Insulin and contraction increase nutritive blood flow in rat muscle in vivo determined by microdialysis of l-[14C]glucose


Corresponding author J. M. B. Newman: Biochemistry, Medical School, University of Tasmania, Private Bag 58, Hobart 7001, Tasmania, Australia. Email:


In the present study, a mathematical model using the microdialysis outflow: inflow (O/I) ratio of the novel analogue l-[14C]glucose has been developed which allows the calculation of the nutritive (and non-nutritive) flow in muscle as a proportion of total blood flow. Anaesthetized rats had microdialysis probes carrying l-[14C]glucose inserted through a calf muscle group (tibialis/plantaris/gastrocnemius). The nutritive fraction of total blood flow was determined under basal conditions and in response to contraction (electrical field stimulation), insulin (hyperinsulinaemic euglycaemic clamp with 10 mU min−1 kg−1 insulin) or saline control from limb blood flow and the microdialysis O/I ratio of l-[14C]glucose. Both contraction and insulin infusion decreased the O/I ratio of l-[14C]glucose and increased total limb blood flow. Calculations based on mathematical models using l-[14C]glucose O/I and limb blood flow revealed that during basal conditions, the nutritive fraction of total flow was 0.38 ± 0.06, indicating that basal flow was predominantly non-nutritive. Contraction and insulin increased the nutritive fraction to 0.82 ± 0.24 (P < 0.05) and 0.52 ± 0.12 (P < 0.05). Thus the increase in limb blood flow from insulin was fully accommodated by nutritive flow, while contraction increased nutritive flow at the expense of non-nutritive flow. This novel method using microdialysis and the O/I ratio of l-[14C]glucose allows the determination of the nutritive fraction of total flow in muscle as well as the proportion of total flow that may be redistributed in response to contraction and insulin.

Total blood flow to muscle increases markedly during contraction and correlates closely with muscle metabolic requirement (Clifford & Hellsten, 2004). In addition, contraction has been shown to increase the number of perfused capillaries (capillary recruitment) (Rattigan et al. 2005). Insulin also increases microvascular perfusion of muscle in vivo by capillary recruitment (Rattigan et al. 1997; Dawson et al. 2002). This effect to recruit capillaries may be a vital control point for the stimulation of glucose uptake (Vincent et al. 2005) by increasing delivery of insulin and glucose to the myocytes. From our previous studies we have proposed that insulin-mediated capillary recruitment results from increased perfusion of capillaries that in the basal state prior to insulin action are receiving little or no blood flow (Rattigan et al. 2005). It is also our view that increased perfusion of these recruited nutritive capillaries can result either from increased bulk flow or from redistribution of flow from a non-nutritive route. The non-nutritive route is a vascular pathway in muscle that is located in connective tissue, and which has poor nutrient exchange capacity with muscle but may supply nutrients and hormones to attendant interfibrillar adipocytes (Clark et al. 2000a). A vasodilatory action of insulin at terminal arterioles controlling flow into the nutritive capillaries may be the key phenomenon in recruiting flow by lowering the vascular resistance of the nutritive route. Alternatively, insulin's vascular action may involve the additional vasoconstriction of non-nutritive vessels to redirect flow to the nutritive capillaries. One candidate vasoconstrictor for which there is evidence in this role is endothelin-1 (Eringa et al. 2005; Kolka et al. 2005).

The two techniques we have developed to determine capillary recruitment are the metabolism of infused 1-methyl xanthine (Rattigan et al. 1997; Zhang et al. 2003) and contrast enhanced ultrasound (Coggins et al. 2001; Clark et al. 2003). However, neither method allows an estimation of the proportion of total flow that is nutritive (or non-nutritive) and thus they cannot provide information on the redistribution of flow that may occur within muscle. Such information may be particularly important as the non-nutritive route in muscle has been proposed to predominate in insulin resistant muscle (Clark et al. 2000a) and to supply nutrient and hormones to attendant interfibrillar adipocytes (Clark et al. 2000b; Yudkin et al. 2005) that form the basis of marbling of muscle.

The microdialysis technique has been in use for several decades for the determination of the interstitial concentrations of biologically important compounds in a variety of tissues (Chaurasia, 1999). This generally requires a knowledge of the recovery of the compound of interest across the microdialysis probe (Henriksson, 1999). In addition, the technique has been applied to the monitoring of tissue microvascular blood flow (Arner, 1999). This second technique involves the addition of a non-metabolized compound to the inflow solution of the microdialysis probe and monitoring the concentration in the outflowing dialysate. The outflow: inflow ratio (O/I) has been shown to vary inversely with blood flow (Hickner et al. 1995; Stallknecht et al. 1999; Newman et al. 2001). The microdialysis recovery (R) of a compound is defined as:


where Cout, Cin and C are the concentrations in the outflowing dialysate, inflowing solution and the interstitial fluid at a distance that is unaffected by the presence of the probe, respectively. From this equation, it is evident that knowledge of both recovery and interstitial concentration simultaneously is not possible. The simplest solution is to estimate the recovery by the addition of a marker for the compound of interest to the inflow solution which is not present in the interstitial fluid (Lönnroth & Strindberg, 1995). In this case, the recovery is also referred to as relative loss and eqn (1) reduces to:


where I and O are the marker concentrations in the inflow solution and the outflowing dialysate, respectively. The simple relationship between the recovery (or relative loss) and O/I ratio of an exogenous compound is readily apparent. Mathematical models for recovery or O/I ratio have been developed, although most have been concerned with recovery (Bungay et al. 1990; Morrison et al. 1991; Stahle, 2000). Recently, we have developed a mathematical model of O/I ratio of non-metabolized compounds that takes into account blood flow, as well as the nutritive proportion of blood flow (Roberts et al. 2005). This steady-state model is based on a cylindrical probe geometry and assumes that diffusion from the probe is only in the radial (r) direction, with none in the axial direction (z) and that the tracer is not metabolized as shown in Fig. 1 (Roberts et al. 2005). This means that there is an implied assumption that the interstitial space is an unstirred compartment. Considering a differential length of probe (dz), a mass flux (M) leaves this probe volume to the outside surface of the probe membrane. At steady state this mass flux must be matched by a mass flux from the interstitial fluid into surrounding capillaries. From the equations of Roberts et al. (2005), it can be shown that:


where Sp, qp and Pp are the probe surface area, flow rate and permeability, respectively, and Pt (the permeability of the tissue) is:


where rp is the probe radius, Qb is the total blood flow around the probe of which γ is the nutritive fraction, Dt is the diffusion coefficient of the tissue and K0 and K1 are modified Bessel functions. Thus, knowledge of the diffusion coefficient and total blood flow around the microdialysis probe allows for the determination of the nutritive fraction of blood flow in the tissue.

Figure 1.

Schematic representation of compartments and fluxes of the microdialysis tissue system
A volume of dialysate of inflowing concentration I moves along the probe in the axial (z) direction with a flow rate of qp. A differential dialysate volume of thickness dz loses a mass (M) of solute across the probe membrane, leaving an outflowing concentration of O. This mass flux is matched (at steady-state) by a mass flux of M into the adjacent blood vessels. Movement through the probe and interstitial fluid is assumed to be solely via diffusion in the radial (r) direction.

Given the importance of knowing the proportion of total blood flow to muscle that is nutritive and non-nutritive, as well as the redistribution of flow that might occur during contraction and under physiological or pathophysiological conditions, we now describe a method based on microdialysis measurements that can be used to address these issues.


Ethical approval

Experiments were conducted on 240–260 g Hooded Wistar rats raised on a commercial diet (Pivot, Launceston, Australia: 21.4% protein, 4.6% lipid, 68% carbohydrate, 6% crude fibre, with added vitamins and minerals) and water ad libitum, housed at 21 ± 1°C with a 12 h–12 h day–night cycle. All experiments were approved by the University of Tasmania Animal Ethics Committee, with animals cared for in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (7th edn; Australian Government Printing Services, Canberra, Australia, 2004).

Microdialysis probe construction

Linear microdialysis probes were constructed by filing the points of two 23-gauge Terumo syringe needles and removing the syringe adaptors. A short length (25 mm) of microdialysis tubing (Bioanalytical Systems; molecular weight cutoff 30 kDa; outer diameter 320 μm) was inserted into the filed ends of the needles so that 10–12 mm of tubing was exposed between each needle and glued in place using Araldite tm.


Anaesthetized rats were used so that direct comparisons between insulin and muscle contraction could be conducted and so that the stimulated test leg could be compared with the contra-lateral control leg for those animals undergoing electrical stimulation. Also, accurate measurement of limb blood flow by positioned flow probe was essential in all animals.

Rats were anaesthetized using Nembutal (50 mg (kg body weight)−1i.p.) and placed on a heated (37°C) water-jacketed platform under a heating lamp. A tracheotomy tube was inserted and the rat allowed to breathe room air throughout the experiment. A polyethylene cannula (PE-50; Intramedic, Becton Dickinson, Parsippany, NJ, USA) was implanted into the left carotid artery for arterial sampling and blood pressure measurement (pressure transducer Transpac IV: Abbott Critical Care Systems, Morgan Hill, CA, USA). Both jugular veins were cannulated for the delivery of anaesthetic and intravenous infusions. Small (1.5 cm) incisions were made in the skin overlaying the femoral vessels of both legs and the femoral artery was separated from the femoral vein and saphenous nerve. A small area of skin over the gastrocnemius and tibialis muscles of the right leg was removed and a microdialysis probe was inserted through these muscles. This was achieved by first inserting an 18-gauge cannula needle through the muscles. One of the 23-gauge needle ends of the probe was inserted into the emergent end of the 18-gauge needle, which was carefully withdrawn until the microdialysis membrane was within the muscle. The final orientation of the probe was transverse with respect to the muscle fibres. The probe was attached to a syringe pump via a length of PE-50 polyethylene tubing and flushed with 200 μl of the inflow solution, before the pump was set at 2 μl min−1 for the course of the experiment. The inflow solution consisted of isotonic saline (0.9% NaCl) supplemented with 6 mm glucose and 3 mm lactate and either 3000 Bq ml−1d-[14C]glucose or 3000 Bq ml−1l-[14C]glucose. For some experiments a combination of 15000 Bq ml−1d-[3H]glucose and 3000 Bq ml−1l-[14C]glucose was used.

An ultrasonic flow probe (VB series 0.5 mm; Transonic Systems, Ithaca, NY, USA) connected to a flow meter (T106 ultrasonic volume flow meter; Transonic Systems, New York, USA) was positioned around the femoral artery of the right leg just distal to the rectus abdominis muscle and the cavity containing the probe was filled with conductive gel (Medical Equipment Services; Richmond, Vic, Australia). Blood pressure and femoral blood flow (FBF) were recorded at a sampling frequency of 100 Hz via Windaq data acquisition software (Dataq Instruments; Akron, OH, USA) running on a PC. Anaesthesia was maintained by Nembutal infusion (0.6 mg min−1 (kg body weight)−1) via the left jugular vein.

For the contraction studies, a microdialysis probe was inserted into each hindleg as well as an ultrasonic flow probe positioned around each femoral artery. In addition, a small section of skin above the knee of one leg was removed and a hook electrode was placed through the fatty tissue in this region and a second electrode underneath the achilles tendon. These were attached to an electrical stimulator (Hugo Sachs Elekronik; March–Hugstetten, Germany).


Protocols for both contraction and insulin or saline infusions are shown in Fig. 2. Blood pressure and femoral blood flow were allowed to stabilize for at least 40 min before collecting any samples or commencing any infusions. Following the collection of a 20 min baseline microdialysis sample, either 10 mU min−1 kg−1 insulin euglycaemic clamp or field stimulation was commenced. In the case of the euglycaemic clamp, 30% glucose was also infused to maintain blood glucose at the preclamp levels and 1 h was allowed to elapse to allow for the stabilization of glucose infusion rate and femoral blood flow before a second 20 min microdialysis sample was collected (Fig. 2A). For the contraction experiments, two intensities of electrical stimulation (10 V and 20 V) of 40 min duration were conducted using 0.1 ms pulses every 2 s. Microdialysis samples were collected during the final 20 min of each stimulation intensity (Fig. 2C). Arterial blood samples were taken from the carotid artery at times indicated in Fig. 2. A femoral vein sample was taken at the end of experiments with an ultra-fine insulin syringe (Becton, Dickinson and Company, Franklin Lakes, NJ, USA).

Figure 2.

Hindleg haemodynamic changes
A and B, rats treated with saline (○, n= 15) or 10 mU min−1 kg−1 insulin (•, n= 17) from time t= 0 as described in Methods. C and D, rats subjected to muscle contraction by field stimulation (FS) at 10 V and 20 V (control leg, □; stimulated leg, ▪, n= 11). Black bars indicate periods over which microdialysis samples were collected. Femoral blood flow (FBF) was determined from 5 s subsamples, while vascular resistance was calculated by dividing blood pressure by femoral blood flow. Arterial blood samples were taken at the same time points as symbols for FBF and vascular resistance. Values are means ±s.e.m.*P < 0.05 versus corresponding time point for saline or control leg.

Analytical methods

Arterial and venous blood samples (25 μl) were assayed for glucose using the glucose oxidase method on a glucose analyser (2300 Stat plus; Yellow Springs Instruments, Yellow Springs, OH, USA). Arterial plasma glucose was also determined at the end of each 20 min microdialysis collection.

Each 20 min microdialysis sample was collected into microcentrifuge tubes to minimize evaporative errors and immediately assayed for glucose (25 μl). From experiments using d-[3H]glucose, the remainder was weighed into 4 ml scintillation tubes and freeze-dried (to remove possible contaminating 3H2O due to direct transfer of the 3H label between glucose and water), then reconstituted in 100 μl of distilled water and 3 ml of scintillant added. The samples from experiments using the other radioactive analogues were weighed directly into a 4 ml scintillation vial to which was immediately added 3 ml of scintillant. In addition, a sample of the inflow syringe solution was assayed for glucose and 15 μl was also weighed into a 4 ml scintillation vial containing 3 ml scintillant (again with a freeze-drying step if using d-[3H]glucose). All vials were counted on a Liquid Scintillation Analyser (Tri-Carb 2800TR; Perkin Elmer, Downers Grove, IL, USA) using a single label or dual label program with crossover correction as required.

The interstitial concentration of glucose was determined in experiments with d-glucose using the internal reference technique (Lönnroth & Strindberg, 1995). The glucose recovery was determined from radioactive d-glucose using eqn (2), which could then be used in eqn (1) to determine interstitial glucose concentration.

Determination of muscle diffusion coefficient

In order to calculate the nutritive fraction of total flow, it was necessary to first estimate the diffusion coefficient for l-[14C]glucose as discussed in the Appendix. To our knowledge there are no reports of the diffusion coefficient of glucose in mammalian muscle. Diffusion coefficient values for glucose have been reported in other tissues (Bashkatov et al. 2003; Kauri et al. 2003; Khalil et al. 2006). It was possible to use an average value from these studies (∼3 × 10−10 m2 s−1) as an estimate of Dt. However, a 10-fold range of water diffusion coefficients between various tissues has been reported (Johnson & Maki, 1991) and it would seem reasonable to assume that a similar effect would occur with glucose. Microdialysis has been used in the past to determine brain diffusion coefficient of ethanol (Gonzales et al. 1998) and mannitol (Chen et al. 2002). The latter study used delivery and receiving probes separated by 1 mm. Diffusion coefficient can be determined from the transient concentration profiles in the receiving probe (Chen et al. 2002). This requires the use of specialized stereotactic equipment to ensure a known distance between probes. Thus, a method for the determination of the diffusion coefficient of ethanol in brain (Gonzales et al. 1998) using our single microdialysis system was tailored for the estimation of muscle Dt for l-[14C]glucose in the current study using a mathematical model we have developed (Roberts et al. 2005) (see Appendix). A by-product of this was the direct measurement of the overall rate constant responsible for the removal of l-[14C]glucose from the probe via the interstitial fluid.

At the end of some experiments using l-[14C]glucose, immediately following killing, the leg (or legs for contraction experiments) containing the microdialysis probe was severed above the knee, immersed in liquid nitrogen to prevent any further diffusion of the tracer, and kept at −20°C until analysed. A muscle sample (approximately 500 mg) was also taken from the thigh, well away from the probes. The lower leg was allowed to thaw sufficiently so that the two needle ends of the probes could be cut off and a section of muscle (∼500 mg) containing the probe in situ excised. Muscle samples were weighed and dissolved in 4 ml of Soluene® at 55°C before adding 200 μl glacial acetic acid (to prevent chemiluminescence) and 17 ml scintillant and counted for radioactivity. The amount of radioactivity in the muscle surrounding the probe (mt) was determined by correcting the total measured amount in the muscle sample for the amount of radioactivity remaining within the probe and the radioactivity in the muscle due to distribution of the tracer around the body over the course of the experiment (thigh muscle sample). The amount of radioactivity remaining in the probe was at least 20% of the total in the muscle sample and has a significant impact on the calculated diffusion coefficient. On the other hand the radioactivity due to distribution around the body was approximately 2%. Thus, although neglecting the radioactivity due to distribution around the body will have little impact on the calculated diffusion coefficient, we have retained it in our calculations for the sake of completeness. Tracer delivery rate inline image was calculated from the O/I ratio at the end of the experiment. The overall muscle rate constant (kt) for l-[14C]glucose was then determined from eqn (A5) (Appendix), assuming an equilibrium partition coefficient (Φt) of 0.2. Equation (A8) was then used to calculate the muscle permeability (Pt) – utilizing the value for Pp determined previously (see Appendix). These two parameters and parameters in Table 1 were used to iteratively solve eqn (A7) for the muscle diffusion coefficient (Dt).

Table 1.  Microdialysis parameters used for determination of muscle diffusion coefficient Dt and nutritive fraction (γ)
  1. Values for L and qp were set during the construction of probes, qp was set at 2 μl min−1 and rp taken from the manufacturer's literature. Probe permeability was determined as described in Methods.

Probe lengthL0.010–0.013m
Probe radiusrp1.6 × 10−4m
Probe permeabilityPp3.33 × 10−6m s−1
Probe flow rateqp3.33 × 10−11m3 s−1
Partition coefficientΦt0.2 

Determination of muscle nutritive fraction

A value for the total blood flow to the muscle surrounding the probe was determined from femoral blood flow data assuming that 8 g of hindlimb muscle received 75% of the measured FBF (Raitakari et al. 1996) and that this was unchanged for both insulin and muscle contraction. The combined average diffusion coefficient and other parameters from Table 1 were used to iteratively solve eqn (A1) for a0 and hence determine the nutritive fraction (γ) from eqn (A3). Absolute nutritive flow was then determined as (FBF × 0.75 × nutritive fraction), while absolute non-nutritive flow was calculated as the difference between total flow (FBF × 0.75) and nutritive flow.

Data analysis

Mean arterial blood pressure and femoral blood flow were determined as the average of 5 s subsamples of the Windaq recording taken just prior to arterial blood sampling. Vascular resistance was calculated by dividing the blood pressure by the femoral blood flow and expressed as resistance units (RU). Heart rate was determined within these 5 s subsamples using the discrete Fourier transform in the data acquisition package. Glucose uptake across the hindlegs (μmol min−1) was calculated as the product of the arterio-venous concentration difference (mm) and FBF (ml min−1) at the end of experiments. Glucose infusion rate (μmol min−1 kg−1) was determined from the infusion rate of a 30% glucose solution and the rat mass.

Statistical analysis

Data are presented as means ±s.e.m. Two-way repeated measures analysis of variance (ANOVA) was used to test the hypothesis that there was no difference among treatment groups. Student's t test (paired or unpaired) was used to analyse glucose uptake and overall rate constant data. Significance was recognized at the P < 0.05 level with Student–Newman–Keuls post hoc test applied to ANOVA tests.


Systemic and hindleg effects

The average heart rate from all experiments during the baseline 20 min period was 343 ± 6 beats min−1 (n= 43). This did not change significantly in response to saline (363 ± 12 beats min−1, n= 15), insulin (346 ± 6 beats min−1, n= 17) or contraction of one leg (358 ± 11 beats min−1, n= 11). Similarly, blood pressure remained constant (baseline, 106 ± 1; saline mmHg, 105 ± 3 mmHg; insulin, 104 ± 2 mmHg; contraction, 107 ± 1 mmHg). Figure 2 shows the femoral blood flow (FBF) and vascular resistance. Femoral blood flow was significantly (P < 0.05) increased and vascular resistance significantly (P < 0.05) decreased by insulin and 20 V contraction (Fig. 2).

Glucose homeostasis data are shown in Fig. 3. Blood glucose was maintained at the baseline level during insulin treatment by glucose infusion at a rate that was essentially constant by 30 min (Fig. 3). Glucose uptake across the hindleg at the end of experiments was significantly (P < 0.05) increased by insulin and contraction.

Figure 3.

Hindleg glucose uptake, blood glucose and glucose infusion to maintain blood glucose during hyperinsulinaemia
Rats treated with saline (○, n= 14) or insulin (•, n= 16) or subjected to muscle contraction by field stimulation (▾, n= 11) as described in Methods. Glucose uptake across the hindleg as determined by arterio-venous difference. Values are means ±s.e.m.*P < 0.05 versus corresponding saline group or control leg.


The in vivo microdialysis recovery for radiolabelled d-glucoses was determined using eqn (2). The in vivo recovery of d-[3H]glucose was not significantly different from d-[14C]glucose, so the data for both labels were combined. Insulin and contraction significantly increased (P < 0.05) the microdialysis recovery of d-glucose (Fig. 4A and C, respectively). Using eqn (1) and the previously determined d-glucose recovery of radiolabelled glucose, it was found that the calculated interstitial glucose concentration was unaffected by any treatment (7.60 ± 0.20 mm (n= 8) for saline; 7.35 ± 0.47 mm (n= 8) for insulin and 7.64 ± 0.39 mm (n= 6) for contraction). All values were, however, significantly lower (P < 0.05) than the average arterial plasma glucose concentration from the corresponding set of experiments: 8.42 ± 0.20 mm (n= 22).

Figure 4.

Microdialysis recovery and O/I ratio
Microdialysis samples (20 min) were taken before (baseline) and then during saline, insulin or muscle contraction at 10 V and 20 V field stimulation (FS). A, recovery (from eqn (2)) of radiolabelled d-glucose for rats treated with saline (n= 9) or insulin (n= 8). B, the microdialysis outflow: inflow (O/I) ratio of l-[14C]glucose for rats treated with saline (n= 8) or insulin (n= 11). C, recovery of radiolabelled d-glucose for rats subjected to contraction (n= 7). D, microdialysis O/I ratio of l-[14C]glucose for rats subjected to contraction (n= 9). *P < 0.05 versus corresponding baseline.

Figure 4 also shows the effect of saline, insulin and contraction on the O/I ratio of l-[14C]glucose. The O/I ratio for l-[14C]glucose was significantly (P < 0.05) decreased by both insulin (Fig. 4B) and contraction (Fig. 4D).

Muscle diffusion coefficient

Table 1 shows the microdialysis parameters used to determine muscle diffusion coefficient and nutritive fraction. The overall muscle rate constant (kt) and the calculated muscle diffusion coefficient (Dt) for l-[14C]glucose are shown in Table 2. Insulin and contraction significantly (P < 0.05) increased kt, but had no effect on Dt. Thus, a combined value for Dt was used to calculate nutritive fraction.

Table 2.  Measured muscle overall rate constant (kt) and diffusion coefficient (Dt) for l-[14C]glucose
 Muscle overall rate constant (l-[14C]glucose) (kt; min−1)Muscle diffusion coefficient (l-[14C]glucose) (Dt; × 10−9m2 s−1)
  1. Rate constant and diffusion coefficient were determined as detailed in Methods and Appendix. *P < 0.05 versus saline, **P < 0.01 versus control leg.

Saline (n= 8)0.032 ± 0.0050.25 ± 0.02
Insulin (n= 8) 0.045 ± 0.006*0.22 ± 0.02
Control leg (n= 7)0.028 ± 0.0040.21 ± 0.01
Contraction leg (n= 7)  0.056 ± 0.008**0.23 ± 0.03
Combined (n= 30)0.23 ± 0.01

Nutritive fraction

Figure 5 shows the calculated nutritive fraction (γ) and absolute values for nutritive flow and non-nutritive flow based on l-[14C]glucose O/I ratio data. It was found that insulin and contraction significantly (P < 0.05) increased nutritive fraction by 42% and 90%, respectively, over baseline (Fig. 5). Contraction at 20 V significantly (P < 0.05) decreased the absolute non-nutritive flow.

Figure 5.

Nutritive fraction of total blood flow and absolute nutritive and non-nutritive flow
Rats were treated with saline or insulin or subjected to muscle contraction (FS) as described in Methods. Nutritive fraction (A and C) was determined using eqn (A1) from FBF and l-[14C]glucose O/I ratio data as well as the calculated diffusion coefficient and other parameters from Table 1. Absolute nutritive and non-nutritive flow (B and D) was determined from FBF and nutritive fraction assuming 75% of FBF supplied muscle of the hindleg. *P < 0.05 versus corresponding baseline value.


The present study represents the first to determine the proportion of total limb blood flow in vivo that is nutritive under basal conditions and as a result of a hyperinsulinaemic clamp or muscle contraction.

Calculations using l-[14C]glucose O/I revealed that during basal conditions, the nutritive fraction of total flow was 0.38 ± 0.06, indicative that under these conditions before the infusion of insulin approx. 62% of the total flow was passing through a vascular route that had little or no exchange with the interstitial fluid surrounding the muscle fibres (the ‘non-nutritive’ route). Insulin at the high physiological dose used increased FBF from 1.2 to 1.8 ml min−1 and based on calculations using l-[14C]glucose O/I resulted in an increase in the nutritive fraction to 0.52 ± 0.12 (P < 0.05). Muscle contraction increased FBF to 2.0 ml min−1 and the nutritive fraction to 0.81 ± 0.14 (P < 0.05). Calculations of the absolute nutritive flow rates (Fig. 5) show that insulin and contraction increased nutritive flow by 113% and 250%, respectively. Conversely insulin did not affect the absolute ‘non-nutritive’ flow rate, but contraction significantly (P < 0.05) reduced it. These data indicate that there was a preferential redirection of flow to the nutritive route as a result of insulin action, which is consistent with the known vasodilatory action of this hormone (Rattigan et al. 1997). In addition to a preferential redirection of flow to the nutritive route, contraction seems to have actively reduced flow to the ‘non-nutritive’ route, suggestive of an accompanying vasoconstriction.

The present finding that 62% of total flow under basal condition is ‘non-nutritive’ underscores the potential importance of the so-called ‘non-nutritive’ (Grant & Wright, 1970) blood flow in resting muscle, particularly since these vessels may be those that provide nutrients and hormones to the interfibrillar adipocytes (Barlow et al. 1961; Grant & Wright, 1970; Newman et al. 1997; Clark et al. 2000a) and constitute marbling of meat. Work in our laboratory has provided evidence for the existence of these two routes, nutritive and ‘non-nutritive’ (Clark et al. 1995, 2000b, 2003) consistent with earlier research by others (Renkin, 1955; Hyman et al. 1959; Barlow et al. 1961; Grant & Wright, 1970). For example, in the isolated pump-perfused rat hindlimb set at constant flow, some vasoconstrictors such as angiotensin II act to increase general metabolism and aerobic muscle performance whereas others such as serotonin decrease general metabolism and aerobic muscle performance (Clark et al. 1995). These metabolic changes were shown to be dependent on changes in the redistribution of flow and not due to direct effects on muscle since vasodilators reversed both positive and negative metabolic effects (Rattigan et al. 1993, 1995). In addition, the two vascular flow routes were found to be dependent on capillaries for final passage as 15 μm microspheres were unable to transit when flow was either predominantly nutritive or non-nutritive. On this basis it was concluded that the ‘non-nutritive’ route could not involve large direct arteriovenous shunts that have been found in other tissues (Newman et al. 1996). From a determination of the distribution of labelled 15 μm microspheres it was also noted that the vasoconstrictors that increased or decreased nutritive flow did not cause a redistribution of blood flow between different muscle types or between muscle and non-muscle tissue. This indicated that the ‘non-nutritive’ and nutritive routes were located within close proximity to each other and within each muscle (Clark et al. 2000b).

It must be emphasized that in our previous reports of insulin-mediated capillary recruitment in muscle in vivo either 1-MX metabolism (Rattigan et al. 1997; Zhang et al. 2004) or contrast enhanced ultrasound (Vincent et al. 2004; Zhang et al. 2004) was employed and comparisons before and after insulin were used to assess change in capillary recruitment. Neither method permits an assessment of the proportion of total blood flow that is nutritive. Thus the present adaptation of microdialysis to achieve this purpose is novel. Earlier application of microdialysis in muscle has been to monitor the interstitial concentrations of biologically important compounds (Chaurasia, 1999; Henriksson, 1999). In addition, it has been used to monitor the blood flow in muscle by the outflow: inflow (O/I) ratio technique (Arner, 1999). This involves the addition of a non-metabolized compound to the inflow solution, usually ethanol, either unlabelled or labelled with 14C (Hickner et al. 1995; Henriksson, 1999), although 3H2O is in regular use (Stallknecht et al. 1999; Rosendal et al. 2004; Ashina et al. 2005) and even the antibiotic gentamycin has been employed (Mand'ak et al. 2004). Changes in blood flow rate lead to alterations in the rate of tracer removal from the interstitial fluid and subsequently decrease the O/I ratio. It has been shown that the O/I ratio is inversely related to blood flow (Hickner et al. 1995; Stallknecht et al. 1999). Indeed, the O/I ratio has been used as a surrogate indicator of total blood flow changes in exercise (Hickner et al. 1994) as well as during hyperinsulinaemia (Rosdahl et al. 1998), even though a change in the O/I ratio must clearly only result from a change in local flow around the microdialysis probe. Based on data from the current study, we propose that l-glucose is superior compared to other freely diffusible compounds for O/I ratio studies due to its relatively small molecular weight and non-evaporative properties. In addition, the fact that it is actively excluded from mammalian cells may be an advantage.

Use of mathematical models in conjunction with microdialysis is not new and there have been a number of mathematical models developed to describe movement of compounds across microdialysis membranes (Bungay et al. 1990; Morrison et al. 1991; Stahle, 2000). The earlier works focused on modelling the recovery of compounds in microdialysis experiments in brain (Kehr, 1993). Wallgren et al. (1995) developed a model for the O/I ratio of microdialysis and this model has been used to quantify total blood flow (Hickner et al. 1997). However, none of these models takes into account the nutritive fraction of blood flow, which may only apply to skeletal muscle. We have recently developed a mathematical model which uses the O/I ratio from microdialysis of a freely diffusible substrate, taking into account the nutritive fraction, and applied it to the isolated perfused rat hindlimb muscle. The nutritive fraction of total flow was determined during infusion of the vasoconstrictors, noradrenaline or serotonin (Roberts et al. 2005), which have been shown previously to either increase or decrease nutritive flow, respectively (Clark et al. 1995).

The geometry of the tissue through which the microdialysis probe passes is an important factor determining tracer removal from the probe and its subsequent distribution in the tissue. Indeed, a tortuosity term has been introduced in some mathematical models to account for some of this tissue geometry (Bungay et al. 1990). In our model the nutritive fraction could be considered to replace the tortuosity term, since changes in either will effectively alter the diffusive path length of the tracer in order to reach a capillary. Although we have not attempted to experimentally determine the diffusive distance of l-[14C]glucose in muscle, it is possible to estimate this using the model. Based on eqn (10) from our previously published model (Roberts et al. 2005), the concentration at radius r in the tissue is proportional to K0[ra], where aQb/Dt and K0 is a modified Bessel function. Using values of γ= 0.3, Qb= 0.0025 s−1 and Dt= 2.3 × 10−10 m2 s−1, then greater than 99% of the l-[14C]glucose is within 2.2 mm of the probe surface which corresponds to a tissue volume of about 175 μl. The volume of muscle sampled to determine radioactivity and hence the diffusion coefficient was greater than 500 μl (effectively a 4 mm radius around the probe). Thus there can be a high degree of confidence that all radioactivity originating from the probe was sampled. This distribution distance compared well with other published data. The distribution of ethanol in brain tissue delivered by microdialysis has been show to be limited to within approximately 2 mm of the probe (Gonzales et al. 1998). In addition, others have monitored the movement of sodium fluorescein from one microdialysis probe into a second probe 1 mm away in skin (Clough et al. 2002). These authors showed that the external concentration sampled by the retrieval probe ranged from 1 to 5% of the external concentration around the delivery probe.

We have assumed that the interstitial fluid is not a well-stirred compartment. A well-stirred compartment (such as that required for the determination of probe permeability in a saline solution) means that the concentration of tracer at the outside surface of the probe is identical to the concentration at all radial distances from the probe. This negates the requirement for a tissue permeability term and in vitro and in vivo O/I ratios would be expected to be the same (which they are known not to be). What may be more relevant to the present study is whether the interstitial space is in fact partially stirred. However, bulk movement of the interstitial fluid away from the probe would in fact be incorporated into the diffusion coefficient term. In addition, any interstitial fluid movement away from the probe is likely to be matched by a similar movement towards the probe from the opposite side of the probe. Finally, one group has studied the lymph drainage of injected 99mTc labelled human immunoglobulin into human forearms (Stanton et al. 2003). These authors found that the distribution volume (i.e. the interstitial fluid) drainage was 0.138% per minute. This corresponds to an effective interstitial fluid flow rate of 2.3 × 10−5 ml s−1 ml−1 interstitial fluid which is of the order of 1/100th the blood flow rate during baseline conditions.

In the current study, we have used the contralateral leg as a control for the field stimulated leg. During exercise, there is an increase in muscle sympathetic nerve activity, which is capable of changing the blood flow to non-working limbs (Fisher et al. 2005), thus invalidating them as a control for the working limb. Others have shown there may in fact be both vasoconstriction and vasodilatation in a non-contracting limb during at least 30% of voluntary maximal exercise (Fisher & White, 2003; Joyner & Dietz, 2003). In the current study, the field stimulation was of a sufficiently low intensity for there to be very little movement of the stimulated leg (to ensure the microdialysis probe remained intact). Since there were no systemic effects (such as heart rate and blood pressure changes), it seems reasonable to assume that there were no changes in muscle sympathetic nerve activity and hence the contralateral leg should be a valid control.

Based on what we have noted with vasoconstrictors in the isolated perfused rat hindlimb and using labelled 15 μm microspheres (Clark et al. 2000b) the bulk distribution of blood flow between muscles or between muscle and non-muscle tissues does not change with insulin and contraction in vivo. In addition, given the assumption that 75% of the femoral blood flow goes to muscle during the baseline period then muscle blood flow is approximately 0.9 ml min−1. Even if the entire increase in femoral blood flow (0.6 ml min−1) was directed to muscle, then the insulin stimulated muscle blood flow would be approximately 1.5 ml min−1. However, to account for the decreased O/I ratio observed during insulin, it would be necessary for the FBF to be greater than 2.5 ml min−1, whereas in fact it only reached approximately 1.8 ml min−1 (Fig. 2). In a similar manner, calculations suggest that the FBF would need to be 2.1 ml min−1 and 4.7 ml min−1 for contraction by 10 V and 20 V field stimulation, respectively, for there to be no change in nutritive fraction. Since insulin was delivered systemically, it seems reasonable to assume that there was no change in the proportion of femoral blood flow supplying muscle. On the other hand there is the possibility that field stimulation has altered this proportion. Due to the nature of the field stimulation, the majority of muscle supplied by the femoral artery was undergoing contraction. Although precise calculations are not possible, for there to be no change in the nutritive fraction during field stimulation, there would need to be essentially a complete shutdown of blood flow to the non-working muscles and its redirection to the stimulated muscles. In addition, the model predicts that providing redistribution of total flow between muscle and non-muscle tissue during contraction is minor, a significant change to the data is unlikely.

To our knowledge there is only one other paper that has used l-glucose in a microdialysis study and was focused on the differing recoveries of l- and d-glucose when insulin was introduced via the microdialysis probe itself (MacLean et al. 2001). The authors found that there was no difference between the two isomers during baseline conditions, but that d-glucose recovery increased with insulin. In this situation, insulin would be expected to exclusively affect the skeletal muscle immediately surrounding the probe, since it is unlikely to obtain a sufficient concentration in the surrounding vasculature to elicit a blood flow response. In the current study, the O/I ratio for l-glucose may be converted into recovery using eqn (2) giving values of 0.181 ± 0.016 (n= 8) during saline conditions and 0.200 ± 0.011 (n= 11) during insulin treatment. These compare to d-glucose recovery values of 0.196 ± 0.008 (n= 9) during saline and 0.222 ± 0.013 (n= 9) during insulin treatment (Fig. 4). It would be expected that the recovery of d-glucose would increase to a greater extent compared to l-glucose during insulin treatment due to the increase in glucose uptake (Fig. 3). This study was statistically powered to detect a difference in the O/I ratio of l-[14C]glucose between insulin and saline and between contracting and non-contracting legs. It was not intended to compare the microdialysis recovery of d- and l-glucose. Although subsequent studies could be designed to make this comparison, the use of l-glucose was for its non-metabolizable and extracellular nature not the fact that it is a stereoisomer of d-glucose. Indeed there is likely to be other non-metabolized extracellular compounds of a similar molecular weight to glucose that could be used instead of l-glucose.

In conclusion, this study shows that the O/I ratio of l-[14C]glucose is a novel analogue for monitoring microvascular blood flow in muscle in vivo. Use of a mathematical model, now allows for the quantification of the nutritive fraction of total flow in the anaesthetized rat. A hyperinsulinaemic (10 mU min−1 kg−1) euglycaemic clamp and muscle contraction increased the nutritive fraction from 0.38 to 0.52 and 0.81, respectively, when calculated using the microdialysis O/I ratio of the non-metabolized stereoisomer l-[14C]glucose. This study therefore paves the way for studies in conscious instrumented animals as a means of monitoring microvascular blood flow and calculating the nutritive proportion in physiological and pathophysiological conditions such as diabetes.


We have previously shown (Roberts et al. 2005) that the equation relating O/I ratio and the nutritive proportion of total microvascular flow in the vicinity of a microdialysis probe is given by:


In eqn (A1)L is probe length (m), Pp is robe permeability (m s−1), rp is probe radius (m), qp is flow rate through the probe (m3 s−1), Dt is the diffusion coefficient in tissue (m2 s−1), K0 and K1 are modified Bessel functions and


where Qb (ml blood s−1 (ml muscle)−1= s−1) is the total blood flow to the tissue surrounding the probe, of which γ is the nutritive fraction. Equation (A2) is valid for tracers that are freely diffusible throughout the tissue. In general, however, it is necessary to include a term for the equilibrium partition coefficient (Φt) (Gonzales et al. 1998), which is the tracer tissue: interstitial concentration ratio, such that:


It is initially necessary to estimate the intrinsic probe permeability (Pp) and muscle diffusion coefficient (Dt). In a well stirred aqueous solution, the term in parentheses in eqn (A1) reduce to unity (due to the Qb input being effectively infinite; Roberts et al. 2005). The probe permeability can then be determined since eqn (A1) reduces to:


A microdialysis probe of known exchange length was immersed in a continuously well-stirred saline and 5 mm glucose solution maintained at 37°C. The flow rate was set at 2 μl min−1 with the inflow solution containing 0.9% NaCl, 5 mm glucose and 3000 Bq ml−1l-[14C]glucose. Probe permeability (Pp) was determined from the O/I ratio of l-[14C]glucose (eqn (A4)).

The estimation of Dt is done by the direct measurement of an overall rate constant for the tissue (kt), which can be used in place of the ΦtγQb term of eqn (A3) and is applicable to any compound (Gonzales et al. 1998). This overall rate constant is determined by:


Where inline image is the steady state rate of delivery of the compound from the probe to the surrounding tissue and mt is the total analyte amount in the surrounding tissue.

Equation (A1) can be rewritten as


where the parameter Pt is known as the tissue permeability (Gonzales et al. 1998). In terms of kt, Pt is:


Rearranging eqn (A6):


Hence, given an estimate of kt (eqn (A5)) and Pt (eqn (A8)), eqn (A7) was iteratively solved for Dt.



This study was supported in part by grants from the NHMRC, ARC and Heart Foundation. J.N. was an Australian Postdoctoral Fellow of the ARC and S.R. is currently a Heart Foundation Career Fellow. Thanks also to Jason Roberts for verifying the mathematics.