Acute Metabolic Stress Induces Lymphatic Dysfunction Through KATP Channel Activation

Abstract Lymphatic dysfunction is an underlying component of multiple metabolic diseases, including diabetes, obesity, and metabolic syndrome. We investigated the roles of KATP channels in lymphatic contractile dysfunction in response to acute metabolic stress induced by inhibition of the mitochondrial electron transport chain. Ex vivo popliteal lymphatic vessels from mice were exposed to the electron transport chain inhibitors antimycin A and rotenone, or the oxidative phosphorylation inhibitor/protonophore, CCCP. Each inhibitor led to a significant reduction in the frequency of spontaneous lymphatic contractions and calculated pump flow, without a significant change in contraction amplitude. Contraction frequency was restored by the KATP channel inhibitor, glibenclamide. Lymphatic vessels from mice with global Kir6.1 deficiency or expressing a smooth muscle-specific dominant negative Kir6.1 channel were resistant to inhibition. Antimycin A inhibited the spontaneous action potentials generated in lymphatic muscle and this effect was reversed by glibenclamide, confirming the role of KATP channels. Antimycin A, but not rotenone or CCCP, increased dihydrorhodamine fluorescence in lymphatic muscle, indicating ROS production. Pretreatment with tiron or catalase prevented the effect of antimycin A on wild-type lymphatic vessels, consistent with its action being mediated by ROS. Our results support the conclusion that KATP channels in lymphatic muscle can be directly activated by reduced mitochondrial ATP production or ROS generation, consequent to acute metabolic stress, leading to contractile dysfunction through inhibition of the ionic pacemaker controlling spontaneous lymphatic contractions. We propose that a similar activation of KATP channels contributes to lymphatic dysfunction in metabolic disease.


Significance Statement
Increasing evidence suggests that lymphatic contractile dysfunction contributes to the pathologies associated with m ultiple chr onic diseases, including a ging, obesity, type 2 dia betes, and meta bolic syndr ome , all of whic h are associated with metabolic stress.Although K ATP channel activation plays a protective role in arterial smooth muscle under ischemic conditions, acti v ation of K ATP channels in l ymphatic m uscle decr eases l ymphatic m uscle excita bility and suppr esses the generation of spontaneous contractions that aid in lymph propulsion.We investigated the possible contribution of K ATP channels to the lymphatic contractile dysfunction associated with metabolic stress.Acute metabolic stress, induced by inhibition of the mitochondrial electron transport chain, led to increased K ATP channel activity in lymphatic muscle, either through a reduction in the intracellular ATP/ADP ratio or increased production of ROS, resulting in impairment of the ionic

Introduction
The lymphatic system plays a major role in tissue fluid homeostasis through the removal of excess fluid and protein filtered fr om b lood capillaries into the interstitium. 1The accumulation of excess interstitial fluid, most often in dependent extremities, leads to the chronic condition lymphedema.Lymphedema may r esult fr om genetic m utations in critical genes contr olling l ymphatic vessel or valve development, 2 surgical disruption of lymphatic networks, 3 altered permeability of lymphatic vessels, [4][5][6][7] and/or impaired spontaneous contractions of the collecting lymphatic vessels that pump lymph centrally. 8 , 9The consequences of lymphatic dysfunction may also be more subtle.Recent studies suggest that altered lymphatic function contributes to subclinical edema that can compromise organ function under certain conditions, such as heart failure with pr eserv ed ejection fraction 10 and diseases associated with the accumulation of tissue sodium. 11 , 12ubtle consequences of lymphatic dysfunction are also beginning to be appreciated in the context of metabolic diseases, including aging, obesity, type 2 diabetes, and metabolic syndrome.][25] Lymphatic v alv e dysfunction, which impairs unidirectional lymph transport, is associated with animal models of obesity, with 6 or without 26 ApoE deficiency, and in models of TNF α hyperactivity. 18 , 27Lymphatic collecting vessel hyperpermeability, whic h counter acts the centr al r eturn of l ymph, has been demonstrated in animal models of obesity 5 , 6 , 26 , 28 and in a model of type 2 diabetes. 29However, the mechanisms by which lymphatic dysfunction contributes to or results from the primary disease are not clear.
A common feature of these disease models is chronic meta bolic str ess.The primar y source of intracellular ATP is the mitochondrial electron transport chain (ETC) and one mechanism of coupling the electrical activity of cells to their metabolic state is through the activity of ATP-sensitive K + (K ATP ) c hannels.K ATP c hannels ar e v olta ge-inde pendent, K + -selecti v e channels that are activated by intracellular ADP and inhibited by intracellular ATP. 30 , 31By sensing the ATP/ADP ratio, they act as "metabolic sensors of the cell." 32Mammalian K ATP channels are hetero-octameric complexes comprised of four inw ard-r ectifying K + channel (Kir6) pore-forming subunits and four r egulator y sulphonylur ea r ece ptor (SUR) subunits.Two pairs of genes ( ABCC8 , KCNJ11 , and ABCC9 , KCNJ8 ) on chromosome 11, each encode one pair of subunits-SUR1, Kir6.2, and SUR2, Kir6.1, r especti v el y. 30 The expr ession of differ ent combinations of Kir6 and SUR subunits results in distinct K ATP channel properties and functional roles in different cell types. 30 , 32In the v asculatur e , K ATP c hannels normall y hav e low acti vity in arterial smooth muscle cells but become acti v ated under ischemic conditions, 30 leading to arterial smooth muscle cell hyperpolarization and dilatation, which in turn incr ease b lood flow and oxygen supply to the ischemic tissue. 32The composition of K ATP channels expressed in lymphatic muscle is similar to that in arterial smooth muscle (Kir6.1 and SUR2B).Although K ATP -de pendent v asodilation plays a pr otecti v e r ole in arteries, 30 , 33 , 34 similar de pr ession of lymphatic muscle excitability by K ATP channel acti v ation could pr omote edema formation and sustain tissue inflammation as a consequence of impaired active lymph transport. 8 , 35n the present study, we hypothesized that acute mitochondrial inhibition would suppress ATP production and result in K ATP channel acti v ation, r educing l ymphatic m uscle excita bility and impairing the acti v e l ymph pump.Furthermor e, gi v en that meta bolic str ess can incr ease R OS pr oduction in mitochondria, 36 we also sought to evaluate the possible contribution of ROS.Our findings suggest that metabolic stress contributes to the contractile component of lymphatic dysfunction observed in metabolic diseases through ROS-dependent and -independent acti v ation of K ATP channels in lymphatic muscle.

Ethical Appro v al
All protocols and procedures were re vie wed and approved by the Institutional Animal Care and Use Committee of the Uni v ersity of Missouri (protocol #9797) and performed in accordance with the National Institutes of Health's Guide for the Care and Use of La borator y Animals (8th edition, 2011).

Mice
Mice were housed and bred under pathogen-free conditions in a controlled environment (22 ± 2 • C, 12/12-hr light/dark cycle) of the animal facility of the Uni v ersity of Missouri School of Medicine.C57BL/6 wild-type (WT) mice w ere pur c hased at 5 weeks of a ge fr om J ackson La borator y (C57Bl/6 J, strain #000 664, Bar Harbor, ME, USA).Kir6.1 −/ − mice were a gift fr om Susum u Seino (Kobe Uni v ersity).Dominant negati v e Kir6.1 ( eGFP-Kir6.1[AAA] ) mice were described previously. 37Mice carrying eGFP-Kir6.1[AAA]were crossed to Myh11-CreER T2 mice (JAX No. 01 979) to generate Myh11-CreER T2 ; Kir6.1-AAA mice with an inducible dominant negative Kir6.1 transgene expressed specifically in smooth muscle.The offspring were injected with tamoxifen (10 mg mL −1 in safflower oil) for consecuti v e 5 days and allo wed to reco ver for 14 days; we confirmed the absence of GFP signal in the smooth muscle cell layer before testing.For R OS ima ging, Myh11-CreER T2 ; tdTomato mice were used.Genotypes were confirmed by PCR with Taq DN A Polymer ase Premix (Intact Genomics Catalog #3249).Mice were studied at 6-8 weeks of age from either sex, depending upon availability; in the case of Myh11-Cr eER T2 mice , only males w ere used, as the tr ansgene is located on the Y-chromosome; for other protocols mice of both sexes were used.For experiments, mice were anesthetized by i.p. injection using a mixture of ketamine/xylazine (100/10 mg kg −1 body weight) and euthanized by intracardiac injection of KCl.

Vessel Isolation, Pressure My ograph y, and Data Acquisition
We hav e pr eviousl y documented the r elia bility and r e pr oducibility of contraction parameter tests for mouse popliteal lymphatic vessels studied ex vivo. 4 , 38 , 39For the isolation of mouse popliteal l ymphatic v essels, a pr oximal-to-distal incision w as made in the skin of the dorso-lateral thigh to expose the superficial saphenous vein.Afferent popliteal lymphatic vessels on either side of this vein were removed and transferred to a Sylgard dissection dish with Krebs buffer containing albumin.After pinning and cleaning, a vessel was cannulated using two glass micropipettes (40-50 μm, outer diameter), pressurized to 3 cmH 2 O and further cleaned of any remaining tissue in order to tr ac k diameter accuratel y.The v essel w as shortened to a length that contained onl y one v alv e. Pol yethylene tubing (PE-190) attached to the bac k of eac h micr opipette holder w as connected to a 2-channel microfluidic device (Elveflow OB1 MK3, Paris) for computer contr ol of pr essur e on the sta ge of an inv erted micr oscope.Input and output pr essur es wer e transientl y set to 10 cmH 2 O immediately after set up and the v essel w as str etched axiall y to appr oximate the in vi v o length, which minimized longitudinal bowing and associated diameter-tr ac king artifacts during subsequent protocols. 40With input and output pr essur e held at 3 cmH 2 O, spontaneous contractions typically began within 15-30 min of warm-up and each vessel was allowed to stabilize at 37 • C for 30-60 min before beginning an experimental protocol.A suffusion line connected to a peristaltic pump exchanged the chamber contents with Krebs buffer at a rate of 0.5 mL/min.A customwritten LabVIEW (National Instruments; Austin, TX, USA) algorithm 41 measured the inner diameter of the v essel fr om video images obtained at 30 fps using a Basler A641fm firewire camera.

Assessment of Lymphatic Contractile Function
Spontaneous contr actions w er e r ecorded with equal input and output pr essur es (3 cmH 2 O) to pr ev ent a pr essur e gradient for forwar d flo w thr ough the v essel during the experiment.The effects of mitochondrial ETC inhibitors or oxidati v e phosphorylation inhibitors on WT, Kir6.1 −/ − and Myh11-CreER T2 ; Kir6.1[AAA] v essels wer e determined by adding the compounds to the perfusate.At the end of the perfusion period (20-60 min), a small, pr edetermined v olume of GLIB (1 μm ) was added to the 3 mL bath, followed by thorough mixing while the bath was stopped.At the end of ev er y experiment, all v essels wer e perfused with Ca 2 + -free Krebs buffer containing 3 m m EGTA for 30 min, and passi v e diameters wer e r ecorded at 3 cmH 2 O pr essure (D MAX ).After an experiment, custom-written analysis progr ams w ere used to detect peak end-diastolic diameter (EDD), end-systolic diameter (ESD), contraction amplitude (AMP), and contraction frequency (FREQ) on a contr action-by-contr action basis at r est befor e addition of the drug (FREQ REST ) and following administration of drug. 42When FREQ was zero, no value of AMP w as r ecorded.F r actional Pump Flow (FPF) is the best estimate of net flow and is a calculated v aria b le due to the lack of flow meters with sensitivities in nL/min r ange .These data were used to calculate sev eral commonl y r e ported parameters that c har acterize l ymphatic v essel contr actile function.Eac h par ameter w as av era ged ov er a 5 min period and used to calculate the following indices of lymphatic contractile function: Ejection wher e FREQ avg r e pr esents the av era ge fr equency (in contractions per minute; cpm) during the baseline period before the addition of a drug to the bath and D MAX r e pr esents the maxim um passi v e diameter (obtained after incubation with calcium-free Krebs solution) at a gi v en lev el of intraluminal pr essur e.

V m Measurement
Popliteal l ymphatic v essels fr om WT mice wer e isolated and pressurized as described above.Wortmannin (1 μm ; Tocris Bioscience, Bristol, UK) was applied to the perfusion bath for 30 min to inhibit m y osin light c hain kinase until contr actions w ere sufficientl y b lunted to maintain impalements into lymphatic smooth muscle using intracellular microelectrodes (200-320 M ) filled with 1 M KCl solution.The pr eserv ation of small contractions ( ≤5 μm amplitude) allowed us to monitor the viability of the pr e par ation.Membr ane potential was sampled at 1-5 KHz using an AxoClamp2A amplifier, digitized through an A-D interface (USB-6216, National Instruments) and recorded using a custom La bVIEW pr ogram.Once impaled, V m w as allowed to stabilize before action potentials and multiple contraction cycles wer e r ecorded for anal ysis.Once r ecordings wer e completed, the electr ode w as r etracted and the r ecording w as corr ected for any offset potential.

Analysis of ROS Production in Lymphatic Muscle
Popliteal lymphatic vessels from tamoxifen-treated Myh11-CreER T2 ; tdTomato or WT mice were dissected and cannulated.To ev aluate R OS pr oduction, the pr essurized v essels wer e loaded with dihydrorhodamine 123 (DHR; Fisher Scientific), a membrane permea b le dye that coverts to cationic rhodamine 123 upon oxidation and then localizes to mitochondria. 43DHR was dissolved in DMSO and diluted to 10 μm in Krebs buffer, perfused and preincubated for 10 min in a bath and remained in the superfusion solution throughout the experiment.All vessels were incubated at 37 • C in a light-protected environment.
Nifedipine (1 μm ) was applied to the bath to completely inhibit any spontaneous contractions that otherwise would have prevented maintaining focus on the smooth muscle layer.).Krebs-BSA buffer was prepared with the addition of 0.5% bovine serum albumin.During cannulation, Krebs-BSA buffer was present both luminally and abluminally, but during the experiment the bath solution was replaced with Krebs solution without albumin.Ca 2 + free Krebs buffer was used at the end of experiment to obtain the maxim um passi v e diameter.All chemicals and drugs (rotenone, antimycin A, CCCP, tiron, PEG-catalase) were purchased from Sigma-Aldrich (St. Louis, MO, USA), with exception of BSA (United States Biochemicals; Cleveland, OH, USA), MgSO 4 , Na-HEPES (ThermoFisher Scientific; Pittsburgh, PA, USA).PEG-catalase w as dissolv ed in distilled water.Antimycin A, CCCP, rotenone and GLIB were dissolved in DMSO and the total amount of DMSO was set below 0.4%, which was determined in separate protocols to be the vasoacti v e thr eshold concentration.

Statistical Procedures
The number n refers to the total number of vessels included per group as stated in the figure legends; N is number of animals.
Values are means ± SD.Statistical analysis was undertaken only for studies where each group size ≥5.Randomization was not a feature of study design.Because baseline values of spontaneous contraction frequency are highly variable among lymphatic vessels, 46 we expressed the data as normalized values of FREQ, AMP, or FPF, with each value normalized to the average of the respecti v e contr ol v alues.Differ ences between FREQ, AMP, or FPF in the presence or absence of treatment were assessed using ANOVA or paired Student's t -tests as stated in the figure legends.All statistical anal yses wer e performed using Prism9 (GraphPad Softw ar e Inc., CA, USA), with significance for all tests set at P < 0.05.

Mitochondrial ETC Disruption Impairs Lymphatic Pacemaking
When pressurized and heated to physiological intraluminal pr essur e and temperatur e, WT popliteal l ymphatics dev eloped spontaneous twitc h contr actions.At 37 • C and 3 cmH 2 O intraluminal pr essur e, the typical lymphatic contraction pattern shown at the start of the recording in Figure 1 A was stable for hours, although in some vessels occasional patterns of burst contractions dev eloped ov er time .In this example , the contr action amplitude was ∼40 μm (41% of maximal passive diameter) and fr equency w as 11 contractions per minute (cpm) until ∼1 min after the application of the ETC complex III inhibitor antimycin A (30 n m ).This concentration w as slightl y higher than the IC 50 (12.2n m ) used to inhibit mitochondrial oxygen consumption. 47n the presence of antimycin A, contractions nearly ceased during the last 5 min of antimycin A treatment [control FREQ = 11 cpm vs 1 cpm (12.2% of control)].Contraction amplitude was largel y unaffected ev en after 20 min exposur e to antimycin A.
The subsequent addition of GLIB (1 μm ), at a concentration that we found pr eviousl y could r ev erse impair ed contractions in v essels from mice expressing overactive K ATP channels, 8 , 35 partially r estor ed contraction frequency.Both 3 and 10 μm GLIB produced mor e complete r ecov er y of fr equenc y (not sho wn), but here we used 1 μm , a concentration that minimizes off-target effects associated with higher concentrations of this inhibitor. 48nder baseline conditions, popliteal lymphatics from Kir6.1 −/ − mice developed similar patterns of spontaneous contractions as WT mice ( Figure 1 B).The lower basal contraction frequency in this particular vessel reflects natural v essel-to-v essel v ariation and w as not a consistent featur e of Kir6.1 −/ − vessels, 35 , 48 as K ATP channels make little contribution to the basal excitability of lymphatic muscle cells (LMCs) of mice under normal (normoxic) conditions.In contrast to the WT v essel, the Kir6.1 −/-v essel w as almost completel y r esistant to the effects of antimycin A [control FREQ = 6.8 cpm vs 5.7 cpm during the last 5 min of antimycin A] and subsequent addition of GLIB (1 μM) produced little further effect (FREQ = 6.2 cpm).
Although we pr eviousl y demonstrated that functional K ATP channels ar e expr essed in smooth m uscle but not endothelium of mouse popliteal v essels, her e we expressed a dominant negati v e construct of Kir6.1[AAA] 37 specifically in the muscle layer to confirm that LMC K ATP channels wer e r esponsib le for pr otection fr om the effects of antim ycin A. The Myh11-Cr eER T2 ; Kir6.1[AAA] contraction pattern was similar to the WT vessel prior to administration of antimycin A, ( Figure 1 C).As with the Kir6.1 −/ − vessel, the Myh11-CreER T2 ; Kir6.1[AAA] vessel was resistant to the effects of antimycin A (control FREQ = 10.3 cpm vs 10.3 cpm after antimycin A) and the subsequent application of GLIB (1 μm ) had little additional effect (FREQ = 9.9 cpm).
These observations suggest that treatment with antimycin A results in the activation of K ATP channels in the muscle layer of mouse popliteal lymphatic vessels to impair pump function.
Summary data for the effects of antimycin A ± GLIB on lymphatic vessels from the three genotypes are presented in Figure 1 D. The frequency of WT vessels was significantly reduced by antimycin A and that de pr ession of fr equency w as largel y r ev ersed by GLIB (1 μm ), as predicted if the mechanism inv olv ed acti v ation of K ATP channels.Antimycin A produced a small but nonsignificant increase in normalized AMP in WT vessels that also was restored to control levels by GLIB ( Suppl. Figure 1 A).The r eduction in FREQ ( Figur e 1 D) w as mor e than sufficient to offset the increase in normalized AMP and produce a significant reduction in calculated fractional pump flow (FPF) ( Suppl. Figure 1 B), which is an estimate of acti v e l ymph tr ansport.The antim ycin A-induced reduction in normalized FPF of WT vessels was significantl y r escued by GLIB ( Suppl.Figur e 1 B).Antimycin A had no significant effects on the normalized amplitude or FPF of Kir6.1 −/ − vessels or Myh11CreER T2 ; Kir6.1[AAA] vessels ( Suppl. Figure 1 A, B).
Next, we explored the actions of a different metabolic stressor, rotenone, which inhibits ETC complex I. 49 Rotenone (100 n m ) caused a pr ogr essi v e decline in the frequency of spontaneous l ymphatic contractions, fr om a contr ol FREQ = 9.9 cpm to 2.2 cpm (22.2% of control) during the last 5 min of rotenone treatment, which was re versed b y GLIB (1 μm ) ( Figure 2 A).Note that this vessel had a bursting contraction pattern, whereby 3-8 contractions occurred in rapid succession followed by a short pause prior to the next contraction burst.In contrast, a popliteal lymphatic from a Kir6.1 −/ − mouse was largely resistant to the effects of rotenone, with control FREQ = 8.1 cpm vs 7.6 cpm during the last 5 min of rotenone treatment; subsequent GLIB (1 μm ) application had little additional effect ( Figure 2 B).A popliteal l ymphatic fr om a Myh11-CreER T2 ; Kir6.1[AAA] mouse (also with a bursting contraction pattern) was also resistant to rotenone (control FREQ = 14.1 cpm vs 13.1 cpm during the last 5 min of r otenone) ( Figur e 2 C).Summar y data for the effects of r otenone (100 n m ) on the 3 genotypes are plotted in Figure 2 D. The only significant effects of r otenone wer e on WT v essels, in which it lower ed fr equency to ∼25% of control, with partial rescue by GLIB (1 μm ).Rotenone had no significant effects on normalized AMP for any of the three genotypes ( Suppl. Figure 1 C).Rotenone significantl y r educed the normalized FPF of WT vessels, but not the normalized FPF of Kir6.1 −/ − vessels or Myh11-CreER T2 ; Kir6.1[AAA] vessels ( Suppl. Figure 1 D).
Next, we tested the effects of the pr otonophor e CCCP (1 μm ), which reduces ATP production by uncoupling mitochondrial oxidati v e phosphor ylation. 50The application of CCCP (1 μm ) to a WT popliteal lymphatic lowered frequency from 9 cpm in the control period to 2 cpm (22.2% of control) during the last 5 min of CCCP ( Figure 3 A), and the effect w as r ev ersed by GLIB (1 μm ).In contrast, popliteal vessels from a Kir6.1 −/ − mouse ( Figure 3 B) and an Myh11-CreER T2 ; Kir6.1[AAA] mouse ( Figure 3 C) were resistant to the effects of CCCP.The data for CCCP on normalized fr equency ar e summarized for the thr ee genotypes in Figur e 3 D.The only significant effects of CCCP were on WT vessels, which on av era ge lower ed fr equency to 22% of contr ol and showed partial rescue by GLIB (1 μm ).There were no significant changes in normalized amplitude in response to CCCP for any of the three g enotypes, sugg esting ther e w as still sufficient ATP for normal actom y osin crossbridge cycling ( Suppl. Figure 1 E).As a consequence of the reduction in frequency, CCCP caused a significant reduction in the normalized FPF of WT vessels (but not Kir6.1 −/ − vessels or Myh11-CreER T2 ; Kir6.1[AAA] vessels) ( Suppl. Figure 1 F).

Complex III Inhibition Suppresses Lymphatic Action Potentials
Gi v en these findings, we then tested the effects of antimycin A on the membrane potential of popliteal l ymphatics fr om WT mice.Pr eviousl y, w e show ed that K ATP c hannel acti v ation leads to inhibition of the ionic pacemaker underlying spontaneous lymphatic contractions, but that overt hyperpolarization is not r equir ed to inhibit action potential generation. 35 , 51Such was the case for antimycin A. Appr oximatel y 100 sec after antimycin A application, spontaneous action potentials ceased, even though the resting V m depolarized slightly ( Figure 4 A), contractions stopped and the vessel dilated.The inhibition of action potential firing despite slight depolarization suggests that the threshold for action potential firing was reset 52 during this period ( Figure 4 B, insert b).The effect of this concentration of antimycin A was similar to that produced by 100-300 n m pinacidil in previous studies. 48 , 51Summary V m data for nine LMCs from nine differ ent v essels showed that spontaneous action potential generation w as a bolished by antimycin A in most cells without significant hyperpolarization (with one exception) ( Figure 4 C).GLIB (1 μm ) application r estor ed AP gener ation in all cells ( F igure 4 D).Although we did not perform Vm measurements during rotenone or CCCP application, there is a 1:1 correspondence of twitc h contr actions with APs, 35 , 53 , 54 and the contraction recordings for rotenone and CCCP suggest that both drugs also inhibited AP firing.

ROS Contribution to Metabolic Disruption of Lymphatic Pacemaking
Next, we measured production of ROS in the muscle layer using DHR.Popliteal lymphatics expressing the td-Tomato r e porter in LMCs under the control of Both 30 n m and 1 μm antimycin A produced significant increases in DHR signal, and the latter was prevented by tiron and catalase pr etr eatment.Neither r otenone nor CCCP tr eatment led to a significant increase in ROS production.
Having esta b lished that antimycin A induces R OS pr oduction in LMCs, we then asked whether pr etr eatment with R OS scavengers could pr ev ent the effects of antimycin A on FREQ and FPF of WT lymphatic vessels.The example in Figure 6 A shows a v essel tr eated with tir on (1 m m ), at a concentration used to effecti v el y scav enge O 2 − in other studies. 55 , 56Tir on w as applied to the bath for 10 minutes prior to the addition of antimycin A. In this case, a transient slowing of FREQ occurred almost immediately after the addition of tiron, but FREQ subsequently returned to normal (within 5 min).The application of antimycin A did not produce any substantial changes in contraction FREQ, nor did subsequent addition of GLIB (1 μm ), which in this case reduced contraction amplitude by ∼40% ( Figure 6 A).The effect of GLIB on amplitude in this v essel w as not a consistent one, as shown in the summary data in Suppl.Figure 2 A. Tiron pretr eatment effecti v el y b locked the inhibitor y effect of antimycin A on normalized FREQ ( Figure 6 B), but again without significant effects on either normalized AMP or FPF ( Suppl. Figure 2 A, B).A similar pr otocol w as used to test the effects of the H 2 O 2 scavenger catalase (250 U/mL), 55 , 56 which also completely blocked the inhibitory effects of antimycin A on normalized FREQ ( Figure 6 C).Summary data for the effects of catalase pr etr eatment on   effects of antimycin A confirm that there were no significant effects on either normalized FREQ ( Figure 6 D) or on AMP or FPF ( Suppl. Figure 3 A, B).These effects of tiron and catalase are consistent with the conclusion that antimycin A is producing ROS, which acti v ates K ATP channels to inhibit pacemaking fr equency.Similar protocols were conducted to test tiron and catalase pr etr eatment on the effects of rotenone.As shown in the example recording in Figure 7 A, pretreatment with tiron failed to prevent the subsequent reduction in contraction FREQ in response to r otenone [contr ol FREQ = 12.5 cpm vs 3.6 cpm (26.9% of control) during last 5 min of rotenone], as confirmed by the summary data in Figure 7 B. Rotenone did not significantly inhibit normalized AMP (AMP actuall y incr eased) in the pr esence or a bsence of tir on ( Suppl.Figur e 2 C) and tir on pr etr eatment failed to block rotenone-induced inhibition of normalized FPF ( Suppl. Figure 2 D).Likewise, pr etr eatment with catalase did not pr ev ent the subsequent rotenone-induced reduction in contraction frequency [ Figure 7 C; control FREQ = 10.6 cpm vs 2.8 cpm (28.2% of control) during last 5 min of rotenone], as confirmed by the summary data in Figure 7 D. Catalase pretreatment did not significantly alter the effect of rotenone on normalized contraction AMP ( Suppl. Figure 3 C) and failed to block the rotenoneinduced inhibition of normalized FPF ( Suppl. Figure 3 D).These results confirm that rotenone does not inhibit lymphatic pumping through the production of ROS.
Tir on pr etr eatment also failed to b lock the inhibitor y effect of CCCP on contraction FREQ ( Figure 8 A; control FREQ = 10.5 cpm vs 4.6 cpm (42.4% of control) during the last 5 min of CCCP], but the inhibition was rescued by subsequent addition of GLIB (1 μm ).The summary data in Figure 8 B confirm that significant CCCP-induced inhibition of contraction FREQ persisted in the pr esence of tir on.Tir on did not alter the effect of CCCP on AMP and failed to pr ev ent the CCCP-induced r eduction in normalized FPF ( Suppl. Figure 2 E, F).Catalase likewise failed to block the inhibitory effect of CCCP on normalized FREQ [ Figure 8 C; control FREQ = 10.2 cpm vs 4.3 cpm (39.3% of control) during last 2 min of CCCP], but the inhibition was rescued by subsequent addition of GLIB (1 μm ).The summary data in Figure 8 D indicate that the reduction in normalized FREQ was significant.Catalase pr etr eatment did not alter the effect of CCCP on contraction AMP ( Suppl. Figure 3 E) but failed to pr ev ent the CCCP-induced r eduction in normalized FPF ( Suppl. Figure 3 F).These results suggest that CCCP does not inhibit lymphatic contraction through the production of ROS.

Discussion
Our study is the first to evaluate the effects of metabolic inhibitors on K ATP channels in l ymphatic m uscle.Thr ee different inhibitors of mitochrondrial ATP production, two that inhibited the ETC and one that inhibited oxidati v e phosphor ylation, each retarded the ionic pacemaker that drives spontaneous lymphatic contractions and acti v e l ymph pumping.The inhibitor of ETC complex III, antimycin A, acted through the production of ROS, as its effects on contraction fr equency wer e pr ev ented by pr etr eatment with either of two R OS scav engers, tir on and catalase.The ETC complex I inhibitor (rotenone) and the oxidati v e phosphor ylation inhibitor CCCP did not incr ease R OS pr oduction and their effects were not blocked by ROS scavengers, consistent with the conclusion that K ATP channel acti v ation in these cases was mediated by a decrease in the intracellular ATP/ADP ratio. 57Three important findings from our study were  that: (1) the effects of all three metabolic inhibitors on contraction frequency and pumping could be reversed by a relatively low and selecti v e concentr ation of the K ATP c hannel inhibitor GLIB; (2) ETC inhibition caused a selecti v e r eduction in contraction frequency without a concomitant inhibition of amplitude; and (3) mice lacking functional K ATP channels in lymphatic muscle ( Kir6.1 −/ − or Myh11-CreER T2 ; Kir6.1[AAA] mice) were completely resistant to the effects of the metabolic inhibitors.Further confirmation that the inhibitors worked thr ough acti v ation of K ATP channels in lymphatic muscle came from direct measurements of V m in LMCs in whic h antim ycin A stopped the generation of spontaneous action potentials, and that inhibition was reversed by GLIB.
The effects of ETC inhibition tested in our study were intended to simulate the chronic metabolic stress associated with metabolic disease.Obesity and insulin r esistance ar e linked to decreased efficiency of ATP synthesis and an increase in mitochondrial R OS pr oduction. 36This imbalance of energy pr oduction vs utilization is a hallmark of metabolic syndrome.While direct measures of ROS production in lymphatic vessels in animal models of metabolic disease in response to metabolic disorders are lacking, metabolic dysfunction induced by high fat or Western-style diets (ie, high fat and processed carbohydrate) are known to increase ROS production in skeletal muscle and cer ebral arter y smooth muscle. 45 , 58A key finding of our study is that while effects of ETC complex I and oxidati v e phosphorylation inhibition are likely reliant upon reduced ATP/ADP, the effects of complex III inhibition appear to be mediated thr ough R OS pr oduction.Scav enging super oxide with tir on or H 2 O 2 with catalase pr ev ented changes in K ATP -dependent lymphatic pacemaker modulation in response to complex III inhibition.It is somewhat surprising that the effects of complex III inhibition could be full y pr ev ented by R OS scav enging as antimycin A also would be expected to have an effect on the ATP/ADP ratio.Nonetheless, modulation of mitochondrial energetics had profound effects on lymphatic contractile function thr ough m ultiple mechanisms of K ATP acti v aton.
ATP is a negati v e r egulator of the K ATP channel, with channel acti vity incr easing as ATP lev els fall. 30But how might ROS activate the K ATP channel?At least two other studies have demonstrated R OS-mediated acti v ation of K ATP channels. 59 , 60Although the exact mechanism is unknown and may v ar y with the ROS species and/or K ATP subunit composition, one study found that H 2 O 2 lowers the sensitivity of the channel to ATP, implying that R OS effects ar e mediated by the SUR channel subunit. 60ROS species can also acti v ate pr otein kinases 61 and R OS-mediated PKA acti v ation could be another potential mechanism for the incr eased acti vity of LMC K ATP channels. 25 , 62Howe ver, in vestigating these mechanisms in lymphatic vessels was beyond the scope of this study.
We have assumed that the K ATP channels controlling lymphatic muscle pacemaking are located on the LMC sarcolemma, based in part on the observation that antimycin A-induced inhibition of ETC inhibited AP generation in LMCs ( Figure 4 ).Howe ver, e vidence suggests that K ATP channels are also expressed on the mitochondrial inner membrane (mitoKATP), where they r egulate ATP pr oduction and alter mitochondrial Ca 2 + uptake by changing the mitochondrial membrane potential. 57The molecular identity of mitoKATP has been r ecentl y described, MitoK ( Ccdc51 ) and MitoSur ( Abcb8 ), 63 but assessment of its role in most intact cell pr e parations is still limited to the use of pharmacological tools such as the inhibitor 5-hydroxydecanoate and the acti v ator diazoxide.8][69] Gi v en the limitations of these modulators and the intact vessel techniques employed in the present study, w e w ould be una b le to definiti v el y discriminate between mitoKA TP and sarcKA TP in the regulation of lymphatic vessel pacemaking.However, this remains an area for further investigation.
A major implication of our results is that acute metabolic stress leads to the activation of K ATP channels in lymphatic m uscle, r etarding the ionic pacemaker and inhibiting the spontaneous contractions that are critical for active lymph pumping.Although K ATP channel activity makes little contribution to the pacemaking frequency of murine lymphatic vessels under normoxic conditions, 35 , 48 acti v ation of this channel acts as a powerful brake on pacemaking 62 by inhibiting the r e petiti v e cycle of diastolic depolarization and AP firing involving anoctamin1 and L-type Ca 2 + channels [for details see recent re vie ws 62 , 70 ].Given these findings, we propose that K ATP channel acti v ation is a possible explanation for the impaired lymphatic contractile function and/or lymph transport observed in a number of animal models of metabolic disease.It is interesting that many of these studies, like ours, show an impairment-often a selecti v e impairment-in lymphatic contraction frequency.Several examples are notable.(1) In mice and rats of advanced age, selecti v e r eductions in basal l ymphatic contraction fr equency 71 and near-elimination of flow-mediated contraction frequency re gulation 13 w ere noted in comparison to vessels from younger animals.(2) In mice fed a high fat diet (HFD) to induce obesity, a reduced packet frequency of dye transport was observed in lymphatic vessels of the hindlimb. 15 , 16(3) In ApoE −/ − mice on a HFD, contraction frequency and FPF of ex vi v o popliteal l ymphatics wer e impair ed at low pr essur es. 6(4) In a mouse model of TNF α ov er expr ession, popliteal l ymphatic v essels showed a reduced spontaneous contraction frequenc y o ver a wide range of intraluminal pr essur e. 18 (5) In a high fructose rat model of metabolic syndrome, contraction frequency and calculated pump flow wer e r educed 40-50%, depending on the intraluminal pr essur e lev el, with no significant change in contr action amplitude , str oke v olume, or ejection fraction. 21 (6)  In a rat model of experimental ileitis associated with increased prostanoid and nitric oxide production, spontaneous lymphatic contr actions w er e nearl y a bolished. 23 , 25Although these findings are consistent with the conclusion that K ATP channels mediated the reductions in contraction frequency, only the latter two studies specifically tested their possible role.In the experimental ileitis model, GLIB (10 μm ) treatment essentially restored the lymphatic contraction frequency to normal levels, 25 and in the meta bolic syndr ome model, GLIB (10 μm ) partiall y r estor ed the de pr ession in contraction fr equency. 22Collecti v el y, these findings and ours point to a nearly selective impairment of l ymphatic contraction fr equency in models of meta bolic str ess, with little or no effect on contraction amplitude, suggesting that a common endpoint of the meta bolic str ess associated with these various disease models is primarily to inhibit the ionic pacemaker driving spontaneous contractions as opposed to interfering with excitation-contraction coupling.][74] It will be important to test the extent to which GLIB treatment [at low er concentr ations that ar e mor e specific for K ATP channels in l ymphatic m uscle 48 ] can r estor e impair ed l ymphatic function in other animal models of metabolic disease and if animals with lymphatic muscle deficient in Kir6.1 channels are resistant to the lymphatic contractile dysfunction associated with those models.
The preceding discussion is not meant to imply that K ATP channels are the sole cause of lymphatic dysfunction in metabolic disease or to downplay the importance of other signaling mechanisms that might also negati v el y impact lymphatic contr actile amplitude .Indeed, tw o of the studies cited a bov e 6 , 18 noted impairments in lymphatic contraction amplitude as well as frequency.Related studies have demonstrated this as well.For example, Liao et al. 20 found that upregulation of nitric oxide (NO) pr oduction thr ough iNOS (by injecting acti v ated macr opha ges or CD11b + Gr-1 + cells expressing iNOS) led to the suppression of lymph transport in the mouse hindlimb.The effect appeared to be mediated both by changes in contraction amplitude and frequency, although (as with other in vivo models) a component of the frequency changes may have been due to systemic compensation in order to maintain fluid balance.In these and other studies, the effects on amplitude are likely mediated in part by nitric oxide (NO).For example, in a TNF α ov er expr ession mouse model, the impairment in both contraction frequency and amplitude of ex vi v o l ymphatic v essels w as partiall y r escued by b loc king NO synthase . 18Likewise , iNOS inhibition with 1400 W largel y r estor ed the l ymphatic contractions lost during experimental ileitis. 25Inhibition of eNOS by L-NAME partiall y r ev ersed the impairment in lymphatic contraction frequency and amplitude in a mouse model of TNF α ov er expr ession, 18 a rat model of aging 71 and a rat model of metabolic syndrome. 22The effects of NO on lymphatic contractile function are complex, but in general, low levels of NO production associated with shear-str ess acti v ation of eNOS r esult in slight enhancement of lymph pump output 75 , 76 whereas higher levels of NO production, often associated with imposed flow or iNOS activ ation, generall y de pr ess both l ymphatic contraction fr equency and amplitude. 18 , 22 , 48 , 62 , 71To complicate matters, NO appears to induce K ATP channel acti v ation in some species 25 but not others, 48 perhaps depending on the r elati v e acti vities of PKA and PKG in lymphatic muscle [for more detail see 62 ].Thus, the extent to which K ATP channels may be acti v ated dir ectl y by metabolic str ess or secondar y to NO and/or pr ostanoid pr oduction is likel y to v ar y with the disease model.
In the context of metabolic disease, lymphatic contractile dysfunction can potentially trigger a complicated sequelae of long-term changes in ov erall l ymphatic function and the interstitium.Suc h c hanges include peri-lymphatic inflammation, CD4 + cell infiltration, lipid accumulation, and fibrosisall of which are associated with R OS pr oduction, iNOS acti v ation, and/or the production of cytokines and prostanoids. 24 , 77-80or example, in addition to an impairment in lymphatic contractile function, 15 , 16 obesity-prone mice show reduced lymphatic transport of macromolecules draining to lymph nodes along with decreased density of lymphatic vessels and reduced expr ession of l ymphatic markers, all suggesti v e of both shortand long-term remodeling of the lymphatic vasculature and surrounding tissue. 17The long-term changes in lymphatic function may negati v el y impact the l ymphatic endothelium, leading to collecting v essel hyperpermea bility 6 , 28 , 29 and compromised integrity of lymphatic valves. 6 , 26 , 27These, together with inhibition of the acti v e l ymph pump thr ough K ATP channel acti v ation and/or NO production would contribute to overall lymphatic system dysfunction.
In summary, acute metabolic stress, either through reductions in the intracellular ATP/ADP ratio or increased ROS production, leads to acti v ation of K ATP channels in lymphatic muscle and inhibition of the ionic pacemaker driving spontaneous lymphatic contractions and active lymph transport.Mice lacking functional K ATP channels in lymphatic muscle cells are resistant to the effects of acute metabolic stress, pointing to a common role for K ATP channels in the impaired lymphatic contractile function observed in a number of metabolic diseases and raising the possibility that K ATP channels in lymphatic muscle may be a via b le therapeutic target.
Fluor escence ima ges wer e acquir ed for 100 ms at 1 min intervals for 15 min with 10 × (Olympus UPlanApo N.A. = 0.40) or 20 × (Olympus UPlanSApo N.A. = 0.75) objecti v es coupled to an EMCCD camera (Photometrics Cascade II) on an Olympus IX81 inverted microscope.Illumination was provided by an Andor/Yokogawa CSU-X Confocal Spinning Disk system with excitation at 472/30 nm and emission at 525/35 nm.Fluorescence intensity was quantified with ImageJ (National Institutes of Health) in a region of interest located in the middle of a vessel following subtraction of background fluorescence.As a positi v e contr ol for generating R OS , the mitoc hondria complex III inhibitor, antimycin A (1 μm ) was added to the superfusion solution.44To verify the sensitivity of DHR to endogenous R OS pr oduction, experiments were repeated following 10 min of preincubation with tiron (1 m m ) in combination with polyethylene glycol (PEG)-catalase 250 U mL −1 .The r especti v e r ea gents wer e pr esent thr oughout the r est of the experiment.Values for DHR fluor escence ar e expr essed in arbitrar y units for the change from baseline within a ROI ( = fluorescence at x min − fluorescence at 0 min, where x r e pr esents 1 min intervals during chemical exposure).45Solutions and ChemicalsKrebs buffer contained (in m m ) 146.9 NaCl, 4.7 KCl, 2 CaCl 2 •2H 2 O, 1.2 MgSO 4 , 1.2 NaH 2 PO 4 •H 2 O, 3 NaHCO 3 , 1.5 NaHEPES, and 5 Dglucose (pH = 7.4

Figure 1 .
Figure 1.Antimycin A reduces lymphatic muscle contraction frequency and the inhibition is re versed b y GLIB.Responses of mouse popliteal lymphatics, pressurized to 3 cmH 2 O, exposed to the mitochondrial ETC complex III inhibitor, antimycin A (30 n m ).Example recordings of vessels from ( A ) wild-type (WT), ( B ) Kir6.1 − / − and ( C ) Myh11-Cr eER T2 ; Kir6.1[AAA] mice .During the last 2 min of the r ecording period, the fr equency of onl y the WT v essel w as r educed by antimycin A and that r eduction w as largel y r escued by GLIB (1 μm ).( D ) Summar y of changes in normalized fr equency for the differ ent tr eatments and genotypes.Gray bars r e pr esent WT v essels ( N = 7; n = 10), red bars are Kir6.1 − / − vessels ( N = 6; n = 5) and green bars are Myh11-CreER T2 ; Kir6.1[AAA] vessels ( N = 4; n = 6).All data are means ± SD. * P < 0.05 between WT vessels before and after treatment of antimycin A and GLIB using a one-way ANOVA with Tukey's post-hoc tests.
Myh11CreER T2 enabled us to focus on the fluorescence signal in the LMC layer per se, while nifedipine (1 μm ) was used to completely block any wall movement associated with spontaneous contractions.The DHR signal in the focal plane of the LMCs is shown in the first 2 columns of Figure 5 (at 0 and 15 min), with the td-Tomato signal in the third column and the merged signal in the fourth column.A contr ol v essel is shown in r ow A , a v essel tr eated with antimycin A (30 n m ) in row B , a vessel treated with rotenone (100 n m ) in row C , a vessel treated with CCCP (1 μm ) in row D , a vessel treated with antimycin A (1 μm ) in row E , and a v essel tr eated with antimycin A (1 μm ) after pr etr eatment with tiron (1 m m ) and Catalase (250 U/mL) in row F .The time courses of the DHR signals under these various conditions are plotted in Figure 5 G.

Figure 2 .
Figure 2. Rotenone r educes l ymphatic m uscle contraction fr equency and the inhibition is r ev ersed by GLIB.Re pr esentati v e examples of the effects of the mitochondrial ETC complex I inhibitor, rotenone (100 n m ) on spontaneous contractions of mouse popliteal l ymphatic v essels fr om ( A ) WT, ( B) Kir6.1 − / − , and ( C ) Myh11-CreER T2 ; Kir6.1[AAA] mice.The frequency of only the WT vessel was reduced by rotenone and this effect was largely rescued by GLIB (1 μm ).( D ) The summary graphs showing normalized frequency for the various treatments and genotypes.Gray bars represent WT vessels ( N = 6; n = 8), red bars are Kir6.1 − / − vessels ( N = 3; n = 6) and green bars are Myh11-CreER T2 ; Kir6.1[AAA] vessels ( N = 3; n = 6).All data are means ± SD. * P < 0.05 between WT vessels before and after treatment of rotenone and GLIB using a one-way ANOVA with Tukey's post-hoc tests.

Figure 4 .
Figure 4. Antimycin A inhibits action potentials in LMCs and GLIB counteracts its effects.Raw traces showing that antimycin A (30 n m ) inhibits the frequency of spontaneous lymphatic contractions ( A ) and action potentials (APs).Membrane potential (V m ) was recorded using an intracellular electrode in a LMC of a WT mouse popliteal lymphatic vessel pressurized at 3 cmH 2 O.The recording was made in the presence of 1 μm wortmannin to minimize vessel wall movement.( B ) Magnification of APs in panel A prior to drug ( a ), in the presence of antimycin A ( b ), and antimycin A and GLIB (1 μm ), which r estor ed AP generation ( c ). ( C ) Summary of resting V m values in lymphatic muscle cells before and after treatment of antimycin A and/or GLIB ( N = 8; n = 9).( D ) Summary of changes in AP (and concomitant contraction) fr equency befor e and after tr eatment of antimycin A and GLIB.All data ar e means ± SD. * P < 0.05 using a one-w a y ANOVA with Tuke y's post-hoc tests.

F igure 6 .
Sca v enging R OS pr ev ents the effects of antimycin A on lymphatic contraction.Raw traces of lymphatic contractions in WT vessels in response to antimycin A (30 n m ) after pr etr eatment with ( A ) tiron (1 m m ) or ( B ) catalase (250 U/mL) for 10 min and after subsequent addition of GLIB (1 μm ).Summary of normalized frequency c hanges to antim ycin A with and without C ) tiron ( N = 6; n = 8) and ( D ) catalase ( N = 4; n = 7).All data are means ± SD. * P < 0.05 before and after treatment of rotenone and GLIB using a one-way ANOVA with Tukey's post-hoc tests.

F igure 7 .
ROS sca vengers do not prevent effects of rotenone on lymphatic contraction.Raw traces showing inhibition of lymphatic contraction frequency in WT vessels in response to rotenone (100 n m ) after pretreatment with m ( A ) tiron (1 m m ) or ( B ) catalase (250 U/mL) for 10 min and after subsequent addition of GLIB (1 μm ).Summary of normalized frequency changes to rotenone with and without C ) tiron ( N = 5; n = 8) or ( D ) catalase pretreatment ( N = 7; n = 10).All data are means ± SD. * P < 0.05 before and after treatment of rotenone and GLIB using a one-way ANOVA with Tukey's post-hoc tests.

F igure 8 .
ROS sca vengers do not prevent effects of CCCP inhibition on lymphatic contraction.Representative traces of lymphatic contractions in WT vessels in response to CCCP 1 μm pr etr eated with ( A ) tiron (1 m m ) or ( B ) catalase (250 U/mL) for 10 min and after subsequent addition of GLIB (1 μm ).Summary of normalized frequency changes to CCCP with and without C ) tiron ( N = 5; n = 8) or D ) catalase ( N = 4; n = 7) pr etr eatment.All data are means ± SD. * P < 0.05 before and after treatment of CCCP and GLIB using a one-way ANOVA with Tukey's post-hoc tests.