Increased tissue transglutaminase activity contributes to central vascularstiffness in eNOS knockout mice

Sung Mee Jung,1,2 Simran Jandu,1 Jochen Steppan,1 Alexey Belkin,3 Steven S. An,4 Alina Pak,6

Eric Y. Choi,4 Daniel Nyhan,1 Mark Butlin,5 Kayla Viegas,5 Alberto Avolio,5 Dan E. Berkowitz,1,6

and Lakshmi Santhanam1,6

1Department of Anesthesiology and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore,Maryland; 2Anesthesiology and Pain Medicine, Yeungnam University School of Medicine, Gyeongsan, South Korea;3University of Maryland School of Medicine, College Park, Maryland; 4Department of Environmental Health Sciences, JohnsHopkins Bloomberg School of Public Health, Baltimore, Maryland; 5Australian School of Advanced Medicine, MacquarieUniversity, New South Wales, Australia; and 6Department of Biomedical Engineering, Johns Hopkins University School ofMedicine, Baltimore, Maryland

Submitted 6 February 2013; accepted in final form 15 July 2013

Jung SM, Jandu S, Steppan J, Belkin A, An SS, Pak A, ChoiEY, Nyhan D, Butlin M, Viegas K, Avolio A, Berkowitz DE,Santhanam L. Increased tissue transglutaminase activity contributesto central vascular stiffness in eNOS knockout mice. Am J PhysiolHeart Circ Physiol 305: H803–H810, 2013. First published July 19,2013; doi:10.1152/ajpheart.00103.2013.—Nitric oxide (NO) canmodulate arterial stiffness by regulating both functional and structuralchanges in the arterial wall. Tissue transglutaminase (TG2) has beenshown to contribute to increased central aortic stiffness by catalyzingthe cross-linking of matrix proteins. NO S-nitrosylates and constrainsTG2 to the cytosolic compartment and thereby holds its cross-linkingfunction latent. In the present study, the role of endothelial NOsynthase (eNOS)-derived NO in regulating TG2 function was studiedusing eNOS knockout mice. Matrix-associated TG2 and TG2 cross-linking function were higher, whereas TG2 S-nitrosylation was lowerin the eNOS�/� compared with wild-type (WT) mice. Pulse-wavevelocity (PWV) and blood pressure measured noninvasively wereelevated in the eNOS�/� compared with WT mice. Intact aortas anddecellularized aortic tissue scaffolds of eNOS�/� mice were signifi-cantly stiffer, as determined by tensile testing. The carotid arteries ofthe eNOS�/� mice were also stiffer, as determined by pressure-dimension analysis. Invasive methods to determine the PWV-meanarterial pressure relationship showed that PWV in eNOS�/� and WTdiverge at higher mean arterial pressure. Thus eNOS-derived NOregulates TG2 localization and function and contributes to vascularstiffness.

endothelial nitric oxide synthase; nitric oxide; tissue transglutaminase;vascular stiffness; S-nitrosylation; pulse-wave velocity; tensile testing

CARDIOVASCULAR DISEASE IS a major contributor of age-relatedmorbidity and mortality, with isolated systolic hypertensionbeing one of the hallmarks of this process (13, 14). It is nowwell established that vascular stiffness directly leads to systolichypertension, impaired ventricular-arterial coupling, and animbalance in myocardial oxygen supply and demand (17, 19).Moreover, multiple studies have conclusively demonstratedthat vascular stiffness, as measured by pulse-wave velocity(PWV) or pulse pressure, is an independent predictor of ad-verse cardiovascular events and can be used to stratify riskpatients in the clinical setting (21, 22).

Tissue transglutaminase (TG2) is a multifunctional proteinthat is ubiquitously expressed in all cell types of the vascula-ture. TG2 catalyzes the formation of ε-(�-glutamyl)lysinebonds in a Ca2�-dependent manner and induces protein cross-links to stabilize the extracellular matrix (ECM). TG2 plays animportant role in vascular function, including remodeling ofresistance vessels (3, 4), increased central aortic stiffness withage (25), and arterial calcification (28). We have previouslyshown that nitric oxide (NO) regulates TG2 cross-linkingactivity and trafficking in the vasculature (18, 24). Specifically,decreased NO bioavailability is associated with decreased TG2S-nitrosylation, increased secretion of TG2 to the cell surfaceand ECM, and increased TG2 cross-linking activity in isolatedendothelial, smooth muscle, and fibroblast cells and in the rataorta. In the previous studies, NO bioavailability was modu-lated pharmacologically using either NO synthase (NOS) in-hibitor NG-nitro-L-arginine methyl ester (L-NAME) to decreaseNO levels or the NO donor S-nitrosoglutathione (GSNO) toincrease NO levels. Alternatively, endothelial cells were cocul-tured with fibroblast and vascular smooth muscle cells in thepresence and absence of L-NAME to alter NO bioavailability.In animal models, L-NAME was administered via osmoticinfusion pumps to diminish NOS-dependent NO production. Inthe present study, using genetic approaches, we tested thehypothesis that a decrease in NO produced by endothelial NOS(eNOS) would lead to decreased TG2 S-nitrosylation, in-creased TG2 activity, and increased vascular stiffness. Specifi-cally, TG2 function was examined using eNOS shRNA-mediatedknockdown eNOS expression in human aortic endothelial cells(HAEC) and using eNOS�/� mice. We demonstrate that eNOS iscritical for inhibiting the activity of TG2 and maintaining centralvascular compliance.

MATERIALS AND METHODS

Animals. Male eNOS knockout mice with their wild-type (WT)background controls (C57BL/6J, 12- to 14-wk-old) obtained fromJackson Laboratories were used in this study. Mice were housed intemperature- and light-controlled conditions and fed and watered adlibitum. All experimental procedures were approved by the Institu-tional Animal Care and Use Committee at the Johns Hopkins Uni-versity School of Medicine.

Cell cultures. HAEC purchased from Invitrogen were cultured inendothelial cell media (ScienCell Research Labs; ECM media sup-

Address for reprint requests and other correspondence: L. Santhanam, 720Rutland Ave., Ross 1150, Johns Hopkins Univ. SOM, Baltimore, MD 21205.

Am J Physiol Heart Circ Physiol 305: H803–H810, 2013.First published July 19, 2013; doi:10.1152/ajpheart.00103.2013.

0363-6135/13 Copyright © 2013 the American Physiological Societyhttp://www.ajpheart.org H803

plemented with 5% FBS, endothelial cell growth supplement, andpenicillin/streptomycin) and used between passages 7 and 10.

eNOS knockdown. Lentivirus-delivered short-hairpin RNA (shRNA;Santa Cruz Biotech) were used to knock down eNOS expression incells. ECM media was replaced with phenol red-free low serum mediacontaining Polybrene, and samples were treated with eNOS shRNAlentivirus or control lentivirus. After 2 h, media was again exchangedto ECM media, and the cells were cultured for an additional 48 h andused. Knockdown of eNOS was confirmed by Western blotting usingeNOS antibody (BD Biosciences).

Protein expression. The expression levels of different proteins weredetermined by Western blotting. Cells were lysed in 1� radioimmu-noprecipitation assay (RIPA) buffer (Upstate) containing proteaseinhibitors (Roche) by mechanical scraping, and total protein concen-tration was determined (Bio-Rad protein assay reagent). Equalamounts of lysate proteins from each sample were resolved by SDSPAGE and electrotransferred to nitrocellulose. Blots were blockedwith 3% nonfat dry milk for 1 h at room temperature. The membranewas incubated with primary antibody followed by HRP-conjugatedanti-rabbit or anti-mouse antibody (1:1,000) for 1 h at room temper-ature. GAPDH was used as loading control. Mouse anti-TG2 (1:5,000;Neomarkers) was used to detect TG2 in HAEC, and rabbit anti-TG2(1:1,000; Neomarkers) was used to detect TG2 in mouse tissue.Rabbit anti-eNOS was from Santa Cruz Biotech (1:1,000), and mouseanti-GAPDH (1:5,000) was from Novus Biologicals. Blots weredeveloped using enhanced chemiluminescence and quantified usingImageJ [National Institutes of Health (NIH)].

Co-immunoprecipitation. Proteins (50 �g) from homogenates ofaorta from eNOS�/� and WT mice were immunoprecipitated usingintegrin �1-antibody (Armenian hamster anti-integrin �; BD Biosci-ence). Samples were eluted in Laemmli buffer, resolved by SDSPAGE, and electrotransferred to nitrocellulose membrane. Co-precip-itation of TG2 was determined by Western blotting. Integrin �1 wasalso determined in the sample by Western blotting, using horseradishperoxidase (HRP)-conjugated goat anti-Armenian hamster secondaryantibody (Pierce). Proteins from eNOS�/� mouse aorta were immu-noprecipitated with isotype IgG as a negative control.

Noninvasive pulse-wave velocity measurement. Mice were anesthe-tized in a closed chamber with isoflurane. Anesthesia was maintainedby mask ventilation of 1.0–1.5% isoflurane (in 100% O2) with acharcoal scavenging system. Animals were positioned supine on atemperature-controlled printed circuit board (Indus Instruments,Houston, TX) with legs and arms taped to incorporated electrocardio-gram electrodes. Body temperature was monitored with a rectal probe(Physitemp, Clifton, NJ) and maintained at 37°C. Doppler spectro-grams of aortic outflow were acquired with a 2-mm-diameter, 10-MHz pulsed Doppler probe (Indus Instruments, Webster, TX). Tho-racic aortic outflow and abdominal aortic flow profiles were captured.The distance separating the probe locations was also measured. AorticPWV was calculated as the quotient of the separation distance and thetime difference between pulse arrivals, with respect to the R-peak ofthe electrocardiogram. DSPW software (Indus Instruments) was usedfor data analysis.

Invasive PWV measurement. We used a high-fidelity dual-pressurecatheter sensor to measure aortic PWV. Mice were anesthetized withan intraperitoneal injection of 1.2% Avertin (2,2,2-tribromoethanol,240 mg/kg) in 0.9% saline. The animal was positioned supine on theheating pad, with water temperature set to 40°C. Anesthesia wasmaintained by mask ventilation with 1.0–1.5% isoflurane (in 100%O2), and the reflex response to hind-paw pinching was assessed tomonitor depth of anesthesia. After midline neck incision from man-dible to sternum, a 1.2-Fr, dual-pressure sensor catheter (Scisense,London, Ontario, Canada) was introduced into the descending tho-racic aorta through the left carotid artery without opening the chestcavity (see Fig. 4A). The distance between two sensors is fixed at 1cm. A 30-gauge cannulation needle connected to polyethylene tubing(10) was inserted into left femoral vein for infusion of fluid/drugs.

After stabilization of the signal for 10–15 min, baseline blood pres-sures were recorded. Mean arterial pressure (MAP) was raised andlowered to obtain a full physiological range of blood pressure usingintravenous infusion of phenylephrine and sodium nitroprusside, re-spectively (see Fig. 4C). PWV at corresponding MAP was calculatedusing the foot-to-foot method, the foot being defined by the peak ofthe second time derivative of two aortic pressures measured simulta-neously during each pulse (see Fig. 4B). PWV was plotted againstMAP to construct phase plots to characterize PWV over a wide rangeof MAPs from 50 to 150 mmHg in the aorta (see Fig. 4D).

In vivo TG inhibition in mice. Mice were randomized into twogroups, and an osmotic pump (Alzet) was implanted and filled with a4-wk dose of either cystamine (40 mg·kg�1· day�1) or vehicle control.

Pressure-diameter analysis. Common carotid arteries were isolatedfrom mice and mounted on the glass pipettes of a pressure myographin a perfusion chamber. The artery was perfused with heated andoxygenated calcium-free Krebs buffer using a peristaltic pump (Cole-Parmer Instrument), which also continuously monitored perfusionpressure. Intraluminal pressure was incrementally increased from 0 to100 mmHg in steps of 10 mmHg, with each step in 30-s intervals, andcorresponding change of vessel outer diameter was simultaneouslyrecorded using microscopic imaging and video dimension analysis(Analog Digital Instruments). Compliance was calculated as previ-ously described (25) using the following equation:

Compliance ��V

�P�

�A

�P�

�d2

�P

where V is the volume, A is the area, d is the diameter of the vessel,and P is the distending pressure.

TG activity assay. A dot blot assay was used to determine TGactivity (25). The assay is based on the incorporation of TG substratebiotin(amido)pentylamine (BPA; Pierce) into proteins. Intact cells/tissues were incubated with 0.1 mM BPA and 1 mM Ca2� at 37°C for4 h in culture media (phenol red-free DMEM supplemented with 2%FBS and penicillin/streptomycin) and then rinsed with PBS threetimes to remove unreacted BPA. Samples were then lysed/homoge-nized to recover proteins. Proteins (0.5–1 �g) were loaded ontonitrocellulose membrane (BioDot Dot Blot apparatus; Bio-Rad). Themembrane was rinsed and blocked in 3% BSA overnight and probedwith HRP-conjugated streptavidin (Amersham Bioscience; 1:10,000dilution in 3% BSA) to determine BPA incorporation. Blots were thenstripped using Restore Plus stripping buffer (Pierce) and reprobedwith GAPDH to determine protein loading. BPA incorporation andGAPDH levels were determined by densitometry analysis usingImageJ software (NIH). For each sample, activity was calculated as aratio of BPA/GAPDH.

Recovery of ECM proteins. Cells were grown to confluence andtreated as indicated. ECM was recovered following established meth-ods (18). In brief, cells were rinsed twice with PBS. Cellular andnuclear materials were extracted by incubation with cell removalsolution (0.05% Triton X-100, 50 mM NH4OH, in PBS) until the cellswere floating. The matrix was then briefly washed once with rinsebuffer (50 mM NH4OH in PBS) and then three times with PBS(Quality Biological). ECM was scraped into 100 �l of lysis buffer,and TG2 expression was determined by Western blotting.

Decellularization of mouse aorta. The aortas were cleaned ofadjacent tissues and fat and decellularized as previously described(18), with minor modifications. Briefly, aortas were incubated indecellularization solution 1 (8 mM CHAPS, 1 M NaCl, and 25 mMEDTA in PBS) for 44 h. Next, samples were incubated in decellular-ization solution 2 [1.8 mM sodium dodecyl sulfate (SDS), 1 M NaCl,25 mM EDTA in PBS] for 44 h. Each decellularization solution waschanged every 22 h with three 15-min washes in PBS. Samples werewashed with and incubated with PBS for 2 days for complete removalof the detergents. All these steps were conducted at room temperature,with continuous shaking under sterile conditions. In the final step,

H804 INCREASED TRANSGLUTAMINASE ACTIVITY CONTRIBUTES TO VASCULAR STIFFNESS

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00103.2013 • www.ajpheart.org

samples were incubated at 37°C for 1 day in endothelial cell media(ScienCell Research Labs) followed by three 15-min washes withPBS to obtain decellularized specimen. Removal of cells was con-firmed by the absence of DNA assayed using the Pico Green assay kit(Invitrogen), and absence of GAPDH (cytosolic protein) was con-firmed by Western blotting.

S-nitrosylation assay. TG2 S-nitrosylation was determined usingthe biotin switch assay in cell lysates/tissue homogenates as previ-ously described (18, 25). Because the activity assay also relies onbiotinylation, the S-nitrosylation assays were performed on a separateset of samples in parallel with the activity assays.

Magnetic twisting cytometry. Vascular smooth muscle cells werefreshly isolated from the aorta using the method of Kobayashi et al.(20) and plated onto fibronectin-coated 96-well strip-well plates. Atthe level of the single living cell, we measured cell stiffness usingforced motions of functionalized beads anchored to the cytoskeleton(CSK) through cell surface integrin receptors using magnetic twistingcytometry (MTC) as described in detail elsewhere (8, 12). For thesestudies, the stiffness of adherent cells isolated from eNOS�/� and WTmice was measured at a frequency of 0.75 Hz. Cell stiffness isexpressed in units of Pascal per nanometer (Pa/nm).

Data analysis and statistics. Unless otherwise noted, all data arepresented as means � SE, with sample size (n) being indicated foreach reported value. All Western blots and dot blots were analyzed bydensitometry using the ImageJ software (NIH). Results are expressedas a percentage change relative to the average value measured in thecontrol group. Statistical evaluation was performed by Student’s t-testfor unpaired observations to compare two means. When more thantwo means were compared, one-way ANOVA with Tukey’s post hoctest was used to identify differences. To satisfy the normal distributionassumptions associated with ANOVA, cell stiffness data were con-verted to log scale before analyses. Vascular response data wereanalyzed with GraphPad Prism data analysis software. Two-wayANOVA with Bonferroni post hoc tests was employed to comparePWV values at each MAP. A value of P � 0.05 was consideredstatistically significant.

RESULTS

eNOS regulates TG2 externalization and activity in HAEC.Decreased expression of eNOS with shRNA treatment wasconfirmed by Western blotting (Fig. 1A). eNOS knockdownresulted in an increase in ECM-associated TG2 (Fig. 1B), adecrease in TG2-S-nitrosylation (Fig. 1C), and an increase inTG activity (Fig. 1D) compared with controls (Fig. 1, A–C).

eNOS�/� mice have altered TG2 activity and localization.Aortae of eNOS knockout mice were examined to determinethe effect of eNOS deficiency on TG2 localization and activity.TG activity was higher in the aortae of eNOS�/� mice com-pared with WT (C57Bl6) mice (Fig. 2A). TG2 abundance in thedecellularized tissue scaffold was higher, and S-nitrosylationlevels were markedly diminished in the aortae of eNOS knock-out mice compared with WT controls, whereas total TG2remained unchanged (Fig. 2, B and C). Immunohistochemicalanalysis showed that TG2 abundance was similar in eNOS�/�

and WT mice, whereas TG2 cross-links were significantly higherin the eNOS�/� mice compared with WT mice (Fig. 2D).

Carotid artery compliance is significantly improved in cystamine-treated eNOS�/� mice. Cystamine is a TG inhibitor that has beenwidely used in animal models to study the role of TG (2, 7, 9,10). In the present study, to examine the role of TG2 inmediating vascular properties, cystamine (40 mg·kg�1·day�1)was delivered to WT and eNOS�/� mice for 4 wk usingosmotic infusion pumps. eNOS�/� mice exhibited lower ca-rotid artery compliance compared with WT mice. In botheNOS�/� and WT mice, cystamine treatment led to a signifi-cant increase in carotid artery compliance compared with theirvehicle-treated counterparts (Fig. 3). Although resting bloodpressures were higher in the eNOS�/� mice, cystamine treat-ment did not result in altered blood pressure (WT control: 66.5 � 5.6

Control NOS3 shRNA0

50

100

150

*

Lentivirus

eNO

S/G

APD

H(N

orm

aliz

ed to

Av.

Con

trol)

Control NOS3 shRNA0

50

100

150

*

Lentivirus

% T

G2-

SNO

(Nor

mal

ized

to C

ontro

l)

Control NOS3 shRNA0

100

200

300

400

*

Lentivirus

ECM

TG

2(N

orm

aliz

ed to

Av.

Con

trol)

Control NOS3 shRNA0

50

100

150

200

250

*

Lentivirus

% T

G2

Activ

ity(N

orm

aliz

ed to

Av.

Con

trol)

TG2-SNO

Total TG2

TG2 activity (BPA)

GAPDH

eNOS ECM TG2 A B

C D

Fig. 1. Effect of endothelial NO synthase(eNOS) knockdown on transglutaminase (TG2)in human aortic endothelial cells (HAEC).A: lentiviral delivery of eNOS short-hairpinRNA (shRNA) resulted in an 50% loss ineNOS protein expression in cultured HAEC.B: TG2 abundance in extracellular matrix(ECM) increased. C: TG2 S-nitrosylation de-creased. D: TG2 cross-linking function in-creased with eNOS knockdown. Graphs showmeans � SE (n 6). *Significant differenceby Student’s t-test (P � 0.05).

H805INCREASED TRANSGLUTAMINASE ACTIVITY CONTRIBUTES TO VASCULAR STIFFNESS

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00103.2013 • www.ajpheart.org

mmHg; WT cystamine: 69.3 � 6.1 mmHg; eNOS control: 96.8 �1.9 mmHg; eNOS cystamine: 98.1 � 2.8 mmHg) or heart rate(WT control: 503 � 45 beats/min; WT cystamine: 505 � 45beats/min; eNOS control: 513 � 18 beats/min; eNOS cysta-mine: 518 � 26 beats/min).

Mechanical properties of the aorta are altered in eNOS�/�

mice. The eNOS�/� mouse is hypertensive and has impairedvasorelaxant responses. Two approaches were used to examineaortic stiffness. First, PWV measurements were obtained usinga Doppler-based device. There was a marked increase in PWVin eNOS�/� mice compared with WT controls (Fig. 4E) and a

concomitant increase in MAP (Fig. 4F). Basal heart rates weresimilar in both groups (Fig. 4G). Given the dependence ofPWV on wall tension and therefore blood pressure, changes inPWV must be compared at similar MAP to accurately assesschanges in arterial stiffness. We therefore measured PWV byvarying MAP over a full physiological range of blood pressureby using an invasive dual-pressure sensor, high-fidelity cathe-ter (Fig. 4H). At lower pressures (MAP � 125 mmHg), PWVsof eNOS�/� mice were similar to WT controls. At higherMAPs (MAP � 125 mmHg), however, PWVs of eNOS�/�

mice were significantly higher than PWVs of WT mice at each

Control eNOS KO0

50

100

150 *

TG2

in D

ecel

lula

rized

Aor

ta(%

of A

v C

ontro

l)

Wildtype ENOS KO0

25

50

75

100

125

150

175 p=.0153N(eNOS)=8N(wt)=4

% o

f Wild

type

tTG

act

ivity

WT

eNO

SK

O

BPA

GAPDH

A B

C

TG2-SNO

TG2 (Total)

GAPDH

WT eNOSKO

D WT eNOS KO

TG2

Cro

ss-li

nks

eNOS KO WT TG2

expr

essi

on a

nd c

ross

-link

s,

Nor

mal

ized

to W

T

Fig. 2. TG2 localization and function in eNOS�/� mice. A: TG cross-linking activity was higher. B: TG2 abundance in the decellularized aortic matrix was higher(n 5). C: TG2 S-nitrosylation levels were lower in the eNOS�/� mice compared with wild-type (WT) controls (n 5). D: TG2 abundance was unaltered,whereas TG-specific cross-links were higher in the eNOS�/� mouse (n 8). *Significant difference by Student’s t-test (P � 0.05).

H806 INCREASED TRANSGLUTAMINASE ACTIVITY CONTRIBUTES TO VASCULAR STIFFNESS

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00103.2013 • www.ajpheart.org

probing MAP. Next, the passive mechanical properties ofintact and decellularized aorta were examined by tensile testingto determine the stress-strain relationship. Intact aortic rings ofeNOS�/� mice were significantly stiffer than age-matched WTcontrols (Fig. 4I); the incremental elastic modulus diverged at0.5 strain. The decellularized aortic tissue scaffolds of theeNOS�/� mice were also significantly stiffer than those of WTmice; the incremental elastic modulus was significantly alteredat 0.8 strain (Fig. 4J).

Cell-matrix interactions and CSK stiffness are altered ineNOS�/� mice. TG2 acts as an integrin-fibronectin bindingcofactor at the cell surface, independent of its catalytic func-tion. As determined by co-immunoprecipitation experiments,TG2-integrin �1 interaction was higher in the aorta ofeNOS�/� mice compared with WT mice (Fig. 5A). Further-more, using MTC, we found that vascular smooth muscle cellsfreshly isolated from the aorta of eNOS�/� mice exhibitedappreciably higher CSK stiffness than those isolated from WTmice (Fig. 5B). Phenylephrine caused 28.97 � 3.90% and37.61 � 7.69% increases in cell stiffness in WT and eNOSknockout cells, respectively. Taken together, these results sug-gest that TG2 might induce vascular stiffening by increasingboth cellular and matrix stiffness.

DISCUSSION

It is well known that the composition and degree of cross-linking of the extracellular matrix influence the mechanicalproperties of the vessel (29). In the vasculature, TG2 is knownto mediate remodeling of the mesenteric artery in response tovasoconstrictive stimuli (3, 4). TG2 secretion, and thus TG2-mediated remodeling, is augmented under reducing conditionsand inhibited by NO (18, 24, 25, 27). A major questionregarding TG2 function is whether increased cross-linkingactivity leads to increased vascular stiffness. In this context, itis interesting to note that TG2�/� mice are born with Mende-lian frequency and do not exhibit any overt vascular phenotype(16). The PWV and MAP of TG2�/� mice are similar toage-matched WT controls; however, TG2�/� mice are pro-tected from the increase in PWV resulting from NOS inhibitionusing L-NAME (25). This suggests that, although TG2 may inpart mediate vascular stiffness resulting from endothelial dys-function due to attenuated eNOS function, it may not be

significant in normal functioning of the aorta. Indeed, underphysiological conditions, with normal levels of NO bioavail-ability, much of TG2 is S-nitrosylated, is confined to thecytosol, and has its cross-linking function held latent, whichsupports this idea. To further test this hypothesis, we studiedthe eNOS�/� mouse model, which has been widely used toexamine the role of eNOS-derived NO in cardiovascular func-tion. These mice are characterized by the absence of eNOSmRNA and enzymatic activity and show the phenotypescaused by total absence of eNOS. This includes lack of endo-thelium-derived relaxation factor (EDRF) activity both in iso-lated arteries and in vivo, as measured by vascular relaxation inresponse to acetylcholine or other endothelium-dependent va-sodilators. These mice are also hypertensive, which under-scores the essential role of eNOS-dependent NO in the phys-iological regulation of blood pressure (15). Again, lack ofeNOS-derived NO had a reciprocal correlation with TG2externalization and activity and coincided with decreased TG2S-nitrosylation. Vascular stiffness in eNOS�/� and WT micewas examined using PWV. This is a well accepted measure ofarterial stiffness, which is typically measured as a single-pointmeasurement and reported together with or corrected for MAP.PWV (measured using a Doppler-based device) and MAP arehigher in eNOS�/� mice compared with WT controls (26).However, wall stiffness also varies with distending pressuresdue to the elastic nature of arteries, which in turn depends onthe composition and structure of the load-bearing components(i.e., elastin and collagen) in the tissue matrix. Thus, to accu-rately compare alterations in vascular stiffness between theeNOS�/� and WT mice, PWV changes must be compared atsimilar arterial pressures or corrected for pressure to accuratelydetermine changes in arterial stiffness, as has been done inprevious studies comparing treatments that affect blood pres-sure (23). Therefore, in this study, we used an invasive ap-proach to determine the pressure dependence of PWV ineNOS�/� and WT mice over a range of physiologically rele-vant pressures. The eNOS�/� mice had PWV comparable toWT mice at low pressures and significantly higher PWV athigher pressures. This suggests that, although compensatorymechanisms may regulate central aortic stiffness in a normalphysiological setting, the eNOS�/� mice will likely havegreater remodeling and elevated stiffness when stressed or in

WT

WT+C

ys

eNOS KO

eNOS K

O + Cys

0

500

1000

Com

plia

nce

(µm

2 /m

mH

g)****

**

*

WT

WT C

ys

eNOS KO

eNOS K

O + Cys

400

450

500

550

600

650

Hea

rt R

ate

(bea

ts/m

in)

WT

WT C

ys

eNOS KO

eNOS K

O + Cys

0

50

100

150

Mea

n Pr

essu

re (m

mH

g)

*

Fig. 3. Effect of TG inhibition on vascular compliance. TG inhibitor cystamine (40 mg·kg�1·day�1) was administered for 4 wk to WT and eNOS knockout miceusing osmotic infusion pumps. Carotid artery compliance, measured by video-dimension analysis, was lower in eNOS�/� mice compared with WT. Cystaminetreatment restored carotid artery compliance toward WT controls. Values are means � SE (n 8). Significant difference by one-way ANOVA with Bonferronipost hoc analysis: *P � 0.05; **P � 0.01; ***P � 0.001.

H807INCREASED TRANSGLUTAMINASE ACTIVITY CONTRIBUTES TO VASCULAR STIFFNESS

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00103.2013 • www.ajpheart.org

L L

Transducer 1

Transducer 2

Catheter

Catheter Drug infusion (PE to 120 mm Hg; SNP to 50 mm Hg)

CC

C

ECG

d2P/dt2

PWV = L/ t

t tttttttttProximal wave

Distal wave

xiProxxiProx

h tthDi

A

C

D

Dua

l BP

1 &

2

(mm

Hg)

2nd D

eriv

ativ

e (m

m H

g/s2

)

B

Ele

ctro

card

iogr

am

Dua

l BP

1

(mm

Hg)

D

ual B

P 2

(m

m H

g)

20 40 60 80 100 1203.0

3.5

4.0

4.5

5.0

MAP (mmHg)

PWV

(m/s

) SNP infusionPE infusion

WT eNOS KO2.5

3.0

3.5

4.0

4.5

***

PWV

(m/s

)

WT eNOS KO400

450

500

550

600

Hea

rt R

ate

(bea

ts/m

in)

WT eNOS KO40

60

80

100

120***

MA

P (m

mH

g)

0 50 100 150 2000

2

4

6

8

WTeNOS KO

* ******

***

MAP (mmHg)

PWV

(m/s

)

E F G

H I JIntact

0.0 0.5 1.0 1.5 2.00

2000

4000

6000

8000

WTeNOS KO

Strain

Stre

ss, m

N/m

m2

**

WT

eNOS K

O0

500

1000

1500

Incr

emen

tal E

last

ic M

odul

us

(Stra

in =

0.5

)

*

Decellularized

0.0 0.5 1.0 1.5 2.00

2000

4000

6000

8000WT

eNOS KO

Strain

Stre

ss, m

N/m

m2

*

WTDec

ell

eNOS K

O Dec

ell0

10002000300040005000

Incr

emen

tal E

last

ic M

odul

us

(Stra

in =

0.8

)

*

Fig. 4. Vascular stiffness measures in eNOS�/� and WT mice. A: simultaneous pulsed-wave velocity (PWV) and mean arterial pressure (MAP) measurementin mice; a 1.2-Fr catheter was used to measure transit time in vivo. B: example waves/second derivative to determine transit time. C: example of beat-to-beatchanges in arterial pressure. D: sample PWV and MAP relationship in a single animal. E: single-point, Doppler-based PWV measurement was higher in eNOS�/�

compared with WT. F: MAP was higher in eNOS�/� mice. G: heart rate was similar in both groups. H: PWV-MAP correlation in eNOS�/� mice diverged fromWT mice at higher MAPs; dotted lines represent baseline pressure and PWV for eNOS�/� (red) and WT (black) mice. I: intact aorta of eNOS�/� mice werestiffer than WT. Inset: incremental elastic modulus of intact vessels at strain 0.5. J: decellularized aortic segments of eNOS�/� mice were stiffer than thosefrom WT mice. Inset: incremental elastic modulus of decellularized vessels at strain 0.8. Values are means � SE (n 8). Significant difference by Student’st-test for two groups or one-way ANOVA with Bonferroni post hoc analysis for more than two groups: *P � 0.05; **P � 0.01; ***P � 0.001.

H808 INCREASED TRANSGLUTAMINASE ACTIVITY CONTRIBUTES TO VASCULAR STIFFNESS

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00103.2013 • www.ajpheart.org

pathophysiological states. Tensile testing revealed that theintact aorta and decellularized tissue scaffold of eNOS�/�

mice are markedly stiffer than those of WT mice. IncreasedTG2 abundance and cross-linking function in the aortic matrixof the eNOS�/� mice may contribute to the increased stiffnessobserved in the decellularized tissue scaffold. Substrates forTG2 in the vascular scaffold could include collagen and fi-bronectin (27); the biochemical analysis to confirm this re-mains the focus of ongoing studies. Indeed, increased TG2-mediated cross-linking results in increased matrix stiffness invitro (6). Interestingly, the tensile properties of intact aorta ofeNOS�/� mice diverged from the WT mice earlier than thedecellularized scaffolds. This suggests that altered cell-matrixinteractions contribute to increased vascular stiffness in theeNOS�/� mouse model. Moreover, TG2-integrin �1 interac-tion is higher in the eNOS�/� mouse aorta, which can directlyfacilitate cell-matrix interaction. Smooth muscle cells fromeNOS�/� mice are stiffer when placed on a fibronectin matrix,which may in part be mediated by increased TG-integrin �1interaction, but their response to phenylephrine remains unal-tered. Taken together, these results underscore the potentialdual role of TG2 in mediating vascular stiffness by alteringmatrix stiffness (6, 11) and by mediating cell-matrix interac-

tion (1). The specific role of TG2 in this context may beexamined by genetic rescue of function experiments whereincatalytically active and C277S catalytically deficient mutantcan be delivered to the aorta of TG2�/� mice followed byassessment of mechanical properties. Such experiments are thefocus of ongoing studies.

Limitations

The simultaneous measurement of PWV with MAP allowsthe comparison of two populations with markedly differentresting MAPs. However, the use of vasoactive drugs (phenyl-ephrine and sodium nitroprusside) may affect intrinsic vesseltone and thus PWV measurement. However, all groups receivethe same drugs, and thus all groups would be prone to the samedrug effects. In addition, this protocol in rats has been shownto have no significant effect on large artery stiffness over andabove the pressure dependence (5). Although the relativelyshort distance (1 cm) between the sensors ensures that thePWV measurement is obtained in the upper thoracic trunk, italso imposes limitations on the resolution of transit timedetection. The estimated resolution for transit time using thesecond differential is within 0.5 ms. For the highest PWVvalues in the pressure range where significance was found, thetransit time is of the order of 2 ms. Hence, the resolution is wellwith the maximum limits. Furthermore, eNOS�/� mice havebasal hypertension. This may by itself alter vessel wall stiff-ness and/or result in TG2 activation mechanisms that were notexhaustively addressed in this study.

In conclusion, this study highlights a critical role for eNOS-dependent NO in regulating TG2 and establishes an inverserelationship between NO bioavailability and TG2 secretion andcross-linking activity. To the best of our knowledge, this is thefirst study to report simultaneous measurement of PWV andMAP in a mouse model, and contributes a new methodology toaddress PWV characteristics over a full range of physiologicalpressures in mice. Divergence of PWV at higher pressures inthe eNOS�/� mice compared with WT mice suggests thatalterations in vascular stiffness are significant in a hypertensivesetting. The response of the eNOS�/� mice to pathophysiolog-ical stresses leading to vascular remodeling are likely differentthan the WT mice. Increase in extracellular TG2 coincides withincreased stiffness of the intact aorta, with only a modestincrease in the stiffness of the vascular matrix, suggesting anoncatalytic role for TG2 in mediating vascular stiffness.Finally, at resting conditions, the altered vascular phenotype inthe eNOS�/� mouse is likely due to endothelial dysfunctionand attenuated smooth muscle tone and function occurring dueto the absence of NO.

GRANTS

This work was supported by a National Heart, Lung, and Blood InstituteGrants 1R01-HL-105296-01 (to D. E. Berkowitz) and HL-107361 (to S. S.An), and an Australian Research Council Grant DP110101134 (to A. Avolio).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: S.M.J., S.J., J.S., A.B., A.P., E.Y.C., M.B., K.V., andL.S. performed experiments; S.M.J., J.S., A.B., S.S.A., and L.S. analyzed data;S.M.J., J.S., A.B., S.S.A., and L.S. prepared figures; S.M.J. and L.S. draftedmanuscript; S.S.A., D.N., A.P.A., D.E.B., and L.S. interpreted results of

IP: Integrin 1 WB: TG2

WB: Integrin 1

WT

eNO

SK

O A

B

IgG

con

trol

WT eNOS KO0

100

200

300

400

TG2/

Inte

grin

(N

orm

aliz

ed to

WT) *

(n = 5)

WT eNOS KO0.0

0.2

0.4

0.6

Cel

l Stif

fnes

s (P

a/nm

)

p = 0.05

Fig. 5. Cell-matrix interactions and CSK stiffness of isolated vascular smoothmuscle cells from eNOS�/� and WT mice. A: proteins (50 �g) of aortichomogenates were immunoprecipitated using integrin �1 antibody; higherlevels of TG2 co-immunoprecipitated with integrin �1 in eNOS�/� mouse(n 6); proteins from eNOS�/� mice were immunoprecipitated with isotypeIgG as control (right lane). *P � 0.05. B: CSK stiffness of freshly isolatedvascular smooth muscle cells from the aorta of eNOS�/� and WT mice wasmeasured using magnetic twisting cytometry. Data are presented as geometricmeans � SE (n 248 cells for eNOS�/�; n 199 cells for WT).

H809INCREASED TRANSGLUTAMINASE ACTIVITY CONTRIBUTES TO VASCULAR STIFFNESS

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00103.2013 • www.ajpheart.org

experiments; S.S.A., D.N., A.P.A., D.E.B., and L.S. edited and revised man-uscript; A.P.A., D.E.B., and L.S. conception and design of research; L.S.approved final version of manuscript.

REFERENCES

1. Akimov SS, Krylov D, Fleischman LF, Belkin AM. Tissue transglu-taminase is an integrin-binding adhesion coreceptor for fibronectin. J CellBiol 148: 825–838, 2000.

2. Alcock J, Warren AY, Goodson YJ, Hill SJ, Khan RN, Lymn JS.Inhibition of tissue transglutaminase 2 attenuates contractility of pregnanthuman myometrium. Biol Reprod 84: 646–653, 2011.

3. Bakker EN, Buus CL, Spaan JA, Perree J, Ganga A, Rolf TM, SoropO, Bramsen LH, Mulvany MJ, Vanbavel E. Small artery remodelingdepends on tissue-type transglutaminase. Circ Res 96: 119–126, 2005.

4. Bakker EN, Pistea A, Spaan JA, Rolf T, de Vries CJ, van Rooijen N,Candi E, VanBavel E. Flow-dependent remodeling of small arteries inmice deficient for tissue-type transglutaminase: possible compensation bymacrophage-derived factor XIII. Circ Res 99: 86–92, 2006.

5. Butlin M, Hammond A, Lindesay G, Viegas K, Avolio AP. In-vitro andin-vivo use of vasoactive agents in characterising aortic stiffness in rats:testing the assumptions. J Hypertension. In press.

6. Chau DY, Collighan RJ, Verderio EA, Addy VL, Griffin M. Thecellular response to transglutaminase-cross-linked collagen. Biomaterials26: 6518–6529, 2005.

7. Davies JE, Rose C, Sarkar S, Rubinsztein DC. Cystamine suppressespolyalanine toxicity in a mouse model of oculopharyngeal musculardystrophy. Sci Transl Med 2: 34–40, 2010.

8. Deshpande DA, Wang WC, McIlmoyle EL, Robinett KS, SchillingerRM, An SS, Sham JS, Liggett SB. Bitter taste receptors on airwaysmooth muscle bronchodilate by localized calcium signaling and reverseobstruction. Nat Med 16: 1299–1304, 2010.

9. Elli L, Ciulla MM, Busca G, Roncoroni L, Maioli C, Ferrero S,Bardella MT, Bonura A, Paliotti R, Terrani C, Braidotti P. Beneficialeffects of treatment with transglutaminase inhibitor cystamine on theseverity of inflammation in a rat model of inflammatory bowel disease.Lab Invest 91: 452–461, 2011.

10. Engholm M, Eftekhari A, Chwatko G, Bald E, Mulvany MJ. Effect ofcystamine on blood pressure and vascular characteristics in spontaneouslyhypertensive rats. J Vasc Res 48: 476–484, 2011.

11. Engler AJ, Griffin MA, Sen S, Bonnemann CG, Sweeney HL, DischerDE. Myotubes differentiate optimally on substrates with tissue-like stiff-ness: pathological implications for soft or stiff microenvironments. J CellBiol 166: 877–887, 2004.

12. Fabry B, Maksym GN, Butler JP, Glogauer M, Navajas D, FredbergJJ. Scaling the microrheology of living cells. Phys Rev Lett 87: 148102,2001.

13. Franklin SS. Hypertension in older people: part 1. J Clin Hypertens(Greenwich) 8: 444–449, 2006.

14. Franklin SS. Hypertension in older people: part 2. J Clin Hypertens(Greenwich) 8: 521–525, 2006.

15. Huang PL, Huang Z, Mashimo H, Bloch KD, Moskowitz MA, BevanJA, Fishman MC. Hypertension in mice lacking the gene for endothelialnitric oxide synthase. Nature 377: 239–242, 1995.

16. Iismaa SE, Mearns BM, Lorand L, Graham RM. Transglutaminasesand disease: lessons from genetically engineered mouse models andinherited disorders. Physiol Rev 89: 991–1023, 2009.

17. Izzo JL Jr. Arterial stiffness and the systolic hypertension syndrome.Curr Opin Cardiol 19: 341–352, 2004.

18. Jandu SK, Webb AK, Pak A, Sevinc B, Nyhan D, Belkin AM,Flavahan NA, Berkowitz DE, Santhanam L. Nitric oxide regulatestissue transglutaminase localization and function in the vasculature. AminoAcids 44: 261–269, 2013.

19. Kass DA. Age-related changes in venticular-arterial coupling: pathophys-iologic implications. Heart Fail Rev 7: 51–62, 2002.

20. Kobayashi M, Inoue K, Warabi E, Minami T, Kodama T. A simplemethod of isolating mouse aortic endothelial cells. J Atheroscler Thromb12: 138–142, 2005.

21. Laurent S, Boutouyrie P, Asmar R, Gautier I, Laloux B, Guize L,Ducimetiere P, Benetos A. Aortic stiffness is an independent predictor ofall-cause and cardiovascular mortality in hypertensive patients. Hyperten-sion 37: 1236–1241, 2001.

22. Mitchell GF, Hwang SJ, Vasan RS, Larson MG, Pencina MJ, Ham-burg NM, Vita JA, Levy D, Benjamin EJ. Arterial stiffness andcardiovascular events: the Framingham Heart Study. Circulation 121:505–511, 2010.

23. Ng K, Butlin M, Avolio AP. Persistent effect of early, brief angiotensin-converting enzyme inhibition on segmental pressure dependency of aorticstiffness in spontaneously hypertensive rats. J Hypertens 30: 1782–1790,2012.

24. Santhanam L, Berkowitz DE, Belkin AM. Nitric oxide regulates non-classical secretion of tissue transglutaminase. Commun Integr Biol 4:584–586, 2011.

25. Santhanam L, Tuday EC, Webb AK, Dowzicky P, Kim JH, Oh YJ,Sikka G, Kuo M, Halushka MK, Macgregor AM, Dunn J, Gutbrod S,Yin D, Shoukas A, Nyhan D, Flavahan NA, Belkin AM, Berkowitz DE.Decreased S-nitrosylation of tissue transglutaminase contributes to age-related increases in vascular stiffness. Circ Res 107: 117–125, 2010.

26. Soucy KG, Ryoo S, Benjo A, Lim HK, Gupta G, Sohi JS, Elser J, AonMA, Nyhan D, Shoukas AA, Berkowitz DE. Impaired shear stress-induced nitric oxide production through decreased NOS phosphorylationcontributes to age-related vascular stiffness. J Appl Physiol 101: 1751–1759, 2006.

27. van den Akker J, VanBavel E, van Geel R, Matlung HL, Guvenc TunaB, Janssen GM, van Veelen PA, Boelens WC, De Mey JG, Bakker EN.The redox state of transglutaminase 2 controls arterial remodeling. PLosOne 6: e23067, 2011.

28. Vanbavel E, Bakker EN. A vascular bone collector: arterial calcificationrequires tissue-type transglutaminase. Circ Res 102: 507–509, 2008.

29. Zieman SJ, Melenovsky V, Kass DA. Mechanisms, pathophysiology,and therapy of arterial stiffness. Arterioscler Thromb Vasc Biol 25:932–943, 2005.

H810 INCREASED TRANSGLUTAMINASE ACTIVITY CONTRIBUTES TO VASCULAR STIFFNESS

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00103.2013 • www.ajpheart.org