Edward F. Adolph Distinguished Lecture. Contemporary model of muscle microcirculation: gateway to function and dysfunction

Abstract

This review strikes at the very heart of how the microcirculation functions to facilitate blood-tissue oxygen, substrate, and metabolite fluxes in skeletal muscle. Contemporary evidence, marshalled from animals and humans using the latest techniques, challenges iconic perspectives that have changed little over the past century. Those perspectives include the following: the presence of contractile or collapsible capillaries in muscle, unitary control by precapillary sphincters, capillary recruitment at the onset of contractions, and the notion of capillary-to-mitochondrial diffusion distances as limiting O₂ delivery. Today a wealth of physiological, morphological, and intravital microscopy evidence presents a completely different picture of microcirculatory control. Specifically, capillary red blood cell (RBC) and plasma flux is controlled primarily at the arteriolar level with most capillaries, in healthy muscle, supporting at least some flow at rest. In healthy skeletal muscle, this permits substrate access (whether carried in RBCs or plasma) to a prodigious total capillary surface area. Pathologies such as heart failure or diabetes decrease access to that exchange surface by reducing the proportion of flowing capillaries at rest and during exercise. Capillary morphology and function vary disparately among tissues. The contemporary model of capillary function explains how, following the onset of exercise, muscle O₂ uptake kinetics can be extremely fast in health but slowed in heart failure and diabetes impairing contractile function and exercise tolerance. It is argued that adoption of this model is fundamental for understanding microvascular function and dysfunction and, as such, to the design and evaluation of effective therapeutic strategies to improve exercise tolerance and decrease morbidity and mortality in disease.

INTRODUCTION

The extraordinary scientist and educator Edward F. Adolph was a pioneer of physiological regulation who won the Presidential Certificate of Merit and served as American Physiological Society’s President from 1953 to 1954. His comparative physiological approach and discoveries revolutionized our understanding of adaptive mechanisms to extreme desert environments. Recognizing the impediment of dogma to scientific progress Adolph’s guiding scientific principle was: “Let not wisdom scoff at strange notions or isolated facts. Let them be explored. For the strange notion is a new vision and the isolated fact a new clay, possible foundations of tomorrow’s science” (59a).

Germane to Adolph’s principle this review utilizes state-of-the-art empirical evidence and logic to construct a contemporary model of muscle microcirculatory function and blood-mitochondrial O₂ transport (Fig. 1). This model is data driven using a range of approaches from anesthetized/conscious animals to exercising humans and modern technological advances to challenge entrenched dogmas some of which are nearing their 100th anniversary. The overarching premise is that mechanisms of dysfunction cannot be recognized (and combatted) effectively until we understand the bases for physiological function in health. It is now evident that major endemic diseases, including heart failure (HF; Refs. 90, 171), diabetes (type I and type II; Refs. 116, 157, 158, 163, 209), and sepsis (39, 133) incur physical disability via microcirculatory dysfunction thereby accelerating morbidity and mortality. An accurate understanding of microcirculatory (dys)function is crucial to designing more effective therapeutics to improve muscle function and patient quality of life.

Fig. 1.

Summary of mechanisms that account for the increase of blood-myocyte O₂ (and substrate) flux, from rest-exercise, within existent red blood cell (RBC) flowing capillaries without the necessity for “recruitment” of previously nonflowing capillaries. A: substantial elevation of blood flow (RBC flux) within originally very low flow capillaries means that capillaries that may have been inconsequential for blood-myocyte flux at rest become important during exercise. B: close proximity of microvessels within 3-dimensional space allows for O₂ flux among vessels, which theoretically may help mitigate the effects of Q̇o₂/V̇o₂ mismatch (illustration by Mal Rooks Hoover, CMI). C: elevation of RBC numbers (hematocrit) within capillaries increases perfusive and diffusive O₂ conductances. Note that this effect can quantitatively account for all of the increased total hemoglobin + myoglobin concentration in the NIRS signal for rest-exercise (141; see Section what evidence is there that these capillary hemodynamics and muscle deoxygenation profiles faithfully represent the physiological response(s) of human muscles during voluntary rhythmic exercise? below). D: increased RBC velocity and greater fractional O₂ extraction mean that additional capillary surface area along the length of the capillary becomes important for blood-myocyte flux (i.e., “capillary longitudinal recruitment”). E: there is the possibility that the endothelial cell surface layer (thought to be crucial in lowering capillary hematocrit below systemic) is modified under high flow conditions. The impact of this effect on blood-myocyte O₂ and substrate flux awaits resolution. F: from rest-exercise, the precipitous fall in intracellular Po₂ at least maintains, and may even increase, the blood-interstitium-myocyte Po₂ gradient and also desaturates myoglobin removing the “functionally depleted O₂ carrier region” (94) elevating its ability to transport O₂, which improves intracellular O₂ diffusing capacity (bar = 1 µm; Ref. 89). Please see text for additional details.

The adult human skeletal muscle capillary bed in toto contains an estimated 9+ billion capillaries with a total length in excess of 8 km and a surface area perhaps greater than the most generous estimates of that in the lung (180 m²). We will consider how access to that vast exchange surface is controlled and inextricably entwined with muscle metabolic function and dysfunction.

A CONTEMPORARY MODEL OF MUSCLE CAPILLARY FUNCTION

The thrust of this paper argues for a revised capillary function model that is based on the compelling weight of direct empirical observations and echoes the prescient words of physics Nobel laureate Richard P. Feynman (56a): “It doesn't matter how beautiful your theory is, it doesn't matter how smart you are. If it doesn't agree with experiment, it's wrong.”

However, if entrenched capillary recruitment theory [see Fig. 2, i.e., ~10% capillaries supporting red blood cell (RBC) flow at rest increasing to >80% with contractions; Refs. 127, 128, 222] cannot be invoked to explain increased blood-myocyte O₂ and substrate flux during contractions, what can account for the up to 100-fold increase in muscle O₂ flux across the physiological range? At least six phenomena (5 supported by empirical evidence and 1 speculative, i.e., endothelial surface layer modifications) must be considered here (see Fig. 1).

Fig. 2.

From 1920 to the present day, scientific discoveries have revolutionized our understanding of skeletal muscle capillary structure and function. RBC, red blood cell. [Reproduced from Poole et al. (168) with permission.]

Most Capillaries Support (at Least) Some RBC and Plasma Flux in Resting Muscle, and It Is Increased RBC Velocity and Flux in These Capillaries That Occurs from Rest to Exercise

In muscle at rest there is a huge variability in RBC flux, velocity, and hematocrit among capillaries. Accordingly, many capillaries sustain such low RBC flux and velocity at rest that they contribute relatively little to blood-myocyte O₂ flux (19, 29, 33, 36, 39, 43, 101, 108, 111, 112, 114, 115, 156, 157, 176, 187, 195, 201, 206, 207, 218). However, upon initiation of muscle contractions, these capillaries evidence increased RBC flux and become important contributors to muscle oxygen uptake (V̇o₂; Refs. 80, 115, 168). Whereas de novo capillary recruitment has been suggested as a mechanism to reduce the decrease in capillary RBC transit times during exercise, sustaining flow in the majority of capillaries at rest and during exercise in and of itself maximizes capillary RBC (and plasma) transit time. Specifically, during exercise, mean capillary RBC transit time depends on the ratio between the volume of flowing capillaries and blood flow (Q̇) and will thus fall to exactly the same point irrespective of the resting conditions. It is also pertinent here that RBC velocity and capillary transit time are inherently dependent on Q̇, and it is the ratio of diffusing capacity (Do₂) to Q̇ that is crucial for achieving a given arterial-venous O₂ difference (a-vO₂). This relationship is presented in Eq. 1 from Wagner and colleagues (196):

$${\text{a-vO}}_{\text{2}}=\dot{\text{V}}{\text{o}}_{\text{2}}/\dot{\text{Q}}{\text{o}}_{\text{2}}\text{=}\dot{\text{Q}}{\text{o}}_{\text{2}}{(1-e}^{-{\text{Do}}_{\text{2}}/\beta \dot{\text{Q}}})$$ (1)

where Q̇o₂ denotes O₂ delivery (i.e., Q̇ × arterial [O₂]) and β represents the slope of the O₂ dissociation curve in the applicable range. Interestingly, even some of the highest muscle Q̇ values measured in humans (~4 l·kg⁻¹·min⁻¹; Ref. 190) that reduce estimated mean capillary RBC transit time far below 0.3 s (193) still permit very high fractional O₂ extractions (i.e., >0.8).

Microvessel Three-Dimensional Geometry

Microvessel three-dimensional geometry allows intervessel O₂ diffusion (40, 68, 142, 144, 162) and enhanced blood-myocyte O₂ transport. Interstitial O₂ partial pressures (Po₂s) are maintained considerably above intramyocyte Po₂ increasing the myocyte diffusional surface area and reducing resistance to trans-sarcolemmal O₂ flux at rest and during contractions (88, 89).

Capillary Hematocrit Increases During Muscle Contractions

Mean capillary hematocrit in resting muscle is extremely low (~15%) and rises toward systemic (~45%) with exercise hyperemia (36, 115, 207; reviews in Refs, 168, 169). One remarkable consequence of this behavior is that even if there were no de novo increases in RBC flux, muscle concentration ([Hb]) and hence Do₂ could potentially increase up to threefold.

Longitudinal Recruitment of Capillary Endothelial Surface Area

Whereas RBCs and plasma transit the capillary length in resting muscle, the relatively high intramyocyte Po₂ (~15–30 mmHg depending on fiber type and Q̇o₂/V̇o₂ ratio) and low fractional O₂ extraction (~0.25–0.50) likely limit the proportion of the capillary length that is important for O₂ flux. The contraction-induced elevation of RBC velocity combined with the steep fall of intramyocyte Po₂ will increase the endothelial surface length and area “recruited” as fractional O₂ extraction rises to 0.8–0.9 (88, 168, 169).

Capillary Endothelial Surface Layer

Often termed the glycocalyx (“sweet husk” in Latin) this plasma restriction/exclusion zone, in combination with differential RBC distribution at arteriolar bifurcations (41), helps lower the capillary hematocrit by establishing differential speeds for the mean plasma (slower) and RBCs (faster) (37). Endothelial layer effects, perhaps via compression when RBC velocity is high, allow capillary hematocrit to increase during hyperemia in health. In disease, degradation of this layer may substantially impair capillary perfusion (139, 215, 240).

Increased Capillary-to-Myocyte Po₂ Gradient

The precipitous fall in intracellular Po₂ (94, 190, 191) combined with a limited decrease of mean capillary (microvascular Po₂, determined by the instantaneous Q̇o₂/V̇o₂ ratio; Refs. 56, 172) preserves a capillary-myocyte Po₂ gradient (94, 190, 191, 221). It is unlikely that capillary-to-mitochondrial diffusion distances are as important as once thought. Elegant physiological experiments have manipulated these intramuscular diffusion distances and dissociated them from Do₂ (85). Moreover, the beautiful electron microscopy of Glancy and colleagues (65) (Fig. 3, bottom) supports the potential for an intramyocyte “power grid” of interconnected mitochondria and myoglobin that distributes O₂ and transmitochondrial potential differences across the myocyte (23, 24, 65).

Fig. 3.

Top: photomicrographs of corrosion casts depicting extreme examples of capillary structure that increase capillary-to-fiber surface contact to facilitate blood-myocyte O₂ and substrate delivery: Left: highly tortuous and branched skeletal muscle capillary bed (pigeon pectoralis muscle, bar = 50 μm; courtesy of Drs. Odile Mathieu-Costello and Rick Potter). Right: tuna red muscle (bar = 10 µm; courtesy of Dr. Odile Mathieu-Costello). Capillary tortuosity is a function of muscle sarcomere length. The degree of branching may be more extreme in highly oxidative muscles of some species. Bottom: 3-dimensional volume rendering of segmented/interconnected mitochondria (individual colors), capillary (red), and nucleus (cyan) from a 25.13 × 16.74 × 4.23 µm focused ion-beam scanning electron microscope volume of an oxidative mouse muscle fiber (courtesy of Dr. Brian Glancy; see Ref. 65 for similar experiments). The concept of an interconnected mitochondrial reticulum that facilitates O₂ utilization in the perisubsarcolemmal mitochondrial matrix and proton transport deep into the intracellular space in and of itself brings into question the importance of intramyocyte diffusion distances.

Foundational tools for interpreting empirical data are as follows: 1) original observations of capillary hemodynamics and O₂ flux across metabolic transitions (i.e., following the onset of contractions) in health and disease; 2) modeling Q̇ and V̇o₂ kinetics from whole body down to the microcirculatory blood-myocyte interface and quantifying the proportionality between Q̇o₂ and V̇o₂; and 3) the eponymous “Wagner” diagram that conflates the Fick principal with Fick’s law to understand the perfusive and diffusive O₂ conductances within contracting muscles.

Intrinsic to construction of Fig. 1 are original observations at the capillary level in muscles at rest and during contractions: a challenging, but not intractable, problem. Figure 2 summarizes how contemporary observations help explain integrative physiological observations and provide insights into the mechanistic bases for physiological function and disease-induced dysfunction: something that previous capillary schema cannot. The reader is referred to the following reviews for additional specifics (90, 166–169, 172, 177). The rationale for Figs. 1 and 2 is best summarized by posing major questions (i.e., 5 subsections in the next section) that link structurally accurate capillary function to the physiology (and pathophysiology) of blood-myocyte O₂/substrate flux and metabolic control.

DESIGNING A CONTEMPORARY MODEL OF CAPILLARY FUNCTION IN SKELETAL MUSCLE: FUNDAMENTAL QUESTIONS

Is Capillary Recruitment (i.e., Initiation of RBC Flux in Previously Nonflowing Capillaries) Present, or Indeed Necessary, in Muscles at Exercise Onset?

The extraordinary physiologist and Nobel laureate August Steenberg Krogh (1874–1949) made sentinel observations in the muscle microcirculation close to a century ago (127–130). Despite a plethora of subsequent investigations utilizing more advanced technology and approaches (see Table 1), the enduring notion is that very few capillaries (~10%; Ref. 222) support RBC and/or plasma flow in resting muscle. This schema holds that, following the onset of contractions, these stopped-flow capillaries are recruited and facilitate enhanced blood-muscle O₂ and substrate flux to support increased energetic demands (Fig. 2). Ever the careful and humble experimentalist Krogh recognized that the conditions required for his data collection were often far from ideal or physiological. Specifically, to permit successful India ink perfusions, to identify which vessels were supporting flow, multiple vessels were occluded, extremely high-perfusion pressures were applied and sometimes several days elapsed between animal euthanization and experimental perfusion. Because the carbon particles of India ink tend to clump together more at low- than high-perfusion rates preferentially denying access to low-flow capillaries especially in noncontracting muscle, these vessels were counted as nonflowing. In contrast, contemporary intravital microscopy of animal muscles at rest reveals RBC flux in the majority of capillaries (Table 1, right column; Figs. 1 and 2). These capillaries support a vast range of RBC fluxes from <5 to over 50 RBCs/s. Of these the lower flow vessels would be the most likely to experience occlusion by infused carbon particles and incorrectly support the notion of no flow. During contractions/exercise the hyperemia (increased RBC flux) will have reduced carbon particle clumping thereby facilitating access to far more of the extant capillary bed. These observations would provide the appearance of more vessels flowing during exercise than at rest and give rise to the “capillary recruitment” hypothesis.

Table 1.

Two opposing views of the microcirculation: exemplar publications

Opening of Previously “Closed” Capillaries	Most Capillaries Support RBC and/or Plasma Flow at Rest
Krogh 1919 (assorted muscles, various species) (127)	Eriksson and Myrhage 1972 (cat, tenuissimus) (43)
Wagner and Latham 1975 (lung) (225)	Burton and Johnson 1972 (cat, sartorius) (19)
Gorczynski and Duling 1978 (hamster, cremaster) (70)	Klitzman and Duling 1979 (hamster, cremaster) (118)
Honig et al. 1980, 1982 (dog, gracilis, indirect) (95, 96)	Klitzman et al. 1982 (hamster, cremaster) (117)
Gray et al. 1983 (chicken, latissimus dorsi) (76)	Renkin et al. 1981 (rabbit, lower leg muscles) (187)
Rattigan et al. 1997 (rat, hindlimb muscles, indirect) (181)	Hudlicka et al. 1982 (rat, extensor digitorum longus) (101)
Fuglevand and Segal 1997 (computer model) (61)	Vetterlein et al. 1982 (rat, myocardium) (218)
Parthasarathi and Lipowsky 1999 (rat, cremaster) (160)	Kayar and Banchero 1985 (rat, gastrocnemius, in vivo) (108)
Rattigan et al. 2001 (rat, hindlimb muscles, indirect) (184)	Snyder et al. 1992 (rat, in vivo, various muscles) (212)
Youd et al. 2000 (human, rat muscles, indirect, 1-MX, contrast-enhanced ultrasound) (236)*	Hudlicka 1985 (review) (99)
Zhang et al. 2004 (239)*	Tyml, 1986 (frog, sartorius) (217)
Vincent et al. 2004, 2006 (219, 220)*	Bosman et al. 1995 (rabbit, tenuissimus) (17)
Wheatley et al. 2004 (232)*	Poole et al. 1997 (rat, spinotrapezius) (176)
Rattigan et al. 1997 (183)*	Kindig and Poole 1998 (rat, diaphragm) (112)
Rattigan et al. 2001 (184)*	Kindig and Poole 1999 (rat, spinotrapezius) (111)
Rattigan et al. 2003, 2005 (182, 185)*	Kindig et al. 2002 (rat, spinotrapezius) (115)
Clark et al. 2008 (review) (25)*	Richardson et al. 2003 (rat, spinotrapezius) (195)
Fry et al. 2013 (hamster, cremaster, modeling) (60)*	Ellis et al. 2002 (direct, rat EDL) (39)
	Padilla et al. 2006 (rat, spinotrapezius) (157)
	Copp et al. 2009 (rat, spinotrapezius) (29)
	Fraser et al. 2012 (rat, EDL) (59)
	Fraser et al. 2012 (model based on direct visualization of rat EDL) (58)
	Bateman et al. 2015 (rat hindlimb muscles) (8)

Note that the majority of those papers supporting red blood cell (RBC) and/or plasma flow in most capillaries in resting muscle (right column) are either direct observations of muscle microcirculation or studies using endothelial staining of capillaries. EDL, extensor digitorum longus.

^*Those studies do not use indirect methods such as 1-methyl xanthine (1-MX) or contrast-enhanced ultrasound where any data and their interpretation are assumption laden (left column). See text for more details.

This, now entrenched, capillary recruitment hypothesis reflects decades of confirmation bias and presumptive data interpretation: including the four following exemplars.

First, as capillary hematocrit, in resting muscle, is highly variable and averages only ~10–15% with often very long inter-RBC distances (Fig. 1; Refs. 29, 115, 168, 169, 201), RBCs are often absent from the lumen of capillaries viewed in capillary/muscle cross sections. Importantly, lack of an RBC does not mean that these capillaries are not supporting RBC flux (i.e., flow). When Honig and colleagues (95, 96) sectioned dog gracilis muscles, they found more capillaries containing RBCs during contractions than in resting muscle and presumed this to constitute capillary recruitment. Following the work of Duling and colleagues (117, 118), it is now known that capillary hematocrit increases from 10 to 15% toward 45% during hyperemia. The lower than systemic capillary hematocrit at rest results from the Fahreaus effect and the endothelial surface layer or glycocalyx retarding mean plasma flow (see Fig. 1E). Specifically, RBCs travel, on average, faster than plasma due to slower plasma velocity in proximity to the capillary wall, and this effect is exacerbated as vessel diameter decreases. Plasma skimming at arteriolar bifurcations also contributes to the lower capillary hematocrits (see bifurcation law; Ref. 208).

Second, from an evolutionary perspective multicellular life on Earth may have evolved, in part, as a means to protect cells from an increasing level of highly toxic O₂ (98, 134). Thus the circulatory system functions not only to help ensure adequate O₂ supply to support cellular mitochondrial energetics but also to help maintain a relatively low microvascular (10, 11, 19, 146, 179), interstitial (88, 89), and intramyocyte (64, 94, 191, 221) Po₂ at rest and during exercise (see the O₂ “cascade” in Fig. 4). Consequently, hyperoxia is a potent arteriolar vasoconstrictor and hypoxia a vasodilator. This consideration becomes important when interpreting experimental observations under altered O₂ conditions where the pertinent question is: “what are the physiologic Po₂s within muscle?”

Fig. 4.

The oxygen cascade from air to skeletal muscle at sea level. Solid and broken lines depict resting and contracting conditions, respectively. Inspired air, alveolar, and arterial values at sea level (barometric pressure: 760 mmHg; water vapor pressure at 37°C = 47 mmHg). Mean muscle capillary and interstitial values are from Ref. 89 (contracting Po₂ corresponds to nadir values). Resting muscle values are from Refs. 70, 94, 230, 231. Contracting muscle intracellular values are from Refs. 94, 191. Note the sustained Po₂ gradient between the microvascular and interstitial compartments resolved recently via dual-probe phosphorescence quenching in healthy skeletal muscle from rest to contractions. Also, that the mean capillary Po₂-to-intramyocyte Po₂ gradient is far lower (~20 mmHg) than that across the blood-gas barrier in the lung (i.e., alveolar-venous Po₂ difference ~60–80 mmHg) places great emphasis on muscle capillary red blood cell (RBC) volume and the ability to recruit capillary surface area along the capillaries (primary determinant of muscle O₂ diffusing capacity) and capillary RBC transit time to maximize blood-myocyte O₂ flux. [Modified from Hirai et al. (88) with permission.] Inset: micrograph is the spinotrapezius muscle (bar = 1 µm). is, interstitial space; p, plasma; w, capillary wall; m, mitochondria; f, muscle fiber. Large arrow denotes minimum RBC-myocyte distance in this 2-dimensional perspective. [Reproduced from Hirai et al. (89) with permission.]

When Parthasarathi and Lipowsky (160) purportedly demonstrated hypoxia-induced capillary recruitment in the rat cremaster by lowering the extant Po₂ from 130 mmHg down to 35 mmHg, they were simply returning the Po₂ toward normal! Had the authors recognized that muscle microvascular/interstitial/intramyocyte Po₂s are all ≤30 mmHg at rest and far lower during contractions (Fig. 4, animal muscles in Refs. 10, 11, 88, 89, 231; human quadriceps in Refs. 190, 191), the mechanistic interpretation of their data would not have supported hypoxia-induced capillary recruitment. Indeed, their so-called hypoxia (actually mildly “hyperoxic”) supports that, at close-to-physiologic Po₂s, the majority of the microcirculation sustains flow and hyperoxia is vasoconstrictive and retards capillary flow. Today, it is widely recognized that hyperoxia induces profound vasoconstriction even within the coronary vasculature of patients with angina or myocardial infarction (148).

Third, not only must investigators using intravital microscopy be careful not to perturb the physiological Po₂ [as with Parthasarathi and Lipowsky (160)], but a plethora of experimental conditions may impact capillary hemodynamics and confound data interpretation. These include the following: 1) anesthesia and dehydration-induced hypotension; 2) sympathetic stimulation (systemic or local); 3) surgical trauma especially as regards stripping fascia to improve optical clarity; 4) mechanical impediments to perfusion such as direct pressure and occlusion of inflowing arteries/arterioles or outflowing venules/veins; and 5) stretching the muscle: transmission intravital microscopy requires a thin muscle, which is why the hamster cheek pouch retractor, cremaster, extensor digitorum longus, spinotrapezius, diaphragm, and mouse gluteal muscles are chosen. There is also the temptation to improve light transmission and thus optical clarity by stretching the muscle. Unfortunately, this practice reduces Q̇ by mechanically impeding flow through the narrowed capillaries (113, 114) and also active sympathetic-induced vasoconstriction (114, 229). Rigorous adherence to measuring and controlling for muscle sarcomere length prevents this problem (111, 113–115, 195). Whereas concern is often expressed that surgical exposure and damage may lead to increased Q̇ across the capillary bed, experimental evidence demonstrates that the opposite is the likely outcome. Indeed, damage- and stretch-induced capillary stoppage may have helped reinforce the notion that the majority of capillaries in resting muscle do not support continuous RBC flux.

Fourth, three contemporary techniques have contributed disproportionally to the persistence of the capillary recruitment hypothesis. Crucially, each technique is assumption-based which obligates caution in interpretation.

The first technique is permeability-surface area product calculated using glucose extraction. Central to interpreting permeability-surface area data is the conviction that increased glucose extraction can only occur by recruitment of more capillary surface area via initiating flow in previously nonflowing capillaries (102, 149, 150). Notwithstanding the massive heterogeneity of individual capillary plasma fluxes (and thus glucose extraction potential) that exists across the capillary bed, this technique ignores principal tenets of blood-muscle glucose flux. The contemporary “distributed control model of muscle glucose uptake” demonstrates that, in addition to vascular glucose delivery, the presence of glucose transporter type 4 and phosphorylation by hexokinase are key determinants of muscle glucose uptake. Even if each capillary served as a set unit for glucose uptake (which it certainly cannot), not knowing glucose transporter type 4 and hexokinase activities across the experimental perturbation invalidates estimation of perfused capillary surface area (227).

The second technique is arteriovenous difference of 1-methyl-xanthine (1-MX): this intriguing technique was developed by the biochemist Michael G. Clark and relies on xanthine oxidase, located primarily on the capillary endothelium, metabolizing 1-MX in proportion to the perfused capillary surface area (181–185, 236). Thus across experimental conditions any increase in total arteriovenous 1-MX removal is interpreted as “capillary recruitment.” This technique lacks any gold-standard calibration and remains to be validated by concurrent intravital microscopy. Sentinel concerns include the following: 1) 1-MX is plasma borne and could not therefore discriminate RBC-perfused capillaries from those that may support just plasma flow, and this is a significant weakness as Kayar and Banchero (108; see also Ref. 212 discussed below) have demonstrated that, at least in the rat, essentially all muscle capillaries (gastrocnemius) support plasma flux at rest; 2) xanthine oxidase activity in the rat is extremely high necessitating prior allopurinol (a xanthine oxidase inhibitor) treatment; 3) there has been no demonstration that 1-MX metabolism is flow or delivery rate independent and biochemically this is unlikely to be the case, and unfortunately, flow/delivery independence is an absolute requirement for estimation of the perfused capillary bed size; and 4) the presence of xanthine oxidase on the capillary endothelia and also in smooth muscle and plasma (84, 152) invalidates the presumption that 1-MX metabolism requires endothelial contact per se.

The third technique is contrast-enhanced ultrasound (CEU): measurement of microvascular volume, capillary RBC transit times, and other important vascular indexes at rest and during voluntary exercise in human muscle(s) would be extremely valuable and that is what CEU proposes. CEU relies on acoustic signals generated by inert gas microbubbles that are infused intravenously and supposedly track RBC behavior and distribution within the muscle microcirculation without impacting hemorheology (25, 147, 226). Unfortunately, to date, there has been no direct in vivo or in situ validation of CEU data. Unless it can be demonstrated that the CEU bubbles, which are far smaller than RBCs, distribute in a quantitatively similar fashion to RBCs across vascular bifurcations in a manner and concentration that is flow rate independent, CEU cannot resolve the volume of perfused capillaries. CEU relies on a disappearingly small ratio of 1 bubble/6,000 RBCs (211) and cannot account for the change in capillary hematocrit that occurs with hyperemia. Specifically, from rest-to-exercise, as capillary hematocrit rises from ~15 toward 45% (systemic), that proportion of the capillary filled with plasma space will decrease. Because the microbubbles are plasma borne, all else being equal, this effect would act to reduce the CEU estimate of vascular volume. Possibly more concerning is that the hyperemic state may impact the microbubble distribution very differently from rest (or the RBCs) concentrating them with respect to vascular volume. Indeed, estimates of vascular volumes exceeding 20% in primate forearm muscle (21) suggest that this is the case. Finally, according to Kusters and Barrett (132), CEU cannot detect any increase in muscle Q̇ from rest to “light” exercise (i.e., 25% V̇o_2max) whereas all other techniques including direct observation (29, 85, 115, 118), microspheres (56, 138), ultrasonography (125), and thermodilution (44–46, 75, 119, 170, 192, 196) can.

Despite concerns noted above, each of these techniques has the potential to reveal novel, and perhaps valuable, insights into microcirculatory function and blood-myocyte transport. However, when data interpretation is filtered through the capillary recruitment theory, correct mechanistic interpretation is hampered.

Whether only 10% (historical perspective) or as much as 90% (contemporary model) of the capillary bed supports RBC and plasma flux at any instant at rest is not a purely academic question. Skeletal muscle is, by mass, the largest organ and determining whether the majority of the capillary endothelial surface is available for blood-tissue O₂, substrate, and metabolite exchange is crucial for understanding these processes and the regulation of substrate fluxes. Moreover, if the common presumption is that no-flow is expected in most capillaries the impact of pathologies such as HF (111, 195) and diabetes (type I in Ref. 116; type II in Ref. 157), that in-and-of-themselves cause capillary hemodynamic stasis, will be missed. The consequences of such an omission include the inability to resolve the mechanistic bases for muscle and metabolic dysfunction and hamstring development of effective therapeutic countermeasures.

Philosophically and practically it is either extremely challenging or not possible to determine the proportion of capillaries that sustain RBC and plasma flux across all skeletal muscles in a resting animal or human. However, there are several analyses and empirical observations that either 1) support the presence of flow in the majority of skeletal muscle capillaries at rest or 2) demonstrate that the fundamental physiological properties of skeletal muscle do not require de novo capillary recruitment following the onset of contractions.

Empirical support for lack of capillary recruitment at exercise onset.

is muscle Q̇ in humans at rest sufficient to ensure most capillaries can support RBC flux?

As presented in Fig. 5 using broadly accepted values for total muscle(s) Q̇ of 1 l/min at rest (197, 198) with a systemic hematocrit of 45% (1) and approximations of skeletal muscle capillarity, 5.4 trillion RBCs/min would be distributed among some 9 billion capillaries. If we consider that 80% of these capillaries support continuous RBC flux, an average of ~13 RBCs/s could pass along each of these capillaries. Empirical data from the rat spinotrapezius muscle, which contains a mixture of slow and fast-twitch fibers, places the mean observed value at 15–20 RBCs/s. As discussed above this mean value conceals extensive heterogeneity among capillaries within a given muscle (111, 115, 176) and is expected to differ as a function of different bulk Q̇s across muscles comprised of different fiber types (56, 92, 93, 178). Given the greater metabolic rate of the rat versus human, the correspondence (i.e., ~13 estimated in the human vs. 15–20 RBC/capillary/s measured in the rat) is remarkable and this calculation supports that it is at least feasible, given the available Q̇, to perfuse the majority of capillaries in human muscle at rest. Propositions to the contrary lack a scientific rationale, defensible mechanistic underpinning, and uncontested supporting data.

Fig. 5.

As demonstrated by intravital microscopy (e.g., Ref. 111) and dye infusion (108) for anesthetized and conscious animals, is it possible that the majority of capillaries support red blood cell flux in resting muscle in humans? This figure uses established values for muscle blood flow (e.g., Refs. 197, 198) combined with good-faith estimates of muscle capillarity (e.g., Refs. 97, 193, 206) to demonstrate that, in humans, ~80% of total muscle capillaries could, in theory, support an average of nearly 13 red blood cells (RBC)·s⁻¹, which is close to that value measured in rat spinotrapezius (15–20 RBC·cap⁻¹·s⁻¹; Refs. 111, 115). In contrast to the leftmost and middle boxes, the italicized capitalized letters in the rightmost box alert the reader that this “80%” value is data taken from the animal literature as such data are not presently available in humans. Illustrated by Mal Rooks Hoover, CMI. See text for more details.

is de novo capillary recruitment required to support Q̇ and V̇o₂ kinetics following exercise onset?

Several decades ago time-honored notions held that the exercise hyperemia was initiated by falling muscle O₂ levels and subsequent metabolite accumulation triggering arteriolar and, perhaps, precapillary sphincter vasodilation. Today it is recognized that, because cardiovascular dynamics are extremely rapid, cardiac output is elevated almost instantaneously. In the absence of a simultaneous vasodilation and increased vascular conductance within contracting skeletal muscles, a profound transient hypertension would be manifested, which certainly has not been observed at the onset of dynamic exercise. Accordingly, it is now known that skeletal muscle Q̇ increases within the first contraction cycle for rhythmic exercise by a combination of muscle pumping action and vasodilation, which is supplemented subsequently by a panoply of vasodilatory mechanisms including sympatholysis, NO, cyclooxygenases, reduced O₂ pressures, and vasoactive metabolites such as K⁺, adenosine, lactate, CO₂, and H⁺ (69, 106, 138, 216). Moreover, evidence that RBCs themselves act as oxygen sensors and sources of vasodilators such as ATP implies a finely tuned degree of temporal and spatial ability to match Q̇o₂-to-V̇o₂ (41, 42). Thus, within the spinotrapezius, following the onset of contractions, capillary RBC velocity and flux increase rapidly such that microvascular (and venous) O₂ levels are either maintained constant or even rise for several seconds despite fast mitochondrial kinetics that increases V̇o₂ with a time constant of 20–30 s (Fig. 6; Refs. 10–12, 115; see also Ref. 194 for human calf intramyocyte myoglobin desaturation kinetics). This characteristic profile corresponds closely with less direct measurements across muscle groups (canine gastrocnemius-plantaris complex, 74), exercising limbs (75), and the whole body (135, 233; reviews in Refs. 172, 199) in animals and humans and, as evidenced by Kindig and colleagues (115), is not dependent on the conventional concept of capillary recruitment. Rather blood-myocyte exchange capacity can be increased by using more capillary length for that exchange i.e., “longitudinal recruitment” (Fig. 1D). Specifically, in concert with the hyperemia-induced elevation of capillary hematocrit, which serves to decrease inter-RBC gap length, greater RBC velocity and also increased fractional O₂ extraction during contractions serve to increase the effective capillary surface area for O₂ flux (68).

Fig. 6.

Data from Behnke and colleagues (9), normalized to 100% response amplitude, showing increase in rat spinotrapezius red blood cell (RBC) flux (solid circles) and microvascular Po₂ (Pmvo₂; triangles) conflated to estimate oxygen uptake (V̇o₂; hollow circles) in response to 1 Hz electrically-induced contractions (initiated at time = 0 s). Capillary RBC flux was measured by high-resolution videomicroscopy analysis in individual capillaries (115). Pmvo₂ was determined using phosphorescence quenching techniques (9–11, 179, 200). V̇o₂ was calculated using the measured RBC flux and arterial O₂ content and Pmvo₂ as an estimate of end-capillary PO₂ and O₂ dissociation curves established for the rat (1). Model fits are shown. Both RBC flux and V̇o₂ (but not Pmvo₂) were fit by a single exponential with no time delay. TD, time delay; τ, time constant. The value for τ resolved is similar to that for human muscle and also phase II pulmonary V̇o₂ kinetics (e.g., Ref. 75).

could surgical exposure and/or pentobarbital anesthesia in experimental animals impair vascular control and increase the proportion of flowing capillaries?

As mentioned above, surgical trauma may potentially change muscle Q̇ and the proportion of capillaries supporting Q̇ with increases or decreases possible. Bailey and colleagues (4) used radioactive microspheres to measure spinotrapezius muscle(s) Q̇ in the anesthetized rat pre- and postsurgery. No significant increase or decrease in Q̇ was evident supporting that surgical trauma did not impact vascular control in this preparation. More directly, with respect to the proportion of flowing capillaries, Snyder and colleagues (212) injected FITC tracer dye into the central circulations of anesthetized and conscious rats and then arrested the circulation after 1, 3, 5, 7, 10, or 15 s and analyzed the proportion of capillaries in the soleus, vastus lateralis, and diaphragm that contained dye. They found that within 3–7 s essentially all capillaries in these three muscles contained dye and this result was not influenced by anesthesia. Previously, Kayar and Banchero (108) made a similar observation in the anesthetized rat gastrocnemius as did Vetterlein and colleagues (218) in the rat heart. Whereas these techniques cannot discriminate between solely plasma flow versus plasma and RBCs in capillaries, these findings cannot be reconciled with the presence of an enormous “reserve” of capillaries to be recruited in resting skeletal muscle.

With regard to whether pentobarbital anesthesia may alter vascular smooth muscle control itself Ferreira and colleagues (56) argued that such an effect, if present, would necessarily alter the established relationship between Q̇ and V̇o₂ of 5–6 l/l (or ml/ml) seen during human voluntary exercise (165, 233). With the use of a combination of microspheres and phosphorescence-quenching techniques across a spectrum of muscle and fiber types at rest and during contractions in the anesthetized rat, the expected physiological relationship was found (Fig. 7). These data are inconsistent with the presence of anesthesia-impaired smooth muscle control and support that the presence of RBC flux in the majority of capillaries in resting muscle is not an artifact of the anesthesia (Table 1, right column).

Fig. 7.

Relation between blood flow and oxygen uptake (V̇o₂) during muscle contractions for various muscles in pentobarbital-anesthetized rats. [Data are means from Refs. 9 and 10 as redrawn from Ferreira (56) with permission.] This figure demonstrates that anesthetized preparations retain vasomotor control such that blood flow increases in the same proportion with V̇o₂ (i.e., 5–6 l/min:1 l/min) as found in conscious preparations (review in Ref. 165) although the range of metabolic rates and blood flows is lower for the animal preparations owing, in part, to the requirement for electrical stimulation.

what evidence is there that these capillary hemodynamics and muscle deoxygenation profiles faithfully represent the physiological response(s) of human muscles during voluntary rhythmic exercise?

Near-infrared spectroscopy (NIRS), especially with the recent development of time-resolved spectroscopy (TRS) sampling deep muscles in the human quadriceps, permits measurement of absolute [Hb] and myoglobin concentration ([Mb]) at rest and across the transition to exercise (i.e., total[Hb + Mb]) and deoxy[Hb + Mb]; Refs. 35, 83; reviews in Refs. 122–125, 141, 155). With respect to the Kroghian concept (i.e., majority of capillaries not flowing at rest) there are three options: 1) those nonflowing capillaries contain stagnant RBCs, 2) they contain solely plasma and other non-RBC constituents, or 3) they are collapsed (i.e., empty). For option 1, capillary RBC transit time would increase relative to the time RBC velocity remained at zero. During this zero-flow duration with no capillary Q̇ (i.e., RBC flux and velocity at zero), fractional O₂ extraction would increase as intracapillary Po₂ equilibrated with intramyocyte Po₂, which itself would be decreasing as a function of the ratio of local O₂ stores/V̇o₂.

Thus, if, as considered by Wagenmakers et al. as recently as 2016 (222), up to 90% of capillaries in resting muscle did indeed contain stagnant RBCs, measurements of deoxy[Hb + Mb] would be extremely high at rest (ratio of Do₂/βQ̇ rising toward infinity as Q̇ approaches zero in Eq. 1) and actually fall precipitously during exercise as these capillaries become recruited and contain “freshly” oxygenated arterial blood (increasing Q̇o₂ and βQ̇, Eq. 1). In direct contrast to this notion, the transition to exercise is actually marked by a pronounced increase in deoxy[Hb + Mb], which rises temporally in concert with the muscle(s) a-vO₂ difference (83, 122–125, 151, 155), and the decrease in intramyocyte Po₂ (191).

The alternative options 2 and 3, that those ~90% of nonflowing capillaries are devoid of RBCs, if true, would necessitate a prodigious increase in total[Hb + Mb] as they become recruited. Moreover, as capillary hematocrit increases with capillary RBC flux (115, 118, 169), an increase in muscle capillary [Hb] of several orders of magnitude would be expected to increase total[Hb + Mb]) greatly. That total[Hb + Mb] increases only modestly (i.e., ~20 µM or 10–30%) in quadriceps from rest-to-exercise can be accounted for by increased capillary hematocrit in capillaries with RBCs already flowing, thus providing unequivocal evidence that substantial capillary recruitment is not a requisite feature of the hyperemia in human muscle(s) (Fig. 8; 35, 62, 83, 122–125, 141, 155). NIRS is becoming a widely employed technique for measuring changes in muscle (de)oxygenation and [Hb + Mb] in exercise laboratories around the world. Across many thousands of observations, the prodigious increase in capillary RBC content predicted by capillary recruitment theory is notably absent.

Fig. 8.

Increase in total hemoglobin + myoglobin concentration (total[Hb + Mb]) from rest to heavy-intensity exercise measured in the human quadriceps (vastus lateralis and rectus femoris) by time-resolved near infrared spectroscopy. Drawn from the data of Fukuoka et al. (62).

Do Precapillary Sphincters Exist in Skeletal Muscle and Provide for Flux Control at an Individual Capillary Level?

Any consideration of the control of Q̇ distribution, and also Q̇o₂-V̇o₂ matching, within the skeletal muscle capillary bed must involve whether or not that control is exerted at the individual capillary level and/or resides within the arteriolar tree. Since Zweifach and colleagues first described precapillary sphincters and metarterioles within the mesentery of frogs and, perhaps, some mammals, beginning in 1937 (e.g., Refs. 22, 241, 242; review in Ref. 202) these structures have been presumed to be homogeneously distributed across the microcirculation of all tissues. Furthermore, in the absence of any compelling structural or functional evidence, precapillary sphincters have been presented as general features of all microcirculatory beds within almost all standard medical school and professional physiology textbooks (32, 164; review in Ref. 202). Despite the scientific imperative for evidence and harsh condemnation by leaders in the microcirculation field for presuming rather than verifying key microvascular control elements (186, 187) precapillary sphincters have been used mechanistically to explain a host of physiological observations. Thus precapillary sphincters in muscle purportedly exhibit vasomotion as a means to improve tissue oxygenation (73, 180), display myogenic regulation, and control transcapillary fluid exchange (78, 140) and capillary hydrostatic pressures (79, 81); constitute the site for neurogenic vasoconstrictor control (3, 26, 87, 188); and play a role in compromising exercise-induced hyperemia in type I diabetic patients (214). As such, the precapillary sphincter is purportedly the site of action for a spectrum of vasoactive mediators including catecholamines, amines (serotonin, histamine), angiotensin, kinins, vasopressin, and glucocorticoids (2, 103) and, while appropriately qualifying “the precapillary sphincter” as “any constrictive activity along the terminal twigs of the arteriolar system,” whole Microcirculatory Society symposia have been dedicated to its structure and function (153).

The superbly researched and posited review by Sakai and Hosoyamada (202) documents the absence of structural and functional evidence for precapillary sphincters in skeletal muscle and all other organs except for the mesentery. In addition, Golub and Pittman (69) as well as Duling and colleagues (34, 70, 71, 117) were unable to find empirical evidence for precapillary sphincter action physiologically. For instance, the action of hyperoxia to arrest RBC flux across large portions of the capillary bed is invoked by vasoconstriction at the arteriolar level and not downstream in individual capillaries (70, 71, 117).

As with the notion of capillary recruitment, discussed in option 1 above, evidence for the existence of precapillary sphincters in skeletal muscle is not compelling (review in Ref. 202). Moreover, it is difficult to reconcile how either concept can be invoked to explain muscle microvascular control including the temporal and spatial matching of Q̇ and Q̇o₂-V̇o₂ during exercise. Notwithstanding the weakness, or absence, of evidence to support the presence of, or necessity for, capillary recruitment and precapillary sphincters in skeletal muscle, respectively, medical and physiology reviews and textbooks continue to extol their central role in capillary function and blood-muscle O₂ and substrate exchange (e.g., Refs. 32, 164, 222).

What Is the Three-Dimensional Geometry of Muscle Capillaries and Mitochondria and Are Intramyocyte Capillary-Mitochondria Diffusion Distances an Important Limitation to O₂ Flux in Skeletal Muscle?

Unlike the structurally unsupported capillaries in the lung, mesentery, or bat’s wing (e.g., Refs. 159, 189, 203, 242) extensively studied in vivo, skeletal muscle capillaries are supported by collagenous struts that serve to affix the abluminal capillary surface to the myocytes and maintain capillary patency even during muscle contractions (15) and when vascular pressures are pathologically low (114). Skeletal muscle capillaries vary widely in length (20–1,000 μm) and also diameter (~2–8 µm), both of which can change as a function of sarcomere length (145, 175). Muscle capillaries are anisotropic and form a branched plexus in alignment to the principal myocyte orientation (Fig. 3, top; Ref. 145) with a capillary volume density that correlates with muscle oxidative capacity (174, 175); this relationship facilitates high perfusive and diffusive O₂ fluxes while constraining the fall in capillary RBC transit time at high Q̇s. In contrast to other tissues, skeletal muscle capillaries are extremely plastic, becoming more tortuous as muscle sarcomeres shorten either actively or passively and straightening and narrowing as the muscle is stretched (especially >2.7 μm; Refs. 113, 114, 145). Spatial relationships among microvessels permit intervessel O₂ diffusion (20, 68; review in Ref. 162), and the tortuous and branched capillary geometry increases capillary-myocyte surface contact (Fig. 3, top; Refs. 40, 144, 145) as does the manner in which capillaries embed into the sarcolemma on specific fibers (66). Rather than capillaries being simply a “point” source of O₂ supply (i.e., Kroghian concept), it is now recognized that the Hill cylinder model (Fig. 3, top right; Refs. 40, 145) better describes how the structure of muscle capillaries provides a more effective means of reducing diffusion resistances and facilitating blood-myocyte O₂ flux. This perspective is consistent with the latest measurements of microvascular and interstitial Po₂s in resting and contracting muscles (88, 89).

Mitochondrial structure.

In keeping with the theory that mitochondria originated from bacteria that were incorporated into unicellular prokaryotic and eukaryotic cells (98, 134), they are often portrayed as distinct bean-shaped organelles distributed within the myocyte subsarcolemmal and intermyofibrillar spaces. This perspective coheres with their profile in thin sections cut transverse to the fiber axis and encouraged the theory that mitochondria more distant from the nearest RBC-flowing capillary might suffer from lack of O₂ within anoxic loci especially during severe intensity exercise. However, when the three-dimensional structure of the mitochondria is appreciated from either serial reconstruction or elegant electron microscopy techniques (e.g., focused ion-beam scanning electron microscopy; Ref. 65) it is revealed that, far from being small isolated structures, the mitochondria form a highly interconnected catenated reticulum (Fig. 3, bottom; Refs. 7, 65, 109, 154). This geometry, in combination with myoglobin, is proposed to negate the importance of intramyocyte “diffusion distances” for O₂ and ATP transport by forming a conductive pathway for energy distribution (23, 24, 65). Such a system helps reconcile the retained exercise capacity in myoglobin-free mice (63, 72), the extremely low (2–5 mmHg) intramyocyte Po₂s in contracting canine gracilis (94) and human (191) muscles, and the independence of muscle Do₂ and V̇o_2max from capillary density after exercise training or immobilization-induced atrophy in the canine gastrocnemius-plantaris complex (85).

To summarize the answer to the question posed in this section: capillary geometry and function as described herein are unique to skeletal muscle and cannot therefore readily be intuited from observations of more easily accessible microcirculations such as skin, nail-fold, or sublingual (110) and certainly not the mesentery, bat’s wing, or lung discussed above. Depending, in part, on muscle sarcomere length, a three-dimensional perspective reveals a tortuous and branched capillary network that provides a greater capillary-myocyte surface area (Fig. 3, top, Hill model) than presumed from modeling individual capillaries, simply as single point delivery system (i.e., Krogh model, 40, 144, 145). This notion is supported by maintenance of a substantial Po₂ in the interstitial space (Po₂is) that envelops the myocyte periphery (88, 89). Within the myocytes, mitochondria form a contiguous reticulum that, in combination with myoglobin, enhances O₂ and energy distribution throughout the myocyte akin to a “power grid” (Fig. 3, bottom; Refs. 23, 24, 65). Such a phenomenon renders capillary-mitochondrial diffusion distances moot (85, 168) at least within these smaller mammalian myocytes.

Can All Capillaries in Skeletal Muscle Be Considered Equal With Respect to O₂ and Substrate Delivery?

Capillary RBC path lengths and transit times are hugely variant at rest and during contractions (115, 205–207). These heterogeneities will impact O₂ and substrate delivery/exchange very differently depending on whether they are transported principally in the RBC (O₂) or plasma (glucose, lactate, and free fatty acid). As O₂ solubility in plasma is extremely low, only that capillary endothelium immediately adjacent to a given RBC is thought to be quantitatively important for blood-myocyte O₂ transport. Accordingly, during exercise, muscle Do₂ is determined principally by the volume of RBCs (i.e., capillary hematocrit × capillary volume) within those capillaries adjacent to the contracting muscle fibers (47, 67, 68, 77). For plasma, versus RBC-borne substrates, especially at rest when capillary hematocrit is ~15%, access to the endothelial surface for blood-myocyte transport is thus far greater.

Across species and muscles, capillary density, volume density and capillary surface-fiber contact correlate with muscle oxidative enzyme activities, mitochondrial volume densities, and V̇o_2max, especially in more oxidative muscles (100, 101, 174). In contrast, in more glycolytic muscles the requirement to remove metabolites during contractions may predicate a larger capillary bed than predicted from the oxidative capacity (review in Ref. 100). However, these broad relationships conceal a substantial heterogeneity of structure and function among individual capillaries. The capacity for a given capillary or unit of capillary surface area to facilitate blood-myocyte or myocyte-blood transport will depend on multiple factors including 1) capillary volume (flowing capillaries only) and functional surface area; 2) RBC flux, velocity, hematocrit, path length, and transit time in a given capillary; 3) characteristics of the exchange molecule, for instance, RBC- or plasma-borne, blood-myocyte concentration or pressure differential, of facilitation by specialized transport processes, e.g., glucose (227). Given these considerations it is not surprising that measurements such as capillary filtration coefficient, for example, do not relate to the number of perfused capillaries (13) and that the presumption of capillaries as quantum units of blood-myocyte exchange is fallacious.

Can Altered Capillary Hemodynamics Help Explain Muscle Dysfunction in Disease and the Therapeutic Benefits of, for Example, Exercise Training and/or Nitrate Supplementation?

Without knowledge or appreciation of muscle microvascular function in health the scientist investigating the mechanistic bases for exercise dysfunction in disease is compromised. When Daniel Boorstin, in his epic book The Discoverers, quipped that “The greatest impediment to scientific progress is not ignorance but, rather, the illusion of knowledge” he might have been referring to this precise scenario (14). If our scientist observes the capillary bed in an animal with HF (195), type II diabetes (157), or sepsis (39) and sees that 40–50%, or more, of the capillaries do not support RBC flux, adherence to “Kroghian” capillary recruitment theory would blind them to one principal basis for compromised muscle function. Thus a potential avenue for therapeutic intervention would be ignored. Conversely, if, as the overwhelming evidence supports [see Is Capillary Recruitment (i.e., Initiation of RBC Flux in Previously Nonflowing Capillaries) Present, or Indeed Necessary, in Muscles at Exercise Onset?], the healthy condition is recognized to be the vast majority of capillaries supporting RBC flux at rest, the presence of pathology is strikingly evident. Moreover, when paired with contemporary models of muscle O₂ delivery that consider perfusive (Q̇o₂) and diffusive (Do₂) O₂ conductances, it is possible to demonstrate that both are impaired even when maximal fractional O₂ extractions and muscle venous effluent O₂ contents are close-to-normal in HF patients (44–46, 88, 90, 107, 171, 195). Exercise training (chronic) and nitrate supplementation (acute) both upregulate the O₂ transport pathway and improve exercise performance. The following brief sections pair, in-so-far as possible, the contemporary capillary function model (Figs. 1 and 2) with the Wagner analysis (Fig. 9, top) to investigate how Q̇o₂ and Do₂ adaptations conflate to improve oxidative function and exercise tolerance in health (exercise training, Fig. 9, middle) and compromise function in disease (HF, Fig. 9, bottom). This approach is also valuable for identifying future directions and innovative hypotheses.

Fig. 9.

Top: schematic showing muscle O₂ uptake (V̇o₂) plotted as a function of the venous or microvascular Po₂ (Pmvo₂). Conflation of perfusive (Q̇o₂; Fick principle, solid curved line) and diffusive (Do₂; Fick’s law, solid straight line from origin) O₂ conductances permits a mechanistic analysis of how these components resolve to achieve a given V̇o₂ (see Eq. 1). Middle: exercise training (dashed lines) acts to increase both perfusive (Q̇o₂, gray arrow 1) and diffusive (Do₂; gray arrow 2) O₂ conductances by elevating maximal cardiac output, muscle Q̇o₂ and Do₂ consequent to capillary neogenesis, which helps to constrain/prevent a reduction in mean capillary RBC transit time. Notice that a substantial increase in Do₂ is required to produce even a small lowering of venous or microvascular Po₂ (i.e., increased arterial-venous O₂ difference, downward pointing arrows on abscissa). Bottom: heart failure with reduced ejection fraction (HFrEF; dashed lines) reduces both perfusive and diffusive O₂ fluxes and hence V̇o_2max. Importantly, for a given submaximal V̇o₂, muscle deoxygenation and microvascular PO₂ fall far more during the transient following the onset of contractions slowing V̇o₂ kinetics as seen in Fig. 10. Resolution of the microvascular impairments that reduce perfusive (Q̇o₂) and diffusive (Do₂) O₂ fluxes constitutes a critical step in the design of effective therapeutic interventions (e.g., exercise training, ↑nitric oxide (NO) bioavailability). These relationships remain to be defined in heart failure with preserved ejection fraction, but indirect measurements suggest a greatly diminished capacity to achieve fractional O₂ extractions seen in healthy individuals (review in Ref. 177). Please see text for additional details.

Exercise training.

healthy.

In healthy humans, exercise training increases V̇o_2max. Based on simple mass-balance considerations, the bulk of that increase is ascribed to training-induced elevation of cardiac output and, therefore, muscle Q̇o₂ (165, 196, 204, 223, 224). As fractional O₂ extraction is capped somewhere below the arterial O₂ content (typically ~20 ml/100 ml) by the presence of a finite Do₂ and the ratio Do₂/βQ̇ (see Eq. 1 above) and increases as a hyperbolic function of V̇o₂ (165, 233), the maximal a-vO₂ difference widens only incrementally after training despite substantial increases in whole body or exercising limb O₂ delivery (i.e., Q̇o₂) according to:

$${\text{a-vO}}_{\text{2}}\text{\hspace{0.17em}difference\hspace{0.17em}}\left(\text{ml/100\hspace{0.17em}ml}\right)=22\dot{\text{V}}{\text{o}}_{\text{2}}\cdot {\left(1+\dot{\text{V}}{\text{o}}_{\text{2}}\right)}^{-1}$$ (2)

where V̇o₂ is expressed in l/min and the numerator (i.e., 22) is determined empirically (119, 170).

If we consider, for example, a healthy young sedentary human with V̇o_2max of 3 l/min who trains rigorously for 12 wk and increases V̇o_2max to 4 l/min.

$${\text{Pretraining:\hspace{0.17em}a-vO}}_{\text{2}}\text{\hspace{0.17em}difference\hspace{0.17em}}\left(\text{ml/100\hspace{0.17em}ml}\right)=22\times 3/\left(1+3\right)=66/4=16.5\text{\hspace{0.17em}ml/100\hspace{0.17em}ml\hspace{0.17em}}$$ (3)

$${\text{Posttraining:\hspace{0.17em}a-vO}}_{\text{2}}\text{\hspace{0.17em}difference\hspace{0.17em}}\left(\text{ml/100\hspace{0.17em}ml}\right)=22\times 4/\left(1+4\right)=88/5=17.6\text{\hspace{0.17em}ml/100\hspace{0.17em}ml\hspace{0.17em}}$$ (4)

Thus, for a substantial 33% increase in V̇o_2max, a-vO₂ difference increases only 6.7%. However, as illustrated in Fig. 9, middle, this requires a substantial increase of muscle Do₂ without which a-vO₂ difference (i.e., Eq. 1) would actually have decreased and V̇o_2max would have only increased ~12% and not the observed 33%. For the training study of Roca et al. (196), muscle Do₂ increased 35% to elevate fractional O₂ extraction by 9.9%. What was remarkable from that investigation was that the increased Do₂ was responsible for 60% of the training-induced V̇o_2max increase! Given that modeling studies support a primary role for the number of RBCs within the capillary bed adjacent to the contracting myocytes as setting Do₂ (47, 77; see also Ref. 142), it would be extremely valuable to determine the amount and behavior (i.e., distribution and hemodynamics) of RBCs within the exercising muscle bed pre- and posttraining. Careful application of TRS-NIRS might address whether training-induced V̇o_2max improvements are accompanied by the expected rise in exercising muscle [Hb].

heart failure.

Impaired exercise performance in HF relates primarily to the reduced capacity of the O₂ transport pathway to increase Q̇o₂ commensurate with V̇O₂ requirements. This creates a supply dependence for V̇o₂ kinetics, increasing the O₂ deficit and reliance on substrate-level phosphorylation, thereby accelerating depletion of finite muscle glycogen reserves and “sowing the seeds” for exercise intolerance (57, 90; reviews in Refs. 171, 172, 199). V̇o_2max is also crippled by reduced maximal Q̇o₂ often coupled with lowered Do₂ (Fig. 9, bottom; Ref. 90). This scenario manifests in other endemic diseases such as diabetes and peripheral arterial disease as well as with aging (173, 199). The severity of HF is classified by the New York Heart Association by V̇o_2max (in ml·kg⁻¹·min⁻¹): Class I: >20; Class II: 16–20; Class III: 10–15; and Class IV: <10. As well as being an excellent indicator of the predations of HF, V̇o_2max (or V̇o_2peak; see Ref. 173) represents the best predictor of survival with cardiac transplantation being safely deferred in patients with values >14 ml·kg⁻¹·min⁻¹ (143, 228). Some therapeutic treatments, such as exercise training, elevate V̇o_2max, reduce patient readmissions, and significantly improve quality of life and longevity, relying, invariably, on elevating both muscle Q̇o₂ and Do₂ (12, 38, 57, 90).

In HF with reduced ejection fraction (HFrEF, systolic failure), there may be a modest capillary rarefaction but by far the biggest contributor to reduced Do₂ is cessation of RBC flux in many capillaries (i.e., down from >80% in controls to 50–60% in HFrEF) in proportion to the severity of cardiac damage and the rise in left ventricular end-diastolic pressure (Fig. 9, bottom, and Fig. 10B; Refs. 111, 115, 171, 195; Fig. 9, middle; Refs. 44–46, 90, 171). HFrEF patients’ muscles retain their angiogenic capacity and increase capillarity and Do₂ after training (45) both of which oppose the predations of HFrEF and contribute to the improved microvascular Po₂, faster V̇o₂ kinetics and increased V̇o_2max and exercise performance (Fig. 10, red arrows).

Fig. 10.

A: constellation of heart failure (HF) responses decrease blood flow (Q̇) and O₂ delivery (Q̇o₂) to skeletal muscle(s) and impair Q̇o₂-to-V̇o₂ matching. B: functional changes in capillary hemodynamics [↓↓proportion of red blood cell (RBC) flowing capillaries and RBC velocity and flux] are far greater than found structurally (↓capillary length, volume, and surface area). C: these functional changes impede the matching of Q̇o₂-to-V̇o₂ by decreasing resting capillary RBC flux and constraining the rapid increase observed in healthy contracting muscles. D and E: this reduces the microvascular O₂ partial pressure (Pmvo₂) at rest and following the onset of contractions thereby decreasing blood-myocyte O₂ flux, slowing V̇o₂ kinetics (D) and markedly impairing sustained exercise capacity (i.e., critical power; E). Text boxes and red arrows indicate directional improvements conferred by exercise training and/or nitric oxide bioavailability/sildenafil. V̇o₂, oxygen uptake; Q̇o₂, muscle O₂ delivery, i.e., perfusive O₂ flux; COPD, chronic obstructive pulmonary disease (here synonymous with emphysema); R, rest; Ex, exercise; CHF, chronic heart failure. [Adapted from Hirai et al. (90) with kind permission.]

HF with preserved ejection fraction (HFpEF, diastolic failure) is highly prevalent, but far less is known about the mechanistic bases for impaired functional capacity especially in terms of altered muscle Q̇o₂ and DO₂. Intriguingly HFpEF patients appear to exhibit lower fractional O₂ extractions than their HFrEF or healthy counterparts (review in Ref. 177). To what degree this relates to altered Q̇o₂ distribution among organs or within skeletal muscle (i.e., Q̇o₂-to-V̇o₂ mismatching) and/or impaired Do₂ has not been resolved and constitutes an important question for future research.

Physiologic and therapeutic effects of increased NO bioavailability.

Both endogenous (NO synthase-derived) and dietary-induced NO bioavailability impact vascular smooth muscle and skeletal muscle function during exercise in health and disease (31). As attenuated NO bioavailability is emblematic of aging and multiple disease conditions, the potential of dietary or supplemental nitrate/nitrite ($\text{N}{\text{O}}_3^{-}$/$\text{N}{\text{O}}_2^{-}$) augmentation to improve cardiovascular and exercise function is great (6, 27, 104, 105, 120, 121, 126, 161, 210, 235).

$\text{N}{\text{O}}_3^{-}$ absorbed into the bloodstream from the diet is concentrated by the salivary glands and facultative oral bacteria such as actinomyces and rothia, which reduce $\text{N}{\text{O}}_3^{-}$→$\text{N}{\text{O}}_2^{-}$. $\text{N}{\text{O}}_2^{-}$ is then resorbed into the blood stream elevating circulating [$\text{N}{\text{O}}_2^{-}$] and forming an available reservoir that can be further reduced to NO by deoxyhemoglobin, myoglobin, and the oxidoreductase system especially under hypoxic and acidic conditions: both of which are found in skeletal muscles performing severe intensity exercise and preferentially in muscles comprised of fast twitch (type II) fibers (11; review in Ref. 105) especially in HF patients (90). The endogenous NO synthase enzymes (endothelial nitric oxide synthase and neuronal nitric oxide synthase) are inhibited in such environments. Thus, raising blood [$\text{N}{\text{O}}_2^{-}$], and thereby NO bioavailability, elevates vascular conductance and improves muscle function specifically in those muscles where the Q̇o₂-to-V̇o₂ mismatch is greatest and microvascular Po₂s are very low (i.e., fast-twitch fibers; Refs. 11, 105, 146).

healthy.

Elevating NO bioavailability via dietary/oral supplementation of $\text{N}{\text{O}}_3^{-}$ or $\text{N}{\text{O}}_2^{-}$ exerts a plethora of physiological effects that improve cardiovascular and muscle function (review in Ref. 105). Germane to this review, muscle Q̇o₂ (50, 51) and microvascular Po₂ (49) are increased by $\text{N}{\text{O}}_3^{-}$ or $\text{N}{\text{O}}_2^{-}$ supplementation especially in muscles or muscle regions comprised of fast-twitch (type II) muscle fibers (28, 53). V̇o₂ is decreased during submaximal exercise (137) via a reduced ATP cost of force production (5) and possibly improved mitochondrial efficiency (136). V̇o₂ kinetics are speeded for exercise recruiting fast-twitch fibers (6, 18), and in vitro muscle contractile function is enhanced (primarily fast-twitch) via improved calcium handling (86), all of which contribute to improved in vivo exercise performance (5, 6, 104, 105). During voluntary exercise, $\text{N}{\text{O}}_3^{-}$ supplementation enhances physical performance more in sedentary or recreationally active individuals than their elite endurance counterparts (review in Ref. 104) making this strategy particularly relevant for HF patients in whom endogenous NO bioavailability is depressed (30, 92, 93; reviews in Refs. 52, 90, 171, 177).

heart failure.

Despite the structural predations of HF such as loss of capillaries in muscle (Fig. 10; review in Ref. 171), exogenous NO application restores the Q̇o₂-to-V̇o₂ balance and microvascular Po₂ profile during muscle contractions in slow- and fast-twitch muscles of HFrEF rats (48, 52, 55). Pharmaceutically targeted enhancement of cGMP function via the phosphodiesterase inhibitor sildenafil raises quadriceps muscle oxygenation (i.e., decreased deoxygenation), speeds pulmonary V̇o₂ kinetics, and improves exercise tolerance in class II/III HFrEF patients (213). Animal studies support that that these effects rely on augmented NO bioavailability increasing both Q̇o₂ and Do₂ in HFrEF patients’ muscles (see Fig. 9, bottom, and Fig. 10, red arrows; Ref. 54; review in Ref. 90). The capacity to increase Q̇o₂ in the exercising musculature coupled with reduction in the O₂ cost of muscle contractions (5, 136, 137) reinforces the potential for $\text{N}{\text{O}}_3^{-}$ supplementation to improve Q̇o₂-to-V̇o₂ matching, V̇o₂ kinetics, V̇o_2peak/max, and exercise capacity in HF patients.

The current prevalence of HFpEF and absence of proven therapeutic options (57, 237, 238) provide great imperatives for resolving mechanisms of dysfunction in HFpEF and exploring putative treatments. Whereas both HFrEF and HFpEF evince markedly reduced maximal cardiac output, fractional O₂ extraction (i.e., a-vO₂ difference) is also reduced in HFpEF whereas it is not in HFrEF (Fig. 9, bottom; Ref. 82; review in Ref. 177). In HFrEF, Do₂ is compromised by microvascular dysfunction that includes cessation of RBC flux in a substantial portion of the capillary bed (Fig. 9, bottom, and Fig. 10; Refs. 111, 195; review in Ref. 177): Whether this occurs in HFpEF and how it may relate to the O₂ extraction deficit are unknown.

Zamani and colleagues (237) employed a single dose of 11 mM $\text{N}{\text{O}}_3^{-}$ (as beet root juice) to improve cardiovascular and muscle function during exercise in class III and IV HFpEF patients. $\text{N}{\text{O}}_3^{-}$ supplementation reduced systemic vascular resistance, elevated cardiac output and increased V̇o_2peak ~10% along with work capacity. Although the $\text{N}{\text{O}}_3^{-}$ treatment may, potentially, have altered myocardial performance it is likely that improved vasodilation within the exercising muscles primarily drove the decrease in systemic vascular resistance and enhanced cardiac output and locomotory muscle Q̇o₂. Whether $\text{N}{\text{O}}_3^{-}$ supplementation increased the proportion of capillaries supporting RBC flux or improved Q̇o₂/V̇o₂ within the exercising muscles remains an important question. As intuitive from Fig. 9, bottom, for HFrEF patients, the presence of unchanged muscle deoxy[Hb + Mb] concomitant with higher V̇o_2max implies a greater DO₂ in HFpEF patients after $\text{N}{\text{O}}_3^{-}$ supplementation (237). More recently, that up to 18 mM/day $\text{N}{\text{O}}_3^{-}$ over 3 wk significantly increases exercise capacity, the economy of exercise (work rate/V̇o₂), and overall quality of life (Kansas City Cardiomyopathy Questionnaire) in HFpEF patients (238) illustrates the pressing need to resolve the underlying mechanisms. It is pertinent that recent investigations in some HFrEF (nitrate; Ref. 91) and HFpEF (nitrite; Ref. 16) patients have not demonstrated improved O₂ transport and exercise capacity with nitrate or nitrite supplementation. Whether patient variability, a facet of concomitant medications, dosing, or some other specific aspect of the study design is responsible for these contrasting findings remains to be determined.

CONCLUSIONS

Valuable theories bring together often disparate findings to explain physiological systems control and help design forward-thinking testable hypotheses. In recent decades, the Kroghian schema of capillary structure and function and the physiological bases for muscle O₂ diffusion have undergone rigorous empirical and theoretical examination. As explored herein, models and technological innovations far beyond those available in previous decades have provided a very different picture of skeletal muscle microcirculatory control. The rapid pulmonary V̇o₂ kinetics observed initially by Krogh and Lindhard (131) and formally characterized by Whipp and Wasserman (234) have been demonstrated in single contracting skeletal muscles where simultaneous capillary hemodynamics and microvascular and interstitial Po₂s and blood-muscle O₂ flux can be measured (10, 11, 115). That most capillaries support RBC flux in healthy resting muscle(s) and that rapid V̇O₂ kinetics can occur following the onset of contractions in the absence of “capillary recruitment” run contrary to current dogma and yet are supported by this and a plethora of evidence in intact animal and human muscles (review in Ref. 171). Contemporary observations of a decreased proportion of RBC-flowing capillaries in muscles of HF animals and the resultant impaired microvascular Po₂s explain mechanistically the slowed V̇o₂ kinetics and exercise intolerance in these and other patient populations (reviews in Refs. 90, 171, 177). Moreover, strategies such as exercise training and augmented NO bioavailability serve to improve capillary hemodynamics and help restore blood-tissue O₂ flux reducing Q̇o₂-to-V̇o₂ mismatching, which speeds V̇o₂ kinetics and enhances exercise tolerance in HF patients (213, 237; review in Ref. 90). However, despite the wealth of evidence to the contrary, some of the latest medical and physiology textbooks (32, 164; review in Ref. 202) retain a tenacious adherence to dated concepts such as capillary recruitment, precapillary sphincters, and intramyocyte O₂ diffusion distances. If, as Thomas Huxley (101a) said, “The great tragedy of science (is) the slaying of a beautiful hypothesis by an ugly fact” how much more tragic is ignoring scientific evidence to preserve “a beautiful hypothesis?” Within the context of rigorous evidence this review has presented the case for integration of contemporary discoveries into our understanding of muscle microcirculatory function and O₂ transport. Let not the jejune “illusion of knowledge” impede scientific progress.

GRANTS

These studies were supported, in part, by National Institutes of Health Grants HL-17731, HL-50306, HL-108328, AG-19228, and AG-11535 and American Heart Association 10 Grant 4350011.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

D.C.P. conceived and designed research, prepared figures, drafted manuscript, edited and revised manuscript, and approved final version of manuscript.