Depressive behavior and vascular dysfunction: a link between clinical depression and vascular...

Depressive behavior and vascular dysfunction: a link between clinicaldepression and vascular disease?

Alexandre C. d’Audiffret,1,2 Stephanie J. Frisbee,1,3 Phoebe A. Stapleton,1,4 Adam G. Goodwill,1,5

Elsa Isingrini,6 and Jefferson C. Frisbee1,5

1Center for Cardiovascular and Respiratory Sciences, Divisions of 2Vascular and Endovascular Surgery and 4ExercisePhysiology, and Departments of 3Community Medicine and 5Physiology and Pharmacology, West Virginia University Schoolof Medicine, Morgantown, West Virginia; and 6Université François Rabelais de Tours, Tours, France

Submitted 30 December 2009; accepted in final form 12 February 2010

d’Audiffret AC, Frisbee SJ, Stapleton PA, Goodwill AG,Isingrini E, Frisbee JC. Depressive behavior and vascular dys-function: a link between clinical depression and vascular disease?J Appl Physiol 108: 1041–1051, 2010. First published February 18, 2010;doi:10.1152/japplphysiol.01440.2009.—As chronic stress and depressionhave become recognized as significant risk factors for peripheralvascular disease in patients with no prior history of vasculopathy, weinterrogated this relationship utilizing an established mouse model ofchronic stress/depressive symptoms from behavioral research. Malemice were exposed to 8 wk of unpredictable chronic mild stress(UCMS; e.g., wet bedding, predator sound/smell, random disruptionof light/dark cycle), with indexes of depressive behavior (coat status,grooming, and mobility) becoming exacerbated vs. controls. In vas-cular rings, constrictor (phenylephrine) and endothelium-independentdilator (sodium nitroprusside) responses were not different betweengroups, although endothelium-dependent dilation (methacholine) wasattenuated with UCMS. Nitric oxide synthase (NOS) inhibition waswithout effect in UCMS but nearly abolished reactivity in controls,while cyclooxygenase inhibition blunted dilation in both. Combinedblockade abolished reactivity in controls, although a significant dila-tion remained in UCMS that was abolished by catalase. Arterial NOproduction was attenuated by UCMS, although H2O2 production wasincreased. UCMS mice demonstrated an increased, although variable,insulin resistance and inflammation. However, while UCMS-inducedvascular impairments were consistent, the predictive power of aggre-gate plasma levels of insulin, TNF-�, IL-1�, and C-reactive peptidewere limited. However, when separated into tertiles with regard tovascular outcomes, insulin resistance and hypertension were predic-tive of the most severe vascular impairments. Taken together, thesedata suggest that aggregate insulin resistance, inflammation, andhypertension in UCMS mice are not robust predictors of vasculardysfunction, suggesting that unidentified mechanisms may be superiorpredictors of poor vascular outcomes in this model.

vasodilation; endothelium-derived factors; peripheral vascular dis-ease; nitric oxide; models of chronic stress; clinical depression

CLINICAL DIAGNOSES of depression and depressive symptoms arebecoming increasingly common among patients with coronaryartery and peripheral vascular disease, with �20% of patientsdisplaying angiographic evidence of vascular disease sufferingfrom major depressive disorders (21). In addition, previousstudies have provided evidence that clinical depression ordepressive symptoms of sufficient severity can serve as inde-pendent risk factors for ischemic heart disease (20). However,

in recent clinical trials, antidepressant therapy failed to im-prove cardiac outcomes among patients with major depression(3, 4), suggesting that a distinct pathophysiological mechanismmay be responsible for the compelling associations betweenclinical depression and the evolution of vascular disease. Pa-tient-based studies have suggested that physiological differ-ences are present between patients suffering from depression/depressive symptoms and control patients (4, 5), with onestudy suggesting that platelet nitric oxide synthase (NOS)activity and plasma nitric oxide (NO) levels were reduced inpatients with depressive disorders (8). However, the validity ofthese reports has been criticized as suffering from a poor levelof control, and heterogeneity in data acquisition procedures. Tofurther interrogate the characteristics linking clinical depres-sion with vascular disease, the contributing mechanisms, func-tional outcomes, and potential for interventional success, it isnecessary to appropriately integrate an animal model for de-pression/depressive symptoms with data collected using humancohorts.

Previously, investigators developed the unpredictable chronicmild stress (UCMS) model for developing depression/depression-like behaviors in rodents (16, 22, 26, 28), with the underlyingassumption that a common source for developing depression isan environment of continual stressors resulting from uncon-trollable exogenous factors, and the genetic component ofdepressive behavior being a function of heritable differences insusceptibility to imposed stressors (16, 19). The UCMS modelis considered to be the most appropriate rodent model forhuman clinical depression, based on its ability to reproduce thedevelopment of many human depressive symptoms, includinganhedonia and learned helplessness (16, 19, 25).

Although the UCMS model of depression in mice has beenwell established in the behavioral sciences, the application ofthis model to interrogate the association between depressivedisorders and peripheral vascular disease has received littleattention. The purpose of this study was to determine therelationships between chronic stress/depressive symptoms andvascular function in a mouse strain that is highly susceptible tothe development of depression/depressive symptoms as a resultof the UCMS protocol, the BALB/cJ (16, 27). This studyevaluated the hypothesis that the UCMS protocol, leading todepressive behaviors in mice, would cause significant impair-ments to vascular reactivity that are consistent with the devel-opment of peripheral vascular disease, including increasedconstrictor reactivity and blunted dilator responses. Resultsfrom this study suggest that chronic depressive symptoms inmice strongly associated with poor vascular endothelial func-tion and that this may have significant implications for the

Address for reprint requests and other correspondence: J. C. Frisbee, Centerfor Cardiovascular and Respiratory Sciences, Dept. of Physiology and Phar-macology, Robert C. Byrd Health Sciences Center, P.O. Box 9105, WestVirginia Univ. School of Medicine, Morgantown, WV 26505 (e-mail:[email protected]).

J Appl Physiol 108: 1041–1051, 2010.First published February 18, 2010; doi:10.1152/japplphysiol.01440.2009.

8750-7587/10 $8.00 Copyright © 2010 the American Physiological Societyhttp://www.jap.org 1041

regulation of vascular tone, thrombosis, atherogenesis, andtissue/organ perfusion.

MATERIALS AND METHODS

Animals. Male BALB/cJ mice (Jackson) were fed standard chowand drinking water ad libitum and were housed in the animal carefacility at the West Virginia University Health Science Center (WVUHSC). All protocols received prior institutional animal care and usecommittee (IACUC) approval. At 9 wk of age, mice were divided into

two groups, control and UCMS (below). After 8 wk duration undereither condition, mice were anesthetized with injections of pentobar-bital sodium (50 mg/kg ip), and a carotid artery was cannulated fordetermination of arterial pressure. Venous blood aliquots were col-lected for biochemical evaluation of plasma markers of dysfunction(below).

UCMS protocol. All mice were singly housed, with the controlgroup in a separate, nearby quiet room. Alternatively, in the UCMSgroup, mice were randomly exposed to the following stressors onmultiple occasions throughout each 24-h period: 1) damp bedding: 10oz of water was added to each standard cage for the next 3 h; 2) water:all bedding was removed and �0.5 in. of water was added to emptycage for the next 3 h; 3) each cage was tilted to 45° with or withoutbedding for 3 h; 4) social stress: each mouse was switched into a cageof a neighboring mouse for 3 h; 5) no bedding lasting for 3 h or, ontwo occasions each week, overnight; 6) succession of light/darkcycles, lasting 30 min throughout a 24-h period; and 7) exposure topredator smells (e.g., cat fur) and/or sounds (e.g., cat growling).

After 8 wk, all mice were subjected to a series of behavioral teststhat were aimed at evaluating the outcomes of the UCMS procedures.

Coat status. This evaluation was done throughout the duration ofthe UCMS protocol. The total cumulative score was computed bygiving an individual score of 0 (clean) or 1 (dirty) to eight body parts(head, neck, dorsal coat, ventral coat, tail, forelimb, hindlimb, andgenital region).

Table 1. Baseline characteristics between mouse groups inthe present study

Control UCMS

Mass, g 31 � 2 29 � 4MAP, mmHg 84 � 5 93 � 4Insulinplasma, ng/ml 1.3 � 0.2 3.9 � 0.6*Glucoseblood, mg/dl 102 � 8 104 � 7Cholesterolplasma, mg/dl 75 � 8 81 � 8Triglyceridesplasma, mg/dl 102 � 8 128 � 10*Nitrotyrosineplasma, ng/ml 17 � 2 27 � 4*

Values are means � SE. All mice are male, aged 17–18 wk. UCMS,unpredictable chronic mild stress; MAP, mean arterial pressure. *P � 0.05 vs.control.

Fig. 1. Depressive symptoms following 8 wk of the unpredictable chronic mild stress (UCMS) protocol in BALB/cJ mice. Data are presented for coat status (A),the latency (B) and frequency (C) of facial grooming following a 10% sucrose solution spray, and the total period of immobility during the tail suspension test(D) for control and UCMS mice. *P � 0.05 vs. control; n � 25 for control; n � 26 for UCMS.

1042 VASCULAR DYSFUNCTION WITH DEPRESSIVE SYMPTOMS

J Appl Physiol • VOL 108 • MAY 2010 • www.jap.org

Splash test. This test was used to evaluate acute grooming behav-ior, defined as cleaning of the fur by licking or scratching. A 10%sucrose solution was sprayed on the dorsal coat of each mouse, andgrooming activity was recorded for 5 min. The viscosity of the sucrosesolution will dirty the coat and induce grooming behavior, withdepressive symptoms characterized by an increased latency (idle timebetween spray and initiation of grooming) and decreased frequency(number of times grooming a particular body part).

Tail suspension test. Mice subjected to the short term, inescapablestress of tail suspension will develop an immobile posture, with longerperiods of immobility in mice exhibiting a depressed behavior. Micewere suspended by the tail on a horizontal bar 35 cm from the baseplatform using adhesive tape. The latency to the first bout of immo-bility and the duration of total immobility were recorded for 6 min.

Measurements of vascular reactivity. In each mouse, the thoracicaorta was removed, rinsed in physiological salt solution, cleared ofsurrounding tissue, and cut in 2- to 3-mm segments. Each ring wasmounted in a myobath chamber between a fixed point and a forcetransducer (World Precision Instruments) and set to 0.5-g tension for45 min to equilibrate. The organ baths contained physiological salt

solution at 37°C and were aerated with 95% O2-5% CO2. Rings werepreconditioned by treatment with 10�7 M phenylephrine for 5 min, atwhich time 10�5 M methacholine was added to the bath to assessendothelial integrity. Any ring that failed to demonstrate both a briskconstrictor response to phenylephrine and viable endothelial functionwas discarded. Subsequently, rings were treated with increasingconcentrations of phenylephrine (10�10 M–10�3 M) to assess con-strictor reactivity. For the assessment of dilator reactivity, rings werepretreated with 10�6 M phenylephrine and exposed to increasingconcentrations of methacholine (10�10 M–10�3 M) and sodiumnitroprusside (10�10 M–10�3 M). To assess the roles of nitric oxide,cyclooxygenase and hydrogen peroxide in modulating vascular re-sponses to the agonist treatments, concentration-response curves werealso conducted following treatment of the rings with N�-nitro-L-arginine methyl ester (L-NAME) (10�4 M), indomethacin (10�5 M),and catalase (10�4 M), respectively.

Measurement of vascular nitric oxide and hydrogen peroxidebioavailability. From a cohort of mice within each group, the aortawas used to assess vascular NO and H2O2 production using ampero-

Fig. 2. Constrictor responses of aortic rings to increasing concentrations of phenylephrine from mice under control conditions (open symbols: Œ/e) and after 8 wk ofUCMS (filled symbols: �/�). Data are presented under control conditions (A; control, upper bound � 3.2 � 0.2 g; UCMS, upper bound � 3.3 � 0.3 g), in responseto NOS inhibition with N�-nitro-L-arginine methyl ester (L-NAME) (B; control L-NAME, upper bound � 3.8 � 0.2 g; UCMS L-NAME, upper bound � 3.5 �0.2 g), cyclooxygenase inhibition with indomethacin (INDO) (C; control INDO upper bound � 3.3 � 0.3 g; UCMS INDO, upper bound � 3.4 � 0.3 g), or both(D; control both, upper bound � 3.9 � 0.3 g; UCMS both, upper bound � 3.5 � 0.3 g). *P � 0.05 vs. responses in untreated vascular rings from control mice;n � 10–11 for control; n � 10–12 for UCMS.

1043VASCULAR DYSFUNCTION WITH DEPRESSIVE SYMPTOMS


metric sensors (World Precision Instruments). Briefly, aortas wereisolated, cleaned, and placed within a sealed chamber filled withphysiological salt solution equilibrated with 95% O2-5% CO2. Withinthis chamber, a nitric oxide sensor (ISO-NOPF 100) and a hydrogen

peroxide sensor (ISO-HPO-100) were inserted and a baseline currentwas obtained. Subsequently, methacholine (10�6 M) was added to thechamber and the changes in current were determined. To verify thatresponses represented NO and H2O2 release, these procedures were



repeated following addition of L-NAME (10�4 M) or catalase (10�4

M) to the chamber.Analyses of endothelial NOS expression. For determination of

endothelial NOS (eNOS) expression, aortas were homogenized andproteins were separated under denaturing conditions on 10% SDS-polyacrylamide gel, which were then transferred to a PVDF mem-brane and blocked. Subsequently, blots were incubated with mouseanti-eNOS/NOS Type III mAb (BD Transduction Laboratories),washed, and incubated with appropriate horseradish peroxidase-con-jugated secondary antibody. GE Healthsciences ECL advance kitswere used to visualize proteins, and all eNOS bands were normalizedto expression of actin.

Biochemical analyses. In all samples, blood glucose was deter-mined using a commercially available glucometer (Freestyle). Lipidpanels for cholesterol and triglycerides were performed using com-mercially available kits (Wako). Additionally, oxidant stress wasassessed through determination of plasma nitrotyrosine (Cayman).Finally, using a multiplexed procedure, plasma insulin and markers ofinflammation were assessed using commercially available kits (Mil-lipore).

Data and statistical analyses. Mechanical responses followingchallenge with methacholine or phenylephrine were fit with thethree-parameter logistic equation:

y � min � � max�min

1 � 10logED50�x�where y represents the isometric tension; min and max represent thelower and upper bounds, respectively, of the change in tone withagonist concentration; x is the logarithm of the agonist concentration;and logED50 represents the logarithm of the agonist concentration (x)where the response (y) is halfway between the bounds. The use of thethree-parameter logistic equation is appropriate for the analysis ofsigmoidal concentration-response relationships, as it simultaneouslyprovides estimates of the curve maximum (upper bound), minimum(lower bound) and the dose at which the dependent variable reaches50% of maximum (ED50).

All data are presented as means � SE. Significant differencesbetween groups were determined using ANOVA. In all cases, Stu-dent-Newman-Keuls post hoc test was used when appropriate andP � 0.05 was taken to reflect statistical significance.

RESULTS

Table 1 presents data describing the characteristics of mousegroups in the present study. At the time of use, plasma insulin,triglyceride concentrations, and plasma levels of nitrotyrosinewere significantly greater in UCMS mice than in controls. Nodifferences were determined between body mass, mean arterialpressure, plasma glucose, and cholesterol levels.

The impact of the UCMS protocol on depressive behaviorsin mice is summarized in Fig. 1. Throughout the duration of theUCMS period, coat status in mice undergoing the stress pro-tocol was consistently poorer compared with that in controls(Fig. 1A). In response to the sucrose spray, control micedemonstrated both a more rapid (Fig. 1B) and a more frequent(Fig. 1C) grooming response compared with that exhibited by

mice following the UCMS protocol. In response to tail suspen-sion, control mice demonstrated a significantly shorter periodof immobility compared with that in UCMS mice (Fig. 1D).

Figure 2 summarizes the constrictor responses of aortic ringsfrom the two mouse groups in response to increasing concen-trations of phenylephrine. As shown in Fig. 2A, there was noevidence of an altered constrictor reactivity of rings fromUCMS mice in response to phenylephrine. However, whentreated with the NOS inhibitor L-NAME (Fig. 2B), the con-strictor responses of rings from control mice increased signif-icantly, a response not present in UCMS. Treatment of vascularrings from either group with indomethacin had no impact onthe constrictor response to phenylephrine (Fig. 2C), and com-bined treatment with both L-NAME and indomethacin was notdifferent from that for L-NAME alone (Fig. 2D).

The dilator responses of vascular rings from control andUCMS mice are summarized in Fig. 3. Rings from UCMSmice demonstrated a significant reduction in methacholine-induced dilation compared with responses in rings from controlmice (Fig. 3A). While pretreatment of rings with L-NAMEnearly abolished methacholine-induced reactivity in controlmice, NOS blockade had only a minor impact on dilation inrings from UCMS mice (Fig. 3B). Further, while methacho-line-induced dilation was only mildly impacted by treatmentwith indomethacin in both groups (Fig. 3C), treatment with

Fig. 3. Dilator responses of aortic rings to increasing concentrations of methacholine from mice under control conditions (open symbols: Œ/e) and following8 wk of UCMS (filled symbols: �/�). Data are presented under control conditions (A; control, lower bound � 10.3 � 2.4%; UCMS, lower bound � 33.6 �4.4%), in response to NOS inhibition with L-NAME (B; control L-NAME, lower bound � 84.5 � 3.2%; UCMS L-NAME, lower bound � 54.6 � 5.4%),cyclooxygenase inhibition with INDO (C; control INDO, lower bound � 21.2 � 4.1%; UCMS INDO, lower bound � 43.4 � 5.2%), or both (D; control both,lower bound � 93.4 � 2.4%; UCMS both, lower bound � 59.2 � 5.6%). Data are also presented for vascular responses to increasing concentrations of sodiumnitroprusside (E; control, lower bound � 2.5 � 1.1%; UCMS, lower bound � 2.9 � 1.3%). *P � 0.05 vs. responses in untreated vascular rings from control mice;†P � 0.05 vs. responses in untreated vascular rings from UCMS mice; n � 8–9 for control; n � 7–9 for UCMS.

Fig. 4. Representative blots describing expression of endothelial nitric oxidesynthase (eNOS) from aortas in control and UCMS mice (A). B presentsdensitometric analysis of band intensity between control and UCMS mice; n �5 for control and for UCMS.



L-NAME and indomethacin produced results that were com-parable to that for L-NAME alone (Fig. 3D). The dilation ofaortic rings to increasing concentrations of the NO donorsodium nitroprusside did not differ between control and UCMSmice (Fig. 3E).

The arterial expression of eNOS in control and UCMS micewas not altered by the UCMS protocol (Fig. 4). These results

suggest that any change in NO bioavailability and the impairedNO-dependent component of methacholine-induced dilation inUCMS mice does not reflect a change in vascular eNOSexpression.

Figure 5 presents data describing the impact of treatment ofaortic rings from both mouse groups with catalase on metha-choline-induced dilation. Treatment of rings with catalase,while having minimal impact of dilator reactivity in control

Fig. 5. Dilator responses of aortic rings to increasing concentrations ofmethacholine from mice under control conditions (open symbols: Œ/e) andfollowing 8 wk of UCMS (filled symbols: �/�). Data are presented undercontrol conditions (A; control, lower bound � 16.6 � 3.9%; UCMS, lowerbound � 37.4 � 4.6%), following pretreatment of rings with the H2O2

scavenger catalase (B; control catalase, lower bound � 23.8 � 4.4%; UCMS catalase, lower bound � 72.3 � 6.1%), or following combined treatment withcatalase and L-NAME (C; control catalase/L-NAME, lower bound � 91.4 �2.8%; UCMS catalase/L-NAME, lower bound � 92.6 � 3.1%). *P � 0.05 vs.responses in untreated vascular rings from control mice; †P � 0.05 vs. responsesin untreated vascular rings from UCMS mice; n � 8 for control; n � 8–9 forUCMS.

Fig. 6. Aortic production of nitric oxide (A) and H2O2 (B) following metha-choline challenge in mice under control conditions and following 8 wk ofUCMS. Data are presented for arteries under control (unstimulated) conditions,following challenge with methacholine and following methacholine challengein the presence of NOS inhibition with L-NAME. *P � 0.05 vs. responses inunstimulated vessels in that group; †P � 0.05 vs. responses in vessels fromcontrol animals following challenge with methacholine; ‡P � 0.05 vs. re-sponses in vessels from that group following challenge with methacholine; n �5 for control; n � 6 for UCMS.



mice, significantly reduced dilator responses in UCMS mice(Fig. 5, A and B). Further, combined treatment of rings withL-NAME and catalase nearly abolished responses of vascularrings to methacholine in both control and UCMS groups(Fig. 5C).

The results of biochemical assays for methacholine-inducedNO and H2O2 production are summarized in Fig. 6, A and B,respectively. Aortas from control mice demonstrated a robustNO production in response to challenge with methacholinecompared with levels in unstimulated vessels, while that fromUCMS mice was significantly attenuated (Fig. 6A), parallelingthe mechanical responses presented in Fig. 3. In contrast,methacholine challenge resulted in a limited production ofH2O2 in aortas from control animals, yet caused a strongincrease in H2O2 release in vessels of UCMS (Fig. 6B),providing further support for the data presented in Fig. 5. Theproduction of H2O2 in response to application of methacholinein either group was unaffected by pretreatment with L-NAME.

Plasma levels of inflammatory cytokines in control andUCMS mice are summarized in Fig. 7. Although mean valueswere elevated for TNF-� (Fig. 7A), IL-1� (Fig. 7B), andC-reactive peptide (CRP; Fig. 7C) in UCMS mice comparedwith control, a high variability in these measurements betweenanimals within the UCMS group prevented differences from

reaching statistical significance. Plasma levels of IL-10 (Fig.7D) were not different between the two groups of mice.

Given the variability in markers of peripheral vascular dis-ease risk in the UCMS mice, it is important to determine theextent to which these substances were predictive of poorvascular outcomes in this model of chronic stress/depressivesymptoms. As presented in Fig. 8, plasma levels of TNF-�(Fig. 8A), insulin concentration (Fig. 8D), and nitrotyrosine(Fig. 8F) were well correlated with vascular reactivity incontrol mice (with r2 0.6 for each), with other risk factors[IL-1� (Fig. 8B), CRP (Fig. 8C), and arterial pressure (Fig.8E)] being less predictive. However, none of the measured riskfactors were strongly predictive of the vascular outcomes in theUCMS mice where r2 values ranged between 0.013 for insulinand �0.3 for both arterial pressure and nitrotyrosine.

However, if relationships in Fig. 8 are separated into tertileswith regard to vascular outcome, these data are more informa-tive. In Fig. 9, vascular dysfunction is separated as such, withthe first tertile representing mild vascular dysfunction (lowerbound of the methacholine-induced dilation �20% of initialtension), the second tertile representing moderate vasculardysfunction (lower bound between 20 and 40% of initialtension), and the third tertile representing severe vasculardysfunction (lower bound 40% of initial tension). When

Fig. 7. Plasma levels of markers of inflammation in control mice and in mice following 8 wk of UCMS. Data are shown for TNF-� (A), IL-1� (B), C-reactivepeptide (CRP) (C), and IL-10 (D); n � 24 for control and for UCMS.



Fig. 8. Correlations between vascular reactivity and individual vascular disease risk factors in control (open symbols, solid line) and UCMS mice (filled symbols,dashed line). Data are presented as the lower bound in the methacholine concentration-response curve for an individual mouse vs. plasma concentrations ofTNF-� (A), IL-1� (B), CRP (C), insulin (D), mean arterial pressure (E), and plasma levels of nitrotyrosine (F). Also presented are lines of best fit through thedata for each group, with the resulting equations and r2 values presented in the legend.



presented in this fashion where the correlation between specificrisk factors and the graded severity of vascular dysfunction inUCMS mice is presented, it is apparent that mild levels ofvascular dysfunction in response to UCMS are not well pre-dicted by any of the risk factors determined. In mice where amoderate reduction in dilator reactivity was present (secondtertile), this degree of vascular impairment was well predictedwhere a significant elevation in arterial pressure was theprimary outcome of the UCMS protocol. The strongest levelsof vascular dysfunction were observed in mice where theUCMS protocol was able to create a more insulin-resistant,hypertensive state.

DISCUSSION

With a growing body of clinical and epidemiological evi-dence suggesting a link between chronic stress/depression andthe progression of vascular disease, it is vital to utilize anappropriate animal model such that experiments can be de-signed that will interrogate these processes at a high level ofresolution, for mechanistic contributors, functional implica-tions, and interventional strategies. The UCMS model for theestablishment of depressive symptoms in rodents has beenrepeatedly validated within behavioral research and has dem-onstrated a remarkably broad utility. Not only is this modelused for general studies involving chronic stress and depres-sion (19, 25), it has also been employed as an excellent meansfor evaluating interventional and potentially ameliorative ther-apies (12, 19, 25), as the depressive state in rodents manifestsbehaviors that mimic those in human subjects afflicted withclinical depression (25, 27). One of the main underlyingassumptions of clinical depression and depressive symptoms isthat the presence of chronic exogenous stressors will initiatedepression/depressive symptoms in all afflicted subjects, butthat underlying genetic, heritable traits are directly relevant tothe severity of these symptoms rather than basic susceptibilityper se (25, 27). The veracity of this fundamental assumptionhas been evaluated previously in multiple mouse strains inresponse to the imposition of the UCMS protocol. Interest-ingly, across numerous mouse strains, the BALB/cJ was con-

sistently demonstrated as being among the most susceptible tothe UCMS protocol, with the most severe development ofdepressive symptoms as a result of the chronic stressors (10,15). Given that previous studies have identified the BALB/cJstrain to be somewhat less susceptible to vascular dysfunction(11, 17) underscores the potential importance of the relation-ships between chronic stress/depression and negative vascularoutcomes under investigation in the present study.

Building on the results from previous studies of impairedvascular function in human subjects with increasing depressivesymptom severity (3), the UCMS mouse model was used tofurther interrogate these relationships. With the success of theUCMS protocol in producing depressive behaviors in mice(Fig. 1), the present results demonstrated that this was associ-ated with significant impairments and alterations to vascularreactivity. Net reactivity of aortic rings to phenylephrine wasnot significantly different between control and UCMS mice, aresponse that is in contrast to the recent work from Neves et al.(18), although that study utilized rats and a much shorterduration of UCMS exposure. However, in the present studyapplication of L-NAME resulted in an increased constrictorresponse in control mice only, suggesting a level of NOproduction that buffers phenylephrine-induced constriction, anobservation that was completely absent in UCMS mice (Fig.2). This lack of an impact of L-NAME on adrenergic reactivitysuggested that one of the effects of UCMS was a reduction invascular NO bioavailability, although the responses of vascularsmooth muscle to exogenous NO appear to be intact as re-sponses to sodium nitroprusside were unaffected.

This concept of impaired endothelial function was furtherexplored through interrogation of dilator responses to theendothelium-dependent agonist methacholine. While metha-choline-induced vasodilation in rings from control animals wasoverwhelmingly a function of NO production, this pattern wasaltered substantially in UCMS mice (Fig. 3). Following theUCMS protocol, methacholine-induced dilation, while bluntedvs. responses from control mice demonstrated a reduced de-pendence on, and role for, NO bioavailability. However, ascombined treatment with L-NAME revealed a residual dilatorresponse to methacholine, these results suggest that a compen-satory dilator mechanism develops in UCMS mice where NObioavailability is compromised. Based on the data presented inFig. 4, it seems unlikely that this reduction in vascular NObioavailability reflects a reduction in NOS expression withinthe aortic endothelium, as expression was not altered betweencontrol and UCMS mice.

As summarized in Fig. 5, treatment of aortic rings fromUCMS mice with the hydrogen peroxide scavenger catalasereduced methacholine-induced dilation, while having minimalimpact on responses in rings from control mice. Further,combined treatment of vascular rings with L-NAME and cata-lase nearly abolished vascular responses to methacholine inboth control and UCMS mice. These results strongly suggestthat the vascular production of H2O2 in response to challengewith methacholine may represent a compensatory mechanismwhich helps to maintain normal vascular mechanical responsesto vasoactive stimuli. This conceptual paradigm received ad-ditional support through the measurements of NO and H2O2

production in aortas of control and UCMS mice followingchallenge with methacholine. Presented in Fig. 6, the reductionin agonist-induced NO bioavailability in UCMS mice was

Fig. 9. Correlations between vascular reactivity and individual cardiovasculardisease risk factors in control and UCMS mice following separation of thelower bound of the methacholine concentration-response curves into tertilesbased on the severity of impairment in dilator responses. Data are presented asthe correlation between a given tertile of vascular impairment severity and aspecific vascular disease risk factor. Please see text for details.



associated with an increased H2O2 production vs. responses inaortas from control mice.

One of the more intriguing elements to the present study wasassessing major contributing mechanisms that could underliethe demonstrated changes in vascular reactivity with chronicUCMS in mice. As evidenced by the data summarized in Fig.8, vascular reactivity in control mice demonstrated a predict-able correlation with markers of inflammation (Fig. 8, A–C), anindex of insulin resistance (Fig. 8D), and a plasma marker ofoxidant stress (Fig. 8F). However, in UCMS mice, thesecorrelations were much less robust as neither these markers,nor arterial pressure (Fig. 8E), were able to effectively predictthe reduced dilator responses in UCMS-mice, with only arterialpressure and plasma nitrotyrosine demonstrating r2 values inexcess of 0.3. This is a particularly interesting outcome in thatUCMS mice developed insulin resistance that, while not se-vere, could certainly contribute to alterations in endothelium-dependent vascular reactivity through multiple mechanismassociated with NO bioavailability, oxidant stress, and chronicinflammation (7, 14, 24). Given this, we elected to analyze thedata after dividing the vascular outcome into tertiles based onthe severity of the impaired dilator reactivity (Fig. 9). Theresults from these analyses were particularly intriguing, as mildvascular impairments as a result of the UCMS protocol oc-curred largely in the absence of alterations to insulin sensitiv-ity, chronic oxidant stress, inflammation, or elevated arterialpressure. However, moderate impairments in vascular reactiv-ity in UCMS mice were associated with the development ofincreased arterial pressure only, were weakly predicted byoxidant stress and inflammation, and were independent ofinsulin sensitivity. It was only at the highest levels of vasculardysfunction as a result of UCMS where both hypertension andinsulin resistance (with the associated elevation in oxidantstress) became robust predictors of the negative outcomes.

Taken together, these data provide for multiple compellingavenues for future investigation. First, a greater understandingof reduced NO bioavailability with UCMS warrants attention,via determination eNOS activity and its regulatory processes.Additionally, future studies should be organized to distinguishchronic stress/depression from a general lack of physical ac-tivity (and the possible role of insulin resistance in this process;Ref. 23) as a result of the UCMS protocol, as the miceexperiencing the stress procedures did manifest a greater de-gree of behavior (delayed/reduced motor responses to stimuli,reduced physical struggling against holding/restraint) that iscomparable to the “learned helplessness” that afflicts humansubjects with clinical depression/depressive symptoms. Fur-ther, the results presented in Fig. 9 suggest that there is a levelof vascular dysfunction inherent in UCMS and the develop-ment of depressive symptoms in mice that is not associatedwith the measured parameters in the present study, and whichmay reflect the impact of currently unidentified processes.What makes this particularly compelling is that it is apparentthat even though all mice undergoing the UCMS protocol weregenetically identical, and each manifest comparable behavioraloutcomes, variability in the vascular outcome, and most cer-tainly in the contributing elements to that vascular dysfunction,is present. As a point of speculation, this suggests that theindividual responses to chronic exogenous stress may be het-erogeneous, even within an inbred strain, and that variability interms of adaptation in response to imposed external stressors

may result in distinct developmental routes through which anintegrated pathophysiological outcome develops. We believethat this warrants future investigation.

One area that requires a brief clarification is that of “vasculardepression,” a term that describes the development of depres-sive symptoms that stems from cerebrovascular dysfunctionimpacting cerebral perfusion distribution (1). In vascular de-pression, the depressive symptoms are the result of poorperfusion to the prefrontal regions or their modulating path-ways owing to an accumulation of vascular lesions that nega-tively impact behavioral outcomes, leading to a state of clinicaldepression (2). Given the novelty of using the UCMS model inmice for the study of vascular dysfunction, one area that willrequire future investigation is the temporal association of thedepressive symptom severity and the alterations to cerebrovas-cular reactivity, as the extent to which these are distinctpathological conditions is presently unclear.

In summary, these results suggest that depressive symptomsof sufficient severity can represent a powerful factor in thedevelopment of negative vascular outcomes in a mouse modelof chronic stress/depression. The vascular dysfunction thatdevelops with depressive symptoms is characterized by anabrogation of vascular NO bioavailability, although net vascu-lar reactivity is still largely intact owing to the emergence ofH2O2 as a compensatory dilator metabolite. We believe thatfuture investigation into the role of depressive symptoms vs.physical activity, the mechanistic bases of the vascular dys-function, the functional implications for the change in endo-thelial function, and the potential for ameliorative interventionsis well justified.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the expertise of Dr. Catherine Belzungof the Université François Rabelais de Tours, France, for assisting us in thedeveloping the UCMS model in our laboratory. Additionally, we acknowledgethe expert technical assistance of Barbara Jackson, Milinda James, and thesupport provided through Center for Cardiovascular and Respiratory Sciencesat the West Virginia University Health Sciences Center.

GRANTS

This study was supported by the American Heart Association (EIA0740129N), the National Institutes of Health (R01-DK-64668), and a WestVirginia University Research Development Grant.

DISCLOSURES

No conflicts of interest are declared by the authors.

REFERENCES

1. Alexopoulos GS, Meyers BS, Young RC, Campbell S, Silbersweig D,Charlson M. “Vascular depression” hypothesis. Arch Gen Psychiatry54:915–922, 1997.

2. Alexopoulos GS, Meyers BS, Young RC, Kakuma T, Silbersweig D,Charlson M. Clinically defined vascular depression. Am J Psychiatry 154:562–565, 1997.

3. Brown AD, Barton DA, Lambert GW. Cardiovascular abnormalities inpatients with major depressive disorder: autonomic mechanisms andimplications for treatment. CNS Drugs 23: 583–602, 2009.

4. Barefoot JC, Brummett BH, Helms MJ, Mark DB, Siegler IC, Wil-liams RB. Depressive symptoms and survival of patients with coronaryartery disease. Psychosom Med 62: 790–795, 2000.

5. Barefoot JC, Williams RB. Antidepressant use and the risk of myocardialinfarction. Am J Med 108: 87–88, 2000.

6. Berkman LF, Blumenthal J, Burg M, Carney RM, Catellier D, CowanMJ, Czajkowski SM, DeBusk R, Hosking J, Jaffe A, Kaufmann PG,Mitchell P, Norman J, Powell LH, Raczynski JM, Schneiderman N;



Enhancing Recovery in Coronary Heart Disease Patients Investiga-tors (ENRICHD). Effects of treating depression and low perceived socialsupport on clinical events after myocardial infarction: the EnhancingRecovery in Coronary Heart Disease Patients (ENRICHD) RandomizedTrial. JAMA 289: 3106–3116, 2003.

7. Busija DW, Miller AW, Katakam P, Erdös B. Insulin resistance andassociated dysfunction of resistance vessels and arterial hypertension.Minerva Med 96: 223–232, 2005.

8. Chrapko WE, Jurasz P, Radomski MW, Lara N, Archer SL, LeMellédo JM. Decreased platelet nitric oxide synthase activity and plasmanitric oxide metabolites in major depressive disorder. Biol Psychiatry 56:129–134, 2004.

9. Davidson KW, Kupfer DJ, Bigger JT, Califf RM, Carney RM, CoyneJC, Czajkowski SM, Frank E, Frasure-Smith N, Freedland KE,Froelicher ES, Glassman AH, Katon WJ, Kaufmann PG, Kessler RC,Kraemer HC, Krishnan KR, Lespérance F, Rieckmann N, Sheps DS,Suls JM; National Heart, Lung, and Blood Institute Working Group.Assessment and treatment of depression in patients with cardiovasculardisease: National Heart, Lung, and Blood Institute Working Group Re-port. Psychosom Med 68: 645–650, 2006.

10. Ducottet CC. Belzung, Correlations between behaviours in the elevatedplus-maze and sensitivity to unpredictable subchronic mild stress: evi-dence from inbred strains of mice. Behav Brain Res 156: 153–162, 2005.

11. Friedman G, Ben-Yehuda A, Dabach Y, Hollander G, Babaey S,Ben-Naim M, Stein O, Stein Y. Macrophage cholesterol metabolism,apolipoprotein E, and scavenger receptor AI/II mRNA in atherosclerosis-susceptible and -resistant mice. Arterioscler Thromb Vasc Biol 20: 2459–2464, 2000.

12. Gass P, Reichardt HM, Strekalova T, Henn F, Tronche F. Mice withtargeted mutations of glucocorticoid and mineralocorticoid receptors:models for depression and anxiety? Physiol Behav 73: 811–825, 2001.

13. Glassman AH, O’Connor CM, Califf RM, Swedberg K, Schwartz P,Bigger JT Jr, Krishnan KR, van Zyl LT, Swenson JR, Finkel MS,Landau C, Shapiro PA, Pepine CJ, Mardekian J, Harrison WM,Barton D, Mclvor M. Sertraline Antidepressant Heart Attack Random-ized Trial (SADHEART) Group. Sertraline treatment of major depressionin patients with acute MI or unstable angina. JAMA 288: 701–709, 2002.[Corrigendum: JAMA 288: 1720, 2002].

14. Kim F, Pham M, Maloney E, Rizzo NO, Morton GJ, Wisse BE, KirkEA, Chait A, Schwartz MW. Vascular inflammation, insulin resistance,and reduced nitric oxide production precede the onset of peripheral insulinresistance. Arterioscler Thromb Vasc Biol 28: 1982–1988, 2008.

15. Mineur YS, Belzung C, Crusio WE. Effects of unpredictable chronicmild stress on anxiety and depression-like behavior in mice. Behav BrainRes 175: 43–50, 2006.

16. Mineur YS, Prasol DJ, Belzung C, Crusio WE. Agonistic behavior andunpredictable chronic mild stress in mice. Behav Genet 33: 513–519,2003.

17. Miyoshi T, Matsumoto AH, Shi W. Paradoxical increase in LDL oxi-dation by endothelial cells from an atherosclerosis-resistant mouse strain.Atherosclerosis 192: 259–265, 2007.

18. Neves VJ, Moura MJ, Tamascia ML, Ferreira R, Silva NS, Costa R,Montemor PL, Narvaes EA, Bernardes CF, Novaes PD, MarcondesFK. Proatherosclerotic effects of chronic stress in male rats: alteredphenylephrine sensitivity and nitric oxide synthase activity of aorta andcirculating lipids. Stress 12: 320–327, 2009.

19. Porsolt RD. Animal models of depression: utility for transgenic research.Rev Neurosci 11: 53–58, 2000.

20. Penninx BW, Beekman AT, Honig A, Deeg DJ, Schoevers RA, vanEijk JT, van Tilburg W. Depression and cardiac mortality: results froma community-based longitudinal study. Arch Gen Psychiatry 58: 221–227,2001.

21. Rudisch B, Nemeroff CB. Epidemiology of comorbid coronary arterydisease and depression. Biol Psychiatry 54: 227–240, 2003.

22. Santarelli L, Saxe M, Gross C, Surget A, Battaglia F, Dulawa S,Weisstaub N, Lee J, Duman R, Arancio O, Belzung C, Hen R.Requirement of hippocampal neurogenesis for the behavioral effects ofantidepressants. Science 301: 805–809, 2003.

23. Venables MC, Jeukendrup AE. Physical inactivity and obesity: linkswith insulin resistance and type 2 diabetes mellitus. Diabet Metab Res Rev25, Suppl 1: S18–S23, 2009.

24. Wiernsperger N, Nivoit P, De Aguiar LG, Bouskela E. Microcirculationand the metabolic syndrome. Microcirculation 14: 403–438, 2007.

25. Willner P. Validity, reliability and utility of the chronic mild stress modelof depression: a 10-year review and evaluation. Psychopharmacology(Berl) 134: 319–329, 1997.

26. Willner P. Animal models of depression: validity and applications. AdvBiochem Psychopharmacol 49: 19–41, 1995.

27. Yalcin I, Belzung C, Surget A. Mouse strain differences in the unpre-dictable chronic mild stress: a four-antidepressant survey. Behav BrainRes 193: 140–143, 2008.

28. Yalcin I, Aksu F, Belzung C. Effects of desipramine and tramadol in achronic mild stress model in mice are altered by yohimbine but not bypindolol. Eur J Pharmacol 514: 165–174, 2005.



Depressive behavior and vascular dysfunction: a link between clinical depression and vascular...

Documents

Transcript of Depressive behavior and vascular dysfunction: a link between clinical depression and vascular...