We describe simultaneous mechanical testing and 3D-imaging of the arterial wall of isolated, live human resistance arteries, and Fiji and Ilastik image analyses for the quantification of elastin and collagen spatial organization and volume densities. We discuss the use of these data in mathematical models of arterial wall mechanics.
The pathogenic contribution of resistance artery remodeling is documented in essential hypertension, diabetes and the metabolic syndrome. Investigations and development of microstructurally motivated mathematical models for understanding the mechanical properties of human resistance arteries in health and disease have the potential to aid understanding how disease and medical treatments affect the human microcirculation. To develop these mathematical models, it is essential to decipher the relationship between the mechanical and microarchitectural properties of the microvascular wall. In this work, we describe an ex vivo method for passive mechanical testing and simultaneous label-free three-dimensional imaging of the microarchitecture of elastin and collagen in the arterial wall of isolated human resistance arteries. The imaging protocol can be applied to resistance arteries of any species of interest. Image analyses are described for quantifying i) pressure-induced changes in internal elastic lamina branching angles and adventitial collagen straightness using Fiji and ii) collagen and elastin volume densities determined using Ilastik software. Preferably all mechanical and imaging measurements are performed on live, perfused arteries, however, an alternative approach using standard video-microscopy pressure myography in combination with post-fixation imaging of re-pressurized vessels is discussed. This alternative method provides users with different options for analysis approaches. The inclusion of the mechanical and imaging data in mathematical models of the arterial wall mechanics is discussed, and future development and additions to the protocol are proposed.
The pathogenic contribution and effects of resistance artery remodeling are documented in essential hypertension, diabetes and the metabolic syndrome1,2,3,4,5. Deciphering the relationship between the mechanical and microarchitectural properties of the microvascular wall is essential for developing mathematical models of this association. Such models will improve understanding the remodeling process and will support the development of in silico models useful for testing pharmacological strategies targeting disease related remodeling of the arterial wall.
Prior studies focused in understanding how the microarchitecture of the arterial wall relates to arterial wall mechanics by incorporating mechanical measures and the microarchitecture of the extracellular matrix (ECM) are almost exclusively performed on large, elastic conduit arteries from mice or swine6,7,8,9,10,11. Imaging of the microstructures of the wall is typically performed using nonlinear optical techniques, taking advantage of the autofluorescence of elastin and second harmonic generation by collagen. This allows spatiotemporal imaging of the two major components of the extracellular matrix, elastin and collagen, without a need for staining. Imaging of the arterial wall in full thickness is a challenge in large conduit arteries due to scatter of the light in the thick tunica media. However, to determine how the microarchitecture of the structural components of the arterial wall relate to the observed mechanical properties, three-dimensional information must be obtained during the mechanical testing. For large arteries like the human aorta, this requires biaxial mounting, mechanical testing and imaging of regions of interest in 1-2 cm2 pieces of the arterial wall7,9,10,12. Only part of the wall can be imaged and mechanically tested.
For smaller arteries of any species (e.g., human pericardial13, pulmonary14 and subcutaneous15 arteries, rat mesenteric arteries16,17,18,19,20, mouse cremaster, mesenteric, cerebral, femoral and carotid arteries21,22,23,24,25,26,27) imaging of the entire wall thickness is possible and can be combined with mechanical testing. This allows simultaneous recording of the mechanical properties and the structural arrangements within the wall. However, a direct mathematical modeling of the relationship between the observed alterations in the three-dimensional structure of the ECM and changed mechanical properties of the resistance arterial wall, has to the best of our knowledge only been reported upon recently in human resistance arteries13,15.
In this work, an ex vivo method for passive mechanical testing and simultaneous three-dimensional imaging of the microarchitecture of elastin and collagen in the arterial wall of isolated human resistance arteries is described. The imaging protocol can be applied to resistance arteries of any species of interest. Image analyses are described for obtaining measures of internal elastic lamina branching angles and adventitial collagen straightness13 using Fiji28. Collagen and elastin volume densities are determined using Ilastik software29 and finally, the inclusion of the mechanical and imaging data in mathematical models of the arterial wall mechanics is discussed.
The goal of describing the imaging and image analyses techniques in combination with mathematical modeling is to provide investigators a systematical approach to describe and understand observed pressure induced changes in the ECM of resistance arteries. The described method is focused in quantifying the changes in the ECM in a vessel during pressurization, by comparing the structure of the ECM at 20, 40 and 100 mmHg. These pressures were chosen for determining the structure of the arterial wall at its more compliant (20 mmHg), stiff (100 mmHg) and intermediate (40 mmHg) state, respectively. However, any process in the vascular wall of live arteries, including changes induced by vasoactive components, hysteresis and flow, can be quantified, depending on the research hypothesis in question by the investigator.
The use of two-photon excitation fluorescence microscopy (TPEM) in combination with a pressure myograph for studying pressure (or other) induced changes in the ECM of live arteries is emphasized. First, because this allows simultaneous acquisition of the overall three-dimensional structure of the arterial wall (diameter and wall thickness) along with three-dimensional label-free acquisition of high quality, detailed images of the collagen and elastin microarchitectures as described13 by taking advantage of the elastin autofluorescence and the collagen second harmonic generation signal (SHG)30. Second, TPEM allows use of low-energy near-infrared excitation light, minimizing photodamage of the tissue and thus, repeated imaging at exactly the same position within the vascular wall is allowed, permitting repeated-measurements analyses of observed changes.
The use of an alternative approach using confocal imaging of pressure fixed arteries is discussed to allow users without access to TPEM an opportunity to use the described method as well. Information on ECM structure and volume densities can also be retrieved from two-dimensional analyses of tissues sectioned in serial, e.g. as described by31,32. However, due to the lack of possibility to retrieve three-dimensional structural information over the length scales of the artery as well as during changing conditions using this method, it is not recommend using this approach for investigations of pressure and treatment induced three-dimensional changes in the ECM.
The minimum requirement for the investigator to apply the herein described method is the access to a setup for cannulation and pressurization of arteries in combination with a confocal or two-photon excitation fluorescence microscope. The setup described in the following protocol is a custom-built pressure myograph with a longitudinal force transducer, built to fit on a custom built inverted two-photon excitation fluorescence microscope.
Collection of biopsies of the human parietal pericardium for use in this work was performed after written informed consent, as previously described33. The study of human tissues conform to the principles outlined in the Declaration of Helsinki34 and was approved by The Regional Committees on Health Research Ethics for Southern Denmark (S-20100044 and S-20140202) and the Danish Data Protection Agency.
1. Collect Tissue and Isolate (Human) Resistance Artery
2. Mount the Isolated Artery in the Pressure Myograph
3. Perform Experiment: Imaging of Arterial Diameter, Wall Thickness and Collagen and Elastin Microarchitecture Using TPEM
4. Calculation of Stress Strain Relationships and Intrinsic Wall Stiffness
5. Image Analysis – (Internal Elastic Lamina) IEL Branching Angles
6. Image Analysis – Adventitial Collagen Waviness
7. Imaging for Collagen and Elastin Volume Densities
8. Image Analysis for Extracting Elastin and Collagen Volume Densities Using Ilastik
The custom-built pressure myograph for imaging used in this work is shown in Figure 1. Special attention for the design of the myograph was paid to i) the chamber with a small volume (2 mL) and ii) the possibility for positioning the cannulae close to, and parallel with the glass bottom (Figure 1B). The bottom of the chamber fits a 50 × 24 mm #1.5 glass coverslip (replaceable). The pressure controller was built from a standard 1 L glass bottle and a sphygmomanometer (cuff removed). For a myograph setup with flow requirements, the use of a servo pressure controller is recommended.
Figure 2 illustrates the recommended positions for obtaining single images and image stacks during the experiment outlined in the protocol. For illustration purposes, a transmission light microscopy image of the mounted artery is included in Figure 2A. Knot-to-knot length of the mounted artery is approximately 1.6 mm. The artery is scanned until the widest diameter is observed (Figure 2B), and at this point, the diameter and wall thickness are determined (Figure 2C). Likewise, a transmission light image of the artery is included in Figure 2D, showing the position and recommended width of the image stacks for determining the microarchitectures of collagen (Figure 2E) and elastin (Figure 2F, small square in Figure 2D) as well as the image stacks for determining the collagen (Figure 2G) and elastin (Figure 2H) volume densities (larger square, Figure 2D). Using this approach, data for constructing pressure diameter curves, followed by the calculation of the corresponding stress-strain curves can be obtained from pressure, diameter and wall thickness measurements (Figure 2C).
A single exponential fit of the stress-strain relationship according to Bloksgaard et al.13 and references herein, provides the β-value, a geometry independent measure of wall stiffness proportional to the incremental elastic modulus, Einc. Calculating Einc at the different pressures at which images for collagen and elastin microarchitectures were obtained, allows a direct analysis of the relationship between the pressure induced changes in the microarchitectures of collagen and elastin and the incremental elastic modulus at different pressures (Figure 3). The measurement of IEL branching angles and collagen straightness is illustrated in Figure 4.
Important for the interpretation of mechanical data comparing two or more groups of interest, e.g. hypertensive vs normotensive patients, is a determination of the content of elastin and collagen. For this, an automatic analysis method in Ilastik, a freeware image analysis software that can be trained to recognize certain image features, was developed. The work-flow in Ilastik is illustrated in Figure 5.
For the automatic detection of elastin and collagen for the volume densities, it is important that there is a good signal-to-noise ratio in the images obtained. In human samples, significant autofluorescence from collagen is often observed in addition to the collagen SHG signal. This decreases the signal-to-noise ratio for the detection of elastin. Figure 6 illustrates how re-training of the Ilastik program can result in better recognition of thin elastin fibers in a sample with significant collagen autofluorescence in the "green"/elastin channel. If bleed-through of the autofluorescence signal into the collagen SHG channel is a problem, obtaining the SHG signal using a longer excitation wavelength may help. If the channel bleed through occurs after staining with eosin, the concentration of the applied solution of eosin should be lowered. Eosin is easily washed out, so repeated washing of the stained sample may decrease the signal too.
Figure 1. Custom built pressure myograph for fluorescence imaging. (A) Perfusion chamber with glass capillaries attached to the micromanipulators, which is mounted on a force transducer (longitudinal force). (B) 2 mL imaging chamber. The 45° bending of the cannulae facilitates imaging using an inverted microscope. (C) Pressure controller. Please click here to view a larger version of this figure.
Figure 2. Combined pressure myography and fluorescence imaging strategy. (A) Transmission light image of the artery with illustration of (B) the scan of the full width of the artery. The collagen SHG is shown in green, and elastin autofluorescence in magenta. (C). Arterial lumen diameter (here 238 µm) and wall thicknesses (here 17 and 18 µm top and bottom, respectively) is determined at the equatorial region (largest diameter observed) of the artery. (D) Transmission light image of the artery with the illustration of the position of the z-stacks obtained at a 10-50 µm scale for determining (E) collagen straightness, (F) branching angles of the internal elastic laminae, and (G) z-stacks obtained at a 50-100 µm scale for obtaining collagen and (H) elastin volume densities. Images B/C: height 300 µm (bar 20 µm), E/F: 50 x 50 µm (bar 10 µm), G/H 130 × 130 × 30 µm (bar = 30 µm). Please click here to view a larger version of this figure.
Figure 3. Microstructural changes in adventitial collagen and the internal elastic lamina (IEL) and their relation to the circumferential incremental elastic moduli at 20, 40 and 100 mmHg. (A) Branching angles of elastin fibers in the IEL. (B) Straightness of adventitial collagen fibers close to the external elastic lamina. p-values and according F-values were determined using repeated measurements one-way ANOVA: p/F-values A: 0.0014/24.18, B: 0.0002/37.38. (C) IEL branching angles correlates nonlinearly with Einc. (D) Collagen straightness correlates linearly with Einc. Data are shown as mean ± SE. Microstructural changes (A and B, and x-axes of C and D) were determined for 6 resistance arteries subjected to vital imaging while Einc (C and D) was calculated for 20 other arteries subjected to pressure myography in a standard pressure myograph. Each artery studied was from a different individual. This figure has been modified from Bloksgaard et al.13. Please click here to view a larger version of this figure.
Figure 4. Illustration of measurements of IEL branching angles and collagen straightness in Fiji. (A) IEL branching angles are manually marked and measured using the Fiji angle tool. (B) Collagen straightness is determined using the Fiji Neuron J plugin as the ratio of the end-to-end length of a single collagen fiber via a straight line (white) divided by the actual end-to-end length of the fiber (orange). Bar size: 20 µm.
Figure 5. Ilastik workflow. (A) Feature selection window with the indication of selected features. (B) Example raw image with (C) labels for elastin fibers, bright spots (undesired) and background highlighted in red, green and blue, respectively. (D) Resulting prediction map for image (B) with labels indicated in (C). (E) Segmentation based on the prediction map in (D) before (E) and after (F) applying a threshold of 0.5. Yellow color shows the connected pixels corresponding to the segregated feature (elastin). Bar size: 25 µm. Please click here to view a larger version of this figure.
Figure 6. Example images before and after the optimization of the Ilastik prediction map showing the effect of re-training Ilastik for recognition of elastin fibers in a sample with high background fluorescence from collagen (sub-optimal signal to noise ratio). (A) Original image, (B) result of first Ilastik analysis, where especially the thin elastin fibers (left side of image B) are not included in the segmentation and (C) final result following the optimization of the prediction map. Bar size: 25 µm. Please click here to view a larger version of this figure.
Supplementary File Please click here to download this file
This work represents our suggestion for a standardized, combined imaging and pressure myography approach, valuable for simultaneous assessment of the mechanical properties of resistance arteries and pressure-related changes in the structure of the arterial wall over a pressure range from 0 to 100 mmHg. The presented approach was developed using custom built equipment, however, any pressure myograph that fits on a two-photon excitation fluorescence microscope can be used, when the design of both equipments allows imaging of the arterial wall. Special attention should be paid to the shape, width and working distance of the objective, and the conditions under which the imaging will be performed. Upright microscopes require the use of water dipping objectives, whereas the use of water immersion objectives is recommended for inverted microscopes to avoid differences in refractive indices between the sample and the objective. Furthermore, attention should be paid to the coverslip thickness, either by applying a correction for a mismatch between the objective and coverslip using the correction collar on the objective (if present) or by matching the coverslip thickness with the objective used.
In the described protocol, advantage is taken of low-energy, near-infrared light for imaging elastin autofluorescence and collagen SHG. This allows imaging for extended periods of time. In the human resistance arteries used in this study, the IEL is comprised of a longitudinally aligned fiber-like network of interconnected elastin fibers13,33, similar to the structure of the IEL of human subcutaneous resistance arteries (hSCA)38,39. Changes in branching angles of the elastin fibers in the IEL were evaluated and used together with measures of the gradual uncoiling of adventitial collagen fibers with increasing pressures37 to assess changes in the microarchitecture of elastin and collagen respectively. Images for assessing elastin and collagen microarchitectures were obtained at the same region of interest within the arterial wall at different pressures. This allows repeated-measurements analyses. When possible, the investigator could aim at obtaining images in at least 3 regions of interest, with image sizes 10-50 µm in each dimension. However, depending on the sample, especially the thickness of the arterial wall, a compromise has to be made between the number of scanned regions of interest and the feasible duration of the experiment. In case of the human pericardial resistance arteries, the arteries remained alive for at least 8-12 h of experimentation. Scanning several regions of interest allows evaluation of intra- versus inter-sample variability and thus supports the conclusion on the obtained findings.
It is important to notice that to avoid inducing phototoxic damage, imaging using high power excitation light should be avoided when working on live tissue specimens. In light of this, it is recommended to obtain the large image stacks for collagen and elastin volume densities at the end of the experimental protocol, alternatively on fixed samples. When absolute quantities of the ECM components are required, any effect of the fixative on the volume densities should be quantified and taken into account. Fixation and permeabilization of the arteries allows furthermore staining of intracellular targets of interest when necessary. In this protocol, staining for elastin is performed using eosin on the live artery. Staining of elastin by specific probes (e.g., alexa 633 hydrazide40) and collagen (e.g., CNA3541), may be applied together with other stains (e.g., phalloidin42) or DNA, to facilitate the assessment of volumetric densities and structural organization of other structures of interest in the arterial wall. This opportunity may provide research laboratories with single-photon confocal microscopes an opportunity to investigate the microstructural changes in resistance arteries. Segments from the same artery may be treated under different conditions, fixed under pressure and imaged at a later stage to compare the effect of the applied treatment. Targets of interest should be visualized by imaging the re-cannulated, re-pressurized and stained vessels. Re-cannulation and re-pressurization is required when structural changes are considered, as elastin cannot be fixed and hence will recoil and alter the 3D structure of the fixed artery43,44. The drawback of determining structural changes in fixed arteries is that several segments have to be fixed for comparison, and repeated-measurement analyses cannot be performed.
The data obtained using imaging can be directly used to calculate wall stresses and strains, and support understanding the mechanics of the vascular wall, as described in Bloksgaard, M. et al13. In this work, a mathematical model was applied to characterize the intrinsic stiffness of the elastin and collagen components of the arterial wall and to estimate the collagen recruitment strains45,46 on the basis of calculated stress strain relationships. Imaging supported the findings from the mathematical modeling, that collagen in human resistance arteries was already recruited at low circumferential strain and wall stress. The quantitative measures of the microarchitecture of elastin and collagen described in this protocol can be further extended, inspired by the extensive work conducted on carotid arteries6,8,47,48,49 and aorta7,9,10,50,51: additional analyses may include assessing the spatial distribution and microarchitecture of collagen and elastin fibers within the different layers of the arterial wall as well as fiber orientation angles (orientation with respect to the longitudinal axis). Bell et al.15 addressed the pressure induced reorganization of the IEL and adventitial collagen including a multi-layer analytical model to calculate the stiffness and stress in each layer of the arterial wall. These authors15 used human subcutaneous resistance arteries obtained from healthy volunteers, and related structure and wall stresses at 3 and 30 mmHg, respectively.
The proposed method in this protocol hopefully motivates further investigations and development of microstructurally motivated mathematical models for understanding the mechanical properties of human resistance arteries in health and disease. With further amendments of the method to include also collagen and elastin orientation angles along the longitudinal axis, smooth muscle cell volume density, spatial distribution and orientation within the arterial wall, collagen and elastin distribution and orientation in tunica media, and elastin organization and distribution in tunica adventitia, the imaging data can feed forward development of mathematical models, facilitating understanding of the remodeling process in hypertension, diabetes and the metabolic syndrome. Similarly, as suggested by Chen and Kassab for the coronary microcirculation52,53, microstructure-based models of human resistance arteries can provide predictions of arterial mechanical responses at the macro- (whole artery) and micro-(structure) mechanical levels for each constituent. Such models must be based on quantitative data of structural parameters and mechanical properties of individual layers and constituents in the arterial wall of arteries originating from both healthy volunteers and patients suffering from different diseases, receiving different pharmacological treatments. Data are collected in a standardized manner and compared across individuals with different risk factor, disease and treatment profiles. The method has the potential to aid understanding how disease and medical treatments affect the human microcirculation.
The authors have nothing to disclose.
The authors thank the Danish Molecular Biomedical Imaging Center at the Faculty of Natural Sciences, University of Southern Denmark, for the use of laboratories and microscopes. Kristoffer Rosenstand and Ulla Melchior are acknowledged for excellent technical assistance with the pressure myography and imaging.
Fine Science Tools | 15401-12 | |
Fine Science Tools | 11251-23 | |
Nikon | SMZ800N | |
Sigma-Aldrich, Brøndby, Denmark. | 761028 | for dissection purpose |
Vitrex Medical A/S, Herlev, Denmark | 1.63, 2.13, 210mm | |
Smiths medical Intl, UK | ||
Ethicon | Ethilon 11-0 | |
Custom built | DK patent number 201200167, University of Southern Denmark, J. Schoubo V. Jensen, F. Jensen. T.R. Uhrenholt | |
Mettler toledo | ||
Sigma-Aldrich, Brøndby, Denmark. | B3259 | |
Sigma-Aldrich, Brøndby, Denmark. | A7030 | |
Sigma-Aldrich, Brøndby, Denmark. | C5670 | |
Sigma-Aldrich, Brøndby, Denmark. | G7021 | |
Sigma-Aldrich, Brøndby, Denmark. | E3889 | |
Merck Millipore, Hellerup, Denmark | 1.00496.9010 | Phosphate buffered (pH 6.9) 4% formaldehyde solution |
Sigma-Aldrich, Brøndby, Denmark. | H3784 | |
Sigma-Aldrich, Brøndby, Denmark. | P9666 | |
Sigma-Aldrich, Brøndby, Denmark. | P5655 | |
Sigma-Aldrich, Brøndby, Denmark. | M2643 | |
Sigma-Aldrich, Brøndby, Denmark. | S2002 | |
Sigma-Aldrich, Brøndby, Denmark. | S5886 | |
Sigma-Aldrich, Brøndby, Denmark. | S5761 | |
Sigma-Aldrich, Brøndby, Denmark. | 1.06462 | |
Gibco, ThermoFisher Scientific | 10010015 | |
Sigma-Aldrich, Brøndby, Denmark. | PHR1423 | |
Sigma-Aldrich, Brøndby, Denmark. | Z370525 | |
Tocris Bioscience, Bristol, UK | 538944 | |
Nikon | Custom built | |
Spectra Physics, Mountain View, CA | ||
Nikon | CFI Plan Apo IR SR 60XWI NA 1.27 | |
Nikon | CFI Plan Fluor 20XMI (multi-immersion) NA 0.75 | |
Hamamatsu, Ballerup, Denmark | H7422P-40 | |
AHF analysentechnik AG (Tübingen, Germany). | ChromaET 460 nm long pass dichroic | |
AHF analysentechnik AG (Tübingen, Germany). | Semrock FF01-520/35-25 BrightLine filter | |
AHF analysentechnik AG (Tübingen, Germany). | Chroma ET402/15X | |
Scotch TM | ||
coverslip thickness should match used objective on microscope (#1 or #1.5), alternatively, set adjustment collar to match coverslip |