GLP-1 Receptors are Enriched in the Lymphatic Endothelium and their Pharmacological Activation with Semaglutide Improves the Pumping Capacity of Lymphatic Vessels

Abstract

Objectives:

Recent clinical studies have suggested that glucagon-like peptide-1 receptor (GLP-1R) agonists may be effective therapies to treat or reduce the risk of developing secondary lymphedema. In this study, we aimed to 1) determine if GLP-1Rs are present in the lymphatic vasculature and characterize their expression, and 2) assess the effects of GLP-1R agonism on the contractile function of collecting lymphatic vessels.

Methods:

We assessed the expression of GLP-1Rs in and around the lymphatic vasculature by single-cell RNA sequencing and fluorescence confocal microscopy, and evaluated the direct effects of GLP-1R agonist, semaglutide, on modulating lymphatic contractility using pressure myography.

Results:

Expression of Glp1r (encoding GLP-1Rs) was detected exclusively in lymphatic endothelial cells. Pharmacological activation of GLP-1Rs led to robust vasodilation and an increase in the pumping capacity of isolated collecting lymphatics from WT, diet induced obese, and hypercholesterolemic ApoE KO mice. The GLP-1R-mediated response was in part facilitated by nitric oxide and its potential interaction with NaV channels; however, additional signaling remains to be elucidated.

Conclusions:

Our results revealed a direct, beneficial effect of GLP-1R agonism on lymphatic pumping capacity mediated by robust vasodilation, allowing lymphatics to accommodate and displace larger fluid volumes, while maintaining strong and highly efficient contractions.

Introduction

Lymphedema is a chronic, debilitating disease affecting >250 million people worldwide with no pharmacological therapies available^1–6. In the U.S. and other developed countries, the largest group of patients afflicted with secondary lymphedema are cancer survivors, who develop this disease following surgery, chemotherapy, radiotherapy, and hormonal therapy, or as a result of cancer progression itself^7,8. Depending on the type of malignancy, anatomical location, and type of treatments, the incidence of cancer-related lymphedema can vary largely^9,10. For instance, the estimated incidence of breast cancer-related lymphedema is approximately 25%; however, multiple factors including high body mass index (BMI) and insulin resistance are known to significantly increase the risk (>3-fold) of developing this disease^11–18. In fact, clinical studies have now demonstrated that obesity alone can cause secondary lymphedema^{11,12,14,16,17,19–21}. Studies using animal models have shown that obesity and metabolic syndrome-induced lymphatic insufficiency is associated with impaired pumping capacity, hyperpermeability, and valve dysfunction in collecting lymphatics, impaired immune cell migration, downregulation of lymphatic endothelial cell (LEC) specific markers, and increased infiltration of immune cells and inflammatory molecules surrounding lymphatic networks among others^22–37. The incidence of obesity-induced lymphedema continues to increase in association with the current obesity epidemic. A recent report by the Centers for Disease Control and Prevention (CDC) indicated that in the U.S., the prevalence of obesity (i.e., BMI≥30) in adults was 40.3%, and 1 in 10 adults were affected by severe obesity (i.e., BMI≥40)³⁸. Therefore, there is a critical need to identify the molecular mechanisms of lymphatic dysfunction in obesity to develop the first pharmacological therapeutics to treat, prevent, or reduce the risk of developing secondary lymphedema.

GLP-1 receptor (GLP-1R) agonists, originally developed to help control blood glucose levels in type 2 diabetes, have gained popularity for their effectiveness to reduce bodyweight, and were recently linked to metabolic improvement in atherosclerosis and dyslipidemia^39–43. Two recent publications by our collaborator and co-author Dr. Joseph Dayan have suggested that GLP-1R agonists may be an effective treatment for lymphedema and may reduce the risk of lymphedema in patients undergoing lymphadenectomy^44,45. While improvement in lymphatic function in these patients could be associated with an overall improved systemic health, i.e., weight management, improved metabolic function, and glycemic control, the case report by Crowley et al. pointed to a potential synergistic effect with lymphatic-specific agonism of GLP-1Rs. That study⁴⁵ reported on the case of a 44-year-old female patient who developed lymphedema following neoadjuvant chemotherapy, mastectomy, and axillary lymph node dissection and radiotherapy. This was a lean patient who, in association with adjuvant chemo and hormonal therapy, experienced a significant increase in bodyweight from 49.9 kg (BMI 19.2 kg/m²) to 66.3 kg (BMI 24.9 kg/m²), still in the healthy weight category. Her weight gain was resistant to diet and exercise intervention, and therefore, started receiving treatment with GLP-1R agonists. Initially with liraglutide, with limited weight loss response, and subsequently with semaglutide. Relevant to the central focus of this study, this patient lost 24% of her bodyweight in a period of 13 months, and more importantly, her lymphedema nearly completely resolved. An additional case report also demonstrated beneficial effects of GLP-1R agonism to reduce body weight, while improving fluid drainage and reducing limb circumference in a case of extreme obesity-induced massive, localized lymphedema⁴⁶. The emergence of promising clinical evidence has prompted a formal prospective study of GLP-1R agonists in the treatment of lymphedema (personal communication, J. Dayan and colleagues December 2025). Of equal and critical importance is obtaining preclinical data to gain insight into the underlying mechanisms of how GLP-1R agonists may treat lymphatic dysfunction and lymphedema. Evidence by recent studies have demonstrated that semaglutide is effective in animal models, including rodents, and its effects are indeed linked to activation of GLP-1Rs in a variety of cell types, including endothelial and smooth muscle cells^47,48. However, hardly anything is known about the expression and distribution of GLP-1Rs in the lymphatic vasculature, let alone the roles that these receptors may play in the regulation of the function of lymphatic vessels. Two important questions surface which underpin the core of this paper: (1) are there GLP-1Rs in the lymphatic vasculature? and (2) if so, is there any observable direct effect on lymphatic function when these vessels are exposed to a GLP-1R agonist?

In this study, we first characterized the expression of GLP-1Rs and their distribution throughout the different cell types that make up the wall of collecting lymphatic vessels and surrounding tissues by means of a single-cell RNA sequencing (scRNAseq) dataset recently generated and published by our group⁴⁹ and fluorescence confocal microscopy; then, we systematically assessed the specific effects of GLP-1R agonism on the contractile function of collecting lymphatic vessels from healthy C57BL/6J (WT) mice; and finally, we determined whether GLP-1R agonism could be employed to restore the contractile capacity of dysfunctional collecting lymphatics from diet-induced obese (DIO) and hypercholesterolemic ApoE KO mice, both models displaying multiple features commonly observed in metabolic syndrome.

Methods

Data Availability

Data and materials are publicly available in the main manuscript and within the Supplemental Materials. The referenced scRNAseq dataset was previously published by our group⁴⁹ and was made available on NIH NCBI Gene Expression Omnibus (GEO accession: GSE294684).

Animals

Mice.

C57BL/6J (Cat. No.: 000664, WT) male and female mice and B6.129P2-Apoe^tm1Unc/J (Cat. No.: 002052, ApoE KO) male mice were purchased from The Jackson Laboratory (Bar Harbor, ME) between 10–12 weeks old. Prox1-GFP mice were a gift from Dr. Young-Kwon Hong (Harvard Medical School). Prox1-CreER^T2;Salsa6f mice were generated by crossing Prox1-CreER^T2 (a gift from Dr. Taija Mäkinen, Wihuri Research Institute, Helsinki) and Salsa6f (expressing the genetically encoded calcium indicator GCaMP6f and a tdTomato reporter, which were purchased from The Jackson Laboratory, Cat. No.: 031968). In these mice, Cre-recombinase was induced by treatment with tamoxifen. Tamoxifen was prepared at a concentration of 10 mg/mL in sunflower seed oil (S5007–250ML; Sigma) with 10% v/v ethyl alcohol (E7023–500ML; Sigma). Reconstituted tamoxifen was administered via a single intraperitoneal injection at a dose of 1 mg. This single injection did not induce 100% recombination, instead only ~60% of the LECs displayed Salsa6f expression, which was sufficient for the purpose of these studies. A 2-week period was allowed before any experimental protocols were conducted on these animals. Mice were kept in a temperature-controlled environment on a 12-hour light/dark cycle, with unrestricted access to standard food and water (Cat. No. 5053; PicoLab Rodent Diet 20 - Irradiated). Mice were given a minimum of 2 weeks before starting experiments for acclimatization. A subset of WT male and female mice were ordered at 4 weeks of age and fed a western diet (WD) for 16 weeks starting at 5 weeks of age (Cat. No. D12079Bi, Research Diets). All mice used in this study were between 3–6 months old. Mice were anesthetized with isoflurane, weighed, shaved, and prepared for terminal procedures and tissue dissection. All animal experiments were conducted in accordance with an approved IACUC protocol No.: 1884.

Solutions and Chemicals

For dissection and cannulation of lymphatics, Krebs buffer was prepared with 146.9 mM NaCl, 5 mM D-glucose, 4.7 mM KCl, 3 mM NaHCO₃, 2 mM CaCl₂·2H₂O, 1.5 mM Na-HEPES, 1.2 mM MgSO4, 1.2 mM NaH₂PO₄·2H₂O, and 0.5% BSA (pH = 7.4). The buffer was sterile filtered and stored at 4°C. A separate Krebs buffer without BSA was used during pressure myography experiments for perfusion of the vessel chamber bath.

The following pharmacological agents were used for pressure myography experiments: GLP-1R agonist, semaglutide (1nM-1μM, Sema), L-NAME (100μM), indomethacin (10μM, Indo), apocynin (100μM, Apo), and tetrodotoxin (1μM, TTX), and exendin (9–39) (100nM, Ex(9–39)).

Chemicals and reagents were purchased from Sigma-Aldrich, TOCRIS, and MedChemExpress and these are listed in a Major Resources Table (Supplemental Materials).

Microdissection of Collecting Lymphatic Vessels and Pressure Myography

Inguinal axillary collecting lymphatic vessels were microdissected from the mouse as previously described^23,49–53. A Sylgard coated dissection dish filled with BSA-containing Krebs buffer was used for pinning down tissues for fine dissection, where the majority of the adipose and connective tissues were removed. The lymphatic vessels were transferred into a custom-made pressure myography chamber with Krebs buffer, cannulated on glass micropipettes, pressurized equally at 3 cmH₂O, and equilibrated with bath perfusion at 37°C for 30 minutes before experimentation. No-flow conditions were used in all pressure myography experiments (i.e., equal inflow and outflow pressures). For fine control of intraluminal pressures, OB1 MK3 microfluidic flow control systems (Elveflow, Paris) were used. Brightfield video recordings of the contractile activity of lymphatic vessels, as well as live tracking of inner and outer diameter changes were automatically recorded at 20 fps using custom-written Python-based programs developed by Dr. Castorena.

Pump Tests

Quantitative assessment of the pumping capacity of collecting lymphatic vessels was achieved by pump tests using lymphatic vessel segments containing 2 valves (i.e., one full lymphangion or pumping unit) as previously described^50,54. Inguinal axillary lymphatic vessels display low density of intraluminal valves which prevented us from obtaining 2-valve segments of manageable length for pressure myography experimentation; therefore, pump tests were conducted in popliteal afferent collecting lymphatic vessels, where valves are commonly found approximately every ~1mm. Each 2-valve lymphatic segment was equilibrated for 30 min under no flow conditions at an intraluminal pressure of 3 cmH₂O and then decreased to 2 cmH₂O for 5 minutes prior to starting the pump test pressure protocol. The contractile activity of each lymphatic vessel was initially recorded with both inflow and outflow pressures set at 2 cmH₂O for 1 minute, then outflow pressure was increased to 3 cmH₂O, creating a sudden adverse pressure gradient of 1 cmH₂O, which normally forced the outflow valve to close and prevent backflow. As the lymphatic continued to spontaneously contract, the propulsive pressure generated within the middle lymphangion transiently opened the outflow valve, allowing fluid to be transported forward. The position of the outflow valve was continuously monitored. Outflow pressure was then progressively increased by 0.5 cmH₂O every minute until the propulsive pressure generated by each contraction was not sufficient to transport fluid against the imposed adverse pressure gradient and the outflow valve remained consistently closed. The maximum outflow pressure that was successfully overcome by the contractile capacity of a lymphatic vessel defined its pump limit. A representative example of this experimental protocol is shown in Figure 3K–M.

Contractile Parameters

Following experimentation, additional custom Python-based tools were used for the automated processing and analysis of contractile activity of lymphatic vessels (i.e., live-tracked changes in diameter as a function of time). As our previous studies^{49–53,55–62} have demonstrated, these analyses include calculating the mean values of contraction amplitude (calculated as EDD-ESD), end-diastolic diameter (EDD), end-systolic diameter (ESD), ejection fraction (EF, calculated as (EDD²-ESD²)/EDD²), fractional pump flow (FPF, calculated as EF multiplied by Contraction Frequency), contraction frequency, and width of contraction. For this study, we have additionally calculated the mean volume displaced per contraction, since the volume of fluid transported is an important metric for lymphatic function, particularly in lymphedema patients, where an improvement in volume transported would be extremely beneficial towards lessening fluid accumulation and increasing the amount of fluid shuttled back into the circulatory system. Volume displaced per contraction is determined via calculating the volume of the vessel during diastole (using an average length of a lymphangion of ~1mm, which is the length of the pumping unit of a lymphatic vessel) multiplied by the ejection fraction. Volume displaced is reported in nL of fluid. An important assumption when calculating volume displaced is valve competence, which allows for the generation of propulsive forward pressure.

Single Cell RNA Sequencing Data Analysis

Analyses were performed using Seurat (4.3.0.1) in R studio (RStudio 2023.06.0+421). For our previously published scRNAseq dataset (see Data Availability section)⁴⁹, we had performed the following quality control, cells with feature counts in the range 1000–6250 and cells displaying less than 5% mitochondrial counts were selected. Following quality control, the dataset underwent pipeline processing that included normalization using VST, multi-dimensional reduction, Harmony-based batch correction (with dims.use=1:30, representing the principal components to be utilized), cell clustering (with cluster resolution=0.8), differential testing framework, and visualization.

Fluorescence Confocal Microscopy

Inguinal axillary lymphatic vessels from Prox1-CreER^T2;Salsa6f were carefully dissected and prepared for pressure myography and live vessel imaging. Lymphatic vessels were pressurized at 3 cmH₂O and equilibrated for 30 minutes at 37°C prior to imaging. Using confocal microscopy, we recorded the intracellular calcium activity of LECs under no flow conditions pre- and post-GLP-1R agonism with a single dose of semaglutide (5nM). For analysis, the fluorescence intensity per frame in 3 regions of interest (ROIs) was integrated. Each ROI usually included 1–2 LECs.

To evaluate the perivascular innervation in collecting lymphatic vessels, lymphatics from multiple anatomical regions were carefully dissected from Prox1-GFP mice, fixed while cannulated and pressurized using a 4% PFA solution and then stained for β3 tubulin (1:300), wheat germ agglutinin (WGA, 1:200), and DAPI (1:1000). See Supplemental Materials for detailed information on these reagents and antibody. Z-stack images were collected with 25X water immersion objective, and max intensity projection images were then generated (Figure 5).

Statistical Analyses

Results were analyzed with GraphPad Prism Version 10.6.1. Paired parametric t-test, unpaired parametric t-test with Welch’s correction (with no assumption of equal variance), two-way ANOVA corrected with Tukey test for multiple comparisons, and one-way ANOVA test that included a correction for multiple comparisons using Dunnett’s test and a Geisser-Greenhouse correction as no equal variances were assumed (i.e., sphericity was not assumed) were used to determine differences in contractile function parameters, with statistical significance set at p<0.05. Statistical tests were specified in the captions of each figure legend. For pressure myography experiments, the number “n” of experiments corresponds to each individually tested collecting lymphatic segment. Depending on the group or protocol, mice numbers ranging from 3 to 10 were utilized; and frequently, multiple lymphatic segments were tested per mouse (See Supplemental Materials).

Results

GLP-1Rs are Expressed in the Lymphatic Vasculature and are Highly Enriched in the Lymphatic Endothelium

We previously performed single-cell RNA sequencing (scRNAseq) using collecting lymphatic vessels from male and female WT mice (n=4 mice per group). Briefly, we prepared single-cell suspensions from individually microdissected lymphatic vessels and surrounding connective tissues. From each mouse, collecting lymphatic vessels from multiple anatomical regions were pooled into a single sample (i.e., inguinal axillary (2 vessels), superficial cervical (4 vessels), popliteal afferent (4 vessels), and mesenteric (4 vessels)) in order to reach the minimum cell-count for scRNAseq experimentation. Automated clustering revealed 29 cell identity clusters, and these were associated with 17 different cell types. We found that Glp1r expression was exclusively detected in a subset of lymphatic endothelial cells (LECs, Figure 1A), with approximately 19.8±2.4% of LECs displaying detectable expression of Glp1r. Furthermore, there were no differences in Glp1r gene expression between lymphatics from female and male mice (Figure 1B). To investigate transcriptomic differences in Glp1r-expressing LECs vs. those that did not express this gene (or its expression was not detectable), we performed a differential gene expression (DGE) analysis. In LECs expressing Glp1r, DGE analysis identified a set of 10 differentially enriched (i.e., upregulated) genes, including Ptgs1, Fabp4, Ptn, Fabp5, Adamts9, Ndufa8, Naaa, Mir100hg, Lsr, and Nedd4l; while only 2 genes (i.e., Ccl21a and Sh3gl3) displayed significant downregulation (Figure 1C). Using the list of 10 differentially upregulated genes, a gene ontology (GO) analysis revealed several important molecular functions and biological processes involving fatty acid binding and transport, and prostanoid synthesis, in addition to roles in barrier integrity and metabolic processes (Figure 1D).

Figure 1. Expression of Glp1r in murine collecting lymphatic vessels.

(A) Percent of Glp1r-expressing cells per cell identity; (B) Average gene expression of Glp1r in LECs from WT female and WT male mice respectively; (C) Differentially expressed genes in LECs expressing Glp1r vs. LECs where expression of Glp1r was not detected; (D) Gene ontology analyses displaying the top molecular functions and biological processes associated with the differentially upregulated genes shown in panel C; (E) UMAP displaying the different LEC subtypes contained within the employed scRNAseq dataset; (F) DotPlot displaying the expression of Glp1r, as well as known markers of various LEC-subtypes (in this panel, the percent of cells expressing a given feature and the average gene expression are encoded in the size and color of dots respectively); (G) Percent of LECs expressing Glp1r in clusters linked to collecting (collLECs), pre-collecting (precLECs), or capillary (capLECs) lymphatics (cluster numbers are indicated under the horizontal axis); (H) Average intensity projection of the GFP signal associated with the expression of the genetically encoded calcium sensor GCaMP6f in a cannulated and pressurized collecting lymphatic vessels from a Prox1-CreERT2;Salsa6f (i.e., GCaMP6f/tdTomato) mouse; and (I) Representative fluorescence intensity traces (from n=4 vessels) of the GCaMP6f signal in 3 different regions of interest (i.e., ROIs) under control conditions and following stimulation with 5 nM semaglutide.

Using our scRNAseq dataset, we then performed sub-clustering of the LECs identity to determine the presence of different LEC-subtypes. Our analysis identified 6 clusters (i.e., 0–5) with unique transcriptomic profiles (Figure 1E). Using markers previously reported to be differentially expressed in the different subtypes of LECs (e.g., capillary, collecting, pre-collecting, valve, etc.)⁶³, we confirmed the presence of LECs from 1) Clusters 0–2: collecting lymphatics (including valve LECs); 2) Cluster 3: initial/capillary lymphatics; and 3) Clusters 4–5: precollecting lymphatics (Figure 1F). Glp1r expression was enriched in LECs from collecting and pre-collecting lymphatics (Figure 1F,G). Interestingly, Glp1r expression was not detected in LECs from the capillary region (Figure 1F,G). The expressions of Foxc2, Gata2, Gja4, and Lama5 suggested that Cluster 1 may encompass LECs from lymphatic valves, which also displayed expression of Glp1r.

To validate the presence of functional GLP-1Rs in the LECs from the mouse, we generated Prox1-CreER^T2;Salsa6f mice expressing the genetically encoded calcium sensor GCaMP6f. Using inguinal axillary lymphatic vessels from these mice (n=4) we assessed the intracellular calcium response of the lymphatic endothelium to stimulation with the GLP-1R agonist semaglutide. Consistent with previous reports from our group⁵⁰, LECs were very quiescent under control conditions, with very rare discrete events being detected; however, upon stimulation with 5nM semaglutide, LECs displayed an increase in both localized events, as well as overall global intracellular calcium (Figure 1H,I).

Pharmacological Activation of GLP-1Rs Modulates the Contractile Capacity of Collecting Lymphatic Vessels in a Concentration-Dependent Manner

Initially developed for the treatment of type 2 diabetes, GLP-1R agonists have been successful for the treatment of obesity and improving weight loss, especially for patients whose obesity is resistant to diet and exercise alone. With the tight and bidirectional association between obesity worsening lymphatic dysfunction and lymphatic dysfunction worsening obesity, there is a critical need to assess the role of GLP-1R agonists in lymphatic function. Impaired pumping capacity of collecting lymphatic vessels contributes to abnormal interstitial fluid drainage and fluid accumulation. Therefore, we assessed the concentration-dependent effects (in the range 1–1000nM) of the GLP-1R agonist, semaglutide, in isolated, cannulated, and pressurized collecting lymphatic vessels. Our results demonstrated that collecting lymphatic vessels are exquisitely sensitive to GLP-1R agonism, as indicative by the robust vasodilation and transient loss of contractions observed even when vessels are exposed to 1nM semaglutide (see representative diameter trace from a concentration response experiment in Figure 2A). Contractions were usually fully restored 30–60 seconds after the initial vasodilation. Contraction amplitude was not affected by the increasing concentration of semaglutide (Figure 2B). Lymphatic vessels continued to vasodilate in a concentration-dependent manner, as shown by the increasing mean end diastolic and end systolic diameters (i.e., EDD and ESD in Figure 2C,D). Similar to contraction amplitude, mean ejection fraction and mean contraction width (at half-maximum) were also unchanged (Figure 2E,G). Contraction frequency and, therefore, the calculated fractional pump flow (FPF) were significantly reduced with each increasing concentration (Figure 2F,H). In this study, we incorporated a new calculated parameter that represents the mean volume displaced per contraction; this parameter was calculated for a 1-mm long lymphatic segment, which approximates the length of a lymphangion or pumping unit, and it is expressed in nL (see Methods section for more information on the calculation of this parameter). Importantly, the mean volume displaced per contraction was significantly increased upon stimulation with semaglutide (Figure 2I). These results demonstrated that while GLP-1R agonism resulted in reduced contractile frequency, lymphatic vessels could accommodate larger fluid volumes (i.e., vasodilated) and maintained strong and potentially more efficient contractions displacing larger fluid volumes.

Figure 2. Concentration response curve of GLP-1R agonist, semaglutide on lymphatic vessels from WT mice.

(A) Representative diameter trace of an inguinal axillary lymphatic vessel displaying the functional response to increasing concentrations of semaglutide in the range of 1nM to 1μM; the contractile activity was assessed for 5 minutes at each concentration; (B-I) Mean contractile parameters as a function of semaglutide concentration (expressed as mean±SEM), including (B) amplitude, (C) end diastolic diameter (EDD), (D) end systolic diameter (ESD), (E) ejection fraction, (F) contraction frequency, (G) width, (H) fractional pump flow (FPF), and (I) volume displaced. A total of n=17 vessels were included and a one-way ANOVA test that included a correction for multiple comparisons using Dunnett’s test and a Geisser-Greenhouse correction as no equal variances were assumed (i.e., sphericity was not assumed) were used to determine differences in contractile function parameters, with statistical significance set at p<0.05.

Acute Treatment with Semaglutide Improves the Pumping Capacity of WT Collecting Lymphatic Vessels

In the previous section, our data demonstrated that the contractile activity of lymphatic vessels was significantly modulated by semaglutide at concentrations as low as 1nM; and we determined that the maximal beneficial effects (i.e., maximum vasodilation and volume displaced), while maintaining rhythmic contractility, were achieved at concentrations in the range of 1–10nM. Therefore, we examined and characterized the functional responses in WT inguinal axillary lymphatic vessels following a single acute stimulus with 5nM semaglutide (while constant superfusion with Krebs buffer was maintained). The contractile activity was recorded and analyzed for 2 minutes under control conditions and following stimulation with semaglutide. A representative trace from a total of n=11 vessels is shown in Figure 3A. Consistent with our concentration response experiments, following a single, acute dose with 5nM semaglutide, the contraction amplitude of lymphatics was not significantly changed, although there was a trend for amplitude to be increased after treatment (Figure 3B); lymphatic vessels displayed robust vasodilation as indicated by the increased end diastolic (EDD) and end systolic (ESD) diameters (EDD 108.9±6.8μm (Ctrl) vs. 124.3±8.0μm (Sema), p<0.0001, and ESD 62.95±4.3μm (Ctrl) vs. 74.53±5.8μm (Sema), p<0.0032)(Figure 3C,D); ejection fraction and contraction width (at half-maximum) remained unchanged (Figure 3E,G); contraction frequency was significantly decreased after GLP-1R agonism with semaglutide (i.e., 19.23±0.98 contractions/minute (Ctrl) vs. 12.13±0.84 contractions/minute (Sema), p<0.0001) and so did the calculated fractional pump flow (Figure 3F,H); however, the calculated volume displaced by each contraction (Figure 3I) was significantly increased after a single 5nM stimulus with semaglutide (i.e., 6.91±1.70nL (Ctrl) vs. 8.63±2.03nL (Sema), p<0.001), which represented an increase in volume displaced of ~25%. To further exemplify this increase in pumping capacity, Figure 3J depicts the calculated volume displaced for each individual contraction in the representative trace being shown in Figure 3A. Note that in the initial 2-minutes under control conditions, each contraction displaced ~7.5nL, and this was significantly increased to >10nL after stimulation with semaglutide. Mean values were then calculated for all contractions within the control period, as well as following treatment with semaglutide, and these are represented as a single datapoint in Figure 3I. In many anatomical regions of the human body, lymph must be transported against an adverse pressure gradient imposed by gravitational forces. To determine if the slower frequency contractions capable of displacing larger fluid volumes following stimulation with semaglutide were indeed indicative of more efficient contractions (i.e., improved pumping capacity), we performed pump tests on 2-valve lymphatic segments, i.e., one lymphangion or pumping unit (see description in the Methods section and Figure 3K–M) to quantitatively determine the maximum outflow pressure that could be successfully overcome by the propulsive pressure generated by each contraction, i.e., pump limit. The pump limits of collecting lymphatics (n=7) were determined under control conditions and following acute treatment with a 5nM single dose of semaglutide. For the representative example shown in Figure 3L, as the outflow pressure was progressively increased, each contraction was capable of transporting fluid against the imposed pressure gradient, as indicative by the transient opening of the outflow valve following each contraction, until reaching its pump limit of 4.5 cmH₂O. At this pressure, the outflow valve was observed to successfully open in 8 out of 13 (~62%) contractions. Once the outflow pressure was increased to 5 cmH₂O, all contractions within 1 minute failed to transiently open the outflow valve. The pumping capacity of collecting lymphatic vessels was consistently and significantly increased following treatment with semaglutide (4.64±0.42cmH₂O (Ctrl) vs. 7.57±0.88cmH₂O (Sema), p<0.0084) (Figure 3M).

Figure 3. Acute stimulation with a 5 nM semaglutide single dose improves the pumping capacity of lymphatic vessels.

(A) Representative diameter trace displaying the contractile activity of an inguinal axillary lymphatic vessel from a WT mouse under control conditions and following stimulation with a single 5 nM dose of semaglutide; (B-I) Summary data of different contractile parameters before (Ctrl) and after a single 5nM stimulus with semaglutide (Sema), including (B) amplitude, (C) end diastolic diameter (EDD), (D) end systolic diameter (ESD), (E) ejection fraction, (F) contraction frequency, (G) width, (H) fractional pump flow (FPF), and (I) volume displaced. These data are presented as means±SEMs. Data includes a total of n=11 vessels and statistical significance (set at p<0.05) was assessed via a paired parametric T test; (J) Volume displaced for each individual contraction in panel A; (K) Diagram depicting the experimental setup for Pump Tests, which directly and quantitatively assess the pumping capacity of a given lymphangion; (L) Representative experimental protocol for a Pump Test, inflow and outflow pressures (upper panel), tracked inner diameter (middle panel), and outflow valve position (bottom panel, C=closed and O=open); and (M) Pump Limit (i.e., maximum outflow pressure at which the outflow valve was detected to successfully open during at least one contraction) assessed under control conditions and following a single 5nM-stimulus with semaglutide. Data includes pump tests in a total of n=7 popliteal afferent lymphatic vessels. A paired parametric T test was used to determine significance, with statistical significance set at p<0.05.

Semaglutide-Induced Increase in the Pumping Capacity of Collecting Lymphatics is Only Partially Mediated by Nitric Oxide Synthase (NOS)-Dependent Increase in Nitric Oxide

To determine the mechanism by which semaglutide improved the pumping capacity of collecting lymphatic vessels, and to determine the potential, suspected involvement of NOS-dependent nitric oxide (NO), we assessed the functional responses of lymphatic vessels from WT mice to 5nM semaglutide following a 20-minute pre-treatment with the NOS inhibitor L-NAME (100μM) (Figure 4). Compared to control conditions, superfusion with a Krebs buffer containing L-NAME resulted in an increase in basal vessel tone, as indicative by the reduced EDD (90.23±3.90μm Ctrl vs. 67.89±3.94μm L-NAME, p<0.0020) and ESD (61.31±4.41μm Ctrl vs. 41.94±2.60μm L-NAME, p<0.0009), which is consistent with removal of basal nitric oxide levels (Figure 4C,D), and a decrease in volume displaced (3.42±0.44nL Ctrl vs. 2.29±0.41nL L-NAME, p<0.0223) (Figure 4I); while contraction amplitude, ejection fraction, frequency, width, and fractional pump flow (FPF) remained unchanged (Figure 4B,E–H). In the presence of L-NAME, semaglutide (5nM) still induced a noticeable response in lymphatic contractility, including increased contraction amplitude, ejection fraction, and volume displaced; while contraction frequency was significantly decreased (Figure 4B,E,F, and I). Lymphatic vessels also continued to display residual vasodilation as shown by the changes in EDD, i.e., 67.89±3.94μm (L-NAME) vs. 75.07±4.51μm (L-NAME+Sema), p<0.0484); however, these were still significantly smaller than those in control conditions, i.e., 90.23±3.90μm Ctrl vs. 75.07±4.51μm L-NAME+Sema, p<0.0303 (Figure 4C). In particular, the observed residual dilation and decrease in contractile frequency despite the presence of L-NAME pointed to the effects of semaglutide extending beyond the activation of the eNOS pathway. An interesting and noteworthy observation was that the initial, transient pause in contractility usually displayed by lymphatics following stimulation with semaglutide was prevented in the presence of L-NAME, suggesting that the transient loss of contractility during the initial phase of the robust vasodilation induced by semaglutide could be associated with transient, localized, and likely high concentrations of NO being diffused and reaching the lymphatic muscle cell (LMC) layer and interfering with its pacemaking activity. Furthermore, while the calculated volume displaced per contraction after treatment with semaglutide was significantly increased from that in the presence of L-NAME (i.e., 2.30±0.41nL L-NAME vs. 2.94±0.41nL L-NAME+Sema, p<0.0102), this was not significantly different from that under control conditions (i.e., 3.42±0.47nL Ctrl vs. 2.94±0.41nL L-NAME+Sema, p<0.2804) (Figure 4I), suggesting a partial but necessary contribution of NO in the semaglutide-induced increase in the pumping capacity of collecting lymphatics.

Figure 4. Role of nitric oxide (NO) in the semaglutide-mediated modulation of lymphatic contractile function.

(A) Representative diameter trace displaying the contractile activity of an inguinal axillary lymphatic vessel from a WT mouse under control conditions, during a 20-minute pre-treatment with L-NAME (100μM), and after stimulation with a single 5 nM dose of semaglutide; and (B-I) Summary data of different contractile parameters under these 3 paired conditions, including (B) amplitude, (C) end diastolic diameter (EDD), (D) end systolic diameter (ESD), (E) ejection fraction, (F) contraction frequency, (G) width, (H) fractional pump flow (FPF), and (I) volume displaced. These data are presented as means±SEMs. A total of n=8 vessels from WT mice were used for these experiments. A one-way ANOVA corrected with a Geisser-Greenhouse correction as no equal variances were assumed (i.e., sphericity was not assumed) was used to determine differences in contractile function parameters, with statistical significance set at p<0.05.

Potential Involvement of Other Paracrine Signals in the Semaglutide-Induced Increase in Lymphatic Pumping Capacity

Our results presented in the previous section suggested that NOS-dependent NO may play a crucial role in partially mediating the semaglutide-induced lymphatic vasodilation; however, these results also pointed to the presence of other signaling components, perhaps other paracrine signaling originating from the lymphatic endothelium and reaching LMCs. Some prostanoids, as well as reactive oxygen species (ROS) are known to play critical roles in vascular physiology, including vasodilation, in health and disease. There is also the possibility that GLP-1R signaling modulates additional NO production in LECs independently of the canonical eNOS pathway. Therefore, to determine the potential contribution of prostanoids and/or NADPH oxidase-mediated ROS, we performed additional experiments on isolated collecting lymphatic vessels exposed to 5nM semaglutide (single dose) following a 20-minute pre-treatment and in the presence of either indomethacin (COX-1/2 inhibitor, 10μM, n=8), apocynin (NADPH oxidase (NOX) inhibitor, 100μM, n=13), or with both inhibitors in combination with L-NAME (i.e., L-NAME+Indomethacin+Apocynin, n=11), to explore the synergistic effects of pharmacological inhibition of NOS, COX-1/2, and NOX-mediated ROS. While some statistically significant differences were identified that suggested that basal levels of prostanoids and ROS may be key modulators of lymphatic contractility, our results were not strongly conclusive regarding the involvement of prostanoid secretion and ROS production as a result of stimulation with semaglutide. Irrespective of the inhibitor (i.e., indomethacin or apocynin), changes in contractile activity remained following stimulation with semaglutide (Supplemental Figures 1 and 2). And in lymphatic vessels pre-treated with L-NAME, indomethacin, and apocynin, a residual response, similar to that observed in the experiments with L-NAME alone, was observed (Supplemental Figure 3). Interestingly, the presence of indomethacin also prevented the transient loss of all contractility during the initial phase of the vasodilation induced by semaglutide, suggesting a potential, subtle involvement of secreted prostanoids, which requires further investigation (Supplemental Figure 1).

Perivascular Innervation and Potential Contribution of Voltage-Gated Sodium Channels in the Semaglutide-Mediated Signaling in Collecting Lymphatics

A known effect of GLP-1R agonists is the decrease in gut motility, which results from the activation of GLP-1Rs in surrounding nerves. A major limitation for our study has been the lack of reliable antibodies to determine the localization of GLP-1Rs at a protein level (e.g., via immunofluorescence). Our scRNAseq, calcium imaging, and functional experiments presented in previous sections validated the presence of functional GLP-1Rs in LECs; however, the extent of perivascular innervation and their potential involvement in the semaglutide-mediated regulation of lymphatic contractility remained unexplored. Nerve cell bodies were not captured in our scRNAseq likely due to their loss during the enzymatic digestion process. Therefore, we isolated, cannulated, pressurized, and fixed inguinal axillary, mesenteric, superficial cervical, and popliteal afferent collecting lymphatic vessels from Prox1-GFP mice (n=5, expressing a GFP reporter in all LECs), and assessed the presence of perivascular nerves by staining with the cell membrane and neuron tracer WGA (wheat germ agglutinin), as well as with tubulin beta 3 (TUBB3) a structural component of microtubules commonly found in neurons and nerves. Our immunofluorescence observations demonstrated a complete absence of perivascular nerves (or at least not detected in the 5 ~2–3mm-long segments included in this study) in inguinal axillary lymphatic vessels (Figure 5A–D), while perivascular nerves were observed consistently in lymphatics from other anatomical regions. Perivascular innervation was particularly robust in mesenteric and superficial cervical lymphatics (Figure 5E–P). In this study, all the contractile responses, with the exception of pump tests (Figure 3K–M), were assessed in inguinal axillary lymphatic vessels. These findings suggested that the residual vasodilatory response observed in inguinal axillary lymphatic vessels in the presence of L-NAME and following stimulation with semaglutide (5nM) was not associated with perivascular nerve activation. Whether GLP-1Rs are expressed in perivascular nerves in other anatomical regions was not assessed at this point but will be the focus of future studies from our laboratory.

Figure 5. Perivascular nerves surrounding collecting lymphatic vessels from various anatomical regions.

Collecting lymphatic vessels from a Prox1-GFP (green) mouse from the inguinal axillary (A-D), mesenteric (E-H), superficial cervical (I-L), and popliteal (M-P) regions. Vessels were stained for tubulin β3 (TUBB3, red), wheat germ agglutinin (WGA, cyan), and DAPI (blue). Although not specific for perivascular nerves, WGA is highly enriched in TUBB3-expressing perivascular nerves. These are representative images from n=5, for each experiment images were collected in 3 regions of interest. Scale bar equals 50 μm.

To further validate our immunofluorescence observations, i.e., suggesting negative involvement of perivascular nerves potentially expressing GLP-1Rs, and given that voltage-gated sodium (NaV) channels are critical for the rapid depolarization of the nerve cell membrane, we assessed the functional responses of lymphatic vessels to semaglutide following a 20-minute pretreatment and in the presence of the NaV antagonist, tetrodotoxin (TTX, 1μM). A representative trace is shown in Figure 6A. Inhibition of NaV channels with TTX did not impact any of the calculated lymphatic contractile parameters (Figure 6B–I); however, as anticipated and consistent with our results from Figure 3B–I, semaglutide induced robust vasodilation (including a transient loss of contractility), with both EDD (99.98±5.89μm TTX vs. 112.20±6.53μm TTX+Sema, p<0.0003) and ESD (70.40±5.99μm TTX vs. 78.21±6.74μm TTX+Sema, p<0.0139) significantly increasing (Figure 6C,D), a significant decrease in both contraction frequency (19.73±0.90 contractions/minute Ctrl vs. 11.90±1.56 contractions/minute TTX+Sema, p<0.0001) and fractional pump flow (9.58±0.86 Ctrl vs. 5.85±0.83 TTX+Sema, p<0.0003), and a significant increase in volume displaced per contraction (3.83±0.39nL Ctrl vs. 5.05±0.71nL TTX+Sema, p<0.0467). Contraction amplitude, ejection fraction, and width were all unchanged (Figure 6).

Figure 6. Role of voltage-gated sodium channels in the semaglutide-mediated modulation of lymphatic contractile function.

(A) Representative diameter trace displaying the contractile activity of an inguinal axillary lymphatic vessel from a WT mouse under control conditions, during a 20-minute pre-treatment with tetrodotoxin (1μM, TTX), and after stimulation with a single 5 nM dose of semaglutide. (B-I) Summary data of different contractile parameters under these 3 paired conditions, including (B) amplitude, (C) end diastolic diameter (EDD), (D) end systolic diameter (ESD), (E) ejection fraction, (F) contraction frequency, (G) width, (H) fractional pump flow (FPF), and (I) volume displaced. These data are presented as means±SEMs. A total of n=12 vessels from WT mice were used for these experiments. A one-way ANOVA corrected with a Geisser-Greenhouse correction as no equal variances were assumed (i.e., sphericity was not assumed) was used to determine differences in contractile function parameters, with statistical significance set at p<0.05.

For completion, we decided to perform additional experiments to assess the contractile responses to semaglutide following pharmacological inhibition of NOS and NaV channels using L-NAME and TTX. Our hypothesis was that based on the results reported in Figure 6, stimulation with semaglutide in the presence of L-NAME+TTX would result in a response similar to that observed in the presence of L-NAME alone (Figure 4). Consistent with data show in Figure 4, a 20-minute pretreatment with L-NAME+TTX resulted in a significant increase in tone, as indicated by the significantly decreased EDD and ESD (Figures 7C,D) and a significant decrease in volume displaced (Figure 7I). Interestingly, the synergistic inhibitory effects of L-NAME+TTX prevented most of the subsequent response mediated by semaglutide, i.e., no significant changes observed between L-NAME+TTX vs. L-NAME+TTX+Sema, except for a decreased in contraction frequency (Figure 7). These unexpected results suggested the interaction of NO and NaV channels as a potential component in the likely complex GLP-1R signaling pathway(s) mediated by semaglutide. To guide future studies, we explored the expression of genes encoding NaV channels using our scRNAseq dataset (Figure 7J). In LECs, where enrichment of GLP-1Rs was identified, Scn5a and Scn3a (which encode NaV1.5 and NaV1.3, respectively) displayed the highest expression. LMCs displayed high enrichment of Scn3a (NaV1.3), followed by Scn2a (NaV1.2).

Figure 7. Combinatory role of NO and voltage-gated sodium channels in semaglutide-mediated modulation of lymphatic contractile function.

(A) Representative diameter trace displaying the contractile activity of an inguinal axillary lymphatic vessel from a WT mouse under control conditions, during a 20-minute pre-treatment with L-NAME (100μM) and tetrodotoxin (1μM, TTX), and after stimulation with a single 5 nM dose of semaglutide; (B-I) Summary data of different contractile parameters under these 3 paired conditions, including (B) amplitude, (C) end diastolic diameter (EDD), (D) end systolic diameter (ESD), (E) ejection fraction, (F) contraction frequency, (G) width, (H) fractional pump flow (FPF), and (I) volume displaced. These data are presented as means±SEMs. A total of n=9 vessels from WT mice were used for these experiments. A one-way ANOVA corrected with a Geisser-Greenhouse correction as no equal variances were assumed (i.e., sphericity was not assumed) was used to determine differences in contractile function parameters, with statistical significance set at p<0.05; and (J) Expression of genes encoding voltage-gated sodium channels in the different cell types within and surrounding the wall of collecting lymphatic vessels using scRNAseq.

Pharmacological Inhibition of GLP-1Rs using Exendin (9–39) does not Completely Inhibit the Lymphatic Contractile Response Induced by Semaglutide

Thus far, our results had implicated NO and a potential interaction with NaV channels; however, most of our additional explored mechanistic components yielded negative or inconclusive data. Therefore, we decided to determine if pharmacological inhibition of GLP-1Rs effectively and completely prevented all the response induced by semaglutide. Consistent with previous protocols, we assessed the contractility of inguinal axillary lymphatic vessels under control conditions, following 20-minute pre-treatment with the competitive antagonist of GLP-1Rs, exendin (9–39) (Ex(9–39), 100nM), and after acute stimulation with semaglutide (5nM). While most of the contractile parameters remained unaffected by inhibition of GLP-1Rs by exendin (9–39), contraction frequency was significantly decreased (19.55±0.71 contractions/minute Ctrl vs. 16.94±0.80 contractions/minute Ex(9–39), p<0.0311) and contraction width was significantly increased (1.67±0.04 seconds Ctrl vs. 1.94±0.10 seconds Ex(9–39), p<0.0274). Suggesting that either basal signaling through GLP-1R modulates the normal contractility of collecting lymphatic vessels, or that Ex(9–39) upon binding to the GLP-1R transiently induces signaling prior to antagonizing the receptor. Importantly, despite the presence of exendin (9–39), subsequent stimulation with semaglutide induced a mild but detectable (i.e., statistically significant) response as evident by the significant increase in EDD (96.88±4.08 μm Ex(9–39) vs. 101.7±3.84 μm Ex(9–39)+Sema, p<0.0013) and ESD (Figure 8C,D), a significant decrease in contraction frequency (16.94±0.80 contractions/minute Ex(9–39) vs. 14.33±0.95 contractions/minute Ex(9–39)+Sema, p<0.0001) and fractional pump flow, and a significant increase in volume displaced (3.97±0.34 nL Ex(9–39) vs. 4.30±0.37 nL Ex(9–39)+Sema, p<0.0360). It is important to note that, while statistically significant, these changes were significantly smaller than those observed in the absence of exendin (9–39). These observations suggest the unexpected and unexplored possibility that, in addition to GLP-1R-specific signaling, semaglutide could modulate lymphatic contractility via signaling in LECs and potentially other cell types, including LMCs, independently of GLP-1Rs. More in-depth studies are required to confirm or disprove these observations.

Figure 8. Effects of pharmacological inhibition of GLP-1Rs to the contractile activity of collecting lymphatic vessels and to the semaglutide-mediated response.

(A) Representative diameter trace displaying the contractile activity of an inguinal axillary lymphatic vessel from a WT mouse under control conditions, during a 20-minute pre-treatment with the competitive GLP-1R inhibitor Exendin (9–39) at a concentration of 100 nM, and after stimulation with a single 5 nM dose of semaglutide; and (B-I) Summary data of different contractile parameters under these 3 paired conditions, including (B) amplitude, (C) end diastolic diameter (EDD), (D) end systolic diameter (ESD), (E) ejection fraction, (F) contraction frequency, (G) width, (H) fractional pump flow (FPF), and (I) volume displaced. These data are presented as means±SEMs. A total of n=10 vessels from WT mice were used for these experiments. A one-way ANOVA corrected with a Geisser-Greenhouse correction as no equal variances were assumed (i.e., sphericity was not assumed) was used to determine differences in contractile function parameters, with statistical significance set at p<0.05.

Semaglutide Improves the Pumping Capacity of Lymphatic Vessels from Diet-Induced Obese (DIO) WT and Hypercholesterolemic ApoE KO Mice

In previous sections, we characterized the effects of a single stimulus of semaglutide to the contractile function of healthy collecting lymphatics from WT mice, while the lymphatic vessels were maintained under constant superfusion with fresh Krebs buffer, allowing for drug washout. We then aimed to 1) characterize the functional response in collecting lymphatics when the concentration of semaglutide was maintained for a longer period of time, and 2) determine the potential beneficial effects of semaglutide to improve the contractile capacity of inguinal axillary lymphatic vessels from diet-induced obese (DIO) and ApoE KO mice. These are both animal models that recapitulate various components of metabolic syndrome and display various aspects of lymphatic system dysfunction. Following a 30-minute equilibration, the contractile activity of lymphatic vessels was first recorded under control conditions, for 2 minutes, and then the concentration of semaglutide in the bath was increased to 5 nM, and the functional response of collecting lymphatics was recorded for 30 minutes. Representative diameter traces of lymphatic vessels from WT (Figure 9A), DIO (Figure 9B) and ApoE KO mice (Figure 9C) are shown following the response to GLP-1R agonism with semaglutide perfusion. The concentration of semaglutide was maintained by constant superfusion with a Krebs buffer containing the same concentration of the drug (i.e. 5 nM semaglutide). In control conditions and compared to lymphatics from WT mice, lymphatic vessels from DIO mice displayed significantly smaller diameters (Figure 9E,F); however, their contractile function was not significantly impaired (although contraction amplitude and volume displaced trended lower, (Figure 9D,K). In contrast, lymphatic vessels from ApoE KO mice had similar basal diameters when compared to WT controls but displayed significantly decreased contraction amplitude (31.45±1.47μm for WT vs. 21.73±2.39μm for ApoE KO, p<0.0104), ejection fraction (0.47±0.02 for WT vs. 0.34±0.03 for ApoE KO, p<0.0117), and volume displaced (4.96±0.27nL for WT vs. 3.37±0.41nL for ApoE KO, p<0.0103) (Figure 9D,G,K); while contraction frequency was significantly higher (Figure 9H). Consistent with our observations reported in the previous sections, the modulation of the contractile function by semaglutide occurred within the first couple minutes of exposure to the drug and these effects remained stable throughout the duration of the recording.

Figure 9. Semaglutide improves and partially restores the pumping capacity of dysfunctional lymphatic vessels from diet-induced obese (DIO) and hypercholesterolemic ApoE KO mice.

(A-C) Representative diameter traces of lymphatic vessels from WT, DIO, an ApoE KO mice during control conditions and during a 20-minute superfusion with semaglutide (5nM); (D-K) Summary data of contractile parameters characterizing the contractile function of lymphatic vessels from WT, DIO, and ApoE KO mice under control conditions and following perfusion with a Krebs buffer containing 5 nM semaglutide. These contractile parameters are expressed as mean±SEM and include (D) contraction amplitude, (E) end diastolic diameter (EDD), (F) end systolic diameter (ESD), (G) ejection fraction, (H) contraction frequency, (I) contraction width (at half maximum), (J) fractional pump flow (FPF), and (K) volume displaced. A total of n=17 (control) and n=10 (sema) vessels from WT mice, n=13 (control) and n=12 (sema) vessels for DIO mice, and n=19 (control) and n=9 (sema) vessels from ApoE KO mice were used for these experiments. A two-way ANOVA was utilized to assess statistical differences in contractile parameters with statistical significance set as p<0.05.

Importantly, the impaired contractile amplitude, ejection fraction, and volume displaced per contraction observed in lymphatics from ApoE KO mice were all improved and restored to values comparable to those of lymphatics from WT mice in control conditions by superfusion with semaglutide (Figure 9D,G,K). In fact, volume displaced, a parameter that we have introduced as an estimate of fluid transport per contraction, was significantly increased with semaglutide perfusion in lymphatics from all groups, i.e., 4.96±0.27nL (Ctrl) vs. 5.56±0.34nL (Sema) for WT, 3.72±0.40nL (Ctrl) vs. 4.70±0.52nL (Sema) for DIO, and 3.37±0.41nL (Ctrl) vs. 4.28±0.72nL (Sema) for ApoE KO (Figure 9K). Irrespective of the group, semaglutide perfusion induced robust vasodilation as indicative by the significant increases in EDD, i.e., 116.47±3.59μm (Ctrl) vs. 126.06±3.49μm (Sema) for WT, 95.35±3.30μm (Ctrl) vs. 107.19±3.82μm (Sema) for DIO, and 112.11±5.58μm (Ctrl) vs. 123.18±6.30μm (Sema) for ApoE KO (Figure 9E). Consistent with our previous observations, semaglutide induced a significant reduction in contractile frequency and calculated FPF (Figure 9H,J); while contraction width (at half-maximum) remained unaffected by group and semaglutide treatment (Figure 9I).

Discussion

Cancer and obesity-related lymphedema make up the largest groups of patients afflicted with this disease, with no pharmacological therapies available. In recent years, GLP-1R agonists, initially developed and approved for treatment of type 2 diabetes, have been shown to be effective for fighting obesity and promoting weight-loss in patients who have adipose tissue that is refractory to diet and exercise alone. A few clinical studies have suggested the potential beneficial effects of GLP-1R agonists on the lymphatic system, specifically in improving lymphedema in patients with various degrees of obesity, including non-obese patients,^45,46 and in reducing the risk of developing secondary lymphedema in patients who underwent axillary lymph node dissection for breast cancer treatment⁴⁴. However, prior to this study, the expression of GLP-1Rs in the lymphatic vasculature and the direct effects of GLP-1R agonists on the function of lymphatic vessels remained unexplored.

The lymphatic system relies on spontaneous, strong, and highly entrained contractions⁵⁰ in order to transport lymph from the interstitial spaces back into the central circulation. Also critical are intraluminal valves, which not only ensure net unidirectional, forward flow, but they also allow for the generation of propulsive pressure during each contraction. Assessment of lymphatic contractile activity in patients with lymphedema is limited; however, a few studies have reported on decreased pumping capacity of collecting lymphatics in the affected lymphedematous limbs^45,64,65. In fact, the study by Crowley et al.⁴⁵ reported on a breast cancer-related lymphedema patient whose lymphatic pumping returned, along with improvement in other lymphedema symptoms, after 13 months of treatment with semaglutide. Therefore, as the lymphatic field continues to move forward and novel therapies emerge, it is crucial to consider the development of pharmacological alternatives that target the improvement and/or restoring of the pumping capacity of lymphatic vessels.

In this study, we aimed to first characterize the expression of GLP-1Rs in the lymphatic vasculature and, because of its importance in fluid drainage and transport, also evaluate the specific effects of GLP-1R agonism to the regulation of lymphatic contractile function.

Our scRNAseq observations showed that in a dataset that included lymphatic vessels and nearby surrounding tissues, expression of Glp1r, encoding GLP-1Rs, was solely detected in LECs. Furthermore, subsequent sub-clustering of LECs led to the identification of 6 unique LEC subtypes. We utilized markers previously reported by Dr. Tatiana Petrova’s group⁶³ to annotate our LEC scRNAseq data and determined that these subclusters displayed transcriptomic profiles consistent with collecting LECs (including valve LECs), pre-collecting LECs, and capillary LECs. Intriguingly, expression of Glp1r was highly enriched in LECs from collecting and pre-collecting LECs but was not detectable in LECs from capillary/initial lymphatics. We confirmed a direct, functional response to pharmacological activation of GLP-1Rs in the lymphatic endothelium by assessing intracellular calcium responses in collecting lymphatics from mice expressing the genetically encoded calcium sensor GCaMP6f specifically in LECs (i.e., Prox1-CreER^T2;Salsa6f mice), and showed that stimulation with semaglutide not only increased global cytosolic calcium, but also increased the number of discrete calcium transients. In- depth characterization of the type, localization (within the cell), and frequency of these events would provide important insights into the underlying GLP-1R signaling pathways. These will be the focus of follow-up studies from our lab.

Our results from ex-vivo assessment of the contractile activity of collecting lymphatics demonstrated, for the first time, a direct effect of GLP-1R agonism to the contractile function of lymphatic vessels. Collecting lymphatics displayed an exquisite sensitivity to semaglutide, even at the lowest tested concentration (i.e., 1nM), this agonist induced robust vasodilation, usually within the 30–60 seconds following exposure to the drug, suggesting that lymphatic vessels are likely sensitive to semaglutide at concentrations even <1 nM. The vasodilatory effects of semaglutide were maximal at concentrations in the range of 1–10nM; in fact, a single stimulus with 5nM semaglutide induced a sustained vasodilatory effect that lasted >10 minutes, even while the preparation was maintained under constant bath perfusion (~0.2 mL/min) with a fresh physiological solution. Other potent vasodilators commonly induce robust vessel relaxation with a concomitant concentration-dependent loss of contractility, including contraction frequency and amplitude. Semaglutide induced a significant reduction in contractile frequency; however, lymphatic vessels maintained strong contractions even at the higher range of concentrations included in this study (i.e., 1μM). This suggested that the signaling cascade induced by semaglutide included not only vasodilatory effects but also included a component of direct or indirect stimulation of LMCs. The significant increase in vessel diameter allowed lymphatic vessels to accommodate larger volumes of fluid resulting in more efficient pumping, as indicated by the significantly increased volume displaced, a new calculated parameter we have introduced in this study to estimate the amount of fluid that would be pumped/transported by a 1-mm long vessel (i.e., the approximate length of a lymphangion).

Importantly, despite semaglutide decreasing contractile frequency, the calculated volume of fluid displaced with each contraction was significantly increased in lymphatic vessels from animal models that recapitulate various aspects present in metabolic syndrome (i.e., DIO and chow fed ApoE KO mice). Previously, studies led by Dr. Véronique Angeli and Dr. Gwendalyn Randolph demonstrated that ApoE KO mice fed a high-fat diet displayed lymphatic vessel degeneration, as well as multiple aspects of lymphatic dysfunction, including severe valve dysfunction and lymphatic contractile impairment^26,66. In this study, our data demonstrated that lymphatic contractile dysfunction is present even in ApoE KO mice fed a regular chow, suggesting that chronic hypercholesterolemia may be sufficient to induce contractile impairment. In fact, our preliminary unpublished observations suggest that lymphatic contractile function in chow-fed ApoE KO mice declines with age. Our lab is currently conducting additional studies, beyond the scope of this manuscript, to characterize contractile dysregulation in control chow-fed and western diet-fed ApoE KO mice to identify the underlying molecular mechanisms leading to severe dysfunction of collecting lymphatic vessels in obesity and hypercholesterolemia, including contractile and valve dysfunction, as well as hyperpermeability. It is important to note that the acute responses to semaglutide in lymphatic vessels from DIO and ApoE KO mice were similar to those observed in lymphatics from control, healthy WT mice, which suggests that impaired NO bioavailability, as well as impairment in other yet-to-be uncovered signaling mechanisms meditated by GLP-1Rs may not be present in these diseased models. Ongoing studies from our lab are currently exploring the effects of long-term treatment with semaglutide to the different aspects of lymphatic function, including contractility, permeability, and valve function.

Although the calculated volume that was displaced by each contraction was significantly increased following stimulation with semaglutide, the concomitant significant drop in contraction frequency raises an important question about the net fluid volume displaced over time (i.e., volume displaced x contraction frequency to estimate volume displaced per minute). An important consideration is that in this study most of our contractile assessment was performed under no-flow conditions, i.e., inflow and outflow pressures set at 3 cmH₂O; however, under physiological conditions and more specifically in most anatomical places of the human body, lymph must be transported against an adverse pressure gradient imposed by the gravitational forces. This is why lymphedema is more commonly observed in the upper and lower extremities. Here, the semaglutide-mediated reduction in contraction frequency also resulted in a significant decrease in the calculated pump flow (i.e., ejection fraction x frequency). Similarly, the estimated net volume transported per minute was ~133 nL/minute (Ctrl) vs. ~105 nL/minute (Sema). These calculated parameters would suggest a decline in overall fluid transport following treatment with semaglutide, when no adverse pressure gradient is present. Therefore, we performed a set of dedicated experiments to quantitatively assess the pumping capacity of collecting lymphatics, i.e., pump tests. When challenged with an increasing adverse pressure gradient, the pump limit of collecting lymphatics was consistently and significantly increased by semaglutide in ~86%. Suggesting that when lymphatics must transport fluid an adverse pressure gradient, treatment with semaglutide would allow them to overcome higher pressure gradients and more efficiently transport fluid.

The strong vasodilatory effects of semaglutide were found to be partially linked to NOS-dependent nitric oxide; however, a residual functional response was consistently present even in the presence of the NOS inhibitor L-NAME. We explored the potential involvement (independent and in synergy with nitric oxide) of secreted prostanoids and NADPH oxidase-mediated ROS using pharmacological inhibitors of COX-1,2, and NADPH oxidase. While some modest effects were observed, these results were mostly inconclusive, suggesting the presence of additional signaling, potentially of paracrine nature, yet to be elucidated. A known effect of GLP-1R drugs related to weight-loss is the decrease motility of the intestines, which is associated with activation of GLP-1Rs in surrounding nerves. We demonstrated that while perivascular innervation was observed in collecting lymphatics from some anatomical regions of the mouse, perivascular nerves were not present (or at least not detected) in inguinal axillary collecting lymphatics, which were the major lymphatic bed utilized in this study. The unlikely involvement of nerve-GLP-1Rs in the regulation of the semaglutide-mediated response specifically in inguinal axillary lymphatics was further validated by demonstrating that the response to semaglutide was not affected by pharmacological inhibition of NaV channels using TTX. Unexpectedly, the combined inhibitory effects of L-NAME and TTX displayed a synergistic effect that mostly prevented all effects induced by semaglutide, suggesting an interaction between nitric oxide and NaV channels as a component of the semaglutide-mediated increase in pumping capacity of collecting lymphatics.

In our experiments, Glp1r expression was detected in ~20% of the LECs, which raises the question about whether GLP-1Rs are indeed solely expressed in this subset of LECs, or GLP-1Rs are expressed in all LECs at a protein and functional level. There is a possibility that gene expression levels in most LECs (i.e., ~80%), as well as in other cell types were in low abundance and below our detection level. The latter alternative would suggest the unexplored possibility that GLP-1Rs are present in LMCs, and/or other cell types, and their activation contributes to the contractile regulatory effects hereby reported. In addition to endothelial expression, GLP-1R expression has been reported in various types of smooth muscle, including vascular smooth muscle, and their activation has been linked to vasodilation and increase in blood flow^67–74. A secondary, unexplored alternative is that semaglutide could induce non-specific signaling in LECs and other cell types independently of GLP-1R activation. This speculation arose after demonstrating that pharmacological inhibition of GLP-1Rs with exendin (9–39) did not effectively prevent all functional responses induced by semaglutide.

In this study, our observations are in agreement with recent clinical studies and support the therapeutic potential of semaglutide to improve lymphatic function in secondary lymphedema including obesity-related and cancer-related lymphedema. Importantly, the exquisite sensitivity of collecting lymphatic vessels to GLP-1R agonism points to semaglutide, and other GLP-1R agonists, as effective therapies to improve lymphatic function independently of their weight-loss effects. The specific lymph concentrations of semaglutide in humans are unknown, and these likely vary between visceral and peripherally located lymphatics; however, direct uptake of GLP-1Rs by the lymphatic system has previously been demonstrated by Dr. Natalie Trevaskis’s group in animal models⁷⁵. Upcoming work by Dr. Trevaskis will soon determine the specific concentrations of semaglutide in lymph.

In conclusion, our results suggest that significantly lower dosages of semaglutide, compared to those commonly prescribed for weight loss, could be employed to restore the pumping capacity of dysfunctional lymphatics in lymphedema. Low dosages (e.g., <100nM) of semaglutide could be directly delivered (i.e., subdermal) into the affected limb minimizing its systemic side effects. While these speculations remain to be tested and validated, unpublished clinical studies presently being conducted by our co-author Dr. Joseph Dayan (shared with permission) have shown that 8 out of 9 patients in a prospective study on GLP-1R treatment for lymphedema have experienced a significant reduction in limb volume (~10% average reduction) and improvement in validated quality of life scores (LLIS) by 36% with the affected limb having a higher volume reduction than the unaffected limb. Noteworthy, these beneficial effects of GLP-1R agonism were observed even after only 3 months of treatment with semaglutide at the lowest prescribed dosage of 0.25 mg/week.

Finally, our results specifically demonstrate a direct beneficial effect of semaglutide to improve fluid transport by allowing lymphatics to robustly vasodilate while maintaining, and even improving, efficient pumping activity. Our findings also demonstrated that GLP-1Rs are highly enriched in LECs, where these channels may play additional important roles, even some potentially regulating lymphatic valve and barrier function (i.e., permeability), which are being investigated in separate studies from our group.

Perspectives

Secondary lymphedema is most commonly associated with 1) cancer-related therapeutic interventions, including chemotherapy, mastectomy, axillary lymph node dissection, radiotherapy, and hormonal therapies, as well as with 2) obesity and metabolic syndrome, with no pharmacological therapies available. In agreement with recent clinical studies suggesting that pharmacological agonism of GLP-1Rs may be an effective therapeutic strategy to treat or prevent secondary lymphedema in these groups of patients; using murine models, this study demonstrated that GLP-1Rs are expressed in the lymphatic vasculature and are specifically enriched in the lymphatic endothelium. Furthermore, collecting lymphatic vessels from healthy mice, as well as from mouse models of obesity and metabolic syndrome displayed an exquisite sensitivity to pharmacological activation with semaglutide, which significantly increased or restored their pumping capacity. In lymphedema, impaired contractile/pumping function of lymphatic vessels is a known key determinant in the development and progression of the disease, and the observations hereby presented (supported by clinical reports), suggest the therapeutic potential of GLP-1R agonists to directly restore the pumping capacity, and likely other key aspects of lymphatic function, of collecting lymphatic vessels in secondary lymphedema.

Supplementary Material

In this section we included a Major Resources table that contains information on animals (i.e., strains used and their source), reagents and chemicals, data availability, and number of animals, as well as their corresponding sex and ages, used to generate the data included in the main figures of this manuscript. We also include Supplemental Figures 1–4.

Abbreviations:

GLP-1R

Glucagon-like Peptide-1 Receptor

scRNAseq

Single-Cell RNA Sequencing

PSS

Physiological Saline Solution

L-NAME

NG-Nitro-L-Arginine Methyl Ester

Indo

Indomethacin

Apo

Apocynin

Sema

Semaglutide

LECs

Lymphatic Endothelial Cells

LMCs

Lymphatic Muscle Cells

WT

Wild Type

DIO

Diet-Induced Obesity

WD

Western Diet

TTX

Tetrodotoxin

Ex(9–39)

Exendin (9–39)

WGA

Wheat Germ Agglutinin

TUBB3

Tubulin β3

NaV

Voltage-Gated Sodium Channels