The activity of rats during 4 hours at the beginning of the light period was behaviorally scored. The mean duration of behaviorally scored sleep in undisturbed rats was 144.6 ± 56.8 minutes. Sleep-deprived rats never slept during the 4 hours of sleep deprivation. Undisturbed rats were not awoken while sleep deprived rats were awoken 80.8 ± 22.5 times during the 4 hours. Figure 1 shows, for consecutive 30-min intervals, the number of interventions which were required to prevent the occurrence of sleep. The number of interventions increased progressively during the first 2 hours and then remained at a high level (except for the 5th interval, i.e. the period from 2h00 to 2h30). A one-way ANOVA for repeated measures revealed a significant effect of the factor "interval" (p < 0.05). Posthoc test (Tukey test) revealed significant increase in the number of interventions between the first half hour and the fourth to eighth ones and between the second half hour and the seventh and eighth intervals.

Rats were sacrificed immediately following the 4 hours of sleep or sleep-deprivation. The whole brain, hippocampus, stomach and adrenals were removed, weighed and stored at -80°C. The stomach was opened and, after careful observation, we were unable to find any ulceration. Adrenal weight was compared between sleep-deprived and undisturbed rats and no significant difference was found (Kruskal-Wallis, p-value > 0.05). Body weight also did not significantly change between the two groups (Kruskal-Wallis, p-value > 0.05).

Hippocampal proteins were extracted, labeled with either Cy3 or Cy5 CyDye and separated by isoelectric focusing between pH 4 and 7 (first dimension). After reduction and alkylation, proteins were then separated following their molecular weight in a second dimension electrophoresis. After scanning the gels, the images of the 2D distribution of hippocampal proteins were analysed. Figure 2 shows a representative bi-dimensional map of the proteome of sleep-deprived rat hippocampi. CyDyes protein labeling combined with automatic spot detection enables visualization of about 4000 individual spots on a 2-D gel. The protein pattern from 4 h sleep-deprived rat hippocampi (n = 6) was globally very similar to that of undisturbed rat (n = 6). However, after matching spots, 87 spots showed significant variation between sleep and sleep deprivation: 64 spots were upregulated by a factor 1.2 to 2.94 following 4 h sleep deprivation, while 23 spots were downregulated by a factor 1.2 to 2.74. From the 16 selected spots sent to the MALDI-TOF mass spectrometry, we unambiguously identified 12 proteins. Their name, accession number, change in relative abundance and other MS data are summarized in table 1.

We next examined the effect on the adrenals proteome following the exposure of rats to 4 h of sleep deprivation. As for the hippocampal proteins, the protein pattern from 4 h sleep-deprived rat adrenals (n = 4) was globally very similar to that of undisturbed rat adrenals (n = 4). Fifty-seven spots were found to display a significant variation in relative abundance after 4 hours of sleep deprivation, compared to undisturbed rats: 18 spots were more abundant (factor of 1.21 to 1.62), whilst 39 spots were less abundant by a factor of 1.20 to 4.72. From the 16 selected spots sent to the MALDI-TOF mass spectrometry, we unambiguously identified 13 proteins. Their name, accession number, change in relative abundance and other MS data are summarized in table 2.

To confirm the validity of the quantitative data obtained in our proteomic analysis, we further performed selected western blots analysis on several samples to verify that our 2D-DIGE analysis indeed reflected true changes in tissue protein expression.

For the hippocampus, we exposed extracted proteins after separation by SDS-PAGE to antibodies directed against α-SNAP and NSF. After spot volume analysis, taking into account variations in sample loading revealed by the actin signal, the ratio of NSF between the sleep and sleep-deprivation conditions was -1.28 (similar to the value -1.33 from the DIGE analysis; figure 3). The ratio of α-SNAP between these two conditions was -1.15 (similar to the value -1.37 found in the DIGE analysis; figure 3).

We separated adrenals proteins by 1D-SDS-PAGE and exposed them to an antibody directed against glucose-6-phosphate dehydrogenase. After incubation with an ECL Plex fluorescent secondary antibody and scanning, the presence and abundance of the protein was analyzed (figure 4). After spot volume analysis taking into account variations in sample loading revealed by the actin signal, the ratio of glucose-6-phosphate dehydrogenase between the sleep and sleep-deprivation conditions was -1.47 (similar to the value of -1.49 from the DIGE analysis).

In the hippocampus of sleep-deprived rats, proteins showing a higher abundance participate in five main functions: cell metabolism, energy pathways, transport and vesicle trafficking, cytoskeleton and protein processing (translation, folding and degradation). Importantly, no protein related to stress response were identified in the hippocampus although cellular stress response genes were upregulated after 3–8 h of spontaneous waking 9.

All the functions attributed to proteins the abundance of which increases after sleep deprivation, require energy. An increase in the expression of genes related to energy metabolism has already been described in the literature 20. Our results support the hypothesis of a fast adaptation of neurons and/or glial cells to the increased metabolic demand of wakefulness relative to sleep 20. They are also consistent with an increase in energy requirement possibly related to a progressive increase in synaptic strength with prolonged wakefulness 21 (although none of the proteins identified here are the products of gene markers for synaptic potentiation 22).

Accordingly, two central enzymes for cellular energy metabolism were identified. Creatine kinase (ratio of -1.26), a major enzyme of the phosphotransfer system in cells, plays a central role in energy transduction in tissues with large, fluctuating energy demands (such as skeletal muscle, heart, brain and spermatozoa) 23. It acts in concert with other enzymatic systems to facilitate intracellular energetic communication (creatine kinase/phosphocreatine "spatial energy transport" system 23) and is responsible for maintaining the phosphocreatine levels, an acute energy reserve 24. Creatine kinase is a cytoplasmic protein also expressed in the mitochondrion whereas NADH ubiquinone oxidoreductase (aka. Complex I, ratio of -1.22) is only expressed in the latter. NADH ubiquinone oxidoreductase catalyzes the oxidation of NADH, the reduction of ubiquinone, and the translocation of 4H⁺ across the coupling, inner-mitochondrial membrane eventually leading to ATP synthesis 25.

The increased abundance of Rab GDI α (ratio of -1.31), vesicle fusing ATPase (NSF) (ratio of -1.33) and the α-soluble NSF attachment protein (α-SNAP) (ratio of -1.37) in the hippocampus suggests an increased regulation of transport and vesicle trafficking, especially the docking and dissociation of vesicles to their target organelles. As proteins interacting with SNARE, α-SNAP and NSF play a central role in the specificity of the docking between a vesicle and its target membrane 26. After vesicle docking, NSF will dissociate SNARE pairs and thus control the time and location of membrane fusion. Finally, Rab GDI α is an important factor in the control of Rab and thus the specificity of vesicular transport. By binding Rab, released after membrane fusion, Rab GDI prevents Rab from releasing its GDP molecule until it has interacted with appropriate proteins in the donor membrane 2728. This increased regulation of vesicle trafficking could underlie a general increase in synaptic activity reflecting the induced motor activity and alertness as well as a reaction to a stressful situation. Interestingly, several authors have found a related increase in gene transcription in the cerebral cortex of undisturbed rats (calcineurin, CaMKIV and the Rab family of genes, see 2930).

α-soluble NSF attachment protein is also the only protein from our study that is related to the β-soluble NSF attachment protein, found by Basheer et al. in the only other proteomic study of the effects of sleep deprivation published so far 11. These two proteins are very similar in sequence and have the same functions in the cell. However, the abundance of the α-soluble NSF attachment protein is increased after sleep deprivation while the expression of the β-soluble NSF attachment protein is decreased after sleep deprivation. This discrepancy could be explained by the fact changes in cytoskeletal proteins after sleep deprivation could go in various directions depending on the precise need of synapses (since we use a different duration and study a different brain area than Basheer et al.).

Sleep deprivation increases at least two proteins related to cytoskeletal function, neural development and synaptic plasticity: α-internexin and GFAP. α-Internexin (a component of intermediate filaments, ratio of -1.52) is a marker of neuronal tissue and a critical protein for neuronal development 31. Thus its increased abundance, after a short-term sleep deprivation in the rat hippocampus, may indicate the presence of new neurons or their morphological modification to fit a possible synaptic potentiation 21. Its importance for axonal caliber and nerve conduction speed, along with its correlation with glutamate transporter levels in the energetically compromised post-injury brain 32 gives α-internexin a potential role in energy homeostasis. Other studies also found it upregulated after voluntary exercise 33 in the same way as expression of LTP-related genes increased during extensive exploration 34. All in all, this higher abundance of α-internexin seems to be consistent with the homeostatic synaptic hypothesis 21. Here, since the duration of the sleep deprivation episode is probably too short to elicit neuron proliferation, the change in α-internexin abundance would simply be due to post-translational modifications related to synaptic plasticity.

Glial fibrillary acidic protein (GFAP) (ratio of -1.35) is a major component of the glial astrocyte cytoskeleton and is generally used as a hallmark of mature astrocytes 35. GFAP controls astrocyte motility and shape by providing structural stability to extensions of astrocytic processes but again the duration of the sleep deprivation episode is probably too short to elicit astrocyte proliferation. The observed increase in the abundance of GFAP in one spot may indicate an effect of sleep deprivation on cognitive function and synaptic plasticity. Indeed as both inhibitory avoidance and habituation alter the in vitro phosphorylation of GFAP 36, the observed change in abundance of GFAP in one spot might reflect such post-translational modification underlying a change in synaptic activity and/or efficacy induced by a stressful situation.

Dihydropyrimidinase-related protein 2 (ratio of -1.64) plays an important role in axon specification and elongation by regulating microtubule assembly, endocytosis of adhesion molecules, reorganization of actin filaments, and axonal protein trafficking 37. In the adult brain, the expression of dihydropyrimidinase-related protein 2 is dramatically downregulated 38. However, it remains expressed in structures that retain their capacity for differentiation and plasticity as well as in a subpopulation of oligodendrocytes. One hypothesis could be that this protein may be involved in neuron plasticity in the hippocampus during sleep deprivation.

Finally, since the transcriptional activator protein Pur-α has been implicated in diverse cellular functions, including transcriptional activation and repression, translation and cell growth 39, its increased abundance (ratio of -1.23) may indicate an increase in DNA transcription and translation, compatible with a general increase of cellular activity. This could indicate a genomic process beginning during the short sleep deprivation event possibly leading to broader DNA transcription in case the sleep deprivation duration increases.

Interestingly, two proteins were more abundant in the hippocampus during sleep, as compared to total sleep deprivation. Protein disulfide isomerase A3 (ratio of +1.83) was found to be more expressed in the hippocampus of undisturbed than of sleep-deprived rat. Protein disulfide-isomerase A3 is expressed in higher amounts in ischemia (another form of stress) in heart 40. It also modulates the redox state of the endoplasmic reticulum(ER), providing control of the ER Ca²⁺ homeostasis 41. But its two main roles are to catalyze disulfide bond formation and isomerisation in proteins and to inhibit their aggregation 42. Disulfide bonds play a major role in the folding and stability of many proteins. As different authors suggested an involvement of sleep in protein synthesis 29, an increased protein disulfide isomerase A3 activity may be involved in the control of the proper folding of newly synthesized proteins. This suggestion deserves two qualifications. First, since the sleep episodes were short in our experiment, protein disulfide isomerase A3 could rather be involved in the control of the release of previously inactivated proteins, a process that could be faster than an activation and translation of genes related to sleep 9. Second, we do not argue that sleep generally promotes protein translation, folding and degradation. Indeed, thiopurine S-methyltransferase (ratio of -1.42) also plays several roles in protein folding depending on its polymorphism and catalyzes the AdoMet-dependent S-methylation of thiopurine drugs such as 6-mercaptopurine 43 and is more abundant in the hippocampus after sleep deprivation.

Also, present in higher abundance in undisturbed rats, YPEL4 (ratio of 1.53) is a protein whose gene is highly conserved and expressed in various eukaryotic organisms 44. This suggests that it must play important roles in the maintenance of life. Moreover, its subcellular localization (in association with centrosome or mitotic spindle) suggests a function in cell division. Unfortunately, its real function has not been uncovered yet.

The number of manipulations which were required to prevent the occurrence of sleep increased with the time spent in the experiment, demonstrating an increase of the sleep propensity. The rate of interventions reached a plateau after 3 hours, at the time when the rats' sleep propensity was at its highest 45. This result is consistent with a previous study in which 3 hours of sleep deprivation triggers the homeostatic regulation of sleep, as indicated by subsequent episodes of sleep rebound 46.

An important issue for the interpretation of these results is to assess whether the higher abundance of some proteins is associated with wakefulness or rather reflects the consequences of stress. To address this issue, we characterized the effects of stress at two levels of description. First, at the macroscopic level, there was no evidence for sleep deprivation being stressful: we could not observe any stomach ulceration, any adrenal hypertrophy or any body weight loss in our rats. This could be expected as the stress duration was probably too short to induce the responses typically observed after 72 hours 47. Second, at a molecular level, we characterized the protein patterns in adrenals of sleep-deprived rats.

Interestingly, we found more proteins with an increased abundance in the adrenals from rat left undisturbed as compared to adrenals of sleep-deprived rats, an opposite trend to what we found in the hippocampi. Most proteins identified in adrenals and more abundant in undisturbed rats belong to cell metabolism. They catalyze the transformation of carboxylic ester (carboxylesterase, ratio of +2.05), aldehyde (aldehyde dehydrogenase, ratios of +1.93 and +2.32), trimethylaminobutyraldehyde (aldehyde dehydrogenase family 9 subfamily A1, ratio of +1.79) and serine (serine hydroxymethyltransferase, ratio of +1.23). One protein is also specifically involved in the synthesis of ATP: NADH dehydrogenase 1 α (ratio of +1.71). This potential increase in ATP production can be advantageous for the increased activity in the metabolic pathways cited above.

Some identified proteins are involved in protein assembly and the regulation of transcription. In the mitochondrion, the steroidogenic acute regulatory protein (ratio of +1.24) enhances the transport of cholesterol and its metabolism into pregnenolone, a steroid hormone involved in steroidogenesis (of progesterone, androgens and estrogens a.o.) 48. In the endoplasmic reticulum, the heat shock 70kDa protein 5 (ratio + 1.28) facilitates the assembly of multimeric protein complexes. And finally, in the cytoplasm, aspartyl t-RNA synthase (ratio of +1.31) assembles L-aspartyl and a t-RNA and consumes ATP. By getting involved in various metabolic steps and in various cell compartments, the increase in abundance of these proteins indicates an increase in protein transcription.

All these identified proteins contribute to the view of sleep as a mechanism of energy repletion 4 and restoration of molecular stocks 2.

Finally, serum albumin (ratio of +1.27), the most abundant blood plasma protein, can be considered here as an experimental artifact.

Another argument in favor of the absence of physiological stress due to sleep deprivation by gentle handling for 4 hours is the relatively low number of proteins the abundance of which increases in the adrenals of sleep-deprived rats. However one important protein identified here is glucose-6-phosphate dehydrogenase (ratio of -1.49). This enzyme catalyzes the first step of the oxidative phase of the pentose phosphate pathway. It is thus involved in the synthesis of NAPDH that may counteract oxidative stress generated following sleep-deprivation 49.

Stress can also be an important factor in the higher abundance of chloride intracellular channel protein 4 (ratio of -1.25) as chloride channels are important for maintaining a proper cell volume and cell resting membrane potential 50. These two important factors can be modified by stress at a cellular level.

α-1-Macroglobulin (ratio of -1.27) is a widely expressed protein able to inhibit proteinases 51. Its presence here could be explained by the fact that it is also an inhibitor of coagulation and could then be treated as an artifact rather than as a protein with a precise physiological function.

Male Sprague-Dawley rats weighing 170–220 g at the start of the experiment were housed by pairs and maintained in a 12-h light-dark cycle with lights on at 08:00, with standard food (PLUS type, SAFE) and water available ad libitum. Rats were accustomed to the housing facilities for at least 7 days prior to the beginning of any experiments. All experiments were approved by the Animal Ethical Committee of the University of Liege.

In the first group, rats (n = 9) were sleep-deprived by gentle handling 54. The animals were continuously observed during 4 hours at the beginning of the light period. Sleep deprivation was achieved by disturbing the cage bedding around the rat, stroking the vibrissae using a small brush and gently stroking the fur with the brush when the rat assumes a sleep posture 55. In the second group, rats (n = 9) were left undisturbed during these 4 hours.

Alertness state was monitored by two means: an automated activity recording setup and program, Gemvid 56, developed in our laboratory, and manual behavioral scoring. For each rat, the alertness level was determined during the light period before the test. Scores below this level during the test were considered as rest/sleep. Rat behavior was also manually scored by selecting alertness states as soon as they occured with software developed in our laboratory. The following alertness states were defined: rest/sleep, movement, eating, washing, foraging, and environment exploration. The software allowed the quantification of the number and duration of rest/sleep periods as well as the number and frequency of manual awakenings for sleep-deprived rats. Statistical analyses (ANOVA, Tukey and Kruskal-Wallis methods) were performed in Statistica 7.1 (Statsoft).

All rats were weighed before and after the observation.

Rats were decapitated immediately after the 4 hour period of sleep or sleep deprivation, and the brains were quickly removed and dissected on ice. The hippocampi were extracted from the brain according to Hortnagl et al. 57, weighed and quickly frozen at -80°C. The rat thoracic cavity was opened, the viscera were removed and adrenals were cut, weighed and quickly frozen at -80°C.

Proteins were extracted by 10 strokes of a Potter homogenizer in a lysis buffer containing 7 M urea (ICN), 2 M thiourea (GE Healthcare), 30 mM Tris at pH 8.5, and 2% ASB-14 (Sigma). The supernatant containing the solubilised proteins was precipitated using the 2-D Clean-Up Kit (GE Healthcare) and protein amounts were determined with the RC DC Protein Assay (Bio-Rad).

The stomach was cut along the smaller curvature, cleaned, pinned on an inspection board and examined for ulcerations.

CyDyes minimal labeling was performed according to the procedure published by Tonge et al. 58 with minor modifications. Briefly, each 25 μg of sample protein was mixed with 200 pmol of CyDyes (GE Healthcare) reconstituted in anhydrous dimethylformamide (Aldrich), according to its labeling group, and left for 30 minutes in the dark, at +4°C. The coupling reaction was then stopped by adding the same volume of 10 mM lysine (Acros) for 10 minutes at +4°C. An internal standard was prepared by mixing equal amounts of all the samples within the experiment. Experimental samples were either labeled with Cy3 or Cy5. Then they were mixed in pairs together with 25 μg of the internal standard labeled with Cy2. For the first dimension (isoelectric focusing), a rehydration buffer was added to each analytical sample. For preparative gels, the rehydration buffer was added to 250 μg of unlabelled proteins. This buffer contained 7 M urea, 2 M thiourea, 2% ASB-14, 0.2% dithiothreitol and 0.5% IPG buffer 4–7 (GE Healthcare). The solution was then poured in a strip holder, covered with a pH 4–7 IPG strip (GE Healthcare) and finally covered with 1 ml PlusOne DryStrip cover fluid (GE Healthcare). The following electrophoresis steps were successively applied: 12 hours of rehydration, 1 hour at 500V (step-and-hold), 1 hour to 1000V (gradient), 3 hours to 8000V (gradient) and 5.5 hours at 8000V (step-and-hold).

Before initiating the second dimension step, proteins in IPG strips were reduced in an equilibration solution (50 mM Tris-HCl, pH 8.8, 6 M urea, 30% glycerol (Bio-Rad) et 1.6% SDS (MP Biomedicals)) containing 1% DTT. They were then alkylated in the same equilibration solution containing 5% iodoacetamide (GE Healthcare).

IPG strips were placed on top of a classical SDS-PAGE gel (12.5% acrylamide (GE Healthcare)).

Electrophoretic migration was performed in an Ettan Dalt six apparatus (GE Healthcare) at 2.5 W per gel during 30 minutes and then 100 W for a maximum of 4 hours.

Once the second dimension step was realized, each gel was scanned at three different wavelengths corresponding to the different CyDyes with a Typhoon 9400 Laser Scanner (GE Healthcare). Gel images were cropped with the ImageQuant 5.2 program (Molecular Dynamics) in such a way that all the images represented the same area.

After scanning the gels, 2-D gel analysis software (DeCyder version 6.5, GE Healthcare) was used in this study for spot detection, gel matching and spot quantification relative to the corresponding spot in the internal standard 59. After matching, spots showing a statistically significant variation in relative abundance of at least 1.2 fold (Student t test, p < 0.05) and a minimal volume of 250,000 were selected for further analysis (Biological Variation Analysis (BVA) module of Decyder software). All spots considered by the software as significantly regulated at a threshold of ± 1.2 were visually checked, possibly selected and a pick list was created containing all spots with proteins sent to mass spectrometry.

All differentially expressed spots were automatically excised from the gels with an Ettan Spot Picker robot (GE Healthcare). The picker head had a diameter of 2.0 mm. Gel pieces were collected in 96-well plates designed for the digestion of proteins.

For hippocampal protein digestion, gel pieces are collected in 96-well plates designed for the Proteineer dp automated digester (Bruker Daltonics). Briefly: gels pieces are washed with 3 cycles of successive soaking in 100% ammonium hydrogenocarbonate 50 mM and a mix of 50% acetonitrile/50% ammonium hydrogenocarbonate 50 mM. Two additional washes are performed with 100% acetonitrile to dehydrate the gel pieces. 3 μl of freshly activated trypsin (Roche, bovine, sequencing grade) 10 ng/μl in ammonium hydrogenocarbonate is used to rehydrate the gel pieces at +8°C for 30 minutes. Trypsin digestion is performed for 3 h at 30°C. Peptide extraction is performed with 10 μl of 1% formic acid for 30 minutes at 20°C. Hippocampal protein digests (3 μl) are then adsorbed for 3 minutes on prespotted anchorchips (Bruker). Spots are washed on-target using 10 mM dihydrogeno-ammonium phosphate in 0.1% TFA-MilliQ water to remove salts. High throughput spectra acquisition is performed using an Ultraflex II MALDI mass spectrometer (Bruker) in positive reflectron mode, with close calibration enabled, Smartbeam laser focus set to medium, and a laser fluency setting of 65 to 72% of the maximum. Delayed extraction is set to 30 ns. Steps of 100 spectra in the range of 860 to 3800 Da are acquired at a 200 Hz LASER shot frequency with automated evaluation of intensity, resolution and mass range. 600 successful spectra per sample are summed, treated and de-isotoped in line with an automated SNAP algorithm using Flex Analysis 2.4 software (Bruker), and subsequently submitted in the batch mode of the Biotools 3.0 software suite (Bruker) with an in-house hosted Mascot search engine (MatrixScience.com) to the NCBI rodent database. The Swiss-Prot 50.5 release and NCBInr_20060605 release databases are used for the rat species. A mass tolerance of 100 ppm with close calibration and one missing cleavage site are allowed. Partial oxidation of methionine residues and complete carbamylation of cystein residues are considered. The probability score calculated by the software was used as a primary criterion for correct identification. For mascot scores slightly above the threshold of p value, measurement errors were carefully watched to detect possible false positive that usually give random mass measurement errors, while relevant identifications give constant measurement errors. Experimental and Mascot results molecular weights and pI were also compared. Adrenal protein digestion and spotting on Maldi targets have been carried out as previously described in Bohler et al. 60. Peptide mass determinations (PMF and MS/MS) were carried out using the Applied Biosystems 4800 Proteomics Analyzer (Applied Biosystems). Calibration was performed with the peptide mass calibration kit for 4700 (Applied Biosystems). Proteins were identified by searching against an in-house database downloaded from NCBi and by using Mascot (MatrixScience.com). The parameters for search in database allowed 2 missed cleavages, a tolerance of 0.5 Da on MS/MS fragments and 100 ppm on precursor mass, as well as carbamidomethylation on cysteine and oxidation of methionine as fixed modifications. For both identification, the probability score (Mowse score 61) calculated by the software was used as a criterion for correct identification.

To confirm the presence of some proteins and the difference between abundance levels, 1D western blots were performed with the same protein samples used in the proteomic analysis. α-SNAP and NSF were detected in the hippocampal protein extracts with mouse monoclonal antibodies (ab16391, concentration 1 μg/ml; ab16681, dilution 1/2000, Abcam). Glucose-6-phosphate dehydrogenase in the adrenals was detected with a rabbit polyclonal antibody (ab993, dilution 1/1000, Abcam).

For western blots, 50 μg of proteins was combined with an equal volume of Laemmli buffer and separated by electrophoresis performed on a 12% SDS-PAGE with a 4% stacking gel (in a Mini Trans-Blot Cell, Bio-Rad). A molecular weight ladder was separated along with the proteins of interest (MagicMark, Invitrogen). Proteins in the gel were then electrophoretically transferred (Trans-Blot SD Semi-Dry Transfer Cell) to a 0.2-μm low-fluorescent PVDF membrane (Hybond-LFP, Amersham). The membrane was blocked with 3% nonfat dry milk in Tween-Tris-buffered saline (TTBS) for 1 h at room temperature and incubated with the corresponding primary antibody in TTBS with 3% dry milk overnight at +4°C with gentle shaking. The membrane was washed with TTBS three times (10 minutes each), incubated in either ECL Plex goat-anti-mouse IgG Cy3 or goat-anti-rabbit Cy5-conjugated secondary antibody (dilution 1/2500, GE Healthcare) for 1 hour at room temperature and washed again three times in TTBS (15 minutes each). Membranes were then washed twice in TBS (15 minutes each). After drying at +37°C, membranes were scanned at the corresponding wavelengths with a Typhoon 9400 Laser Scanner (GE Healthcare). Membrane images were analyzed with the ImageMaster 1D software (GE Healthcare).