Inflammatory responses in primary muscle cell cultures in Atlantic salmon (Salmo salar)

Cell culture and stimulation

Primary muscle cell cultures were assessed for differentiation and purity by light microscopy at 4× and 10× magnification (Data not shown). Nine grams of white skeletal muscle pooled from six fish provided sufficient cells to reach confluence when evenly split between two 6 well plates. Prior to performing the microarray analysis, confirmation that the cells responded to rIL-1β was carried out by real time PCR using IL-1β itself as a marker gene since it is known to increase in expression in response to rIL-1β stimulation. IL-1β expression was significantly increased (541 fold) in the stimulated samples compared to the control samples.

Microarray analysis

In order to better understand the changes in whole cell transcriptomic output, gene ontology analysis was used to indicate the biological processes that were modulated by the IL-1β stimulation. From the 2504 features retained for analysis, 2196 (88%) were annotated to a functional protein and 1945 (78%) were assigned at least one gene ontology (GO) identifier for biological process, enabling further assessment of biological function. These proportions reflect the annotation of all features on the microarray slide. Statistical analysis for enrichment for biological processes resulted in 1195 biological process GO terms being identified. The nature of GO analysis means that many of these are overlapping and only the non-redundant major groupings are presented (Figure 1). Observation of both the GO analysis and manual assignment identification of functions was used to assign genes to functional groups. The differentially expressed genes could be defined as belonging to a number of distinct functional classes, especially immune response, proteolysis, growth regulation and structural proteins, cell cycle and lipid metabolism. The directional expression changes indicated how these processes were being affected, with a general increase in the expression of immune related and protein metabolism genes, whereas growth, structural proteins and cell cycle showed a negative trend, with a majority of genes being down regulated in expression. A complex response was found for genes encoding lipid metabolism proteins, indicting major transcriptional changes relating to lipid mobilisation.

Immune response genes

Proteolysis

Growth regulation and structural proteins

Cell cycle and DNA metabolism

Lipid and sterol metabolism

Temporal response and interaction of IGF and IL-1β

To assess the effect of time of rIL-1β stimulation on primary myocytes on gene expression, rIL-1β stimulation was performed at 6, 24 and 48 h and four key marker genes from the microarray analysis (IL-1β, TNFα, MyF5 and IGFBP-6) were examined by real time PCR (Figure 2). IL-1β was highly increased in expression at all-time points but it was at 48 h that the highest increase in expression was found. TNFα also showed the greatest fold increase at 48 h however this was more due to a reduction in the control expression seen at 48 h, than an increase in the stimulated cells. MyF5 was consistently down regulated at all time points with no increase in effect seen after 6 h. Finally IGFBP-6 was increased at all 3 times, but with a maximum fold increase at 24 h and 48 h.

To assess the interaction between rIL-1β and rIGF-I primary myocyte cultures were stimulated with rIL-1β (25 ng/ml), rIGF-I (100nM), rIL-1β (25 ng/ml) + rIGF-I (100nM) or maintained as control. These stimulations were carried out for both 6 h and 24 h to determine if rIL-1β interfered with early effects that IGF-I may have on the cell cultures. The genes analysed were chosen to represent the immune response (IL-1β, TNFα and hepcidin) and protein metabolism/growth (atrogin-1, MyF5, IGFBPs-4, 5 & 6).

At 6 h co-stimulation of cells there was an up regulation of IL-1β and TNFα expression in response to rIL-1β stimulation, and this was not significantly altered by co-incubation with rIL-1 β + rIGF-I (Figure 3). Hepcidin was also found to be up regulated in response to rIL-1β (5.8 fold), with co-incubation with rIL-1β + rIGF-I reducing the magnitude of this increase ~30% (Figure 3). Regarding the expression of the IGFBPs, there was no effect of any treatment on the expression of IGFBP-5. IGFBP-6 was up regulated in response to rIL-1β (13.0 fold) and this effect was not altered by co-incubation with rIGF-I. However, stimulation with just rIGF-I led to a significant reduction in the expression of IGFBP-6 (Figure 3). Curiously IGFBP-4 was found to be significantly down regulated in response to co-incubation with rIL-1β + rIGF-I (-2.4 fold) but not by either treatment alone. MyF5 was found to be down regulated in response to rIL-1β (-7.8 fold) and this effect was not significantly altered by co-incubation (-10.7 fold) (Figure 3). Lastly, atrogin-1 was found to be significantly down regulated in response to stimulation with rIGF-I (-4.9 fold) but unaltered by rIL-1β treatment (Figure 3). Co-incubation with rIL-1β + rIGF-I however ablated the rIGF-I effect.

At 24 h co-stimulation of cells with rIL-1β + rIGF-I significantly reduced the expression of IL-1β relative to cells only stimulated with rIL-1β, from 654 fold to 427 fold (Figure 4). No significant effect of co-incubation of rIL-1β + rIGF-I was found on the expression of TNFα or hepcidin (Figure 4). Additionally co-incubation did not alter the expression of MyF5 or any of the IGFBPs. While rIL-1β alone significantly increased the expression of atrogin-1 (2.8 fold) this increase was not found in cells co-incubated with rIL-1β + rIGF-I (Figure 4). However the co-incubated cells had significantly increased expression of atrogin-1 compared to cells stimulated with just rIGF-I. rIGF-I alone also significantly reduced the expression of hepcidin (-1.9 fold) but had no effect on the other genes. All the genes tested that were also hybridised with sufficient intensity on the microarray showed the same direction and similar magnitude of response in this cell culture experiment.

Immune response

The immune genes up regulated included several proinflammatory cytokines such as TNF-α, IL-1β and IL-8, indicating that stimulated myocytes are capable of synthesising these cytokines and are undergoing a proinflammatory response. The response to IL-1β is extremely rapid in other cell types in fish [57, 58] and it is likely that within 24 h these molecules will have been secreted into the medium. Several genes in the inflammatory signalling cascade were induced including NFκB subunits p100 and p105, and the NFκb inhibitor (IκB), as seen during inflammation in other cell types [58]. Under normal conditions IκB binds to NFκB to inactivate it but IκB is phosphorylated by IκB kinase (IKK) and subsequently ubiquitinated and destroyed by the proteasome [59, 60]. A related key signalling molecule up regulated was MAP kinase-interacting serine/threonine kinase 2, central to the MAPK pathways involved in IL-1β signalling [61], and with additional roles in the regulation of IGF signalling [26]. Another important transcription factor up regulated was the MAPK activated jun-B which increases transcription of IL-1β responsive genes generally at AP-1 responsive sites [62]. Interestingly, although jun-B may be associated with inflammatory signalling, it also has a role in maintaining muscle mass and its over expression in mammals can induce hypertrophy [63], indicting complex regulation of transcriptional machinery. In parallel to this, several genes encoding proteins that have roles as anti-inflammatory factors were activated; these include two suppressors of cytokine signalling (SOCS 1 and 3), IL-10 and an IL-10 receptor chain. SOCS proteins are often co-regulated during inflammation to prevent cellular damage and are negative regulators of cytokine signalling and function that interferes with signal transduction from cytokine receptors. The SOCS genes have been characterised in salmonid fish [64] and are increased in expression following stimulation with several different cytokines including IL-1β, TNFα and IL-6. Other immune related genes such as hepcidin, ferritin, C type lectin and the complement system were also significantly increased in expression. Both hepcidin and ferritin control iron availability and have antimicrobial actions with ferritin sequestering iron to reduce availability to microbes [65], whereas hepcidin also has direct antimicrobial properties and is often described as an antimicrobial peptide [66–68]. C-type lectins recognise carbohydrate moieties and are often induced by proinflammatory signals [58, 69], to regulate a variety of immune processes including the complement system [70–72]. There was also activation of some genes that are components of the adaptive immune system, such as major histocompatibility complex (MHC) class I and CD4-like protein, but at the time point we examined the predominant immune gene response was by molecules of the innate defences.

Protein catabolic processes

A major proteolytic pathway in muscle is the ubiquitin proteasome pathway, which in mammals is believed to be responsible for the majority of muscle protein degradation initiated by a number of different stimuli including inflammation in mammals [30]. This pathway has also been seen to be activated in salmonid fish during muscle atrophy induced by food deprivation [45, 73, 74], hormonal changes [75], with some evidence of several components being modulated during immune responses [45, 76]. The end product of proteolysis is the release of free amino acids for de novo protein synthesis or for the oxidation of the amino acids and gluconeogenesis. Following the inflammatory stimulus, several components of the UBP were increased in expression in myocytes. Several ubiquitin E3 ligases, which initiate the targeting of proteins for degradation and a number of proteasome subunits from the catalytic core of the proteasome (β subunits 6, 7 and 8 and α subunit 6), were increased in expression. We hypothesise that these changes would result in increased protein degradation and reduced muscle growth releasing free amino acids, which in vivo would be reallocated to other organs, such as the liver as occurs in mammals [77, 78]. Although the predominant proteolytic genes modulated were related to the UBP system, cystatin B, an inhibitor of the acidic lysosomal cathepsins was down regulated, possibly indicating an increase in cathepsin bioavailability and activity [79]. In addition the calcium dependant protease calpain subunit 1 was down regulated following the IL-1β stimulation. This protease has roles in positive regulation of myofusion inhibiting the differentiation of myocyte cells [80, 81] and this may indicate a reduction of muscle cell differentiation.

Other proteases observed to be increased included collagenase 3, that is increased in expression in NFkB mediated inflammation in mammals [82–84] and during vitellogenesis induced muscle atrophy in salmonids [43]. Angiotensinogen, the precursor of both angiotensin I & II, was also increased in expression, and is known to interfere with the actions and production of IGF-I, which in mammals is mediated by the NFκB pathway in collaboration with protein kinase C [85, 86].

In general there was a clear effect of rIL-1β on the expression of genes related to catabolism as evidenced by a transcriptomic shift towards muscle catabolism by the increase in mRNAs related to protein degradation and the down regulation of protein degradation related genes that have positive effects on growth.

IGFBPs

The IGF system is instrumental in the control of protein synthesis and growth in both mammals and fish [87]. The activity of IGF is under tight control, often by a family of IGF binding proteins (IGFBPs), which have recently been characterised in salmonid fish [88]. They function by either stabilising the IGF or by competitively binding the IGF to prevent attachment to the IGF receptor [87] and thus reducing the anabolic effects of IGF on the cells.

We found several IGFBP encoding mRNAs were modulated by the proinflammatory stimulus. IGFBP-6 is thought to have a binding preference for IGF-II but also binds IGF-I [89]. These direct effects on the activity of both IGFs might drive the cells away from high levels of protein synthesis and anabolism towards a state of catabolism [90, 91]. Previous studies indicate IGFBP-6 expression is associated with the inhibition of cell proliferation in both fish [12] and mammals [89, 92]. Additionally IGFBP-6 expression is reduced during resumption of growth following starvation [20, 93]. These findings tend to indicate that IGFBP-6 expression has a negative relationship with growth due to the ability of IGFBP-6 to act as a negative regulator of IGF-I & II activity, thus making an increase in the expression of IGFBP-6 a potential marker of inflammation induced catabolism in salmon muscle.

Other IGFBPs 4, 5 and rP1 were all decreased in expression following the inflammatory stimulus. In salmonids IGFBP-4 expression in muscle is increased by anabolic stimuli such as refeeding after starvation [20, 93] and is positively related to the expression of the promyogenic transcription factors MyoD and MyF5 in vitro[12]. IGFBP-5 can potentiate the effects of IGF-I especially with regard to bone [94] and muscle differentiation [90]. In rainbow trout IGFBP-5 increased in expression in muscle during refeeding after starvation [93] and, in Atlantic salmon primary myocytes, the expression of IGFBP-5 decreased during cell proliferation suggesting this protein is associated with entry to cell cycle [12]. Together these results suggest the IGFBPs are responding in a coordinated fashion to reduce IGF signalling and altering the balance between anabolic and catabolic pathways.

Growth regulation and structural proteins

Many transcription factors involved in growth regulation were altered. CCAAT/enhancer binding protein delta was increased, and is a transcription factor with multiple functions, that is positively related to myostatin expression in mammals [95]. In rainbow trout muscle it is increased during energy reallocation caused by vitellogenesis [43] indicating a blocking of muscle growth. A second key transcription factor, NFκB, is often associated solely with immune function but also negatively regulates myogenesis via the transcriptional repressor YY1 [34, 96]. Both of these molecules were increased in this experiment by IL-1β. YY1 is likely to be a mediator of NFκB induced muscle growth inhibition, achieving this by silencing myofibrillar promoters in myoblasts [34, 96]. MyF5, a muscle specific transcription factor, regulates muscle cell differentiation [3, 97] and a reduction in its expression level in this experiment fits with our anticipated reduction of muscle growth markers in response to rIL-1β stimulation. Additionally we found a general decrease in expression of mRNAs coding for muscle structural proteins such as collagens, myosins, actin and keratin, consistent with the hypothesis that the muscle cells are undergoing a reduction in growth in response to immune stimulation, as previously shown in mammals [98].

Cell cycle

The cell cycle is largely mediated through the actions of cyclin/cyclin dependent kinase complexes [99]. In the salmon myocytes multiple cyclins were modulated by IL-1β stimulation strongly suggesting cell cycle activity is being altered. For example, cyclin D1 expression was increased and functions in combination with cyclin dependent kinases to initiate and progress through the G1 phase of the cell cycle [99, 100]. The increase of cyclin D1 may be related to NFκB mediated arrest of muscle growth by preventing myocyte differentiation [101]. Cyclin G2 was also increased and may inhibit entry into the cell cycle [102, 103].

The remaining cyclins A2, B (1 and 2) and E (1 and 2) were all decreased in expression. Cyclin A2 is a rate limiting factor during S-phase and DNA synthesis and entry to mitosis [104], whereas cyclins E1 and E2 are responsible for the transition from G1 to S phases and initiation of DNA replication [105]. Cyclins B1 & B2 have roles during the S-phase and the M-phase and are crucial for maintenance of the mitotic state [106]. Several other cyclin related kinases, cell division proteins and minichromosome maintenance complex components were generally down regulated indicating a major reduction in cell cycle activity and DNA metabolism in these primary muscle cells under an inflammatory stimulus.

Lipid and sterol metabolism

A final group of genes found to be altered were those related to lipid and sterol metabolism, here several cholesterol transport proteins were increased in expression including Apo L3, glycolipid transfer protein and lipoprotein lipase. Apo-L3 is known to be a TNF-α inducible protein and its expression is known to be involved in the activation of the NFκB signalling pathway activated by cytokines in mammals [107]. The increase in the lipoprotein lipase could reflect an increased breakdown of lipoproteins for immune or cellular processes; this gene is under the control of many different signals in mammals including insulin, nutritional state and cytokines [108]. Prostaglandin-endoperoxide synthase 2, a gene known in mammals to be inducible by a variety of inflammatory substances [109], was also increased as a result of the rIL-1β stimulation. Apo A1 binding protein and Apo B were reduced in expression as well as several other sterol synthesis proteins. These results indicate that lipid metabolism is being actively changed in these myocytes under the inflammatory stimulus, resulting in complex changes in transcription of their mRNAs. Many of these changes could be mediated through intracellular crosstalk with the IGF/insulin pathway(s) [26, 110].

Interaction between IGF-I and IL-1β

Results from the microarray clearly indicated genes involved in the IGF regulation were being modulated by IL-1β, especially a number IGF binding proteins. We performed additional experiments to address if transcripts altered by inflammation would be modulated by IGF-1, a hormone which drives cells towards an anabolic status. Atrogin -1 a key gene involved in protein degradation was reduced in expression by incubation of cells with rIGF-1 as previously reported, conversely it is increased following incubation with IL-1β as occurs in mammals [31]. When cells were co-incubation with rIL-1β and rIGF-I an almost total inhibition of the atrogin-1 down regulation was found, suggesting the proinflammatory signal is blocking the anabolic effect of IGF-1. IL-1β alone results in an increase in atrogin -1 expression at 24 h as found in mammalian cells stimulated by proinflammatory cytokines [111]. The co-stimulation also decreased the magnitude of the response of the antimicrobial peptide hepcidin, highlighting an alternative allocation of resources depending on the signalling the muscle cells are receiving. Together these results show how anabolic signals may attenuate transcription of immune defence molecules and that proinflammatory signals can increase catabolic effects in the cells.

Myosatellite isolation and stimulation

Atlantic salmon (mean weight of 25 g and mean length of 12 cm) were used for skeletal muscle myosatellite cell extraction, as previously described [112–114]. For each muscle extraction 6 fish were used (~1.5 g tissue from each fish), this was to remove any individual fish effects. No experimental procedures were carried out on the fish and fish maintenance was in line with national ethical guidelines in an experimental facility at University of Aberdeen, UK. Fish were maintained in freshwater and fed a commercial diet at 1.5% body weight per day. Fish were killed using a schedule one method and muscle tissue from above the mid line of the fish was removed sterilely with scalpel and forceps. This pooled muscle (approx 9 g) was placed into a pre-weighed flask containing 30 ml of Leibovitz L15 medium (Gibco) + penicillin/streptomycin 1% (Pen/Strep, Gibco, Penicillin 10,000 units/ml, streptomycin 10,000 μg/ml) (L15 + P/S). The muscle was diced into small blocks (2 mm³) using sterile procedures and the diced muscle then centrifuged (300 g, 5 min) and the supernatant removed. The tissue was digested in collagenase (0.2%) at 11°C for 1 h. Following digestion the cell suspension was centrifuged (300 g, 5 min) and washed before being centrifuged again (300 g, 5 min). This pellet was digested in trypsin (0.1%) at 11°C for 30 min. The cell suspension was again centrifuged (300 g, 40 sec) and the remaining supernatant was added to L15 + P/S plus 10% foetal calf serum (FCS, Sigma) before being passed through a 200 μm nylon mesh. The suspension was centrifuged (300 g, 15 min), the supernatant was removed and 12 ml of L15 + P/S + 10% FCS were added. Finally the contents of this tube were added to two 6 well plates. Prior to plating laminin (mouse laminin, Sigma-Aldrich) was applied to the well surfaces 24 h before the cells were plated out, at a concentration of 1 mg/ml. Cell cultures were then left for 1 h for microsatellite cells to bind to the surface before the medium was first changed and cells allowed to differentiate at 22°C, with the medium being changed every 2 days.

Stimulation of cells

Cells were cultured for 4 days to allow cellular differentiation, this was observed using light microscopy. Morphology typical of satellite cells and time taken to reach confluence in our system was approximately 6 days. The cells were found to exhibit the typical growth pattern previously observed for muscle satellite cells as described in Bower & Johnston (2010) [12]. Initially cells were mono-nucleic before beginning to proliferate and differentiate into spindle shaped cells, a small number of which were beginning to fuse together by day 4. For the microarray experiment stimulations, the medium was removed and 1 ml of new medium (with 0.5% FCS) containing either 10 μl recombinant trout IL-1β protein (rIL-1β) to achieve a concentration of 25 ng/ml or a non-stimulated control with 10 μl PBS. The concentration of IL-1β has previously been determined to be optimal in salmonid cell lines [58]. The cells were then stimulated for 24 h before RNA extraction was carried out.

Subsequent experiments were carried out to further investigate the effects of rIL-1β at different time points and to investigating the effects of the anabolic hormone IGF-I on rIL-1β actions. For these experiments the same procedure was carried out with the only alterations being the length of time the cells were stimulated and in some cases the addition of 100nM of recombinant human IGF-I (rIGF-I) (Sigma). For these experiments the cells were cultured in the stimulant for either 6, 24 or 48 h as specified.

Study design and sample replicates

Cells cultures were generated from six individual fish, this allowed purification of sufficient cells for two six well plates. For the microarray four of these wells were used as biological replicates and stimulated with rIL1β and the remaining four were mock stimulated as described above. RNA extractions were performed and the stimulated samples were kept separate whereas the control unstimulated samples were pooled to have a single common reference prior to mRNA amplification and labelling (Figure 6).

RNA extraction

For microarray experiments RNA was isolated using the RNAeasy extraction kit (Qiagen) as per the manufacturer’s instructions. For all other samples RNA was isolated using Trizol (Sigma) as per the manufacturer’s instructions. In both cases the RNA was resuspended in 50 μl of nuclease free water and concentration measured by Nanodrop ND1000 (LabTech). The quality of the RNA was assessed using an Agilent Bioanalyzer RNA 6000 Nano Kit as per the manufacturer’s instructions.

Microarray hybridization and analysis

Microarray analysis of the samples was carried out using a custom-designed Agilent microarray platform with 4 × 44 K probes per slide (Salar_2; Agilent Design ID:025520) as previously described [115]. The microarray platform design is available at array express accession number A-MEXP-2065.

To produce labelled template for hybridisations aRNA was generated using a MessageAMP II aRNA Amplification Kit (Ambion) as per the manufacturer’s instructions. Briefly 1 μg of total RNA was reverse transcribed to create first strand cDNA. This was then used in the synthesis of second strand cDNA and this product was purified using the supplied reagents and columns. Finally the in vitro transcription to synthesise amino allyl modified aRNA was carried out to incorporate amino allyl dUTP in to the aRNA after a 14 h incubation at 37°C and the product purified using the supplied reagents and columns.

For incorporation of flouresence dye 3 μg of aRNA in a volume of 10 μl was denatured at 70°C for 2 min, and to this was added 3 μl of NaHCO₃ and 2 μl Cy dye (Cy3 or Cy5 mono-reactive dye pack, Amersham, resuspended in DMSO). The dye was allowed to incorporate for 1 h in the dark before excess dye was removed using a DyeEx spin column purification kit (Qiagen). Confirmation of dye incorporation and aRNA recovery was by nanodrop spectrometry. Agilent microarrays were hybridised according to the manufacturer’s instructions as described by [115]. Briefly 825 ng cDNA of each labelled template was fragmented in the dark and made up to a final volume of 20 μl with nuclease free dH₂O. After fragmentation, 57 μl of 2XGEx hybridisation buffer (Agilent) was added to each sample which was then briefly mixed and spun down before being stored on ice in preparation for loading 103 μl onto each microarray. Four biological replicates of rIL-1β stimulated cells aRNA were labelled with Cy3 dye and a control consisting of four biological replicates of control cells aRNA was labelled with Cy5 dye. Each rIL-1β stimulated aRNA was hybridised against the control. Hybridisations were carried out in a microarray hybridisation oven (Agilent) overnight (18 h) at 65°C. Following hybridisation the slides were washed in the gene expression wash buffers 1 and 2 (Agilent) following the manufacturer’s instructions. The slides were scanned using a GenePix personal 4100A scanner (axon Instruments) at a resolution of 5 μm. Files saved as ^*. TIF files were extracted using feature extraction software v9.5.3 (Agilent) and background correction and normalization were carried out within this program. Statistical analysis was performed using the Genespring GX analysis platform (version 9.5, Agilent Technologies). Significant differences between IL-1β stimulated cells and control cells were established by t-test analysis (P < 0.05) followed by correction for multiple tests (Benjamini Hochberg FDR post hoc test). Further filtering was carried out to maintain only those genes that showed a ≥2 fold difference in expression as a result of the stimulation. The raw hybridisations data have been deposited at ArrayExpress under accession number (http://www.ebi.ac.uk/arrayexpres, E-MTAB-1692).

Gene ontology (GO) enrichment was carried out on all features with associated GO identifiers using GOEAST software [116]. Fishers exact test was used within the GOEAST program to determine if the GO identifiers occurred significantly more often in a group than would be expected by chance. The output from GOEAST was entered into the software REVIGO [117] to remove redundant GOs.

Gene expression analysis by real time PCR

Availability of supporting data

The microarray data is submitted to Array express public archive (E-MTAB-1692). Other supporting data is as supplementary files attached to this paper.