CrossTalk opposing view: De novo capillary recruitment in healthy muscle is not necessary to explain physiological outcomes


Several thousand capillaries are small enough to fit comfortably on the period at the end of this sentence and yet, within skeletal muscle, thousands of miles of capillaries combine to provide a vast surface, some hundreds of square metres, that is essential for blood–myocyte flux of O2, glucose and free fatty acids. As dictated by muscle metabolic demands and a complex array of neural, humoral and local controllers, muscular arterioles function as ‘gatekeepers’, regulating capillary red blood cell (RBC) and plasma flow. The issue considered here is whether or not most capillaries support RBC and/or plasma flux in resting skeletal muscle. If such is the case, this dismisses de novo recruitment of capillaries as a requirement for increased blood–myocyte flux either at rest or during muscle contractions. Resolution of this problem is fundamental to understanding muscle metabolic function in health and the mechanistic bases for dysfunction in diseases such as diabetes, sepsis and heart failure (Lemkes et al. 2012; Poole et al. 2012; Eskens et al. 2013).

The noted historian, Daniel J. Boorstin (Boorstin, 1983), remarked that: ‘The greatest impediment to scientific progress is not ignorance, but the illusion of knowledge.’ In science, construction of models is valuable. Models form the bases for construction and testing of hypotheses and, to a degree, establish the limits of what is known. Science progresses when results differ from expectations and the model must be either elaborated or dismissed. However, when a model becomes entrenched dogmatically and proponents simply ignore the compelling weight of opposing data, progress is impeded. As early as the 13th Century, Roger Bacon's Idola theatri (Idols of the Theatre) cautioned against rigid adherence to convention over free thought and observation (Rees & Wakely, 2004). Seven centuries later, Karl Popper continued this theme by extolling the power of well-designed experiments to refute hypotheses (‘black swan’ approach) as opposed to the simple accumulation of confirmatory data (‘white swan’ approach; Popper, 2002).

In this CrossTalk, the reader is asked to consider the perceived validity of the ‘capillary recruitment’ model, attributed originally to August Krogh (Krogh, 1919a,b), in the face of a plethora of structural, observational and physiological data demonstrating that capillaries in muscle: (1) cannot contract (as originally thought) and possess collagenous struts from the abluminal surface to adjacent myocytes structurally resisting collapse (Borg & Caulfield, 1980); (2) are not controlled individually by any ‘precapillary sphincter’ (Gorczynski et al. 1978; reviewed by Poole et al. 2013); (3) mostly (>80%) support RBC and/or plasma flux at rest and during contractions (e.g. Kindig et al. 1999, 2002); (4) at blood flow rates reported for humans can feasibly sustain an RBC flux of >10 RBCs per capillary s−1, which is close to that observed in rats (Kindig et al. 2002; Poole et al. 2011, 2013); (5) exhibit a wide range of RBC fluxes and velocities at rest and during/following contractions (Klitzman & Duling, 1979; Klitzman et al. 1982; Hudlicka et al. 1982; Kindig et al. 1999, 2002; Fraser et al. 2012); (6) can increase blood–myocyte O2 flux rapidly without recruitment of previously non-flowing capillaries (Klitzman & Duling, 1979; Klitzman et al. 1982; Hudlicka et al. 1982; Behnke et al. 2002; Kindig et al. 2002); (7) do not represent quantum units of exchange but, rather, have rates of substrate delivery dependent on their individual RBC/plasma flux(es) (Poole et al. 2008, 2011, 2013); and (8) have an endothelial surface layer, the modification of which may be crucial for increasing substrate delivery and blood–myocyte flux (reviewed by Eskens et al. 2013).

The strength of any scientific contention rests upon the data and the methods used in their collection. Robust models and theories (Poole et al. 2008, 2011, 2013) garner broad support across a breadth of experimental approaches, which may ideally encompass in situ and in vivo approaches using animals and humans, as appropriate. Pursuant to questioning the validity of the capillary recruitment hypothesis, it is useful to review a few of the observations that support points 1–8 above.

The first group of investigations involves intravital muscle microscopy in anaesthetized animal preparations with intact vascular/neural inflows. These investigations demonstrate that capillaries remain open (non-collapsed) when inflowing mean arterial pressure is reduced from normal (>100 mmHg) to 20–30 mmHg (Kindig & Poole, 1999). The RBC flux in certain capillaries may cease at very low inflowing pressures (e.g. haemorrhagic hypotension, arteriolar vasoconstriction) in the absence of discernible luminal restriction, implying arteriolar rather than capillary control per se.

Following the onset of contractions, RBC fluxes increase in individual capillaries, and no marked elevation in the proportion of flowing capillaries occurs (Burton & Johnson, 1972; Kindig et al. 2002; reviewed by Poole et al. 2011, 2013). Capillary haematocrit is hugely variable among capillaries but averages ∼20% (Klitzman & Duling, 1979; Klitzman et al. 1982; Sarelius & Duling, 1982; Kindig et al. 1999, 2002; reviewed by Poole et al. 2008). This reduction from systemic haematocrit (∼45%) occurs principally because the presence of a capillary endothelial surface layer (sometimes called the glycocalyx) reduces mean plasma velocity far below that of the RBCs (Desjardins & Duling, 1990). As metabolic rate (O2 uptake) and RBC flux/velocity increase with contractions, capillary haematocrit rises towards systemic (Klitzman & Duling, 1979; Kindig et al. 2002; Richardson et al. 2003).

In the absence of de novo capillary recruitment, muscle O2 uptake (math formula) increases without delay, evincing the exponential kinetics profile predicted from pulmonary gas exchange measurements in humans and the similarity of the muscle(s) oxidative capacity between species (Behnke et al. 2002; Kindig et al. 2002). Crucially, the ∼6:1 blood flow: math formula ratio is preserved during these electrically stimulated muscle contractions (Ferreira et al. 2006).

Evidence amassed from investigations using conscious animals also supports points 1–8. Specifically, rats injected intravascularly with fluorescent tracers exhibit staining of essentially all muscle capillaries examined within a single passage of the cardiac output (Kayar & Banchero, 1985; Snyder et al. 1992). This technique cannot determine the presence/absence of RBCs but demonstrates that all capillaries are patent and at least support plasma flow. The presence of complete capillary bed perfusion has also been demonstrated in myocardium of resting anaesthetized rats with an estimated <5 s transit time (Vetterlein et al. 1982).

Observations from investigations in conscious humans are also revealing. Near-infrared spectroscopy measures haemoglobin and myoglobin concentrations ([Hb] and [Mb]) and changes thereof within muscles at rest and during exercise. Given that [Mb] in the sampled region is not expected to change with exercise, this technique provides a sensitive measure of muscle microvascular (mostly capillary) [Hb]. Were there to exist a substantial fraction of the capillary bed (e.g. 80%), not supporting RBC flux (i.e. empty of RBCs) at rest, in order to initiate RBC flux during exercise a severalfold increase in [Hb] would be expected. A plethora of near-infrared spectroscopy studies report that [Hb] increases only ∼10–40% from rest to heavy- or severe-intensity exercise (e.g. Lutjemeier et al. 2008; Davis & Barstow, 2013; reviewed by Poole et al. 2013). This is entirely consistent with the elevation of capillary haematocrit described above but not with de novo recruitment of capillaries en mass.

In the absence of substantial de novo capillary recruitment, there are at least five mechanisms that can account for the increased blood–myocyte O2 flux during contractions. These are as follows: (1) elevated RBC flux, especially in capillaries where RBC flux was very low at rest; (2) increased capillary haematocrit; (3) recruitment of more capillary surface area for O2 flux along the length of already flowing capillaries (so-called ‘longitudinal recruitment’); (4) reduced intramyocyte O2 partial pressure, which decreases the O2 carrier-free zone inside the myocytes and raises the diffusing capacity; and (5) modification of the capillary endothelial surface layer (Eskens et al. 2013; reviewed by Poole et al. 2011, 2013).

The wealth of evidence presented above argues convincingly that de novo capillary recruitment is not requisite to explain physiological behaviour and, indeed, may not occur substantially. In this light, is it appropriate for investigators using very indirect techniques to interpret their data in presumption of its occurrence? Take, for instance, contrast-enhanced ultrasound. This is a relatively recent technique with the advantage that, like near-infrared spectroscopy, it can be used non-invasively in humans. This method relies on small bubbles behaving like RBCs (never proven) and their ‘concentration’ in the capillary bed being independent of flow but determined solely by microvascular volume (Rattigan et al. 1997; Wheatley et al. 2004; Meijer et al. 2012; reviewed by Poole et al. 2011, 2013). This is flawed logic. For instance, as RBC flux increases so does the capillary RBC content (haematocrit; see above). At the crude resolution and low signal-to-noise ratio of this technique (one bubble per 600 or so capillaries; Sjøberg et al. 2011), how then is it possible to separate an increased bubble density due to flux (or other flow-redistribution phenomena) from that resulting from more intravascular volume (i.e. capillary recruitment) as contended? Also, contrast-enhanced ultrasound cannot discern between increased capillary volume due to recruitment of flow in more vessels and a modification of the endothelial surface layer by the action of insulin. Specifically, insulin elevates capillary volume and blood–myocyte glucose flux not by de novo capillary recruitment but by altering the function and thickness of the endothelial surface layer (Eskens et al. 2013). Removal of this layer with hyaluronidase treatment abolishes this effect; a finding that opposes diametrically the conclusions of the actions of insulin resulting from the contrast-enhanced ultrasound technique.

The very motto of our Royal Society, ‘Nullius in verba’ (Take nobody's word for it), establishes a mandate, wherever possible, to observe physical phenomena directly. To rely on dated notions or models when a compelling weight of contrary evidence exists, which is either dismissed cursorily or ignored altogether, impedes the opportunity for correct data interpretation and thus scientific discovery.

Call for comments

Readers are invited to give their views on this and the accompanying CrossTalk articles in this issue by submitting a brief (250 word) comment. Comments may be submitted up to 6 weeks after publication of the article, at which point the discussion will close and the CrossTalk authors will be invited to submit a ‘Last Word’. Please email your comment to

Additional information

Competing interests

None declared.


This work was supported by grants from the National Institutes of Health (HL-50306 & HL-10832) and the American Heart Association.


The assistance, encouragement and expertise of Drs. T.I. Musch, C.A. Kindig, B.J. Behnke, S.W. Copp, L.F. Ferreira, D.M. Hirai, D.J. Padilla, T.J. Barstow, R.N. Pitman, B.R. Duling, O. Hudlicka, and C.G. Ellis is greatly appreciated.