Volume 99, Issue 2 p. 368-380
Research Paper
Free Access

Exogenously applied muscle metabolites synergistically evoke sensations of muscle fatigue and pain in human subjects

Kelly A. Pollak

Kelly A. Pollak

Departments of Anesthesiology

Department of Anesthesiology and Pain Medicine, University of Washington, Seattle, WA, USA

Search for more papers by this author
Jeffrey D. Swenson

Jeffrey D. Swenson

Departments of Anesthesiology

Search for more papers by this author
Timothy A. Vanhaitsma

Timothy A. Vanhaitsma

Exercise and Sport Science

Search for more papers by this author
Ronald W. Hughen

Ronald W. Hughen

Departments of Anesthesiology

Search for more papers by this author
Daehyun Jo

Daehyun Jo

Anesthesiology and Pain Medicine Department, Daejeon St Mary's Hospital, The Catholic University of Korea, Daejeon, Korea

Search for more papers by this author
Kathleen C. Light

Kathleen C. Light

Departments of Anesthesiology

Search for more papers by this author
Petra Schweinhardt

Petra Schweinhardt

Alan Edwards Centre for Research on Pain, McGill University, Montreal, Québec, Canada

Search for more papers by this author
Markus Amann

Markus Amann

Departments of Anesthesiology


GRECC, Veterans’ Affairs Medical Center, Salt Lake City, UT, USA

Search for more papers by this author
Alan R. Light

Corresponding Author

Alan R. Light

Departments of Anesthesiology

Neurobiology and Anatomy, University of Utah, Salt Lake City, UT, USA

Corresponding author A. R. Light: University of Utah, Department of Anesthesiology 3C444 SOM, 30N 1900E, Salt Lake City, UT 84132, USA. Email: [email protected]Search for more papers by this author
First published: 23 October 2013
Citations: 158


New Findings

  • What is the central question of this study?

    Can physiological concentrations of metabolite combinations evoke sensations of fatigue and pain when injected into skeletal muscle? If so, what sensations are evoked?

  • What is the main finding and its importance?

    Low concentrations of protons, lactate and ATP evoked sensations related to fatigue. Higher concentrations of these metabolites evoked pain. Single metabolites evoked no sensations. This suggests that the combination of an ASIC receptor and a purinergic P2X receptor is required for signalling fatigue and pain. The results also suggest that two types of sensory neurons encode metabolites; one detects low concentrations of metabolites and signals sensations of fatigue, whereas the other detects higher levels of metabolites and signals ache and hot.

The perception of fatigue is common in many disease states; however, the mechanisms of sensory muscle fatigue are not understood. In mice, rats and cats, muscle afferents signal metabolite production in skeletal muscle using a complex of ASIC, P2X and TRPV1 receptors. Endogenous muscle agonists for these receptors are combinations of protons, lactate and ATP. Here we applied physiological concentrations of these agonists to muscle interstitium in human subjects to determine whether this combination could activate sensations and, if so, to determine how the subjects described these sensations. Ten volunteers received infusions (0.2 ml over 30 s) containing protons, lactate and ATP under the fascia of a thumb muscle, abductor pollicis brevis. Infusion of individual metabolites at maximal amounts evoked no fatigue or pain. Metabolite combinations found in resting muscles (pH 7.4 + 300 nm ATP + 1 mm lactate) also evoked no sensation. The infusion of a metabolite combination found in muscle during moderate endurance exercise (pH 7.3 + 400 nm ATP + 5 mm lactate) produced significant fatigue sensations. Infusion of a metabolite combination associated with vigorous exercise (pH 7.2 + 500 nm ATP + 10 mm lactate) produced stronger sensations of fatigue and some ache. Higher levels of metabolites (as found with ischaemic exercise) caused more ache but no additional fatigue sensation. Thus, in a dose-dependent manner, intramuscular infusion of combinations of protons, lactate and ATP leads to fatigue sensation and eventually pain, probably through activation of ASIC, P2X and TRPV1 receptors. This is the first demonstration in humans that metabolites normally produced by exercise act in combination to activate sensory neurons that signal sensations of fatigue and muscle pain.


The sensation of fatigue that occurs during and after muscle contractions or in certain pathological conditions (Kos et al. 2008) is the perception of tiredness or heaviness, which can substantially reduce physical activity and performance (Amann & Dempsey, 2008; Amann et al. 2013). Muscle pain in undamaged skeletal muscle can also be evoked by ischaemic exercise (Alam & Smirk, 1937). At present, the metabolites evoking and the molecular mechanisms detecting and leading to the sensations of fatigue and muscle pain remain largely unknown.

In skeletal muscle, nociceptive afferent neurons are mostly group III and IV neurons corresponding in conduction velocity to cutaneous Aδ and C afferents. Studies have shown that mechanical stimulation can activate at least some of these neurons (Mense & Schmidt, 1974; Mense, 1977; Kaufman et al. 1983, 1984; Kaufman & Rybicki, 1987; Bove & Light, 1995; Adreani et al. 1997; Graven-Nielsen, 2006). Additionally, intramuscular infusion of supraphysiological concentrations of various chemicals can activate these neurons (Mense, 1977; Kniffki et al. 1978; Hill et al. 1992; Decherchi et al. 1998; Khan & Sinoway, 2000).

More relevant to our understanding of the perception of fatigue is the fact that a variety of exercise-induced intramuscular metabolites, including lactate, ATP and protons, also stimulate a subset of these muscle afferents (Bangsbo et al. 1993, 2001; Hellsten et al. 1998; Li et al. 2003, 2005; Mohr et al. 2007). Animal experiments have demonstrated that at least three different molecular receptors, namely a heteromer of Acid Sensing Ion Current family member 1 (ASIC1) and Acid Sensing Ion Current family member 3 (ASIC3) or ASIC1 + ASIC2 (Gautam & Benson, 2013), a P2X (P2X5, P2X4 or P2X1/3 heteromer) Purinergic receptor detecting receptor family member X4 and transient receptor potential cation channel subfamily V member 1 (TRPV1), co-localized on cardiac and skeletal muscle afferent endings (Benson et al. 1999; Immke & McCleskey, 2001a,b; Sutherland et al. 2001; Connor et al. 2005; Molliver et al. 2005; Smith et al. 2005), co-operate to detect precisely physiological levels of a combination of these three metabolites (Naves & McCleskey, 2005; Yagi et al. 2006; Light et al. 2008; Birdsong et al. 2010; Jankowski et al. 2013). These metabolite receptors are reported to be densely localized near blood vessels immediately below the muscle fascia (Molliver et al. 2005); however, group III/IV afferent endings have also been localized to other parts of skeletal muscle, including veins and the adventitia of lymphatic vessels (Andres et al. 1985).

We and others have recently shown that these three metabolites (lactate, ATP and protons) act synergistically to activate at least two significant populations of dorsal root ganglion group III/IV afferent neurons innervating skeletal muscle of mice (Light et al. 2008; Jankowski et al. 2013). One population responds to the levels of metabolites produced by non-painful muscle contractions and may contribute significantly to the exercise pressor reflex (the increase in blood pressure that occurs in response to skeletal muscle contraction; Alam & Smirk, 1937; Mitchell et al. 1968, 1983; Secher & Amann, 2012) and, probably, to the sensory experience of exercise-induced fatigue, as was noted initially by Alam & Smirk (1937). The other population begins to respond at levels of metabolites that only occur during muscle ischaemia, and could contribute to acute muscle pain as experienced during ischaemic exercise or during muscle inflammation and injury. Additional mouse experiments verified that no individual metabolite or pair of these metabolites had the effectiveness of the proper combination of all three metabolites (Light et al. 2008).

Strong evidence in favour of a significant involvement of group III/IV muscle afferents in the perception of fatigue comes from recent human studies. When the inputs from these afferents were partly blocked by lumbar intrathecal injection of the μ-opioid agonist fentanyl, subjective ratings of perceived exertion and fatigue during lower limb exercise were significantly attenuated (Gandevia, 2001; Amann et al. 2009, 2010, 2011a,b). These same studies also suggest that activation of group III/IV muscle afferents limits central motor drive in exercising individuals, which protects against excessive development of peripheral fatigue (fatigue caused by the reduction in the ability of muscle fibres to contract; Amann et al. 2011a). However, it is unclear whether human afferents use the same molecular receptors to detect the same metabolites as animal models. Therefore, the aims of this study were twofold: (i) to create a method by which metabolites could be infused directly into muscle interstitium; and (ii) to determine whether humans also use a unique combination of muscle contraction-induced protons, lactate and ATP to activate group III and IV skeletal muscle afferents evoking sensations of fatigue and pain.



Ten healthy volunteers (six men and four women, aged 30–60 years) participated in these studies. Written informed consent was obtained from each participant. All procedures were approved by the University of Utah Institutional Review Board and conformed to the standards set by the Declaration of Helsinki.

Metabolite solutions

Solvent-free ATP was obtained from Sigma-Aldrich; all other metabolites and electrolytes were USP grade. All mixing procedures were conducted by an anesthesiologist and supervised by a licensed pharmacist.

A series of metabolite solutions containing increasing concentrations of lactate, protons and ATP (Table 1) were used in the present experiments. Details of the composition of the injected solutions are in the Supporting Information Supplemental File 1. The lactate and pH levels in these solutions were derived from studies where blood samples, muscle biopsies and direct measurements of muscle interstitial pH were obtained from healthy humans during exercise of varying intensity (Allsop et al. 1990; Bangsbo et al. 1993; Mohr et al. 2007). The ATP levels were derived from studies in which interstitial muscle concentrations of ATP were measured during mild to intense exercise in humans, cats and rats (Hellsten et al. 1998; Li et al. 2003, 2005).

Table 1. Metabolite concentrations used in the study
Metabolite chart
pH ATP (nm) Lactate (mm) Exercise level Order of presentation
7.4  300  1 Resting (neutral solution) 1
7.3  400  4 Mild 2
7.2  500 10 Moderate 3
7.0 1000 15 High 4
6.8 2000 20 Very high 5
6.6 5000 50 Ischaemic (painful) 6

The values of metabolites in the solution at pH 7.4 approximate those found in interstitial muscle fluid when muscles are at rest. Generally, the low metabolite values associated with pH > 7.0 are likely to be non-noxious, whereas higher metabolite values at pH < 7.0 may be associated with pain. The metabolite concentrations used here (ranging from pH 7.4, 1 mm lactate and 300 mm ATP to pH 6.6, 50 mm lactate and 5000 nm ATP) bracket the physiological range of metabolites found in skeletal muscle interstitium at rest and during exercise and ischaemic contractions (Steinhagen et al. 1976; Graham et al. 1993; Bangsbo et al. 2001; Reeh & Kress, 2001; Li et al. 2003, 2005).

The solutions were mixtures of sodium, calcium, magnesium, potassium and glucose buffered with 30 mm sodium phosphate and adjusted to normal interstitial levels to balance the osmolarity at between 280 and 300 mosmol l−1. Given that ATP degrades quickly when in solution at and above room temperature, ATP was thawed immediately before use, and added by an anesthesiologist in a sterile hood to obtain the values in Table 1 at the time of the experiment. Once mixed, solutions containing ATP were used within 20 min. Once combined, all solutions were kept at muscle temperature (35°C).

Injection placement

The desired metabolite solution was drawn into a 5 ml syringe and placed in a clinical grade syringe pump attached to sterile catheter tubing mating with a sterile 33 gauge needle.

An anesthesiologist, using ultrasound guidance, inserted 33 gauge needles bilaterally under the fascia of the abductor pollicis brevis (APB) muscle of the thumb (see Fig. 1). This permitted intramuscular infusion of the series of six sterile metabolite solutions of increasing concentrations in the APB of one hand (e.g. the left hand), and in five of the subjects a series of six solutions corresponding to only the resting neutral (pH 7.4) metabolite levels (see Table 1) in the APB in the other hand (e.g. the right hand). This served as a control for volume effects of the injections. Four other subjects (see ‘Individual Metabolites’ section below) received injections of single metabolites. The dead space in the needle was filled with warmed (35°C) neutral (pH 7.4) metabolite solution prior to placement in the muscle, and the catheter tubing was purged of air before mating with the cannula. A sensation of aching pain was often (but not always) perceived when the needle penetrated the superficial muscle fascia. This sensation was allowed to diminish before solutions were injected. To ensure that positioning would not itself produce muscle fatigue, the forearms of volunteers were rested on pillows to maintain a neutral position of the APB muscles.

Details are in the caption following the image
Figure 1. Ultrasound image of needle placement under muscle fascia of abductor pollicis brevis (ADP) muscle of the thumb

The fascia layer is indicated, as are the muscle fibres of ADP. The tip of the needle is indicated by the bright spot with the arrow pointing to it.

Metabolite application

Two hundred microlitres of each solution in Table 1 was infused over 30 s in an ascending series in one thumb. At least 5 min was allowed between all infusions. For later infusions (from pH 7.0 to pH 6.6) the time between infusions averaged 10 min. Between each infusion in the series, a neutral solution (pH 7.4) of the same volume was infused over 30 s in the opposite thumb. This allowed for the control of any sensations related to volume effects. These volumes and rates have been used with hypotonic solutions to evoke pain in the same muscle in past experiments (Coppieters et al. 2006; Graven-Nielsen, 2006).

Pain and fatigue reporting

Subjects were asked to report and describe non-pain and pain sensations every 30 s during and after each metabolite infusion until the subjects could not detect any evoked sensations (an average of 3.4 min, range 0.2–4.8 min). Subjects were not cued or briefed on possible descriptions of these sensations. They were simply asked if they felt anything and whether it was painful or non-painful, and for their description of what it felt like. The duration of sensations was estimated by subtracting the time of the last report from the time of the first report.

Subjects were also asked to quantify the magnitude of non-pain and pain sensations using a Verbal Response Scale (VRS) ranging from 0 to 100, with 0 being no sensation and 100 being the strongest non-pain or pain sensation they could imagine. These scales have been used for human experiments to assess noxious sensation and have been proved to be valid (Ferreira-Valente et al. 2011). A 2 min rest period was given upon cessation of all subject-reported sensations before administering the next solution in the series.

At the conclusion of the series of infusions, ultrasound was again applied to the APB. Needle position was examined to look for possible migration. In addition, the muscle and subcutaneous tissues were closely examined to look for evidence of fluid build-up, observable as a ‘pocket’ of fluid, or leakage of injected fluid out of the muscle and into the overlying tissues.

Individual metabolites

To test for the effects of the maximal amounts of individual metabolites and for potential residual effects of injected metabolites that might be affecting subsequent metabolite injections, we conducted the following experiment.

Four healthy volunteers received infusions over 30 s of 200 μl of each of the three individual metabolites (protons, lactate or ATP) at maximal levels (see Table 2), each in the same phosphate-buffered solution with normal interstitial levels of sodium, calcium, magnesium and potassium (see Supporting Information Supplemental File 2 for details of the composition of these solutions).

Table 2. Control metabolite solutions
pH ATP (nm) Lactate (mm)
Maximal lactate 7.4 0 50
Maximal protons 6.6 0  0
Maximal ATP 7.4 5000  0

For this procedure, the four subjects were first infused with 200 μl of ‘neutral solution’ (pH 7.4 + 300 nm ATP + 1 mm lactate). Next, subjects were infused with each of the three individual metabolites (protons, lactate or ATP) 10 min apart at maximal levels. Solutions administered were counterbalanced across subjects to control for order effects. As a control for volume effects after the individual metabolite administrations, three of the four subjects received two infusions (10 min apart) of the neutral solution (pH 7.4 + 300 nm ATP + 1 mm lactate). After each infusion, non-pain and pain sensations were again recorded, and the same VRS was used to quantify the magnitude of sensation.


Data from each subject were recorded by a researcher during the metabolite infusions. Data containing the number of subjects reporting sensations, onset times, sensation duration, sensation descriptors and VRS score for both non-pain and pain sensations for each metabolite series were later transferred to an Excel spreadsheet for further analysis. Mean values and SEM for each of the interval values collected were calculated, and comparisons were made between values for non-pain and pain sensations for each metabolite series. Verbal Response Scale scores were analysed first using one-way repeated-measures ANOVAs (d.f. = 5,45). For the non-pain data, F = 3.915, P < 0.005. For the pain data, F = 15.13, P < 0.00001. Individual contrasts between baseline and other solutions were analysed using one-tailed Student's paired t tests with the a priori hypotheses being that the first three non-baseline solutions (pH 7.3, 7.2 and 7.0) would evoke increased non-pain sensations, while pain sensations would not be evoked until pH 7.2, and would increase in intensity as the metabolites were increased (pH 7.0, 6.8 and 6.6). Likewise, VRS scores for maximal combination metabolite infusions were compared with VRS scores for the maximum of each individual metabolite using Student's paired t tests with the a priori hypothesis being that the combination of metabolites would have a higher VRS score than the mean of each of the VRS scores evoked by individual maximal values of metabolites. Owing to the floor effect created by the many zero VRS scores reported, VRS scores were not normally distributed. To create normality for parametric statistics, a constant (1) was added to each measurement, and this value was log transformed. No corrections were made for multiple comparisons because only a priori hypotheses were tested. Statistical tests were not done on the duration data because no a priori hypotheses were created and the variance in these values was high, thus, only descriptive values are shown.


Of note, the adjectives, ‘sharp’, ‘stinging’, ‘prickling’, ‘pins and needles’, and ‘soreness’, referred to the skin above the muscle were reported by three individuals. Ultrasound examination at the completion of the infusion series in these three individuals revealed extravasation of metabolite solutions into subcutaneous tissues superficial to the muscle in these individuals. In all other subjects, infusions were entirely restricted to the muscle, and the only descriptions concomitant with pain were ‘ache’ and ‘hot’ (See Table 3). These descriptions are reported in Table 3, but for clarity, only data with the descriptors ‘ache’ and ‘hot’ were used in graphs and statistical analyses of pain.

Table 3. Descriptions of non-pain and pain sensations evoked by metabolites
Descriptors (in order of how many times described)
Non-pain descriptors Pain Descriptors
Pressure Related:
pressure (8) Ache (27)
full (4) (modifier-dull (4))
heavy (3) (modifier-throbbing (1))
strip (2)
puffy (1)
swollen (1)
(total = 19)
Movement Related:
shaking (9)
twitching (2)
effervescent (1)
pulsing (1)
flowing (1)
shooshing (1)
vibration (1)
tingly (1)
(total = 17)
warm (9) Hot (3)
cool (5)
(total = 14)
Fatigue Related:
fatigue (4)
tired (2)
exhausted (1)
exercised (1)
well used (1)
(total = 9)
*pins and *needles (2) *Sharp (3)
*raw (1) *stinging (3)
(total = 3) *soreness (4)
*pins and needles (1)
(total = 11)
  • * (For pain sensations, only ache and hot were NOT associated with back leakage of the metabolites into the skin. In all cases, sensations of pins and needles, stinging, raw and soreness were referred to the skin, not the injected muscle, and leakage of metabolites into the skin was observed with ultrasound imaging!)

Non-pain sensations

As illustrated in 2-4, the infusion of metabolites at levels found in muscles at rest (pH 7.4 + 300 nm ATP + 1 mm lactate) evoked no sensations in nine subjects and very mild (VRS score = 1) non-pain sensations in one of the 10 subjects. This lasted only 6 s.

Details are in the caption following the image
Figure 2. Percentage of the 10 subjects (six men and four women) reporting non-pain and pain sensations during and/or after infusion of indicated levels of metabolite combinations

Two hundred microlitres of metabolites was infused over 30 s.

Details are in the caption following the image
Figure 3. Quantitative amounts of non-pain and pain sensations evoked by combinations of, and individual metabolites

A, average of verbal reports of amount of non-pain and pain sensations with combinations of metabolites as indicated (n = 9, five men and four women). One subject (male) was removed because of Verbal Response Scale (VRS) scores for non-pain sensations 10 times greater than any other subjects for the highest three metabolite infusions. B, average verbal reports of amount of pain sensations caused by 200 μl infusions of individual metabolites indicated below each bar (n = 4, two men and two women). *P < 0.05, **P < 0.01 increase from baseline (pH 7.4 + 300 nm ATP + 1 mm lactate), Student's paired t test

Details are in the caption following the image
Figure 4. Duration of non-pain and pain sensations evoked by 200 μl of metabolites (same subjects as in previous figures)

For this graph, all sensations (including ‘sharp’, ‘stinging’, ‘prickling’, ‘pins and needles’, and ‘soreness’) were included. Removal of these sensations would eliminate the pain sensations at baseline and would very slightly reduce the variance in pain duration at other metabolite levels.

Metabolites at pH 7.3 + 400 nm ATP + 5 mm lactate (found with minimal muscle contraction) evoked fatigue-related descriptions (fatigue, warming, flowing, pulsation; see also Table 3) in four subjects (average VRS = 1.33 ± 0.6, P < 0.049 when compared with baseline; Figs 2 and 3) and evoked tremor in the APB muscle in two subjects.

Metabolites at pH 7.2 + 500 nm ATP + 10 mm lactate evoked non-pain sensations in five subjects, with ‘pressure’ being the most common descriptor. Average VRS was 2.90 ± 1.08 (P < 0.01, when compared with baseline).

At pH 7.0 + 1 μm ATP + 15 mm lactate (found with moderate exercise), metabolites evoked non-pain sensations in five subjects, with an average VRS of 2.44 (±1.25) (P < 0.03) and lasting 276 (±195) s (for this VRS score, a single outlier was removed). This male outlier reported a non-pain VRS score of 65 (the highest other score was 10) for this metabolite administration; if this outlier had remained in the analysis, the mean and SEM would have been 8.70 and 6.69, respectively.

‘Warm’ and ‘tired’ were the most common descriptors (others included ‘full’, ‘fatigued’, ‘strip-like’, ‘shaking’ and ‘vibration’). In most cases, the non-pain sensations were referred to large regions of the thumb, sometimes including the entire region from the origin of the muscle to the insertion (this resulted in the description of ‘strip-like’). The sensations waxed and waned in a pulse-like fashion. However, the increases and decreases in sensation were not related to either the heart rate or respiratory rate as determined by manual palpation of pulse and observation of respiratory movements by the anesthesiologist.

Higher levels of metabolites evoked no greater non-pain sensations (VRS score 2.33 ± 0.9 for pH 6.8 + 2 μm ATP + 20 mm lactate and VRS score 2.67 ± 1.0 for pH 6.6 + 5 μm ATP + 50 mm lactate), although VRS scores at both of these levels were different from the baseline (P < 0.02 and P < 0.01, respectively). The most common specific non-pain sensations described by participants for these last two metabolite levels were ‘shaking’, ‘pressure’ and ‘warm’; see also Table 3).

Figure 4 indicates that the duration of non-pain and pain sensations increased in a similar pattern to the pain VRS scores as metabolite concentrations were increased.

It should be noted that none of the ‘pressure’-related descriptions were accompanied by swelling of the thumb. In addition, ultrasound examination of the thumb indicated that no ‘pocket’ of metabolites was formed with the injection that could have been perceived as pressure or swelling when these descriptors were used. Likewise, when warm or cooling were described, the thumb was not warmer or cooler. Movement descriptors were also not accompanied by movement of the solutions in the muscle. In all cases, the solution administration had ended long before the subjects reported any movement-related sensations. ‘Shaking’ and tremor-related descriptions, however, were accompanied by tremor of the ADP muscle of the thumb.

Pain sensations

2-4 also show that metabolites at levels found in muscles at rest (pH 7.4 + 300 nm ATP + 1 mm lactate) evoked no sensations of ache or hot. Higher levels of metabolites at pH 7.3 + 400 nm ATP + 5 mm lactate (found with minimal muscle contraction) also evoked no pain sensations. The solution of pH 7.2 + 500 nm ATP + 10 mm lactate evoked some pain sensations in four subjects (ache, VRS = 6.8, P < 0.05 when compared with baseline). However, pH 7.0 + 1 μm ATP + 15 mm lactate (found with moderate exercise) evoked stronger pain sensations in seven of the 10 subjects (VRS = 10.9, P < 0.01 when compared with either baseline or pH 7.2 metabolites). Higher levels of metabolites (corresponding to extreme exercise levels) evoked more and stronger pain sensations in all 10 subjects (VRS = 15.4 and 16.9 for pH 6.8 and 6.6, respectively, maximal VRS = 50, P < 0.001 for both). While VRS at pH 6.6 and 6.8 were not different from each other, pH 6.6 was different from VRS at pH 7.0 (P < .00003).

Consistent adjectives corresponding to pain sensations used by subjects were the words ‘ache’ and ‘hot’ (see also Table 3). Sensations of ache and hot were most often referred to the base of the thumb, a site remote from the location of the insertion or the tip of the infusion needle. In some cases, referral extended to the back of the thumb and to the adjacent index finger. As with non-pain sensations, pain sensations waxed and waned in a manner that was described as ‘throbbing’ by several subjects. Again, the throbbing frequency was not related to either the heart rate or respiratory rate.

As with the non-pain descriptors, sensations of ache were not accompanied by swelling of the ADP or by pockets of metabolites building up in the muscle. Sensations of ‘hot’ were also not related to heating of the ADP muscle by the administered solutions.

Individual metabolites

Four healthy volunteers received infusions of 200 μl over 30 s of individual metabolites at concentrations corresponding to levels that accumulate in muscle interstitium during extreme exercise (pH 6.6 or 5 μm ATP or 50 mm lactate; see Table 2). As shown in Fig. 3B, infusions of pH 6.6 with no ATP or lactate (protons only) evoked no non-pain or pain sensations, as measured by VRS scores and was significantly less than VRS evoked by the combination of all three metabolites at pH 6.6 in the opposite thumb of these same four subjects (P < 0.02, Student's paired t test). Likewise, infusions of 5 μm ATP at resting values of pH and no lactate (ATP only) produced a mean VRS score of 0.25, significantly less than the same concentration of ATP when combined with protons and lactate in the same four subjects (P < 0.02). Lactate at 50 mm at resting values of pH with no added ATP (lactate only) evoked an average VRS score of 0.5, again significantly less than the same concentration of lactate when combined with the other two metabolites (P < 0.01). This compares with the average VRS score of 16 from all 10 subjects when all three of the same metabolites were administered together.

Control-side injections

Controls for volume effects on evoked sensations included injections in the thumb on the side opposite to the active metabolite injections in five subjects. This consisted of a series of six infusions of neutral resting solution (pH 7.4 + 300 nm ATP + 1 mm lactate; all solutions buffered with 30 mm phosphate), each infusion being 200 μl over 30 s. One subject reported the non-pain sensation of warmth (VRS = 2) after the sixth presentation. None of the other three subjects reported any non-pain sensations. Three subjects reported no pain sensations during or after the first three presentations of the neutral solutions. One subject reported ‘ache’ and ‘sharp’ around the needle site during the first three injections. This subject and two others (total three subjects) reported a mild ache or soreness (average VRS 6.25) after the fourth presentation. However, two of these three subjects reported no sensations with the fifth presentation. The other subject continued to report ‘sharp’ localized to the needle tip with the fifth presentation. With the sixth and final presentation, two subjects reported no sensations, while one reported a mild ‘twinge’ (VRS = 2) around the needle and the other subject reported mild ache (VRS = 2) referred to the thumb. Overall, minimal pain sensations were induced with these infusions (mean VRS = 3.30, average duration = 86.22 s).


Our results suggest at least two populations of sensory neurons in humans that are activated by intramuscular metabolites; one mainly responding to lower metabolite concentrations and causing the perception of fatigue, the other mainly responding to higher metabolite concentrations and causing the sensation of pain. As shown in 2-4, fatigue-related descriptions and duration of sensations increased linearly as the concentrations of the infused metabolites increased from resting values to levels corresponding to moderate- to high-intensity exercise. Higher levels of infused metabolites produced no further increases in fatigue sensations. The results for ache and hot pain sensations differed substantially from this pattern. Low levels of metabolites evoked no pain sensations; pH 7.2, 1 μm ATP and 10 mm lactate elicited mild pain sensations in four subjects, and as metabolite levels increased beyond this, pain sensations continued to show a linear increase to the maximal levels applied, pH 6.6, 5 μm ATP and 50 mm lactate (see 2-4).

Agreement with previous animal studies

The present findings are consistent with calcium responses previously seen in isolated mouse dorsal root ganglion neurons and in responses of identified group III and IV fibres innervating the mouse foot muscles (Light et al. 2008; Jankowski et al. 2013). These previous experiments also showed two distinct populations of skeletal muscle afferents that respond to metabolites. One population responded to non-painful levels of metabolites, whereas the other population only responded to levels of metabolites that are associated with ischaemic and painful muscle contractions. In addition, our mouse experiments also documented that no individual metabolite in the physiological range activated calcium responses in dorsal root ganglion neurons.

Two subsets of metabosensitive muscle afferents in humans?

Subjects’ sensation descriptors showed two sets of adjectives consistently used. One set was elicited by metabolites at low concentrations, the other by higher concentrations of metabolites. The adjectives used at lower concentrations were non-pain descriptors. These included eight traditional postexercise fatigue adjectives, i.e. ‘tired, fatigued, tremor, twitching, shaking, pressure, heaviness and exhausted’ (Blanchard et al. 2001; Semmler et al. 2007), as well as other descriptors consistent with non-painful muscle activity, i.e. ‘pulsing, puffy, swollen, flowing, effervescent, cool and warm’. We believe all of these adjectives were related to sensations of muscle fatigue similar to that experienced during and immediately after moderate to heavy exercise.

We also found a distinct set of adjectives used by subjects during infusion of higher concentrations of metabolites. In this case, adjectives were synonymous with pain and included ‘ache’ and ‘hot’. Interestingly, unlike previous muscle infusion experiments that used hypertonic saline (Graven-Nielsen et al. 2003) or very low pH using phosphate-buffered saline (Frey Law et al. 2008), ‘cramp’ was not described by any of the subjects in the present study. Likewise, microneurographical stimulation of group III mechanoreceptors evoked the sensation of ‘cramp’, not ache or hot (Marchettini et al. 1996). Whether this is due to quantitative differences in the amount of pain evoked in the present study, or if this represents qualitative differences in the types of sensory afferents activated with the combination of metabolites used here is unknown. Taken together, these findings suggest that low levels of metabolites activate muscle sensory neurons that contribute to the perception of fatigue, while high levels of muscle metabolites activate muscle sensory neurons that contribute to the perception of pain.

Combinations of metabolites are necessary to evoke sensations of fatigue and ache

A critical finding of our experiments is that individual metabolites are ineffective at evoking the perception of fatigue or pain when applied at concentrations occurring during normal and even ischaemic exercise (Fig. 3). This is consistent with previous findings in animal models, which showed that individually applied ATP, lactate or protons were ineffective at low concentrations and only evoked responses when applied at very high concentrations corresponding to vascular and muscle injury (Rybicki et al. 1985; Graham et al. 1986; Kaufman & Rybicki, 1987; Thimm & Baum, 1987; Rotto & Kaufman, 1988; Sinoway et al. 1993; Reeh & Kress, 2001; Li & Sinoway, 2002; Reinöhl et al. 2003; Hanna & Kaufman, 2004; Hoheisel et al. 2004; Kindig et al. 2006; Light et al. 2008).

Conversely, the combination of at least three of the metabolites produced by muscle contraction (protons, lactate and ATP) activates a large proportion of the sensory neurons innervating skeletal muscle and evokes sensations of fatigue and pain in humans. This synergy of metabolites activating muscle innervating sensory neurons was suggested in previous animal studies (Immke & McCleskey, 2001b, 2003; Naves & McCleskey, 2005; Yagi et al. 2006; Birdsong et al. 2010). These investigations also suggested that ATP, via a P2X receptor, enhanced ASIC3 responses to protons. TRPV1 receptors may also play a role in this synergy (Light et al. 2008). Given that the present study showed that these same metabolites evoke sensations of fatigue and pain when injected into human skeletal muscle, it is possible that ASICs, P2X and TRPV1 receptors also mediate the response of muscle afferents to intramuscular concentrations of contraction-induced metabolites in humans as well.

Several other metabolites increase with muscle contraction and may also enhance the sensory responses to muscle contraction in humans. These include bradykinin, potassium and arachidonic acid metabolites, such as prostaglandin E2 and various cytokines (Mense & Schmidt, 1974; Mense, 1977; Kaufman et al. 1983; Rotto & Kaufman, 1988; Hoheisel et al. 2005; Cui et al. 2007). It is also possible that these metabolites either directly stimulate or enhance the sensitivity of metaboreceptors in a variety of conditions, such as metabolic derangements and disease states.

Implications of findings

Some of the many interesting implications of the present findings are listed below.
  1. Given that the same combination of metabolites that evokes sensations of fatigue at low concentrations evokes pain at higher concentrations, disease or stress might change the relative ratio of these metabolites to differentially affect muscle pain versus fatigue. Furthermore, this implies that disorders altering sensory fatigue could also alter muscle pain.
  2. The synergistic action of metabolites explains why most conditions that cause large increases in the individual metabolites (for example, metabolic acidosis) do not cause sensations of fatigue and muscle pain.
  3. The temporal and spatial differences reported for ache versus hot imply that different sensory neurons may signal hot versus ache. Previous studies showed that only 48°C evoked sensations of pain in human muscle (Graven-Nielsen et al. 2002). We found that combinations of metabolites in muscle evoked the non-pain sensation of ‘warmth’, and higher levels evoked ‘hot’. This suggests that metabolites may be a better stimulus for sensations of warm and hot in muscle than real increases in muscle temperature.
  4. The ‘throbbing’ nature of aching pain may not be related to muscle movement, heart rate or ventilation. Instead, this phenomenon may be related to the mechanisms of sensory receptor activation or brain processing (Mo et al. 2013).
  5. Metabolites evoked sensations of mechanical movement, including pressure, heavy, flowing and vibration in the absence of any mechanical stimulus. This suggests that metabolites activate at least some mechanoreceptive neurons in skeletal muscle.

Experimental considerations

We recognize several potential limitations of our experiments. Metabolite infusions were not blinded or randomized, and subjects knew when an infusion was occurring. Given that we were unsure of the potential intensity of evoked sensations, we administered metabolite solutions in an ascending series from resting levels to levels associated with ischaemia. Although this could have led to a progressive accumulation of metabolites within the muscle, this potential scenario is unlikely, because we waited, before administering the next dose, for at least 5 min after the sensations of the previous infusion has ceased. Furthermore, because we administered all components of the interstitial fluid with the metabolites, the resulting final concentration of metabolites should not exceed the concentration injected, and is likely to be less. We also ruled out volume effects in these experiments using six infusions of resting metabolites in the contralateral APB, which evoked minimal sensations.

The average VRS values were low (average 16 out of 100 maximum for pain). However, two subjects’ values were 30 and 40. It is unclear whether low pain VRS values resulted from the small amount of solution, and therefore area of the muscle affected, or from the dilution of the metabolites by interstitial fluid.

One problem was the often long delay between the infusion and sensations of fatigue and pain. This delay implied that the sensory neuron receptors were not near the administration site. Thus, it is possible that our assumption that receptors for the metabolites were on the superficial fascia of the muscle was incorrect.

The APB muscle is fairly specialized. It is highly vascular and highly innervated. For these reasons, it is likely that at least some of the activation characteristics for sensations in other skeletal muscles will differ from those reported here.

We used an on-demand method to determine subject's descriptions of fatigue and pain sensations. This was necessary because of the short time between changes in pain and fatigue sensations we observed in pilot studies and because of the lack of data on previous descriptions of fatigue-related sensations. Using data generated from this study, a short list of descriptions from which subjects may choose may be viable in the future.

Finally, we eliminated the VRS scores resulting from reports of ‘sharp’, ‘stinging’, ‘prickling’, ‘pins and needles’ and ‘soreness’ from the statistical analyses and Figs 2 and 3 because they were associated with extravasation of metabolites into the skin and because these sensations were referred to the skin overlying the muscle. We inferred that these sensations were being generated by skin receptor neurons, not those found in muscle. These sensations occurred in three of the subjects and, in two of these, began with the resting levels of metabolites and reoccurred with most of the other metabolite injections, with the VRS scores diminishing as the metabolite concentrations increased. Removing these sensations from the analyses had no effect on significant differences reported here but did cause a small decrease in the variance, and removed the appearance of pain sensations at rest in Figs 2 and 3, which is shown in the uncorrected duration graph in Fig. 4.


Our findings suggest that physiological levels of metabolites produced by contracting muscles, specifically protons, lactate and ATP, activate human muscle afferent neurons via metabolite receptors similar to those found in animal skeletal muscle. These receptors are likely to include ASICs, at least one P2X and TRPV1. Furthermore, all three metabolites are required for the sensations of fatigue and pain, though it is possible that other metabolites and receptors are also involved.

In addition, the muscle afferents in humans responding to these metabolites appear to consist of at least two separate populations. One population responds to lower levels of metabolites and is associated with the perception of exercise-induced fatigue. The other population responds to higher levels of metabolites that are seen with ischaemic exercise or during post-exercise muscle ischaemia, and is associated with the perception of pain. This is consistent with prior animal experiments.

Additional Information

Competing interests

None declared.


These studies were funded by a grant from the University of Utah, Department of Anesthesiology and by NIAMS AR060336 to K.C.L. M.A. and A.R.L. were supported by grants from NHLBI (HL-103786 and HL107529, respectively).