). PFT-RP27M2 mice were indistinguishable from WT and 100% survived until at least 10 days old (Fig. 6a and data not shown). However, PFT-RP27M4 pups had an overall edema that was considerably reduced, a slightly looser skin, and excess skin in the area around the neck (Fig. 6a). H&E staining of dorsal skin indicated that the skin looked normal in both mutants with no evident subcutaneous edema (Fig. 6b). 100% of PFT-RP27M4 mice died did shortly after birth. Lyve-1 staining of PFT-treated RP27M2 and RP27M4 skin at postnatal day 0 (P0) showed that lymphatic vessels reverted to a normal size (Fig. 6c). We speculate that the extent of phenotypic rescue at the low PFT dose used may be determined by the severity of the presentation. Therefore, increasing the amount of PFT given to the RP27M4 carrying mothers may allow for RP27M4 pups to survive.
In conclusion, genetic inhibition by single allele loss of p53 or drug-driven control of p53, reversed cutaneous lymphatic defects, edema and embryonic lethality. This is a powerful proof of concept demonstration that elevated p53 contributes to lymphatic defects. The data also suggest that pharmacological modulation of p53 in mid-to-late gestation is clinically pertinent to treatment of lymphedema where p53 is overexpressed in lymphatic endothelium.
p53 is overexpressed in lymphatic defects and NOT in normal lymphatics
To test for the involvement of p53 overexpression in cases of human lymphedema, we checked p53 levels by immunostaining in seven lymphatic diseases associated with severe edema. We observed high p53 positivity in the lymphatic endothelium in six out of eight samples. Interestingly, venous and arterial endothelium were negative in the same p53-positive patient tissues (SFig. 7a). Staining for p53 in the endothelial cells of normal neonatal and adult human skin was also negative (Figs. 6d, SFig. 7b).
In lymphatic diseases, more often females are affected than males (National Organization of Rare Diseases). Therefore, we checked for gender differences in the disease presentation of our mutants. Based on edema and hemorrhage scoring following defined criteria (Fig. S3a), we noted that at E15.5, RP27M2 females were on average two-fold more significantly affected by both hemorrhaging and edema than males, while RP27M4 females suffered from more severe hemorrhaging than males with a tendency for worsening of edema (Fig. 6e). Sex differences for edema in RP27M4 mice were likely masked by the extensive magnitude of this phenotype in the mutants. Taken together, p53 upregulation plays a central role lymphatic defects and our murine models seem like a good representative model of human lymphatic defects.
Discussion
We show here for the first time a link between the transcription factor p53 and the lymphatic system. The characterization of two mouse models of high p53 demonstrated that p53 overexpression leads to lymphatic defects, in particular lymphedema, during embryonic development. As such, RP27M2 and RP27M4 mice that express elevated p53 triggered by ribosomal stress display pronounced cutaneous edema and hemorrhaging during late gestation. These mice also exhibit reduced cutaneous lymphatic vasculature that becomes enlarged and filled with blood, and a stark delay in formation of mesenteric lymphatics.
The lymphatic phenotypes in the mutants were first observed at E14.5 (Figs. 1, 2), coinciding with the onset of the lymphatic proliferation that establishes the lymphatic network throughout the body. Lymphatic structures continue to develop in the edemic embryos but with a clear delay compared to WT mice, likely due to the considerable drop in Prox-1 and its target Vegfr-3 in these tissues as seen in other lymphedema mice28,46 (Fig. 3 and SFig. 4f). Interestingly, we observed a huge decline in c-Kit expression in both mutants. Stanczuk L. et al.37 also noticed a reduction of c-Kit in mice heterozygous for Vegfr-3 and p110α, the catalytic subunit of PI3K, a key downstream effector of Vegfr signaling in endothelial cells. A part of the lymphatic vasculature of the mesentery in these mice developed from a non-venous c-Kit lineage cells of hemogenic endothelial origin, which is contrary to the long held doctrine that mammalian lymphatic vessels sprout from veins. Therefore, the depletion of c-Kit in our mutants that may be reflective of the loss of hematopoetic c-Kit+ stem cells in Rpl27alow mice13, may also contribute to lymphatic phenotype. Further investigation into the origins of dermal lymphatic vessels is ongoing and is a prerequisite for restoring function in diseases with lymphatic deficiencies. In agreement with the confocal microscopy of E14.5 and E16.5 skin, FACS sorting of E15.5 panendothelial dermal cells (CD31+:CD45-) revealed a drastic change in the LEC subpopulation, with a decrease in putative initial lymphatics (subpopulation IIIB, Lyve-1high). A compensatory response of increasing putative collector lymphatics (subpopulation IIIA, Lyve-1low) appears to have ultimately failed when faced with the large volumes of blood mixed interstitial fluid. Analyzing the transcriptional profiles of the 4 endothelial clusters is important as it may lead reveal additional diagnostic markers of lymphatic disease and ultimately unveil how p53 orchestrates lymphatic disorders.
Mouse models of high p53 typically show p53-induced Caspase-dependent apoptosis in affected tissues13,33,35,47-49. Despite an overexpression of Puma and Noxa mRNA in our mutants (SFig. 6a), surprisingly Caspase-3 was not detected (data not shown). This observation does not preclude a Caspase-independent cell death, but the increase in cell cycle arrest due to p21 upregulation (Fig. 2e) points to a preferential mode of action of p53 in the lymphatic system. Considering the anti-proliferative function of p53 in tissues, it was not surprising that p53 overexpression induced a p21-dependent cell cycle arrest in mutant lymphatic endothelium (Fig. 2d and 2e), resulting in an insufficiency of lymphatic networking. Hence, the inability to efficiently drain the lymph from the interstitial tissues led to extreme cutaneous edema and vessel leakage in the mutants (Fig. 2c). The phenotypic concordance of both models and the near complete penetrance of their manifestations strengthens our conclusion that wild type p53 is the common culprit of these manifestations.
We observed that abnormalities associated with the loss of Mdm4 were much more accentuated than those with loss of Mdm2 (Figs. 1-3), which is the reverse of what is seen with Mdm2 and Mdm4 gene deletion models33-35. Mdm2, the main p53 inhibitor, is an E3 ubiquitin ligase that degrades p53, while Mdm4 does not38. Therefore, Mdm2 generally exerts a more powerful control on p53 stability and activity than its family member Mdm4; thus lack of Mdm2 typically affects health much more seriously than Mdm4 loss6,7,33,47,50. The increased severity of Mdm4 associated lymphedema and hemorrhaging may be due to intensified p53 expression at E13.5 in RP27M4 skin, a day earlier than in RP27M2 mice (Fig. 4a). Remarkably, Mdm2 levels in RP27M4 skin were low compared to WT skin, which may have further disrupted the negative regulation by Mdm4 and its interaction with Mdm2 to efficiently downregulate p53. An alternate explanation for this unexpected observation is a potential additive contribution of p53-independent functions of Mdm4 to the severity of the lymphatic anomalies. Nevertheless, both models support a relationship between p53 activation and lymphatic defects, largely through impeding proliferation via growth arrest (Figs. 2, 3).
Hirashima et al. (2008)51 identified the Apoptosis Stimulating Protein of p53 (Aspp1) as an endothelial-specific gene functioning during mouse embryonic development. Aspp1-null embryos exhibited an impaired cutaneous lymphatic drainage and edema that resolved in adulthood, suggesting that Aspp1, a promoter of the apoptotic activity of p53, plays a role a lymphangiogenesis. However, the function of Aspp1 in this system seem to be entirely p53-independent. In contrast, the phenotypes of our mutants are completely p53-dependent. Genetic deletion of one copy of p53 in both RP27M2 and RP27M4 mice reversed symptomatic lymphedema and hemorrhaging. RP27M2:p53+/- and RP27M4:p53+/- were born at a Mendelian ratio and lived normally with no overt pathologies (Fig. 5). Expression of p53 targets and lymphatic markers in RP27M2:p53+/- and RP27M4:p53+/- mice were restored comparable to WT levels (Fig. 5 and SFig. 6).
More importantly, pharmacologic control of p53 levels in both mutants successfully reversed symptoms of edema and hemorrhaging and mice were born following a Mendelian ratio. However, RP27M4 mice were born with a loose skin around their neck and did not survive past the first day perhaps since RP27M4 phenotypes were much more severe than those of RP27M2 mice. Starting PFT injections a day early or testing an increased range of PFT may replicate the same results we gained from treated RP27M2 mice. Nevertheless, attaining a symptomatic reversal in a genetic model of lymphedema with pharmacological means has not yet been established and the outcomes with PFT treatment appear in this context extraordinary and very promising for future translational research52,53.
Our findings were further corroborated in lymphedema-associated lymphatic disorders. p53 was detected at high levels specifically in the lymphatic endothelium and not in the venous or arterial endothelium of the same patient skin. From 8 cases, 6 disease tissue tested positive for p53, while normal neonatal and adult skin were negative for p53 in any endothelial tissue (Fig. 6 and SFig. 7). This observation once again corroborated findings from our mutants that show a striking resemblance to multiple mouse models of lymphedema and human lymphatic disease (Fig. 6e). Of note, Rpl27alow/+ embryos had some low level of cutaneous hemorrhaging and edema along with less dense lymphatics. Therefore, we cannot completely exclude the contribution of ribosomal stress to the pathogenesis of lymphatic defects. Coincidently, hydrops fetalis associated with the inherited Diamond-Blackfan anemia disorder (DBA) was highly linked to mutations in genes coding for ribosomal proteins. A subsequent p53 stabilization triggered by ribosomal stress was detected and a role for p53 in the pathogenesis of DBA was demonstrated54 suggesting that p53 is the common factor in lymphatic malformations associated with hydrops fetalis or lymphedema. Keeping p53 at bay is then imperative for the proper formation of the lymphatic network. After all, having no p53, as in p53-null mice, in general did not affect gestation8,55.
Together, our findings may ultimately lead to novel therapies of lymphedema in the absence of any current pharmacological avenue. Additionally, understanding the molecular determinants of normal lymphangiogenesis may have a bearing on other related diseases such as obesity, inflammation and cancer. This study, the first to highlight a role of p53 in lymphatic disease, indicates that the lymphatic system is particularly sensitive to high levels of p53 and that p53 overexpression in a subpopulation of endothelial cells disrupts the proper progression of lymphangiogenesis. Therefore, while p53 is not required for normal formation, p53 needs to remain restricted during development to circumvent its immediate response when p53 passes a certain threshold of cellular stressors. p53 would then trigger its typical anti-proliferative effects that promote symptomatic lymphedema. We think we know a lot about p53, yet remarkably, we still have much to learn about its functions in tissues, and in particular the lymphatic system. At the time when no medicinal therapy exists to treat lymphedema, p53 may once again come to the rescue.
Materials and Methods
Mice
All mice were maintained on a C57BL6/J background. For timed pregnancies, the first day of observed plug were recorded as day 0.5 post-coitum or embryonic day 0.5 (E0.5). Mice haploinsufficient for the ribosomal protein L27a (Rpl27low/+) express 20% lower levels of Rpl27a in the skin13. Rpl27alow/+ mice are crossed with Mdm27 or Mdm433 heterozygous mice (Mdm2+/- or Mdm4+/-) to generate WT, Rpl27alow/+, Mdm2+/-, Rpl27alow/+:Mdm2+/- (RP27M2), Mdm4+/-, and Rpl27alow/+:Mdm4+/- mice (RP27M4). To generate mice on a p53-deficient background, we crossed Rpl27alow/+ mice to Mdm2-/-:p53-/- or Mdm4-/-:p53-/- mice. We therefore obtained RP27M2:p53+/- mice and RP27M4:p53+/- mice. Genotypes were determined by PCR analysis of extracted DNA from tails using published primer sets for Mdm27, Mdm433, and p5356. The sex of embryos was determined by detection of Sry and Raspn genes by PCR57. We followed animal care and euthanasia guidelines of the Colorado Institutional Animal Care and Use Committee for all animal work.
Pifithrin-α drug Treatment
A 10mM stock of Pifithrin-α (PFT-α, Selleck Bio, cat. S2929) was diluted 1:10 in 1X PBS, protected from light and used instantly. Pregnant mice were injected intraperitoneally from E11.5 to E16.5 and subcutaneously from E17.5 to the day before delivery at 2.2 mg/kg of weight. Animal were monitored daily post-treatment and weights were recorded after delivery.
Histology and Immunostaining
Tissues were fixed in 4% neutral buffered paraformaldehyde, processed, and embedded in paraffin by the UCD Research Histology core. Sagittal sections (5µm) were subjected to immunofluorescence staining (IF) as previously described13 and according to the manufacturer’s recommendations. The primary antibodies used for IF were monoclonal mouse Prox-1 (1:50, P21936, Thermofisher Scientific, Massachusetts, USA), monoclonal rabbit Lyve-1 (1:100, ab14917, Abcam Inc., California, USA), polyclonal rabbit Ki-67 (1:1000, VP-K451, Vector Laboratories, California, USA) and mouse monoclonal p21 (1:100, sc-6246, Santa Cruz Biotechnology, California, USA) antibodies. We used anti-rabbit or anti-mouse Alexa Fluor 594 or Alexa Fluor 488 conjugated secondary antibodies (1:1000, Invitrogen, California, USA), captured the images on a Nikon Eclipse 90i, and quantified using the ImageJ software. Pediatric lymphatic edema was categorized using the ISSVA Classification of Vascular Anomalies and confirmed by PdPn (D2-40 antibody, Ventana or RnD Systems AF3670) staining. IHC for p53 on human tissues were performed using the D-07 antibody (Ventana) on automated strainers (Ventana Ultra) following the manufacturer’s recommendations and Clinical Laboratory Improvement Amendments (CLIA) certified procedures. For the visualization of the three-dimensional lymphatic vasculature, murine skin were fixed in 4% Paraformaldehyde (PFA) for 2-4 hours at room temperature for whole mount staining. Tissues were then subjected to procedures as published58. H&E was performed following Harris protocols for staining.
RNA Extraction and quantitative real time PCR
Mouse embryonic skins were placed in RNA Later (Sigma, cat. R0901) overnight at 4°C, then stored at −80°C until RNA was extracted. 5 mg of skin was homogenized in 20% 0.4M DTT in RLT Lysis Buffer (Qiagen, Hilden, Germany). The RNA was isolated using the RNeasy Plus Micro Kit (cat. 73404 and 74004, Qiagen, Hilden, Germany) and their corresponding protocol. Samples with an RNA concentration greater than 500 ng/μL and A280/260 ratio 1.8-2.0 were used for cDNA synthesis. cDNA of 100 ng/μL concentration was synthesized using the SuperScript III First Strand Synthesis SuperMix Kit (cat. 18080-051, ThermoFisher Scientific, Massachusetts, USA). qPCR was performed with Apex Probe Master Mix (cat. 42-116P, Genesee, California, USA), TaqMan Gene Expression Probes (Thermofisher Scientific, Massachusetts, USA), and mouse Gapdh (ref. 4352339E, ThermoFisher Scientific, Massachusetts, USA) used as reference. The reactions were run on a BioRad CFX96 Real Time C1000 Touch ThermoCycler and the gene expression fold change was determined via the ΔCT method59. The probes we used were designed for the following target genes: Lyve-1 (Mm00475056_m1), Prox-1 (Mm00435969_m1), c-Kit (Mm00445212_m1), Trp53 (Mm01731290_g1), Mdm2 (Mm01233136_m1), Bbc3 (Puma, Mm00519268_m1), and Pmaip1 (Noxa, Mm00451763_m1).
Flow Cytometry
Embryonic skins of E12.5-E15.5 dpc were harvested using a stereoscope and placed in EHAA media without L-glutamine (Irvine Scientific). Skin was cut into 1mm sized pieces and digested for 45 minutes at 37°C by 0.25 mg of Liberase DL (Roche) per mL of EHAA media and DNAse (Worthington). An equal volume of 0.1 M EDTA in Hank’s buffered saline solution without calcium or magnesium was added to the digested cells and incubated for 5 min at 37°C. Digested skin was passed through a 100µm strainer and washed with 5mM EDTA, 2.5% FBS in EHAA. Stromal cells were stained with CD45 (clone 30-F11), PdPn (clone 8.1.1), CD31 (clone 390), and Lyve-1 (clone 223322). Stromal cell subsets were identified by the expression of PdPn and CD31 and the lack of CD45 expression. Blood endothelium populations were classified as CD31mid or high PdPn− CD45−. In contrast, lymphatic endothelium cells were categorized as CD31+ PdPn+ CD45−. Cells were run on the DakoCytomation CyAn ADp flow cytometer (Fort Collins, CO) or BD FACS Canto II, acquired using Summit acquisition software and analyzed with FlowJo software (Tree Star, Ashland, OR).
Statistics
Statistical differences were analyzed using t tests, Chi-Square or one-way ANOVA on GraphPad Prism 8 software. A P value of 0.05 or lower was considered significant.
Acknowledgements
TT was funded by the NIH/NIAMS K01-AR 063203-01 award, the University of Colorado School of Medicine Dermatology Department, CCTSI (Child and Maternal Health Program) supported by the NIH/NCATS CTSA Grant UL1 TR002535 and the Dermatology Foundation. NB was supported by the NIH/NIAMS 1R03AR066880-01A1 and the Daneen and Charles Stiefel Investigative Science Award from the American Skin Association. Additional funding came from the Cancer Research Summer Fellowship and the Cancer League of Colorado to RM. We thank Dr. Ellen Elias for clinical advice and critical reading of the manuscript, Dr. Chiping Day for his meticulous review of the paper and Dr. Michael Detmar for his input and critical feedback. We would also like to thank Rachel Maxwell and Ellie Mackintosh for technical assistance. We are grateful to Drs. Tracy Lyons and Veronica Wessells for excellent technical assistance with the staining of normal human tissues, advice and crucial input on the manuscript. Histology of murine histology tissues was prepared by the Cancer Center Histology Core supported by the P30CA046934 grant funding. The MRI was performed by the University of Colorado Animal Imaging Shared Resource (AISR). We thank Dr. Jerrold Ward for detailed histopathological reading of the embryos. We apologize to those whose work we have not cited their work due to journal citation restrictions.
Footnotes
CONTACTS: Rohan Mylavarapu. Email: rohan.mylavarapu{at}colorado.edu., Molly Kulikauskas. Email: molly.kulikauskas{at}unc.edu., Cathrin Dierkes. Email: cathrin.dierkes{at}mpi-muenster.mpg.de., Nema Sobhani. Email: nema.sobhani{at}ucdenver.edu., Jeffrey Finlon. Email: jeffrey.finlon{at}cuanschutz.edu., Michelle Mangette. Email: michelle.mangette{at}cuanschutz.edu., Wanida Stevens. Email: wanida.stevens{at}cuanschutz.edu., Neil Box. Email: neil.box{at}cuanschutz.edu., Farinaz Arbab. Email: farinazarbab{at}netscape.net, Carrie Shawber. Email: cjs2002{at}cumc.columbia.edu., Ajit Muley. Email: am4289{at}cumc.columbia.edu., Mark Lovell. Email: mark.lovell{at}childrenscolorado.org, Friedemann Kiefer. Email: fkiefer{at}gwdg.de., Beth Tamburini. Email: beth.tamburini{at}ucdenver.edu., Tamara Terzian. Email: Tamara.Terzian{at}cuanschutz.edu.