Perfusion Staining Methods for Visualization of Intact Microvascular Networks in Whole Mount Skeletal Muscle Preparations

Capillaries are a critical component of the skeletal muscle vascular network, vital for oxygen delivery and nutrient exchange to cells of tissues within the body and, more recently, being proposed as instrumental in the coordination of the blood flow response during skeletal muscle contraction [1]. This role proposes that capillaries can sense changes in the cellular environment of metabolically active skeletal muscle fibres and can signal distant levels of the microvasculature in order to coordinate enhanced perfusion of capillaries feeding the metabolically active cell. This coordination requires specific architectural relationships between capillaries and other microvascular elements. A better understanding of how capillaries achieve this intervascular coordination would be facilitated by understanding capillary network architecture and their structural connectedness within the microvascular network (Fig. 1a).

To visualize connected microvascular networks requires viewing whole, intact tissues. Further, the tortuosity of capillary networks in skeletal muscle will depend on the contractile state of the skeletal muscle and the length of the muscle fibres [3‒5]. Staining techniques that cause muscle contraction, such as fixation or freezing, must be avoided. We surmised that intravascular perfusion of dye followed by microscopy of excised, fresh, whole-mount tissues with skeletal muscle fibres set to near optimal length for force production (L_o) will allow us to visualize the intact microvascular network close to its in vivo state, but the methods and tools for this are not clear.

Different techniques and stains have been used to label the vascular endothelium in order to visualize capillaries (for review see [6]). Lectins have been successfully used to label vascular endothelial cells (ECs) of tissues given their binding affinity to the carbohydrate components of the glycocalyx [7] on ECs and the ability to visualize the bound lectin using fluorescent tags or biotin conjugates. Intravascular injections of lectins in vivo, followed by a short period of in vivo circulation has been shown to produce bright staining within the smallest elements of the vasculature [8‒11]. Among the lectins that have been used in skeletal muscle to observe the vasculature, isolectin GS-IB4 (ISO), wheat germ agglutinin (WGA) and lycopersicon (LYCO) have proven to be most effective [11‒14] although the labeling of the capillaries [15] and other levels of the microvasculature is not always consistent [16, 17]. The variable labelling of the vasculature may be attributable to differences in the lectins binding affinity and techniques used to label the vasculature. The lectins optimal for vascular labeling in skeletal muscle using intravascular perfusion have not been established.

Therefore, we sought to develop a methodology for intravenous injection and perfusion labeling of the capillary network architecture and the connected arteriolar and venular networks in fresh, whole mounted skeletal muscle preparations where skeletal muscle length was set to L_o. We used fluorescently labeled lectins: ISO, WGA and lLYCO, as well as a FITC-dextran labeled gel (GEL), in order to find the effective labelling conditions for visualization of intact skeletal muscle microvasculature networks, including the capillaries.

A total of 20 male CD-1 mice (age 8–12 weeks; Charles River Ltd, QC, Canada) were used in this study. All mice were housed socially on wood-chip bedding, had access to food and water ad libitum on a 12:12 hour light dark cycle. Following all experimental protocols, animals were euthanized with an overdose of sodium pentobarbital (0.26 mg/mL iv to effect).

Three fluorescently labelled lectins were used to visualize the vasculature: ISO GS-IB4 from Griffonia Simplicifolia, Alexa Fluor 568 conjugate (ISO; λ Ex 579 nm; λ Em 603 nm) (ThermoFisher Scientific, USA), Wheat Germ Agglutinin, Alexa Fluor 488 (WGA; λ Ex 495 nm; λ Em 519 nm) (ThermoFisher Scientific, USA), and Lycopersion Esculentum (Tomato) Lectin, DyLight 488 (LYCO; λ Ex 493 nm; λ Em 518 nm) (ThermoFisher Scientific, USA). ISO was diluted to 166.7 μg/mL in 0.9% saline, WGA and LYCO were diluted to 1 mg/mL in 0.9% saline. The solutions were divided into 200 μL aliquots and frozen until the time of use. The efficacy of a FITC-dextran labelled gel (GEL) was also tested, a 2% gelatin (Knox, Lipton, Canada) solution was mixed in warmed 0.9% saline solution and 1% FITC-dextran (Sigma, USA) [18], and continuously stirred on a heated magnetic stirrer so that the solution remained liquid and homogenous. Once mixed, the gel solution was kept in a heated water bath to maintain liquidity until the time of infusion.

Mice were anesthetized using sodium pentobarbital administered via intraperitoneal injection (75 mg/kg). Depth of anaesthesia was assessed by monitoring the absence of a withdrawal reflex in response to a toe pinch. Once in the surgical plane mice underwent a tracheotomy and the insertion of a breathing tube. A catheter was placed in the right jugular vein for further anesthetic administration and fluorescent lectin infusion. For ISO/WGA/LYCO experiments either 200 μL ISO alone or 200 μL ISO + 200 μL WGA or 200 μL ISO + 200 μL LYCO were administered to the mouse through the jugular vein catheter. The dyes were allowed to circulate in vivo in the bloodstream for approximately 15 min before the animal was euthanized and skeletal muscles dissected out. For GEL experiments, 200 μL ISO was administered to the mouse through the jugular vein catheter and allowed to circulate for 15 min. Mice were administered 2,000 U/L heparin in 0.9% saline which was allowed to circulate for 10 min. Mice were then euthanized, the abdominal aorta was canulated and the abdominal vena cava was cut. 60 mL of the FITC-labelled gel was infused into the animal until the outflow of the abdominal vena cava ran clear. The animal was then submerged in ice to set the gel.

In the ISO/WGA/LYCO experiments five different skeletal muscles were dissected out for inspection of the microvascular networks: cremaster (CRE), gluteus maximus (GM), diaphragm (DIA), extensor digitorum longus (EDL), and soleus (SOL). The DIA was not dissected out in the GEL experiments as the DIA vasculature would not have been perfused with gel given that the abdominal aorta was the point of infusion and the abdominal vena cava was cut. Excised muscles were pinned out in a silicone-lined petri dish and submerged in room temperature physiological salt solution (PSS) containing (in mmol/L) 131.9 NaCl, 4.7 KCl, 2.0 CaCl₂, 1.2 MgSO₄, 30 NaHCO₃ and 0.3 mg/L tubocurarine hydrochloride pentahydrate aerated with 5% CO₂ and 95% N₂ gas to attain a physiological pH (7.35–7.45). Each muscle was pinned out to set the skeletal muscle at its approximate optimal length, confirmed by determining sarcomere length of muscles under ×80 magnification. Sarcomere length was set to between 2.3 and 2.8 μm (Fig. 1b).

The fluorescent labelling of the vasculature was detected using an upright Olympus BX51WI microscope (Olympus Canada Inc., Richmond Hill, ON, Canada) using 4× (na 0.13), 10× (na 0.3) or 20× (na 0.4) objectives. Vessels were visualized using a 100 W mercury arc lamp and 480 ± 20 nm narrow band pass filters for fluorescent excitation of FITC-dextran labelled gel, WGA and LYCO. Emissions were collected using a 500 ± 20 nm long pass filter (Chroma, Brattleboro, VT). Excitation of ISO was achieved with a 560 ± 20 nm narrow band pass filter and emissions were collected using a 580 ± 20 nm long pass filter (Chroma, Brattleboro, VT, USA). Monochromatic images were collected using a cooled CCD camera (CoolSNAP HQ, Roper Scientific, now Teledyne Princeton Instruments, NJ, USA) and RSS Image software on a PC running a Windows 7 operating system. All images were taken under the same collection conditions, i.e., 500 ms capture duration, same mercury arc lamp, etc. The CoolSNAP HQ camera is a high resolution camera that autogains (i.e., the camera automatically adjusts brightness) therefore quantitative comparisons of brightness required background subtraction. Images were pseudo-coloured and merged using FIJI Image (https://imagej.net/software/fiji/) and, other than pseudo-colouring, images analyzed and presented were not altered (i.e., brightness and contrast were not altered).

20 animals were used for these experiments, and some animals were double stained: 4 mice were infused with GEL, 4 with LYCO, 15 with ISO and 8 with WGA. For each experiment 2 SOL, 2 EDL, 1 GM, 1 CRE and the whole diaphragm was removed and multiple areas of each muscle were examined with the exception of the GEL experiments where only CRE and GM were examined.

To compare the dyes quantitatively, we measured the brightness within a region of interest of each label for arterioles, capillaries and venules from each image studied and subtracted the background from each brightness measure to account for camera autogain functions. Data are reported as mean ± SE. Data were compared using a two-way ANOVA. When the ANOVA identified significant differences, a protected least square difference test was used post hoc to determine differences between vascular levels using a given dye and between dyes at a given vascular level. Differences were considered significant when p < 0.05.

WGA labelled each level of the microvasculature clearly and brightly. Figure 2a shows the cremaster microvasculature where arterioles and venules are clearly visible as well as capillaries along their lengths. WGA labelling of small arterioles, capillaries along their length and venules was clear in all muscles excised (Fig. 2b–e).

In ISO-perfused preparations the entire microvascular network was not visible (Fig. 3, 4). ISO clearly labelled all levels of the arteriolar side of the microvasculature and early capillaries that stemmed from the last ramification of the arteriolar tree, the terminal arterioles, but brightness quickly faded across the capillary towards the venous vasculature which was not visible under fluorescence. Figure 3 shows part of the CRE network where an arteriole and venule are clearly in view when viewing the section using brightfield illumination (Fig. 3b), but under fluorescence only the arteriole is visible (Fig. 3a). Figure 3c, d show that in cremaster, small arterioles and early capillaries are visible using ISO but not the full length of the capillary, nor smaller venules. This phenomenon is further exemplified in Figure 4, in GM, DIA, EDL and CRE, where animals were double-labeled with ISO and WGA. Figures 4a, c, e, g show ISO staining alone where Figures 4b, d, f, h show both ISO and WGA in the same tissue section where the capillary and venous sections of the microvasculature unstained by ISO can be seen using WGA.

Lower magnification images of cremaster (Fig. 5a, c) show that LYCO labelled every level of the microvascular network. Figure 5b, d show the same sections double-labeled with LYCO and ISO to help discern the arteriolar from the venous vasculature. LYCO labelling capillaries along their length is evident at higher magnifications and in different muscles (Fig. 5e–l) Again, double-labelling LYCO-labelled networks with ISO allowed for the distinction between the arteriolar and the venular side of the network.

GEL perfusion made arterioles, capillaries and venules visible but not all branches of each level were reliably stained. Large arterioles and venules were brightly labelled (Fig. 6a) but the smaller vascular elements and the capillaries were not as consistently evident. This can be seen clearly when comparing double-labeled ISO and GEL sections of the same network where arteriolar elements labeled by ISO are not labeled with the GEL (Fig 6).

Assessment of the pattern of the brightness of the dye at different levels of the microvasculature did not differ between muscles so data for each dye from the muscles were pooled. The data used for the quantitative analysis for WGA included 8 mice using 64 images, ISO: 15 mice, 93 images, LYCO: 4 mice, 44 images, and GEL: 4 mice, 30 images. In support of qualitative observations, quantitatively, we found that WGA brightness, although significantly dropping from arterioles to capillaries, was the most consistent across the different levels of the vasculature (Fig. 7a). ISO dropped significantly from arterioles to capillaries and was not detectable in veins (Fig. 7a). LYCO stained arterioles and veins similarly to WGA, but WGA labeling of capillaries was significantly brighter (Fig. 7a). GEL labeled arterioles and veins the brightest but WGA staining of capillaries was significantly brighter than GEL. Further, WGA stained capillaries significantly brighter than ISO, LYCO and GEL (Fig. 7b).

Interestingly, the appearance of the vessel walls differed dramatically depending on the label used (Fig. 8). This was most evident in the vasculature of cremaster muscle given the thinness of the tissue and the ability to focus clearly on larger vessels. ISO clearly labeled the EC periphery making the morphology of the ECs apparent (Fig. 8a). GEL perfusion labels the lumen of the vessel brightly but the morphology of the ECs is not distinguishable and the vessel wall is harder to focus on directly (Fig. 8a). Both WGA (Fig. 8b) and LYCO (Fig. 8c) labelled ECs in a more punctate manner which did not allow for the distinction of EC morphology as with ISO. The patchy appearance of WGA and LYCO staining may also be due to the staining of cells within the blood as movement of stained elements could often be visualized in the lumen of vessels.

We sought to develop a methodology to label intact microvascular networks via intravascular perfusion in fresh, whole-mount skeletal muscle preparations such that the skeletal muscle fibres could be set to near L_o. We infused fluorescently-labeled lectins and a fluorescently-labelled gel to determine which would allow for visualization of all levels of the microvascular network, including capillaries. We found that WGA and LYCO labeled the entire microvascular network but WGA more brightly labelled the capillaries. FITC-labelled GEL was excellent at making the larger microvasculature visible but smaller vascular elements, such as small arterioles and capillaries, were not always apparent. Finally, ISO labelled arterioles and early capillaries stemming from the terminal arterioles but did not label the entire length of the capillaries nor the venous side of the microvasculature. These labelling patterns were consistent across all skeletal muscles tested. Therefore, WGA was the most effective for visualizing the entire connected microvascular network structure through perfusion, in fresh, whole-mount skeletal muscle preparations in a variety of muscle fibre types. Additionally, double labelling muscles with WGA and ISO allowed for the entire network to be visualized and differentiation of the arteriolar side of the network from the venous side of the network, making capillary modules more distinct.

We chose perfusion to maximize the delivery of the dyes to all levels of the vasculature and used fresh, wet mounts to reduce tissue disruption through tissue processing techniques that can be associated with fluorescent labelling, i.e., freezing and fixing tissue. Other groups, using dye perfusion in skeletal muscle, followed by fixation/freezing and sectioning steps have also found LYCO [11, 14] effective at staining different levels of the vasculature. Further, studies using fixed/frozen tissue, sectioned and incubated with dye from many different species found WGA [12, 15, 19, 20] and LYCO [13] effective at labeling small vessels including capillaries in skeletal muscle with only one report of WGA not labelling capillaries in pig semitendinosis [15]. Therefore, WGA and LYCO appear to be effective for visualizing the vasculature in a variety of methodological protocols and tissue processing techniques. Our observations expand on this and add that when using a perfusion technique and fresh tissue, both LYCO and WGA were effective at labeling the entire vascular network, but WGA stained capillaries brighter than LYCO.

Conversely, the effectiveness of ISO at staining the microvasculature varies dramatically with the methodology. Studies using fixed/frozen tissue, sectioned and incubated with dye found that ISO stained all capillaries strongly [13, 15‒17, 20, 21] with reports that vessels larger than 20 μm were not well stained [16, 17]. Immersion of whole, fresh tissue in ISO showed that capillaries stained brightly as well as staining of arterioles and venules [16]. In contrast, our methodology, using perfusion of ISO, yielded brightly stained arterioles regardless of size and no venous labelling. Further, capillary staining was limited to early segments stemming from the arterioles.

The differences in ISO staining effectiveness as a result of methodology may be due to the side of the EC, luminal or abluminal, that is exposed to the dye. Intravascular perfusion of ISO exposes the luminal surface of the EC [11] therefore we expect that the ISO binding sites on the EC glycocalyx are limited to the luminal side of arterioles and early capillaries. Studies using fixed/frozen tissue sections incubated with dye [13, 15‒17, 20, 21], or whole muscle preparations incubated with ISO [16, 17] will expose both the luminal and abluminal side of the blood vessels to ISO and these methodologies stain the entire vascular network, indicating that the EC has binding sites for ISO on the abluminal side of the ECs throughout the vascular network or binding sites on the basement membrane [17].

Infusing dye/gel into the lumen of the vasculature could be considered more of a gold standard for identifying microvascular structures as it does not rely on binding to cellular structural elements but simply fills the lumen of each vessel. This technique has been successfully used to map the vasculature of skeletal muscle by infusing a carbon ink followed by fixation [22, 23], but this technique does not allow for observations to be made at optimal skeletal muscle length. Injection of FITC-dextran into an anaesthetized animal allows for exteriorized skeletal muscle preparations to be observed using intravital microscopy (for example see [24, 25]) but limits the vascular beds observed to muscles that can be exteriorized in the live animal. Adding FITC-labeled gel allows for any muscle to be removed and observed, post-labelling. Further, this technique allows for fresh skeletal muscle tissue to be studied at optimal length once the muscle is brought back up to temperature following the cooling phase to set the gel. The important component with this technique is getting a gel consistency liquid enough to get into capillaries but viscous enough to stay in the vessels once the tissues are removed. We used a 2% gel mixture [18] with a cooling step added to the protocol to set the gel and were able to identify larger vessels but were not confident that the smallest components of the vasculature, were fully and consistently perfused. Atrill et al. [26] have used a 1.25% gel mixture to visualize the vasculature with similar results. A lower concentration of the gel in the perfusate may have allowed for more consistent labeling of smaller vessels. Further, pre-infusion of the animal with a vasodilator may aid in more complete vascular perfusion. The gel perfusion does have the added advantage of being able to control the brightness through the addition of more or less FITC to the perfusate. Therefore, the brightness of the perfusion solution can be adapted for different tissue densities, imaging conditions and equipment available.

The staining patterns were similar across all muscle types tested. We used five muscles of differing fibre types: mouse SOL being approx. 70% type I, EDL 50% type IIb and 45% type IIa/x and DIA approx. 85% type IIa/x (CD-1 mice; [27]), GM predominantly expresses type IIb fibres (C57/BL/6J mice; [28]) and cremaster is comprised of between 60 and 80% type IIb fibres (hamster and rat [29]). The ubiquitious staining pattern of WGA, LYCO and ISO between muscles of differing fibre types and function indicates that they are consistent markers of the vasculature in skeletal muscle and may be useful across a wide variety of muscle vascular beds in mice.

The consistency of staining may hold across skeletal muscles in one species but may vary across species. Alroy et al. [12] tested a wide variety of tissues, including skeletal muscle, from a wide variety of species (cat, cow, dog, goat, horse, mouse, pig, rat, sheep and human) using a wide variety of lectins in fixed, sectioned, incubated tissue samples and found species differences in staining. For example they observed that ulex europaeus agglutinin I (UEA or UEA-I) only stained human vascular ECs but, interestingly, they found that WGA was effective across all tissues and species. A similar cross species study was performed by Hansen-Smith et al. [16] using different tissue incubation techniques (frozen sections or whole tissue incubation) in muscles of mice, rats, hamsters, rabbits, dogs and monkeys using ISO and found that ISO consistently stained the microvascular beds of all tissues. Therefore, there is a ubiquitous nature of binding of WGA across species, and with the addition of our observations using perfusion, a ubiquitous nature of WGA binding across different methodologies. There is also a consistency of ISO binding across species, but not across methodologies, given that we show perfusion staining results in a different staining pattern than fixed, sectioned tissue.

While perfused WGA was optimal for labelling the entire vasculature, when the entire network is labelled, the density of the network of capillaries (like those found in GM and DIA) makes it complicated to orient the arteriolar and venous sides of the network or to track capillary modules from the arterial to the venous side. We found that the characteristic partial labelling of the network by ISO can be used as a tool to help orient these dense networks. Double-labeling the network with WGA and ISO allowed for visualization of the entire vascular network while differentiating the arteriolar side of the network from the venular side of the network making the capillary modules much easier to identify and visualise the connectivity of the network elements.

Interestingly, we found that the clarity and the pattern of staining of the wall of the microvessels differed depending on which labelling procedure was used. ISO appeared to label the periphery of the ECs, allowing for the visualization of individual ECs themselves while WGA and LYCO showed a more punctate staining pattern. This may be because of the more diffuse location of the binding site for WGA and LYCO on the EC membrane and possibly due to fluorescent interference with staining other blood elements in the vessel lumen. LYCO has been reported to bind to leukocytes [30‒32]. Lectin staining patterns were in contrast to the GEL labelling where ECs were not visible. Therefore, there may be advantages to using one label over another depending on exactly what components of the vasculature need to be visualized.

Whole mounts of skeletal muscles were used to visualize intact, connected microvascular networks in skeletal muscle. The muscles chosen for study were chosen for two distinct reasons: (1) muscles were comprised of different fibre-type compositions, different microvascular structures and different vascular densities (DIA, SOL and EDL) and (2) muscles were chosen to optimize network visibility, with flat, sheet-like muscle fibre orientation (CRE, GM and DIA). In each of these muscles we were able to visualize multiple levels of arterioles, capillaries and multiple venular levels with different degrees of success. Each element of the microvascular network was discernible in CRE and GM including larger 1A paired feed arterioles and venules. Much of the connected network was visible in DIA, although the larger arterioles and venules deeper to the tissue were difficult to discern, and their connectivity to the smaller elements of the microvasculature were not visible. We were most limited in visualising the connected network in SOL and EDL due to the thickness and the curvature of these muscles. The visible parts of the network in these muscles were limited to 3A and 4A arterioles, capillaries and 3V, and possibly 2V venules. Regardless of these limitations we were able to determine that the key differences in the lectin labelling patterns were consistent between all muscles tested.

We have developed a methodology and a protocol that allows for the visualization of the intact microvascular network in fresh tissue and a wide variety of skeletal muscles where muscle fibre length can be set to L_o. WGA was found to effectively label the entire vascular network and was able to label capillaries the brightest. And although ISO labeling of the vasculature was limited to the arteriolar side of the network and early capillaries, when used in conjunction with WGA, it allows for the orientation of complicated, dense networks from the arteriolar to the venular sides of the network, allowing for clear orientation of capillary modules. Further, this technique will allow for intact vascular networks to be studied in a wide array of skeletal muscles and set to optimal length, therefore visualizing the network of different muscles at in a more physiologically relevant state.

All experimental procedures were reviewed and approved by the Animal Care Committee at the University of Guelph, Approval No. AUP4562. All protocols and procedures are in accordance with the guidelines set out by the Canadian Council on Animal Care.

No conflicts of interest, financial or otherwise, are declared by the authors.

This work is funded by Natural Sciences and Engineering Research Council of Canada grant RGPIN-2019-05146.

Barbara M. Hyde-Lay, Mackenzie E. Charter and Coral L. Murrant were each involved in every aspect of this work including conceiving and designing the research, performing experiments, analyzing data interpreting results of experiments, preparing figures, drafting, editing and revising the final version of the manuscript.

Barbara M. Hyde-Lay and Mackenzie E. Charter shared first authorship.

The data that support the findings of this study are not publicly available due to the lack of an openly available public repository for these types of data. Data will be made available from the corresponding author Coral L. Murrant upon reasonable request.