Lipopolysaccharide-induced cytokine secretion from in vitro mouse slow and fast limb muscle

Abstract

Skeletal muscles play important roles in innate immunity. However, in vitro, their sensitivity to lipopolysaccharide (LPS) is low. In other tissues, LPS sensing is facilitated by the presence of plasma, LPS binding protein (LBP) or soluble CD14 (sCD14). This study addressed whether these are critical for LPS sensitivity in skeletal muscle and whether LPS responsiveness is different between slow vs. fast muscle. Soleus (SOL) or extensor digitorum longus (EDL) muscle from adult male C57bl/6 mice were mounted in 1 ml oxygenated baths containing: 1) buffer only; 2) buffer+1% mouse plasma; 3) buffer+1 μg/ml LBP; or 4) buffer+1% plasma from sCD14^−/− mice. In each condition, muscles were exposed to LPS from 0–1.0 μg/ml. Bath samples were collected at 0, 1 and 2 h, and analyzed using cytokine multiplex arrays. In both SOL and EDL the predominant responding cytokines/chemokines were KC(CXCL1), IL-6 and MCP-1(CCL2) and their average responses were amplified by ~10 fold in the presence of 1% plasma. Overall, SOL and EDL exhibited similar secretory responses in the presence of 1% plasma, with a lower limit of sensitivity to LPS of 0.01 μg/ml. LBP supplementation did not augment secretion; however, 1% plasma from CD14^−/− mice suppressed cytokine/chemokine secretion from EDL muscle. In conclusion, intact slow and fast mouse muscles have similar cytokine/chemokine responses to LPS but depend on the presence of low levels of plasma constituents. Though sCD14 plays some role in EDL muscle, neither sCD14 nor LBP can fully account for the strong effects of plasma on LPS sensitivity.

Introduction

Skeletal muscles secrete cytokines and chemokines in response to a variety of stimuli including muscle contraction, heat, catecholamines, and exposure to pathogens (1–6). The physiological importance of skeletal muscle responses to immune stimuli in the whole organism has only recently been demonstrated using muscle fiber-specific knockout of the toll-like receptor (TLR) adapter protein, MyD88 (7), and by knockout of muscle fiber IL-6 (8). In these transgenic models, whole body immune cell trafficking is altered and accumulation of a variety of plasma cytokines is suppressed during sepsis.

In most cell types, the primary pathways for sensing pathogen-associated molecular patterns (PAMPS) like lipopolysaccharide (LPS) arise from activation of TLRs on cell membranes. There is considerable evidence for TLR protein and mRNA expression in skeletal muscle fibers (9–12) and both myotubes and intact isolated limb muscle respond to LPS by secreting cytokines (2, 5, 13). However, there are numerous remaining questions regarding the nature of pathogen recognition and signaling in skeletal muscle fibers that have yet to be resolved. For instance, it is not clear why mouse limb muscle requires relatively high levels of LPS to induce cytokine secretion when the LPS is simply dissolved in buffer (e.g. 1 μg/ml) (5). In addition, the sensitivity of immune cells to LPS is greatly enhanced in the presence of low concentrations (0.2–1%) of autologous plasma (14). Whether similar effects are seen in skeletal muscles is not known. Another aspect to consider is the underlying mechanism for the effects of plasma on the sensitivity to TLR activation in immune cells is known to reflect interactions of LPS with LPS binding protein (LPB) and with membrane-bound or soluble forms of CD14 (mCD14 and sCD14, respectively) (14, 15). Though mCD14 is not known to be expressed extensively on skeletal muscle sarcolemma (16), the roles of sCD14 and LBP have yet to be established in muscle. Moreover, cytokine secretion in response to other stimuli are thought to be fiber type dependent (17, 18); however, it is unknown whether slow oxidative and fast fibers express similar cytokines or whether they exhibit similar sensitivity to LPS stimulation, even though soleus muscles have a greater abundance of TLRs (19).

Therefore, in this study we tested the following hypotheses: H1) LPS sensitivity of intact skeletal muscle is amplified by co-exposure with low concentrations of sterile, cytokine free, mouse plasma. H2) The effects of plasma on skeletal muscle LPS sensitivity is dependent on the influence of available LBP and sCD14. H3) Muscles with a predominantly oxidative fiber type will have a greater responsiveness to LPS in terms of net cytokine secretion per mass. Our results demonstrate that isolated skeletal muscles secrete an abundance of a variety of cytokines in response to LPS, that the response is amplified in the presence of 1% mouse plasma and that this may be partially due to the influence of sCD14 receptor availability in the plasma.

Methods

Animal and Tissue Procedures

All protocols were approved by the University of Florida Institutional Animal Use Committee (IACUC#201609372). Male C57BL/6 male mice (12–15 weeks old) were maintained in groups until they were transported to the laboratory from the animal vivarium the night before muscles were collected. All animals were kept on a 12:12-h light-dark cycle at 20–22°C/30–60% relative humidity (RH) and on a standard chow diet (LM-485 m Envigo;Teklad, Madison, WI).

Mice were anesthetized under isoflurane and their hind limbs removed. Soleus muscles (SOL) or extensor digitorum muscles (EDL) were excised rapidly under oxygenated (95% O₂, 5% CO₂) Krebs Ringers solution containing (in mmol/L) 0.45 Na₂SO₄, 0.6 Na₂HPO₄, 1.0 MgCl₂, 5.9 KCl, 2.0, CaCl₂, 21.0 NaHCO₃, 121.0 NaCl, and 11.5 glucose. Muscles were then placed in temperature controlled 2 mL tissue baths (Radnoti, Covina CA, #158303) at 35°C. The bath volumes were further reduced to 1 mL by placing glass rods around the internal perimeter of the bath. Muscles were set at 1 g of tension, which approximates the preload required to achieve optimal length in these muscles (5). Electrical stimulation of the muscles was avoided at all times because of the effects of contraction on cytokine production (20, 21). The muscles were first allowed to equilibrate in the baths for a 30 min acclimation period, then the buffer was changed and muscles exposed to fresh Krebs and either 1μg/ml LPS (Escherichia coli O127:B8, Sigma-Aldrich, St. Louis, MO) or combinations of other mediators. Bath samples at baseline, after the acclimation period (T0), 1 hr (T1) and 2 hr (T2) were taken, combined with a protease inhibitor cocktail (Bimake, Houston, TX) flash frozen in liquid nitrogen, and stored at −80°C for later multiplex analysis. Of the 25 analytes evaluated with Luminex cytokine kit on a Magpix System, 14 were identified in previous experiments to be responsive and were selected for statistical analyses: GCSF, GM-CSF, IL-1α, IL-6, IL-9, IL-10, IL-12p40, IL-12p70,IL-13, IP10(CXCL10), KC(CXCL1), MIP-1α(CCL3), MIP-1β (CCL4), and MIP-2(CXCL2). The Luminex tests were performed according to the manufacturer’s protocols.

Specific Experiments:

• An initial group of 8 mouse SOL muscles were tested for their cytokine response to exposure to 1 μg/ml LPS when only in Krebs buffer.

• Tests of SOL muscles were repeated in response to 1μg/ml LPS in the presence of 1% C57BL/6 cytokine-free plasma (IGMS-C57-N, Innovative Research, Novi, MI), 1μg/ml LPS + 1μg/ml LPS binding protein (LBP; 6635-LP/CF, R&D systems, Minneapolis, MN) or Krebs buffer alone. Greater than 1% plasma concentrations were attempted but discontinued because of uncontrollable bubbling in the bath and loss of sample. The commercially obtained C57BL/6 plasma was tested using Luminex MILLIPLEX MAP Mouse cytokine/chemokine premixed 25-plex assay kits (Sigma Millipore, St. Louis, MO) and confirmed that all cytokines were below the detection threshold. (n = 11–12 muscles per group).

• Tests of EDL muscle cytokine responses compared to SOL using 1μg/ml LPS in Krebs with 1% cytokine free plasma (N = 10 for SOL, N = 17 for EDL).

• Dose response curves for LPS in SOL and EDL. The muscles (n=40 males; 10 per dose tested, 12–15 weeks old) were first allowed to equilibrate for a 30 min acclimation period, then the buffer was changed and muscles exposed to either 0.0001 μg/ml, 0.001 μg/ml, 0.01 μg/ml, or 0.1 ng/ml LPS and 1% C57BL/6 cytokine-free plasma.

• Effects of sCD14 in plasma. SOL and EDL (n=7) from male C56BL/6 mice were tested for LPS responsiveness using plasma lacking sCD14 vs. the standard cytokine free plasma approach. The plasma was obtained from B6.129S4-Cd14tm1Frm/J mice purchased from Jackson Labs, Bar Harbor Maine. The sCD14^−/− plasma was also tested via Luminex, and values of 2 cytokines were slightly above the low-end cutoff and accounted for statistically by subtracting from measured T0-T2 values. Bath samples at baseline at the conclusion of the acclimation period (T0), 1 hr (T1) and 2 hr (T2) were taken as above prior to Luminex multiplex analysis.

Statistical Approach:

Group distributions of cytokine values were assessed for normality using the Shapiro-Wilks test. Nearly all groups were non-parametric, so nonparametric paired comparisons (Wilcoxon Signed ranks), nonparametric ANOVA (Kruskal Wallis) and nonparametric post hoc tests were used where appropriate. P<0.05 was considered statistically significant for all evaluations. In some experiments more than one muscle from the same mouse was included in a group (e.g. experiments on EDL). In this case, additional nested statistics were performed to account for non-independence between groups (22, 23). Specific statistical approaches are described in the figure legends and Tables.

Results

The cytokine secretory responses for 3 representative cytokines from isolated SOL muscles in response to 1 μg/ml of LPS dissolved in Krebs buffer only are shown in Fig. 1. In general, the responses during the T1-T2 interval were more robust compared to intervals T0-T1. The secretory response of the same cytokines to the addition of 1% sterile plasma to the buffer for both SOL and EDL are shown in Fig. 2. In the presence of 1% plasma, the concentrations of secreted cytokines at T2 were increased substantially for most cytokines. Because of the elevation in the secretory response, experiments were matched using SOL muscle with Krebs buffer, Krebs buffer + 1% sterile serum, and Krebs buffer + LBP (plasma protein believed to facilitate LPS-TLR signaling). The comparative results of significantly altered cytokines and chemokines measured within the Luminex panel are illustrated in Table 1. The secretory responses of KC(CXCL1), IL-6, MCP-1(CCL2), MIP1(CXCL2), GCSF and IP10(CXCL10) were greatly amplified in the presence of 1% plasma. However, the addition of LBP to Krebs buffer alone did not account for the effect of plasma supplementation and usually reduced responsiveness to LPS. Interestingly, IL-12p40, IL-13 and IL1α exhibited nearly complete inhibition of secretion in the presence of 1% plasma.

Figure 1.

Representative cytokines secreted from isolated SOL in response to 1 μg/ml LPS dissolved in Krebs buffer (n = 8 per group), Stats are Wilcoxon Signed ranks for matched pairs *P <0.05, ** P<0.01. T0 = the beginning of LPS exposure, T1 = 1 hour after LPS exposure, T2= 2 hours after LPS.

Figure 2.

Representative cytokine secretory responses in isolated mouse SOL and mouse extensor digitorum longus (EDL) in response to 1 μg/ml LPS in Krebs buffer + 1% serum. Stats used Wilcoxon-Signed ranks for matched pairs from time 0 to T1, and T1 to T2. ** = p < 0.01, *** p < 0.001, n = 10 for SOL, n = 17 for EDL. T0 = the beginning of LPS exposure, T1 = 1 hour after LPS exposure, T2= 2 hours after LPS.

Table 1.

Effects of 1% plasma and lipopolysaccharide binding protein on cytokine secretion from mouse soleus

Cytokine	Media	Mean+/− SD (pg/ml)	p ANOVA	p-post hoc from Plasma
Positive effect of 1% plasma
KC(CXCL1)	Krebs	86 ±118		<0.004
	Krebs + 1%P	1546 ±1469	<0.0001
	Krebs + LBP	26 ±14		<0.0001
IL-6	Krebs	17±23		<0.003
	Krebs + 1%P	357±299	<0.0001
	Krebs + LBP	5.3±4		<0.0001
MCP-1(CCL2)	Krebs	10±18	<0.0001
	Krebs + 1%P	95±55		<0.001
	Krebs + LBP	2±6		<0.0001
MIP2(CXCL2)	Krebs	43 ±22		ns
	Krebs + 1%P	169±138	<0.0001
	Krebs + LBP	7.4±5.5		<0.0001
GCSF	Krebs	6.9±10		<0.02
	Krebs + 1%P	34 ±29	<0.0001
	Krebs + LBP	1.8 ±0.74		<0.0001
IP10(CXCL10)	Krebs	1.8±0.58		<0.001
	Krebs + 1%P	9±13	<0.0001
	Krebs + LBP	1.6±0		<0.001
Negative effect of 1% plasma
IL-12p40 B	Krebs	20±8.9		<0.0001
	Krebs + 1%P	2.1 ±0	<0.0001
	Krebs + LBP	20±19		<0.002
IL-13	Krebs	43±35		<0.0002
	Krebs + 1%P	6.9 ±0.63	<0.0003
	Krebs + LBP	36 ±27		ns
IL-1a	Krebs	157±97		<0.0001
	Krebs + 1%P	4.9 ±6	<0.0001
	Krebs + LBP	86 ±52		<0.005

No significant effects of LPS treatment: TNFα, IL-12p70, RANTES, IL-1β, IL-15, or IL-10

ANOVA for non-parametric data, Kruskal-Wallis. Post hoc analysis was Dunn’s against Plasma + Krebs (n = 11–12/group)

To test for the possible role that sCD14 might play in the effects of plasma in cytokine secretion, the SOL and EDL muscles were tested when using plasma obtained from sCD14 knockout mice (sCD14^−/−). Loss of sCD14 had no significant effect on most cytokines. However, in the EDL muscle (Fig. 3A) the use of plasma from sCD14^−/− mice suppressed the responses of GCSF, IL-12p70, IP10(CXCL10), KC(CXCL1) and MIP-2(CXCL2) secretion accumulated by T2, suggesting an important role of sCD14 in fast muscle. In contrast, for SOL muscle (Fig 3B) cytokine secretion for GM-CSF, IL-6, IL-12p70 and MIP1α were modestly but significantly increased when plasma from sCD14^−/− mice was used.

Figure 3.

Effects of 1% sCD14−/− plasma vs 1% sCD14+/+ plasma on cytokine secretion in A: EDL and B SOL. Statistics are Wilcoxon Signed Ranks test. * =p <0.05, ** = p < 0.01, *** p < 0.001; Medians +/− 25–75% quartiles; (n = 10 for SOL, n = 17 for EDL).

Comparisons were made between the quantity of cytokines produced by SOL (a predominantly slow oxidative fiber limb muscle) and the EDL (a predominantly fast fiber limb muscle), shown in Fig. 4. Interestingly, the concentrations of specific cytokines produced in the tissue bath was nearly the same for both muscle types at T2. Only GM-CSF and IL-12p70 were significantly elevated in the SOL buffer compared to EDL; though the differences were small. When the results were normalized to dry weight, only IL-12p70 remained significantly higher in SOL. The lack of effect of muscle weight on the comparisons between SOL and EDL is not surprising since the average dry or wet weight of SOL in the C57bl/6 mouse is within 5–6% of the EDL (24, 25).

Figure 4.

Comparisons of total cytokine secretion (log scale) of SOL vs. EDL at T2. Muscles were incubated in Krebs + 1% plasma. T2 = accumulated cytokines at the end of the two hour exposure to LPS. Means ± SD, n = 11 for SOL, n = 16 for EDL. Comparisons are two sample t for parametric and Mann-Whitney for nonparametric samples. * = P<0.05, ** P<0.01

A separate series of SOL and EDL muscles were studied to determine the dose response relationships of LPS exposure for significant cytokine secretion. In Table 2, a summary of these responses is shown for three of the most predominant cytokines. Doses as low as 0.01 μg/ml induced significant cytokine release from isolated muscle in the presence of 1% plasma. Interestingly, the peak response was usually at 0.1 μg/ml and the responses plateaued or decreased at the highest LPS dose (1 μg/ml).

Table 2.

Dose response of isolated SOL and EDL muscle to LPS

	Median Cytokine Concentrations (pg/ml)
	IL-6		KC(CXCL1)		MCP-1(CCL2)
LPS DOSE (μg/ml)	Sol	EDL	Sol	EDL	Sol	EDL
0.001	23	101	97	186	26	56
0.01	360	214*	963*	723*	89	113
0.1	389*	361*	1336*	1349*	119*	131*
1	198*	423*	897*	1436*	89	85*

Median values (N = 10) for secreted cytokines in 1 ml muscle baths after 2 hour incubation (T2) in response to varying doses of LPS in Krebs buffer containing 1% plasma.

^*= p < 0.05 nonparametric post hoc analysis for differences from 0.001 dose.

Discussion

These results illustrate the functional capacity for limb muscle to sense and respond to pathogen-associated molecular patterns (PAMPs) such as LPS by secreting significant quantities of a variety of cytokines and chemokines. The cytokines and chemokines presumably function as the efferent arm of a myoimmune axis that communicates with the classic innate immune system during severe infection (7). This is consistent with recent findings in the whole organism in which knockout of specific components of the pathogen response in skeletal muscle fibers (i.e. MyD88 and IL-6) affects immune cell trafficking and circulating cytokines during sepsis (8, 26). Skeletal muscle LPS responsiveness in vitro was greatly amplified in the presence of a low concentration of plasma, a response similar to that described for immune cells (14, 15). Manipulation of two important variables present in the plasma, LBP and sCD14, could not fully account for the effects of plasma in both muscles, but removing sCD14 from the plasma in EDL experiments significantly suppressed secretion of most cytokines and chemokines. The results also demonstrate surprisingly equivalent cytokine responsiveness to LPS in both predominantly slow (SOL) and predominantly fast (EDL) fiber populations even when muscle mass was accounted for.

The sensitivity of in vitro skeletal muscle to LPS was amplified in the presence of 1% plasma but is still somewhat below the sensitivity of in vitro immune cells (14, 15). It is likely that the in vivo sensitivity of intact muscle would be higher because inherent diffusion barriers in a tissue bath setting would limit accessibility of LPS micelles to the interior of the muscle. The lowest sensitivity of muscle to LPS that we observed (0.01 μg/ml) is within a high range of values reported in the blood of patients with gram negative bacterial sepsis (e.g., 0.005 ug/ml (27), 0.025 ug/ml (28)), making the sensitivity to LPS potentially relevant to sepsis in intact organisms.

The specific array of cytokines secreted by muscle fibers is not well defined. Using an in situ wick method for sampling muscle interstitial fluid between the female rat adductor magnus and gracilis muscles in response to systemic LPS exposure, Borge et al. (3) found significant elevations in IL-1β, IL-6, KC(CXCL1), MCP-1(CCL2) IL-10 and Eotaxin (CCL11) within 3 h of initial exposure to LPS given systemically. They did not see significant LPS-responses of IL-12, IL-13, IL-18, INFγ, TNFα, MIP1α or RANTES. Our results in male mice for the highest secretory products (IL-6, KC(CXCL1) and MCP-1(CCL2)) were very similar. However, depending on the muscle type and specific environment, at times we also saw significant secretory responses of GCSF, GM-CSF, IL-12p70, IL-12p40, IP-10 and MIP-2. The wick method samples from the interstitial compartment and like this study represents the exudate from a mixed cell population. We and others have studied cytokines released from isolated myoblasts and myotubes in response to LPS (5, 13), but whether the cytokine responses of fibers or cells grown in culture can be considered normal representatives of the intact muscle remains debatable.

It is likely that cytokines secreted from an intact muscle reflect the response of multiple phenotypes which differ between the two muscles evaluated here. The EDL muscle is considered a fast glycolytic muscle, containing <2% oxidative Type I and IIA fibers, with high glycogen content, whereas the soleus contains ~37% Type I and 39% Type IIA oxidative fibers with high mitochondrial content (29). There are many other cellular phenotypes besides muscle fibers that could potentially contribute to cytokine secretion, such as resident macrophages, mast cells, adipocytes, mesenchymal stromal cells, satellite cells, endothelial cells and fibroblasts (30). A prime candidate is the resident macrophage, which resides in a relatively high population in muscle, making up >10% of the total nuclear material (31, 32). Interestingly, the resident macrophage density in a mouse gastrocnemius muscle (a fast muscle) is only 1/6th the macrophage density in the SOL (31). In rat limb muscle there is a similar elevation in the overall number of macrophages ~6900/mm³ in highly oxidative soleus muscle vs. the plantaris ~4500/mm³ or the tibialis anterior ~4200/mm³, both of which are fast glycolytic muscles (33). In the soleus, there is also a two-fold elevation in the number of resident proinflammatory, type M1 activated macrophages compared to comparable fast muscles (33).

The observed differences in responsiveness to sCD14 between SOL and EDL may reflect these differences in immune cell populations. Since fewer macrophages per mass of muscle are available in EDL and we observed that LPS responsiveness was facilitated by available sCD14, it implies that a greater proportion of cytokines in EDL may be produced by muscle fibers or other non-CD14 expressing cell types that require sCD14. A similar interaction has been described for smooth muscle, which also does not express mCD14 and responds weakly to LPS in the absence of sCD14, but strongly when sCD14 is present (34). In contrast, SOL muscle was less sensitive or even negatively affected by sCD14, which could reflect a predominant contribution to cytokine secretion from resident immune cells containing constitutive mCD14 (35). Consistent with our observations, cytokine secretion from immune cells containing mCD14 in response to LPS have previously been shown to be attenuated by supplementation of sCD14 (36).

These results could also be affected by different relative expression levels of TLRs and CD14 in both the muscle fibers and their resident immune cells in slow oxidative vs. fast muscle. However, we have very little specific information about the expression of these proteins in specific muscle phenotypes. Using a mouse model of muscular dystrophy, it has been shown that muscle-committed progenitor myoblasts from fast muscle (gastrocnemius) contain only 1/2 to 2/3 of the TLRs seen slow oxidative soleus or diaphragm (37).

Addition of 1 μg/ml LBP to Krebs buffer failed to augment cytokine secretory responses to 1 μg/ml LPS (Table 1) and in most cases tended to decrease the response seen in Krebs alone. The lack of an effect of LBP may be a concentration effect. Our experiments aimed to mimic extracellular concentrations of LBP in normal plasma, which are ~1–2 μg/ml. (14). However, based on work in macrophages, at low levels of LPS stimulation, nearly all concentrations of LBP facilitate greater TNFα release but at high concentrations of LBP, as we used in this study, cytokine release is attenuated in response to LPS (38) suggesting that LBP could introduce reactions with LPS that limit the responsiveness of the responding cells (38). The role of LBP in TLR signaling is complex. TLRs require LPS monomers as ligands but LPS exists in solution in the form of lipid micelles. LBP binds to the micelles and facilitates extraction of monomers by coordinated actions with CD14, which then delivers monomers to the TLR complex (39). So, the effectiveness of these co-factors depends on the state of LPS in the buffer system and the accessibility of micelles or other forms of LPS in the tissue microenvironment. CD14 also functions to internalize the CD14-TLR4 complex into endosomes, which is necessary for activation of the non-canonical, Myd88 independent pathway of signaling (39).

Both plain Krebs and the addition of LBP in Krebs resulted in a distinct pattern of response compared to that of the Krebs+1% plasma. Specifically, with either Krebs or Krebs+LBP, increases in IL-1α, IL-12p40, and IL-13 were substantial. In contrast, the responses were completely absent in the presence of 1% plasma. The difference in IL-1α is of particular note, averaging 24-fold greater at T2 than was seen when 1% plasma was included in the experiments. The loss of an IL-1α and IL-13 response in the plasma-treated samples could be due to the presence of soluble IL-1 receptor (sIL-1r) and soluble IL-13 receptor (sIL-13r). Both are considered plasma decoy receptors for these cytokines. They inhibit the activity of their ligands in several ways. The concentrations of sIL-1r and sIL-13r can be very high in plasma, e.g. exceeding 1 ng/ml for sIL-1r in humans (40) and >8 ng/ml for sIL-13 in the mouse plasma (41). Though IL-12p40 is not a soluble receptor, it can bind with multiple protein and peptide partners in plasma to form biologically active cytokines that could be missed in our Luminex cytokine assay (e.g. IL-12 and IL-23) (42). Therefore, the effects of plasma on reducing these three cytokines in the bath could be due to binding reactions with other proteins, receptors and peptides in the plasma-containing buffer.

Other factors within plasma could have affected the response to LPS, that we did not study, including soluble MD-2, which stabilizes TLR4 expression on the cell surface in other non-immune cells like epithelium (43) and greatly enhances TLR4 mediated NFκB activation (44). There are also many other peptides, amino acids, lipids and proteins within the plasma that have the potential to affect immune responsiveness and may have been responsible for our observations. Even simple amino acids or metabolites like glutamine, cysteine, branched chain amino acids or lactate can have important impacts on innate responses of immune cells and possibly on skeletal muscle fibers (45).

The potential for skeletal muscle cytokines to contribute to total circulating cytokines during systemic infection is substantial. As an organ system, it makes up ~40% of lean body mass in both mouse and man (46). The total amount of a cytokine such as KC(CXCL1) secreted by a SOL muscle in a 1 ml bath over 2 hours was ~1500 pg (Table 1). To illustrate, since mice have only a 2 ml blood volume, this secretory capacity of a single SOL muscle could account for all the circulating KC(CXCL1) accumulated in the blood at 6 hours into sepsis exposure (8, 50).

The specific cellular phenotypes within an intact skeletal muscle responsible for cytokine secretion remain poorly understood. The assumption is often made, based on the cytokine responses to exercise, that the primary source is the muscle fibers themselves. However, this is difficult to prove, particularly since direct regional comparisons using microdialysis in humans are consistent with much greater release in areas surrounding the peritendinous tissues compared to the release from the body of the muscle (47). Many macrophages in muscle are found in the perimysium (32, 48), i.e. the sheath surrounding the muscle bundle, but they are also found in the endomysium and in the spaces between muscle fibers (32, 49). Bioavailability to endotoxin dissolved in the muscle bath would be greatest in macrophages located at the perimysium, whereas macrophages buried deep within the muscle bundle would likely exhibit a delayed or suppressed response. An important area of future research will be to discover how the resident immune cells and other secretory phenotypes in the muscle interact with the muscle fibers and influence each other’s response in within the muscle microenvironment.

Previous reports of the influence of knockout of critical elements of pathogen responsiveness within skeletal muscle fibers (8, 26) have led us to view the integrated skeletal muscle-resident immune cell organ as an “armed reserve” of the innate immune system. It is unlikely that this system represents the “front lines” of infection control but is instead activated only when infection is sufficiently intense to cause the presence of high levels of pathogens or PAMPS to be released into in the circulation that can diffuse to muscle tissue microenvironment. Once activated, however, its potential for contributing to the network response of the whole organism could approach that of circulating immune cells. The responses of this myoimmune axis to infection remain an important frontier for research at the interface of immunology and physiology. In conclusion, intact in vitro skeletal muscle is sensitive to exposure to relatively low levels of LPS and responds by secreting a variety of cytokines and chemokines that are likely to play important roles in immune cell recruitment, trafficking, and enhancement of the innate immune responses to pathogens in the intact organism.