Exercise but not Prostanoids Enhance Levels of Vascular Endothelial Growth Factor and other Proliferative Agents in Human Skeletal Muscle Interstitium



In the present study we examined whether exercise and prostanoids have an effect on the muscle interstitial concentration of vascular endothelial growth factor (VEGF) and on the proliferative effect of muscle interstitial fluid. Dialysate from resting and exercising human skeletal muscle, obtained either during control conditions or during cyclooxygenase inhibition, was examined for its content of VEGF and for its effect on endothelial cell proliferation. Microdialysis probes with high (960 kDa) and low (5 kDa) molecular-mass cut-off membranes were placed in the vastus lateralis muscle of healthy young males. The subjects performed one-legged knee extensions (20 W). The concentration of VEGF in the 960 kDa dialysate was greater (P < 0.05) during exercise compared to at rest (67 ± 28 vs. 230 ± 22 pg ml−1). The rate of endothelial cell proliferation was 2.7-fold higher (P < 0.05) with the 960 kDa dialysate from resting muscle than with perfusate and was 5.8-fold higher (P < 0.05) than the perfusate value with dialysate from exercising muscle. VEGF was not enhanced with exercise in the 5 kDa dialysate, yet the exercise dialysate induced a 1.9-fold higher (P < 0.05) proliferation than the resting dialysate. Cyclooxygenase inhibition did not affect the VEGF concentration or the proliferating effect of the dialysates (P > 0.05). This study demonstrates for the first time that VEGF is present in the interstitium of human skeletal muscle and that exercise enhances the interstitial concentration of VEGF and of other, as yet unidentified, angiogenic compounds. Products of cyclooxygenase do not appear to have an effect on the release of VEGF or other proliferative agents in human skeletal muscle.

It is well known that endurance training produces adaptive changes in skeletal muscle, such as increases in oxidative enzymes and in the number of capillaries (Andersen & Henriksson, 1977; Brodal et al. 1977). Angiogenesis, the development of new capillaries, is a process of several steps involving dissolution of the extracellular matrix underlying the endothelium, cell migration and endothelial cell proliferation (Iruela-Arispe et al. 1991). Angiogenesis can be induced by a number of naturally occurring growth factors, hormones and cytokines (Folkman & Klagsbrun, 1987; Patterson & Runge, 1999), of which one is vascular endothelial growth factor (VEGF).

VEGF is a predominantly endothelial-cell-specific, 45 kDa homodimeric glycoprotein that has been shown to be of major importance in angiogenesis (Leung et al. 1989). In human skeletal muscle, expression of VEGF mRNA has been found to increase in response to acute exercise (Gustafsson et al. 1999; Richardson et al. 1999), and in serum, VEGF protein levels have been shown to be elevated after 10 min of wrist flexion exercise (Nemet et al. 2002). Protein levels of VEGF have, moreover, been found to increase in human skeletal muscle after 10 days of endurance training (Gustafsson et al. 2002) and in rat hindlimb muscle after 3 days of electrical stimulation (Annex et al. 1998). Thus, there is reason to believe that VEGF may be involved in exercise-induced angiogenesis in skeletal muscle. It is not known, however, to what extent VEGF protein levels are increased in the skeletal muscle in response to acute exercise or whether exercise enhances the proliferative effect of skeletal muscle interstitial fluid.

The products of cyclooxygenase, prostanoids, have a variety of functions in skeletal muscle, including regulation of blood flow and angiogenesis (Messina et al. 1975; Boushel et al. 2002). Inhibition of cyclooxygenase with indomethacin attenuates angiogenesis in the rabbit cornea and rat sponge implants (Ziche et al. 1982; Majima et al. 2000) and in rat skeletal muscle (Pearce et al. 2000). Furthermore, prostanoids have also been shown to induce VEGF expression in cell cultures (Harada et al. 1994). Thus, as prostanoids are released from skeletal muscle during dynamic exercise (Frandsen et al. 2000), they may be involved in the regulation of angiogenesis in human skeletal muscle, potentially by increasing VEGF levels. To our knowledge, no previous studies have investigated the effect of prostanoids on VEGF levels in human skeletal muscle.

We hypothesised that exercise and prostanoids are regulators of interstitial VEGF and endothelial cell proliferation. To test this hypothesis, the microdialysis technique was applied to determine the concentration of VEGF in the interstitial fluid of human skeletal muscle at rest and during exercise. The proliferative effect of the muscle microdialysate was examined by adding the dialysates to human vascular endothelial cells in culture and measuring the rate of proliferation. Microdialysis probes of two different molecular-mass cut-offs, were used to allow for discrimination between the proliferative effect of dialysate with and without VEGF. Acetylsalicylic acid (ASA), an irreversible inhibitor of cyclooxygenase, was used to examine whether prostanoids regulate the release of VEGF and other angiogenic compounds in skeletal muscle.



Eleven healthy young men with a mean age and weight of 26.5 ± 0.4 years and 82.1 ± 4.4 kg, respectively, participated in the study. The average leisure time physical activity of the subjects amounted to 3 h (range 1–5 h) of moderate-intensity physical activity per week. The subjects were informed of the experimental procedure, the potential risks, and that they could withdraw from the experiment at any time. The subjects gave their informed consent to participate prior to the start of the experiment. All procedures used conformed to the Declaration of Helsinki and the study was approved by the Ethical Committee of Copenhagen and Frederiksberg.

Microdialysis probes

The semipermeable fibres used to construct the microdialysis probes had a molecular mass cut-off and inner/outer diameter of either 5 kDa and 0.20/0.22 mm (Gambro, Lund, Sweden), respectively, or 960 kDa and 0.34/0.44 mm (Asahi Medical, Tokyo, Japan), respectively. The 960 kDa cut-off probe was made by gluing each end of a fibre 2 cm into a hollow nylon tube (0.50 mm inner diameter and 0.63 mm outer diameter; Portex SIMS, Kent, UK). A suture, 6-0 (Vicryl, Ethicon, Denmark), was advanced inside and glued to the ends of the probe to provide tensile strength. The distance between the two nylon tubes was 4 cm. The 5 kDa cut-off probe was made by gluing each end of a fibre to a 6-0 suture. The fibre and suture were then inserted and glued to an 11 cm inlet and 6 cm outlet nylon tube (0.50 mm inner diameter and 0.63 mm outer diameter; Portex SIMS), so that the fibre went through the outlet tube; the distance between the two tubes was 4 cm, exposing 4 cm of the fibre.

Microdialysis probe insertion

Prior to the insertion of the microdialysis probes, the skin, subcutaneous tissue and fascia close to both the insertion and exit points were anaesthetised with lidocaine (Xylocaine; 20 mg ml−1). Using a 17G/45 mm Venflon I.V. catheter, five microdialysis probes, two with a 5 kDa cut-off membrane and three with a 960 kDa cut-off membrane, were inserted into the vastus lateralis muscle of the quadriceps femoris muscle group of either the right or left leg (randomly chosen). The direction of the microdialysis probes was aligned with the direction of the muscle fibre. The perfusate was Ringer acetate containing (mM): 130 Na+, 2 Ca2+, 4 K+, 1 Mg2+ and 30 acetate, with 3 glucose and 1 lactate added. The microdialysis probes were perfused via a high-precision syringe pump (CMA 102, Carnegie Medicine, Solna, Sweden) at a rate of 5 μl min−1. Suction was connected to the outflow tube of the 960 kDa probes to prevent fluid loss from the probe during the experiment. The weight of the sample tubes was determined before and after sampling in order to validate the perfusion rate. Samples were immediately frozen to −80 °C until analyses were done.

Effect of ASA on dialysate concentrations of prostaglandin (PG)E2 and PGI2

To verify the inhibitory effect of ASA, dialysate from two subjects was analysed for PGE2 and prostacyclin (PGI2), two products of cylcooxygenase. Sixty minutes after probe insertion, dialysate was obtained during a 45 min resting period; the subjects then received 1000 mg ASA and after 15 min dialysate was collected for 2 × 45 min. PGI2 is unstable and converts to 6-keto-PGF-1α; measurements of the 6-keto-PGF-1α concentration in the dialysate were therefore used as a marker for PGI2. PGE2 and 6-keto-PGF-1α were analysed by immunoassays (R&D Systems, Abingdon, UK).

Experimental design

After insertion of the last microdialysis probe, a restitution period of at least 60 min was allowed to minimise the tissue response to the insertion trauma (Langberg et al. 1999). To make sure that probe insertion elicited no effect on VEGF or proliferation levels, dialysate was collected during two 30 min periods, 1 h after probe insertion in three subjects before beginning the protocol, and no effect on the levels of either VEGF or on proliferation were observed. The subjects performed one-legged knee-extensor exercise (kicking frequency 60 s−1) on a modified Krogh ergometer (Andersen et al. 1985). The exercise bouts were performed for 60 min at a work rate of 20 W, corresponding to an approximate intensity of 25–30 % of the leg maximum oxygen uptake. Dialysate from the probes was collected over 30 min periods during a protocol that comprised two exercise bouts. Sampling was performed during 30 min of rest, twice during 60 min of knee extensor kicking and during 30 min of recovery after exercise. Seven of the subjects then ingested two tablets, each containing 500 mg (1000 mg per person) ASA (Nycomed, Denmark) to inhibit cyclooxygenase, before continuing the protocol. After 1 h, dialysate was collected during 30 min of rest, twice during 60 min of exercise at 20 W and then once during 30 min of recovery after exercise. Four control subjects who did not ingest any tablets performed the same protocol as described above to evaluate the effect of ASA.

VEGF measurements

Dialysate obtained during the experimental protocol was analysed for VEGF protein by a Quantikine enzyme-linked immunosorbent assay kit (R&D Systems) according to the manufacturer's protocol. The coefficient of variation for the VEGF assay performed on duplicate dialysate samples was 14.3 %. The VEGF assay has the potential to recognise all variants of VEGF-A, but it has only been tested with VEGF 121 and 165 and recombinant human VEGF.

Measurement of endothelial cell proliferation

Human umbilical vein endothelial cells, supplemented in medium 200 with low serum growth supplement containing fetal bovine serum, fibroblast growth factor, heparin and epidermal growth factor, intended for use in the culture of endothelial cells (Cascade Biologics, Portland, OR, USA), were grown on 96-well plates for 24 h before replacing the medium with 50 μl of microdialysate, perfusate or supplemented medium 200. The dialysate was diluted 1:1 with perfusate to obtain values that did not exceed those of the positive control (supplemented medium 200) and did not reach the maximum level of proliferation. After an additional 24 h of incubation, bromodeoxyuridine (BrdU) was added and then incubated for 12 h. Incorporation of BrdU into the DNA was detected by an immunoassay (Roche, Mannheim, Germany) according to the manufacturer's recommended methods. All measurements were made in duplicate. In two experiments, recombinant human VEGF (293-Ve, R&D Systems) was added to all of the dialysates in concentrations double and triple that of measured dialysate VEGF concentrations to determine whether additional VEGF would further enhance proliferation.

Determination of probe recovery

In order to assess whether the effect of exercise on the interstitial VEGF concentration and on cell proliferation could be explained by an exercise-induced enhancement in recovery, compounds of different molecular weights were used to determine the in vivo relative recovery by the internal reference method (Scheller & Kolb, 1991). Radioactively labelled VEGF was not commercially available, and instead [3H]adenosine (2 kBq ml perfusate−1), [14C]methylated albumin (1.48 kBq ml perfusate−1) and [14C]casein (7.4 Bq ml perfusate−1) were included in the perfusate in five different experiments, and 5 μl of each dialysate (outflow) sample was counted in a Packard 2300TR liquid scintillation counter in order to determine fractional loss. Albumin was used because it has a molecular weight that is not very different from that of VEGF. The relative recovery was calculated for each microdialysis probe as (Cp - Cd)/Cp, where Cp is disintegrations per minute (d.p.m.) in the perfusate and Cd is d.p.m. in the dialysate. It was assumed that relative loss from the perfusate to the interstitial fluid of labelled metabolite equals relative diffusion from the interstitial fluid to the perfusate of unlabelled metabolite.


All data are presented as means ±s.e.m. Only values from microdialysis probes that had a perfusion flow of 4.5–5.0 μl min−1 throughout the entire experiment were used (for the 5 kDa probes an average of 1.6 (range 1–2) probes per person were used and 2.0 (1–3) probes of the 960 kDa probes). For each subject, within the different sampling periods, an average value for the included 5 kDa and the 960 kDa microdialysis probes, respectively, was determined. An overall average was then calculated from the individual means. Changes in VEGF concentration and endothelial cell proliferation were analysed using a two-way repeated-measures ANOVA comparing changes in cyclooxygenase inhibition or probe size (5 kDa vs. 960 kDa) to changes over time (rest vs. exercise and first vs. second exercise bout). A one-way repeated measures ANOVA was used to determine changes in PGE2 and 6-keto-PGF-1α. When significant changes were found, a Student-Newman-Keuls Method for multiple comparisons was used to determine where the significant changes occurred. Differences were considered significant if P < 0.05.


PGE2 and 6-keto-PGF-1α

PGE2 and 6-keto-PGF-1α concentrations in dialysate from the skeletal muscle were lower as a result of cyclooxygenase inhibition (P < 0.05; Fig. 1). At the third collection, which was taken 1 h after ASA intake, only 13.6 ± 0.5 % of PGE2 and 1.1 ± 0.6 % of 6-keto-PGF-1α remained in the dialysate.

Figure 1.

Effect of oral intake of ASA on skeletal muscle microdialysate PGE2 and PGI2 concentrations

Concentrations of PGE2 (A) and 6-keto-PGF-1α (B) in dialysate before (−45 to 0 min; −45–0), and 15–60 min (15–60) and 60–105 min (60–105) after ASA intake. *P < 0.05, vs. before ASA intake; n= 2.

VEGF protein

VEGF was not detectable in any dialysates from probes with 5 kDa cut-off or in the perfusate. The dialysate concentration of VEGF from the 960 kDa probes was 50.4 ± 19 pg ml−1 at rest, and the concentration was higher during the first and the second sampling period of the first exercise bout (P < 0.05; 193.9 ± 24 pg ml−1 and 173.9 ± 16 pg ml−1, respectively). In recovery after the first exercise bout, the concentration was lower (P < 0.05; 107 ± 15 pg ml−1) than during the exercise (Fig. 2). For the second exercise bout, the dialysate concentration of VEGF (960 kDa probe) increased from 90.9 ± 17 pg ml−1 at rest to 452.9 ± 24 pg ml−1 in the first sampling period of the exercise bout (P < 0.05). In the second sampling period of the second exercise bout, the VEGF concentration decreased (P < 0.05) to 330.8 ± 26 pg ml−1 and was further lowered after exercise to 140.4 ± 18 pg ml−1 (P < 0.05). The dialysate concentration of VEGF was approximately twofold higher during the second compared to the first exercise bout (P < 0.05). Since there were no differences (P > 0.05) in dialysate VEGF concentrations at any time during the experiment between the control group and the group that ingested ASA (Table 1), Fig. 2 includes the means of both groups.

Figure 2.

VEGF concentrations in human skeletal muscle dialysate from 960 kDa molecular-mass cut-off microdialysis probes

Concentration of VEGF in skeletal muscle dialysate at rest (Rest), during knee extensor exercise at 20 W (Exercise) and after exercise (Recovery). Means include data from both the ASA and the control group, as there were no differences between the two groups; n= 11 *P < 0.05, vs. rest and recovery. †P < 0.05, vs. rest. ‡P < 0.05, vs. first exercise bout. §P < 0.05, vs. first sampling period of exercise bout.

Table 1. The effect of exercise on the VEGF content and endothelial cell proliferative effect of mucsle dialysate without (Control) and with (ASA) cyclooxygenase inhibition
 First partSecond part
Exercise protocolRestExerciseRecoveryRestExerciseRecovery
  1. In the ASAgroup cyclooxygenase was inhibited after the first part of the protocol. Data from dialysate obtained during work are means of first and second sampling period of the exercise bout. *P < 0.05, vs. rest (mean ±s.e.m.).

VEGF (pg ml−1)
  Control28.5 ± 11.3126.6 ± 27.8+86.0 ± 21.766.6 ± 19.5363.2 ± 46.5*115.6 ± 22.4
  ASA59.4 ± 27.5216.5 ± 18.1*118.1 ± 21.0101.2 ± 22.4408.2 ± 31.1*154.6 ± 25.8
Endothelial cell proliferation, dialysate from 960 kDa probes (% of perfusate)
  Control189.6 ± 56.2413.7 ± 97.5191.1 ± 48.9328.0 ± 43.1521.9 ± 135.8310.8 ± 88.8
  ASA310.1 ± 36.0675.9 ± 54.0*403.4 ± 56.8367.9 ± 39.3654.4 ± 54.2*412.6 ± 65.9
Endothelial cell proliferation, dialysate from 5 kDa probes (% of perfusate)
  Control143.8 ± 50.1311.5 ± 73.5310.0 ± 73.5274.9 ± 30.4392.6 ± 99.9280.1 ± 69.2
  ASA210.8 ± 27.7396.2 ± 68.5*410.7 ± 58.2302.3 ± 39.24445.5 ± 56.1*358.1 ± 42.7

Endothelial cell proliferation

The endothelial cell proliferation with dialysate obtained during both rest and exercise with the 960 kDa probes was 140 % higher than that obtained with the 5 kDa cut-off probes (P < 0.05). Dialysate obtained at rest from 5 kDa cut-off probes increased proliferation by 186.4 ± 25.9 % above perfusate values (P < 0.05), and the 5 kDa dialysate obtained during the first and the second exercise bout increased proliferation by 356.7 ± 55.6 % and 374.1 ± 45.8 %, respectively, above perfusate values (P < 0.05). In recovery after exercise, the proliferation (292.3 ± 26.6 % above perfusate) was not significantly lower than during exercise P > 0.05; Fig. 3B). In dialysate obtained from the 960 kDa cut-off probes, endothelial cell proliferation was 266.3 ± 34.3 % above perfusate values, and increased significantly (P < 0.05) to 548.6 ± 62.5 % and 612.5 ± 61.1 % above perfusate values during the first and second sampling period, respectively. Proliferation with the dialysate obtained after exercise (326.2 ± 50.4 %) was lower than that observed with dialysate obtained during exercise (P < 0.05; Fig. 3A). There was no difference in the proliferative effect between dialysates obtained before, during and after the first versus the second exercise bout (P > 0.05). In all assays, the addition of supplemented medium 200 yielded a higher response than dialysate (results not shown) and acted as a positive control. Endothelial cell proliferation was increased (P < 0.05) by addition of recombinant VEGF compared to dialysate with no VEGF added in samples obtained during rest, exercise and recovery (data not shown). When comparing endothelial cell proliferation elicited with dialysates from the control group and the ASA group, no differences were found either with high or low cut-off probes (Table 1; P > 0.05).

Figure 3.

Effect of human skeletal muscle dialysate on endothelial cell proliferation

Proliferation of endothelial cells after the addition of skeletal muscle dialysate from 960 kDa (A) and 5 kDa (B) molecular-mass cut-off probes, obtained at rest (Rest), during knee extensor exercise at 20 W (Exercise) and after exercise (Recovery). Means include all subjects, as there were no differences between the ASA and control groups; n= 11. *P < 0.05, vs. rest and recovery. †P < 0.05, vs. rest.

Relative recovery

The in vivo determination of probe recoveries for casein, albumin and adenosine showed a 0 %, 32 % and 25 % increase, respectively from rest to exercise. The results are combined for 5 kDa and 960 kDa probes as there were no differences. In recovery after exercise, the relative recovery was similar to pre-exercise.


The present study demonstrates that VEGF is present in the interstitium of human skeletal muscle and that exercise enhances the interstitial concentration of VEGF. Furthermore, muscle dialysate obtained at rest and during exercise has a proliferative effect on endothelial cells that appears partly to be due to the presence of VEGF, but also to other, yet unidentified, angiogenic compounds. In addition, prostanoids do not seem to play an important role in regulating the interstitial concentrations of VEGF or other proliferative compounds.

Regulation of interstitial VEGF

In the current study, VEGF was found to be present at rest and to a greater extent during exercise in the muscle dialysate. It is in the interstitium that VEGF elicits its effects on the receptors, as these are localised predominantly to the membrane of endothelial cells (Jakeman et al. 1992) and therefore measurements of VEGF in interstitial fluid are highly relevant. The origin of and factors responsible for the observed increase in interstitial VEGF concentration during exercise can only be speculated on. Studies on animal muscle have suggested that the mechanical effects of muscle contraction, intracellular hypoxia or increased blood flow, which elevates capillary shear stress, are possible regulators of the induction of VEGF expression (Breen et al. 1996; Cherwek et al. 2000; Milkiewicz et al. 2001). The results of these studies, however, have not been confirmed in human studies (Gustafsson et al. 1999; Richardson et al. 1999). These possible stimuli for the upregulation of VEGF may be mediated by metabolites, as muscle contraction, hypoxia and shear stress result in the production of various metabolites that can induce VEGF expression, such as hypoxia-inducible factor (HIF)-1α (Gustafsson et al. 1999), heavy metals (Eyssen-Hernandez et al. 1996), adenosine (Takagi et al. 1996) and nitric oxide (NO) (Gavin et al. 2000). Several cell types could be responsible for the synthesis of these metabolites. For example, NO can be formed in skeletal muscle cells as well as smooth muscle and endothelial cells (Reid, 1998) and muscle interstitial adenosine has been reported to largely originate from extracellular nucleotide degradation via an ectonucleotidase present on the muscle cell membrane (Hellsten, 1999; Lynge et al. 2001), but may also be formed in endothelial cells (Deussen et al. 1986). Thus, it appears likely that more than just one compound and more than just one cell type contributes to the regulation of VEGF concentrations in the interstitial fluid of human skeletal muscle.

The increased VEGF concentration in dialysate from exercising muscle compared to resting muscle may not solely originate directly from the synthesis and release of VEGF from muscle and interstitial cells. Intracellular synthesis of VEGF is not always followed by a release of VEGF to the interstitial fluid. The two shorter isoforms (VEGF121 and VEGF165) are released efficiently to the interstitium, whereas the longer isoforms (VEGF189 and VEGF206) need protease activity and cleavage to be released from the cell membrane (Ferrara et al. 1991; Park et al. 1993). Thus, the availability of proteases could be a mode of regulating VEGF concentration in the interstitial space of skeletal muscle. During exercise there is an increased protein turnover (MacLean et al. 1994), and the proteases responsible for this may be released and could account for part of the exercise-induced increase in interstitial VEGF.

In the present study the second exercise bout elicited a significantly greater increase in dialysate VEGF than the first exercise bout, both with and without ASA ingestion prior to the second exercise bout. Knee-extensor exercise has been found to elevate VEGF mRNA in skeletal muscle 75–90 min after the onset of exercise (Gustafsson et al. 1999; Richardson et al. 1999). The increase in VEGF mRNA, which would be expected to occur shortly after the first exercise bout, might increase the production of VEGF which, if not released, could be either saved in intracellular pools or secreted and bound to the cell surface (Park et al. 1993) ready for release during the second bout of exercise. This extra accumulation of VEGF could account for the increase observed during the second exercise bout and may also be responsible for the higher VEGF release observed during the first 30 min vs. the last 30 min of the second exercise bout. An alternative explanation could be that a signalling substance accumulates when two exercise bouts are performed; however, since resting values of VEGF were not increased after the first exercise bout, this explanation does not appear likely. Nevertheless, it could be that exercise is necessary to activate the signal that increases VEGF levels.

The possibility that muscle fibre damage during the first exercise bout may be responsible for the further increase in VEGF protein concentrations measured during the second exercise bout cannot be excluded; however, in this case the release of proliferative agents would have been expected to follow the same pattern as VEGF release, and this was not observed. In addition, measurements of VEGF concentrations in the dialysate at 60–90 min after probe insertion revealed that the VEGF level was not higher than at 90–120 min after insertion, indicating that the initial cellular damage due to probe insertion did not cause a release of VEGF.


When comparing endothelial cell proliferation obtained from low- and high-molecular-mass cut-off probes, dialysate from high cut-off probes had the greatest proliferative effect, both at rest and during exercise, suggesting that some factors, too big to diffuse through the 5 kDa probes, induced the proliferation of endothelial cells. One likely contributor to the enhanced proliferation is VEGF, as this was present in the interstitium and is known to be an important angiogenic compound (Leung et al. 1989). Nevertheless, an observation that argues against an important proliferative role for VEGF is that the greater VEGF concentration in the dialysate collected during the second exercise bout than during the first bout did not result in a further enhanced rate of endothelial cell proliferation. Addition of extra VEGF to the dialysate obtained during the first exercise bout was, however, observed to result in a further enhancement of endothelial cell proliferation. It appears, therefore, that an explanation for this observation must be found in a concomitant decrease in the concentration of other proliferative compounds or in an increase in VEGF inhibitory compounds.

Other angiogenic compounds with molecular masses above 5 kDa, such as basic fibroblast growth factor (bFGF) and transforming growth factor β (TGF-β), could also have contributed to the proliferative effect of the 960 kDa dialysate. It has been shown in studies on laboratory animals that exercise induces an increase in both bFGF and TGF-β mRNA and protein (Morrow et al. 1990; Breen et al. 1996). It should nevertheless be mentioned that studies on human muscle show no upregulation of bFGF and TGF-β mRNA with exercise (Richardson et al. 1999; Wagner et al. 2001). Thus, the role for these compounds in human muscle angiogenesis remains unresolved.

Muscle dialysates from the 5 kDa cut-off probes obtained both at rest and during exercise were found to induce endothelial cell proliferation, although these dialysates did not contain either VEGF or any other high-molecular-mass compounds such as bFGF and TGF-β. Thus, our data indicate that the muscle interstitium appears to contain either one or several proliferative compound(s) of relatively low molecular mass. Furthermore, as the proliferative effect of the 5 kDa muscle dialysate obtained during exercise was greater than that obtained at rest, such proliferative compounds are seemingly increased in response to exercise. The present study did not unveil the identity of these compounds, but cyclooxygenase inhibition did not influence the effect of the 5 kDa probe dialysate on endothelial cell proliferation, suggesting that prostanoids were not responsible for the proliferative effect of the dialysate. Several other compounds such as NO (Raychaudhury et al. 1996), endothelin-1 (ET-1; Salani et al. 2000) and adenosine (Ethier et al. 1993) are regulators of endothelial cell proliferation and have been shown to be present in the human muscle interstitium at rest, and to increase in concentration in response to exercise. Thus, one or several of these compounds are potential contributors to the proliferative effect of the 5 kDa dialysate.

In the present study, probe recovery was not possible to assess for VEGF levels, as radioactively labelled VEGF is not commercially available. Without an estimation of probe recovery a calculation of interstitial concentrations is not possible, and a precise comparison between dialysate concentrations at rest and during exercise cannot be achieved, as a fraction of the change in the dialysate concentrations observed will be due to an enhancement in probe recovery during exercise compared to rest (Radegran et al. 1998; Frandsen et al. 2000). In addition, as the identity of proliferative compounds other than VEGF in the interstitium was unknown, it was not possible to obtain probe recovery for the proliferative effect. Nevertheless, in order to obtain an estimate of probe recovery we determined the relative recoveries for several metabolites of different molecular size (albumin, adenosine and casein). It was found that the increases in relative recovery seen during exercise compared to rest for the metabolites examined (from 0 to 32 %) were all less than the increases in VEGF concentration (> 380 %) and the elevation in proliferation rate (> 170 %) observed with exercise in the present study. Thus, our findings of increased VEGF concentrations and of enhanced proliferation with exercise cannot be explained solely by an enhancement of probe recovery during exercise.

Cyclooxygenase inhibition

Cyclooxygenase is, like VEGF protein and mRNA, expressed in skeletal muscle cells and in nonendothelial and endothelial cells between the muscle cells (Symons et al. 1991), and products of cyclooxygenase, such as prostaglandins, are known to induce synthesis of VEGF (Harada et al. 1994). Thus, we hypothesised that the intracellular VEGF production, and consequently VEGF release, would be attenuated by cyclooxygenase inhibition, and that the effect of cyclooxygenase inhibition on VEGF release would be greater during exercise, as cyclooxygenase activity is known to be increased by mechanical stimulation of skeletal muscle cells (Vandenburgh et al. 1995) and by static hindlimb contraction (Symons et al. 1991). However, it was observed that effective inhibition of cyclooxygenase did not affect the interstitial concentration of VEGF at rest or during exercise. This observation was not due to poor inhibition of cyclooxygenase, as ASA elicited 86.4 % and 98.9 % decreases in PGE2 and PGI2 concentrations in the dialysate, respectively, at 1 h after ingestion, which was the time when the second part of the experimental protocol was initiated.

The present study is to our knowledge the first to investigate the influences of prostanoids on VEGF protein levels in skeletal muscle and indicates that inhibition of cyclooxygenase does not affect VEGF release. Even though many studies on cell cultures have revealed that prostanoids do increase VEGF synthesis (Harada et al. 1994), the observation in the present study is partly in agreement with a study in rat, showing that prostaglandin infusion does not affect VEGF mRNA levels, although prostacyclin infusion decreased VEGF mRNA levels (Benoit et al. 1999). It could be speculated that during cyclooxygenase inhibition, other substances that are capable of inducing VEGF were produced or released to a greater extent in response to contraction, thereby compensating for diminished prostanoid levels. Even though this could be true, it is not very likely, since most known substances that induce VEGF also induce endothelial cell proliferation, and in the present study cyclooxygenase inhibition did not cause a change in proliferation, suggesting that such compounds were not produced to a greater extent during cyclooxygenase inhibition.

In conclusion, this study demonstrates for the first time that VEGF is present in the interstitial fluid of skeletal muscle and that VEGF release is increased during exercise. Although VEGF is likely to be an important angiogenic compound in muscle, other, yet unidentified endothelial cell proliferative compounds with a molecular mass below 5 kDa are present in the muscle interstitium and are released during exercise. Products of cyclooxygenase do not seem to play an important role in regulating the release of VEGF or other proliferative compounds in human skeletal muscle.


The work was supported by the Danish National Research Foundation (504–14) and Team Danmark.