Nicotine Affects Murine Aortic Stiffness and Fatigue Response During Supraphysiological Cycling

Increased stiffness and reduced distensibility of arteries are known contributing factors for cardiovascular conditions, such as stroke, hypertension, or myocardial infarction [1]. Increased aortic stiffness is also considered a predisposing factor for the formation of abdominal aortic aneurysms (AAA), and an understanding of aortic biomechanics is essential to facilitate more accurate models of stress distribution that can assist in predicting the risk of AAA rupture [2]. Most AAA patients (>90%) have a history of smoking [3]. Smoking has been identified as an independent risk factor for arterial stiffening [4,5], with nicotine as the main component in tobacco smoke in particular [6,7]. In this context, we recently examined increased aortic stiffness in abdominal and thoracic murine arteries that were subjected to nicotine exposure for 40 days [8]. The stiffness increased in the abdominal segment and thoracic segment at 10 days and 40 days, respectively, which was associated with enhanced activity of matrix metalloproteinases and elastin degradation. Of note, the stiffness of thoracic aortic segments was not significantly impacted by nicotine exposure for only 10 days suggesting that other mechanical measurements may be necessary to obtain insights into early pathomechanisms.

Several experimental techniques have been applied to estimate the mechanical properties of arteries. Direct methods of measuring elasticity ex vivo include uniaxial tension [9,10], biaxial tension [11], wire myography [12], and pressure dilation [13]. Noninvasive methods of measuring arterial stiffness include pulse wave velocimetry [14,15] and ultrasound imaging [16,17]. Equibiaxial tension testing is important and informative for discerning anisotropy and orthotropic mechanical properties, but such testing has typically been limited to larger tissue specimens (e.g., porcine pulmonary arteries [18]). Very small tissues such as mouse aortic valve leaflets have required approaches such as adhesion to carrier membranes [19]. Cyclic uniaxial tension tests on human atherosclerotic plaques have revealed differences in mechanical strength depending on tissue type, with the lowest fracture stress among intimal tissues occurring in the circumferential direction of the fibrous cap [20].

Several material models, both phenomenological and structural, have been applied to describe arterial wall mechanics based on data from uniaxial extension experiments [21]. Although anisotropic material models are inherently more thorough in structural modeling, isotropic material models have also shown to have utility for arterial wall modeling under supraphysiological loadings [22]. Structurally motivated stored energy functions using four-directional fibers (axial, circumferential, and two diagonals) have been used to describe murine inferior vena cava subjected to inflation testing [23].

In this study, we apply multiple modes of mechanical testing to examine the effects of nicotine exposure on the stiffness, viscoelasticity, and fatigue behavior of murine thoracic aortas. In addition to conducting stress–strain and stress relaxation tests, we take special interest in how the effects of nicotine can be revealed over hundreds of cycles of repetitive stretching and relaxation. We hypothesize that nicotine exposure will exacerbate strain softening of aortic tissue at supraphysiological cyclic stretching.

Murine arteries were harvested from C57BL/6 male mice, aged 10–12 weeks, purchased from The Jackson Laboratory (Bar Harbor, ME). All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) for the Palo Alto Veterans Institute for Research (PAVIR) at the VA Palo Alto Health Care System (VAPAHCS). Mice were treated with nicotine in sterile phosphate-buffered saline through subcutaneous mini-osmotic pumps, which was placed into a skin pouch in the neck (ALZET model 2004, DURECT Corporation, Cupertino, CA) and set to a concentration of 25 mg/kg/day for a total of 28 days. Implantation of the osmotic mini pumps is minimally invasive, requiring an approximately five-minute procedure to introduce a small subcutaneous pouch and closure with one to two staples [24]. Control mice received phosphate-buffered saline only. Following explantation, the aortas were cut longitudinally to achieve an open-circumferential configuration for imaging and mechanical testing. The circumference of each test specimen was approximately 3 mm and the nominal length along the longitudinal axis was approximately 5 mm. The cross-sectional area for each specimen was determined before testing by measuring thickness and length, while saturated in physiological Tyrode's solution. The thickness of each specimen was measured between two glass slides, using a digital micrometer with a rated accuracy of  ± 1 μm. After each respective specimen was gripped between clamps, the length was measured using a calibrated 5-megapixel microscope camera (MU500, AmScope, Irvine, CA), within an uncertainty of approximately  ± 0.1 mm.

Five aortas were tested from each treatment group (i.e., five nicotine-treated and five untreated controls). Each aorta was stretched uniaxially between aluminum grips in an in-house tensometer system (Fig. 1). The system uses a 113 g load cell (LCL-113G, OMEGA Engineering, Norwalk, CT), a programable motorized nanopositioner (MP- 285, Sutter Instrument, Novato, CA), and a 24-bit digital acquisition unit (Loadstar DI-1000 U, Loadstar, Fremont, CA). In contrast to pressure dilation, this configuration allows us to apply supraphysiological loading to accelerate fatigue observations. This open uniaxial configuration does not require the lumen to be sealed, avoiding the problem of fluid leakage that can occur in pressurization methods.

Mechanical testing was performed within 6 h of harvesting, while tissues were stored in physiological Tyrode's solution at 4 °C until ready for testing. Each aortic segment was stretched uniaxially under prescribed displacement in the circumferential direction. In contrast to arterial tissue specimens from larger animals, the small size of murine aortas (nominally 3 mm × 5 mm) made uniform suturing and biaxial stretching prohibitive. During testing, the gripped aorta was immersed in a stationary container filled with approximately 30 mL of Tyrode's solution. To ensure consistent clamping force, the screws for the grips were secured with 71 N·mm (10 in·oz) of torque. After immersion, the load cell was attached using a quick-release spring-loaded pin. The clamps began in flush contact for each test, establishing an initial gage length of 1.0 mm before stretching. To account for clamp weight and buoyancy after immersion in the physiological fluid, the relative position between the grips was inspected under a microscope and adjusted if necessary to reestablish consistent gage length. The time required for inspection, alignment, and procedural confirmation of machine scripts and data logging was at least 3 min. The corresponding time that the tissue was fully immersed in the room-temperature testing vessel ensured ambient temperature conditions before commencement of testing. The total elongation of the system is the sum of aorta elongation and cantilever tip deflection. Accordingly, the deflection of the cantilever tip was subtracted from the total displacement of the motorized actuator, to determine the actual elongation and strain for each aorta.

Three stages of testing were conducted: (1) a sequence of five loading and unloading cycles, by applying actuator extension of 1000 μm at a gentle rate of 200 μm/s, and returning to gage length at the same speed; (2) one abrupt actuator extension of 1000 μm at a rate of 5000 μm/s, with a hold for 30 s to measure viscoelastic response; and (3) 500 cycles of repetitive loading with a mean displacement of 1500 μm and a superimposed alternating displacement of ±500 μm (i.e., actuator extension between 1000 μm and 2000 μm). The choice of five preconditioning cycles was consistent with precedent for testing of porcine pulmonary arteries [25] and rabbit coronary arteries [26]. Actuator speed was set to 5000 μm/s for fatigue loading. The actual speed required deceleration and acceleration when reversing directions, such that the time to complete 500 cycles was approximately 12 min.

Force measurements during stress relaxation were recorded at a sampling rate of 100 Hz. Force measurements for preconditioning and fatigue cycling were recorded at a sampling rate of 10 Hz. These sampling rates ensure that even at a maximum speed of 5000 μm/s over a travel distance of 1000 μm, the Nyquist– Shannon sampling criterion is satisfied (i.e., the sampling interval of 100 ms is half of the fastest possible actuation time of 200 ms).

To compare the effect of nicotine on the mechanical response of the aortic tissue, stress–strain data were fit to a Fung-type phenomenological model [27] based on the strain energy density of the general form ρW = C₀ exp(a₁E₁² + a₁E₁² + 2a₁₂E₁E₂) for biaxial loading, where ρW is the strain energy density with respect to initial volume, E_ij is Green strain components, and C₀ and a_ij are material constants.

For uniaxial loading in the circumferential direction of the sectioned aorta, only E₁, and its corresponding material constant a₁ were used for fitting to the experimental stress–strain data. This highly idealized simplification requires that perpendicular strain (i.e., in the longitudinal direction of the sectioned aorta) be negligible. Accordingly, the perpendicular strains were inspected by microscope imaging during pull testing, and the maximum perpendicular strain (i.e., narrowest extent of necking in the middle) was measured to be less than 4.3% of the strain along the direction of applied extension. The thicknesses of all specimens (when held flat between glass slides) were less than 60 μm, such that deformation along the thickness direction was also considered negligible relative to the applied extension. The tensile stress σ = 2 C₀a₁ε exp(a₁ε²) is the partial derivative of the strain energy density with respect to uniaxial strain ε. The stress–strain data for each of the control and nicotine-treated aortas were fit by nonlinear least-squares curve-fitting using the scipy.optimize.curve_fit() function in Python [28] to determine parameters C₀ and a₁ for each aorta.

Multiphoton images of the murine aortas were acquired using two-photon excited fluorescence (2PEF) and second-harmonic generation (SHG) to image elastin and collagen, respectively [29,30]. The images were acquired using an upright multiphoton confocal platform (Leica DM6000, Buffalo Grove, IL) at the Cell Sciences Imaging Facility at Stanford University. Open circumferential tissue specimens were placed on a microscope slide and wetted with a droplet of Tyrode's physiological solution. A z-stack was generated for each aorta, with images sliced at every 2 μm with a field of view of approximately 1.5 mm × 1.5 mm. Images were taken at two different locations along the length, approximately 2 mm apart, recognizing that aortas (and those of C57BL/6 in particular) have substantial variability in morphology and strain distribution along their length [31]. Intensity for each SHG and 2PEF z-stack was normalized to the maximum intensity in the stack. Since the exact geometric boundaries for the beginning and end of wall tissue were difficult to discern objectively, the normalized origin for each z-stack was set to where 2PEF intensity reached 5%, and the normalized position of 1.0 was set where the elastin signal diminished to 0.1%.

Results were analyzed by a Student's t-test for differences between means for control and nicotine-treated aortas, following testing for normal distribution. A condition of p < 0.05 was considered significant. Statistical significance tests were performed using the statistics and machine learning toolbox in matlab.

Figure 2 shows the stress–strain response of control and nicotine-treated aortas upon the fifth loading cycle (i.e., after four cycles of preconditioning). Best-fit values for parameters C₀ and a₁ are summarized in Table 1, showing the mean values, standard errors of the mean. Quality of fit for the average is shown in terms of root-mean-square-error (RMSE), showing the higher quality of fit for the control aortas. The local modulus of elasticity (i.e., the slope of the fitted stress–strain curve) at 50% strain was significantly higher (p = 0.03) for the nicotine-treated aortas (377 kPa ± 165 kPa) than for the control counterparts (191 kPa ± 65 kPa).

Normalized stress relaxation curves were calculated for control and nicotine-treated aortas (Fig. 3). Both cases decline to approximately 92% of maximum stress after 30 s. Power-law fits of the form σ₀ = βt^−α were applied to the experimental data, in which σ₀ is normalized stress, t is the time in seconds, and β and α are fitting parameters (Table 2). The exponential parameter α values are less than unity and consistent with fractional calculus models that describe uniaxial stress relaxation of arteries [32]. These relaxation tests do not indicate statistically significant differences between control and nicotine-treated aortas (p > 0.05) for the fitting coefficients. There was an associated amount of simultaneous creep that occurred as the relaxation was coupled with elastic springback of the in-line load cell, but at 50% applied strain the contribution from creep was within 3%.

In contrast to the relatively modest strain magnitude for preconditioning and stress–strain measurement, fatigue tests were conducted at elevated stress to reveal longer-term effects rapidly under accelerated testing [33]. Fig. 4(a) shows a close-up of stress during extension and relaxation cycling for the first 25 cycles of testing for one representative (control) specimen. Figure 4(b) compares stress S versus the number of cycles N for the five nicotine-treated and the five control specimens. The slopes and intercepts a fitted equation S = A − BN^C, adopted from a fatigue degradation model used for composite materials [34] and high-cycle loading of arterial elastin [35]. Raw data without log transformation is shown in Fig. S1 available in the Supplemental Materials on the ASME Digital Collection. The fitting parameters A, B, and C can be described as the peak stress, acute loss of tension, and degradation slope, respectively. A fourth variable D describes the vertical span of the fatigue data (i.e., the difference between the maximum and minimum stresses as the data reached a plateau). Effects of each of these parameters on the shape of the cyclic loading plot are explained in Fig. S2 available in the Supplemental Materials on the ASME Digital Collection. Values for the parameters were found using the least-squares curve fitting in matlab. The vertical span D was quantified as the difference between maximum and minimum stress values near the end of cycling, arbitrarily but consistently at 400 cycles. Compared to control aortas, nicotine-treated aortas exhibited a 108% higher value of parameter A, 118% higher value of parameter B, 10.4% higher value of parameter C, and 107% higher value of vertical span D (Fig. 5). Overall, these data revealed a significant degradative effect with an increasing number of cycles, with quantitatively distinct behavior caused by nicotine supplementation.

Figure 6(a) shows a sectional view of elastin (2PEF, lower layer) and collagen (SHG, upper layer) distributions of a representative control aorta, with the positive z-axis along the radially outward direction. Elastin and collagen are characteristically localized to medial and adventitial sections, respectively. Figure 6(b) shows intensity as a function of normalized radial position (i.e., from the luminal side of the aorta wall to the outer side) for the SHG images of collagen and 2PEF images of elastin. These distributions represent the average of ten z-stacks (i.e., two locations on each of five aortas for the control and nicotine treatment, respectively). The proportions of SHG and 2PEF signals are approximately even, consistent with compositional analysis using cross-sectioning and histological staining of thoracic aortas from C57BL/6 mice [36].

The shapes of the elastin intensity distributions for control and nicotine-treated aortas are similar. For the collagen, however, at approximately 25% from the luminal side, a conspicuously higher proportion of the SHG intensity is observed. When comparing the proportional SHG area below z = 0.5 (shaded for nicotine in Fig. 6(b)), the nicotine-treated case at 23.5% ± 2.1% was significantly higher (p = 0.01) than the control counterpart at 19.6% ± 2.4% (mean ± standard deviation). A plausible explanation for this difference is that the locally high SHG intensity may correspond to the deposition of compensatory collagen [37]. The higher average stiffness observed for nicotine-treated arteries is consistent with other studies that have combined micromechanical testing of arteries with multiphoton imaging [38], for which it has been observed that a decrease in the elastin-to-collagen ratio corresponds to higher arterial stiffness.

The severity of residual stress is typically evidenced in the opening angle of open-circumferential artery sections, and residual stresses are known to be significant for transmural circumferential stress distribution in larger tissue specimens [39]. For these very thin mouse tissues, however, residual stresses were minimal, as evidenced by the high degree of flatness (Fig. 6(a)) that was achieved. Although some curling was evident in the free state, the magnitude of residual stress was small enough such that surface tension was sufficient to achieve flatness against the glass substrate when saturated in the physiological fluid.

In this study, we characterized the effect of nicotine exposure on arterial stiffness, viscoelastic response, and fatigue cycling of murine thoracic aortas. The stress–strain curves exhibit strain stiffening that is characteristic of systemic arteries [27] and stress magnitudes are in a range similar to observations in prior studies using mouse thoracic aortas. Specifically, our observed values for stress near 50 kPa at 50% strain (Fig. 2) are consistent with circumferential mechanical testing of mouse aortas in Ref. [40]. Our measured (normal) stress values are also in a similar kilopascal range as “wall stress” measurements on mouse aortas [41], where wall stress values are expectedly lower because of the larger area used in computation (i.e., inside circumference times length, rather than cross-sectional area perpendicular to the direction of tension). Although inflation-based testing has been relatively well established, our approach of direct pulling offers much faster cycle time. Standard protocols for inflation testing recommend increasing pressure gradually by 10 mmHg every five minutes [42]. The physiological pressure difference between diastolic and systolic pressures is approximately 40 mmHg, which would require several hours to complete fatigue cycling tests. Fluid leakage through arterial branch points further limits the rate of cycling for inflation-based methods. In contrast, our direct-pull experiments typically completed 500 loading cycles in 12 min (i.e., faster than 1.5 s per cycle and over 40 cycles per minute). Although our cycling rate is not as fast as the normal heart rate for mice (between 310 and 840 beats per minute [43]), direct mechanical testing is substantially closer to physiological heart rate than inflation-based testing. Cycling rate may make a critical difference in evaluating tissue response. When collagen substrates are mechanically cycled under much gentler conditions of less than 30% strain and cycle period of 50 min (i.e., ∼2000 slower than our tests), fatigue stiffening–rather than softening–has been observed [44]. Fatigue stiffening during slow cycling suggests that stabilization occurs within or between fibrils, but an opportunity for similar stabilizing changes may not be present during rapid oscillations. Such data can be useful for validating cyclic stress softening models [35] as well as constitutive damage models specifically for biological tissues, as has been applied to rat tail tendon tissues under supraphysiological uniaxial loading [36].

In this work, we observed a modest increase in the arterial stiffness at physiological strains due to nicotine treatment, which is consistent with our previous observations [8]. This small increase suggests that elastin in nicotine-treated aortas has similar overall behavior as in control aortas, and suggests that there is no remarkable degradation of elastin, at least for the exposure duration in these experiments. In the physiological range up to approximately 50% strain, the aorta exhibits relatively high extension with low stress, with elastin highly recruited at low strain [45]. Crosslinked collagen fibers have elastic modulus as high as 7500 MPa and extensibility up to 16%, while these values for elastin fibers are 1 MPa and 150%, respectively [46]. This large difference in mechanical properties thus motivated the use of supraphysiological loading to reveal significant differences between nicotine-treated and control aortas. Our data show how the effect of nicotine-induced damage on collagen is manifested under supraphysiological testing conditions, which may not have become apparent in physiological testing conditions, either in vitro or by in vivo methods such as using pulse wave velocimetry [47]. Supraphysiological cyclic loading has special relevance for pathological conditions in which tissue may be subjected to higher-than-normal stress concentrations in the vicinity of aneurysms or other structural abnormalities [48]. This could imply that nicotine-induced damage, although not always detectable in experimental settings, may still result in impaired functionality of the artery.

Changes in arterial stiffness can be observed even without substantial changes in elastin distribution, as measured by our 2PEF imaging data. At supraphysiological loading (>80% strain), collagen is the major load-bearing component in arteries [49]. The changes in the SHG intensity profiles of collagen suggest that changes to collagen or its distribution can provide a plausible explanation for the observations in our fatigue experiments. Attribution to changes in collagen is supported by several studies associating nicotine addiction or smoking with impaired vessel collagen architecture [50–52] and inflammation-induced with collagen remodeling tissue damage [53,54]. Some evidence also suggests that the mechanical degradation is not necessarily attributed to collagen fiber realignment. For example, circumferential loading of rabbit carotid arteries showed that fiber directions remained unchanged even beyond 50% strain [26].

The results in this study are presented as a relative comparison between nicotine-treated and control specimens under identical test conditions. However, there are two notable limitations. First, we have evaluated the contributions of elastin and collagen from a phenomenological approach. Although we believe that our conclusions present a likely scenario, similar mechanical testing in transgenic mouse models would be needed for obtaining detailed mechanistic insights into the relative contribution of collagen and elastin at physiological and supraphysiological strains. Second, we performed multiphoton imaging using only unstretched test specimens. Compared to tissues from larger animals (e.g., rabbit or porcine) it is very difficult to ascertain real-time structural details for mouse aortas under mechanical strain [55]. As imaging resolution and capabilities continue to develop, it will be valuable to investigate microstructural changes in greater detail to develop an informative constitutive model based on wall composition and fiber orientation.

Our study does not include unoperated controls (i.e., aortas from mice that were not subject to saline-filled pumps with no nicotine) and, therefore, does not conclude that nicotine-treated aortas have different behavior than aortas from (unoperated) mice in general. Rather, our results show a direct comparison between nicotine exposure and saline controls, consistent with similar studies in the field of aortic aneurysm, dissection, and atherosclerosis using saline-filled pumps [56–58].

Our investigation using murine aorta under open-circumferential, uniaxial loading showed higher stiffness after 28 days of nicotine exposure. In addition, rapid cyclic loading revealed changes in mechanical behavior with softening effects (e.g., acute loss of tension and larger changes of stress) potentially resulting from collagen redistribution in the media due to nicotine exposure. These new observations from cyclic loading offer additional insights into the role of collagen on the pathomechanisms of nicotine-mediated vascular remodeling beyond changes in elastic and short-term viscoelastic behavior alone. Supraphysiological loading and fatigue cycling may thus help to detect phenotypic differences sooner than possible with quasi-static, physiological loading.

The authors express appreciation to Wilson Eng for designing the first-generation tensometer apparatus that was adapted for this study, and also to Sue-Mae Saw for help with mechanical testing experiments. The authors further acknowledge the Cell Sciences Imaging Facility at Stanford University for help with multiphoton microscopy.

• American Heart Association (Grant No. 18AIREA33960524; Funder ID: 18AIREA33960524).

• California State University Program for Education & Research in Biotechnology (CSUPERB) (Funder ID: 10.13039/100008134).

• National Institutes of Health (Grant No. NHLBI 1R56HL135654; Funder ID: 10.13039/100000050).

• Veterans Affairs Office of Research and Development, U.S. Department of Veterans Affairs (Grant No. 1I01BX002641; Funder ID: 10.13039/100000738).

• University of California Tobacco-Related Disease Research Program (Grant No. T29IR06360; Funder ID: 10.13039/100005188).

• German Research Society (Deutsche Forschungsgemeinschaft) (Grant Nos. MU 4309/1-1 and WA 33533/2-1; Funder ID: 10.13039/501100001659).

A =

fitting parameter for peak stress in fatigue cycling

B =

fitting parameter for acute loss of tension in fatigue cycling

C =

fitting parameter for degradation slope in fatigue cycling

C₀, a_ij =

material constants for nonlinear elasticity

D =

span in fatigue cycling

E_ij =

Green strain components

n =

number of experimental specimens

N =

number of fatigue cycles

p =

probability of rejecting null hypothesis

S =

nominal stress for fatigue cycling

t =

time

z =

vertical scanning coordinate for multiphoton imaging

β, α =

fitting parameters for power-law fit

ε =

uniaxial tensile stress

ρW =

volumetric strain energy density

σ =

uniaxial tensile stress

σ₀ =

normalized stress (for stress relaxation)