Effects of chloramphenicol on brain energy metabolism using ³¹P spectroscopy: influences on sleep-wake states in rat

Animals

Male Wistar rats (3–5 months old, weighing 250–300 g) were obtained from IFFA-Credo (L’Arbresle, France). All experiments were performed in compliance with the relevant decree of the French Agriculture Ministry (No. 03-505). At the end of each experimental section, the animals were killed with a lethal dose of anesthetic (Chloral hydrate) administered i.p.

Animal preparation for spectroscopic measurements

The method used for chronic probe implantation was crucial to the reproducibility of the measurements. Animals were anesthetized with chloral hydrate (400 mg/kg in saline, i.p.; Merck, Darmstadt, Germany). After the full induction of anesthesia, they were mounted in a stereotaxic frame (David Kofp Instruments, Tujunga, CA, USA) and their body temperature maintained at 36.5–37.0°C using a thermostatically controlled blanket. After a median sagittal incision, the skull was carefully cleaned and four non-magnetic screws (Molex Electronics, Bievre, France) were inserted into the bone through drilled holes (a pair in each frontal and cerebellar position). A home-made radiofrequency probe (copper wire, 1.25 mm in diameter, shaped into three loops including a microcapacitance for phosphorous resonance frequency tuning) was fixed on the rat skull between the bregma and lambda, according to the atlas of Paxinos and Watson (1998). Using depth pulses (Bendall and Gordon 1983), the surface coil could detect signals at a maximal distance of 3–4 mm, a distance well suited for brain measurements. At both ends of the probe, non-magnetic plugs were also soldered for connection to the final tuning and matching setup. To ensure robust anchorage to the skull, the entire assembly, including the non-magnetic screws, was covered with dental resin (Ivoclar, Lyon, France). Finally, two nuts were embedded in the resin on the left and right side to ensure stable head immobilization during NMR ³¹P spectroscopic measurements. Afterwards, the animals were housed individually and left undisturbed for recovery for about 15 days [food and water ad libitum; light/dark cycle (12-hour/12-hour); 24 ± 0.5°C].

Restraining system

For the collection of NMR spectroscopic data in anesthetic-free conditions, the head of the animal had to be comfortably immobilized and the body movements prevented using the following home-made restraining set-up: (i) a fixed plastic frame with screws positioned in a way allowing to be screwed into the nuts on the skull prosthesis and (ii) body movements were limited with a semicylindrical polymethylmethacrylate mould that extended beyond the frame holding the head of the animal. The animals were trained to the immobilization procedure for 8 days before the recording began (head and body restrained; one time/day during the light period).

Experimental magnetic resonance spectroscopy

After the habituation period, the rats were carefully positioned in the restraining set-up and the NMR probe was connected to the tuning and matching circuit with non-magnetic plugs. Under local anesthesia (Xylocain, Astra Zeneca, Reims, France), a catheter (21 gauge, Vialon; Becton-Dickinson, Franklin Lakes, NJ, USA) was also introduced into the tail vein to inject vehicles [saline: NaCl 0.9%, 1 mL; dimethyl sulfoxide (DMSO) + saline, 1 mL containing 2/3 DMSO + 1/3 saline; Sigma, Lyon, France] or substances (CAP: 400 mg/kg in 1 mL of saline; TAP: 400 mg/kg in 1 mL of DMSO + saline; i.p., Sigma). Moreover to maintain the temperature of the animals when placed in the NMR experimental position, they were covered with a small warming blanket forming part of a warm water circuit. The whole set-up was then introduced into the bore (diameter: 17 cm) of a 2 T superconducting magnet (85/310, Oxford Instruments, Oxford, UK). The field homogeneity was adjusted on the water signal acquired from the rat brain. This procedure was considered sufficient when the half width of the water peak was 18 Hz. ³¹P spectra were collected every 6 min using a one-pulse sequence (pulse duration, 60 μs; repetition time, 650 ms; numbers of averages: 600). To follow the evolution of the ³¹P metabolites (mainly PCr and ATP), the acquisition program used allowed us to acquire recurrently free induction decays. The effects of CAP and TAP were compared using the methods described above. For each substance, 40 spectra were acquired as follows: 15 spectra were acquired after vehicle injection (97 min) and 25 spectra after CAP injection (162 min); the same procedure was performed for TAP.

Quantification of the spectra and processing of the kinetics

Spectra were quantified using the ‘AMARES’ algorithm available in j-MRUI software (Naressi et al. 2001). The signals of phosphomonoesters, phosphodiesters, inorganic phosphate (P_i), and PCr were represented by an exponentially decaying sinusoid, γATP and αATP by doublets, and βATP by a triplet. All the lines in the doublets and in the triplet had an identical line width, which was optimized by AMARES (Vanhamme et al. 1997) (Fig. 1). The intensity ratios were 1 : 1 and 1 : 2 : 1 for the doublets and the triplet, respectively. Because of the experimental difficulties of operating on a non-anesthetized animal at 2 T (low sensitivity, residual motions blurring the static field, limitation of the experiment duration), it was not realistic to introduce other a priori knowledge than the number of expected spectral lines (phosphomonoesters, P_i, phosphodiesters, PCr, three ATP) and their spectral relative positions (chemical shift). Prior to analysis, omission of two initial data points was performed to reduce the detrimental influence of compounds belonging to membranes (phospholipids and other macromolecules), including ³¹P (Fig. 1). Results were normalized with respect to the peak intensity of PCr observed before CAP or TAP injection, and the proposed curves (39 points: points 1–14, vehicle injection, 97 min; points 15–39, CAP or TAP, 162 min) correspond to a mean of six independent measurements performed in different animals (Figs 2 and 3).

Determination of intracellular pH

The intracellular pH (pH_i) was calculated from the chemical shift difference between P_i and PCr in the ³¹P spectra as described by Petroff et al. (1985).

where δ represent the chemical shift of P_i relative to PCr. By convention, the PCr resonance is used as an internal reference assigning a chemical shift of 0.00 ppm.

Determination of the free cytosolic ADP

Absolutes concentrations of phosphorylated metabolites were calculated after correction for partial saturation. Free Cytosolic [ADP] was calculated from the pH and [PCr] using a creatine kinase equilibrium constant (K_eq) of 1.66 × 10⁹ M (Lawson and Veech 1979) and assuming that the concentration of total creatine is 10.3 μmol/g wet weight (Fitzpatrick et al. 1989).

Determination of the phosphorylation potential G

G, phosphorylation potential, was calculated from the free cytosolic [ADP], [ATP] and [Pi].

Statistics

The effects of CAP and TAP on PCr and ATP were evaluated by a variance analysis (anova; Statgraphics, Manugistics, Rockville, MD, USA) using two factors (animals, and drug duration and dose) and relative interactions. When the F ratio of a factor was significant (p < 0.05) and the interactions were not, comparisons between the mean factors levels were investigated by a post hoc multiple least significant difference (LSD) range test at p = 0.05. The same statistical methods were used for EEG power bands (expressed as a percentage of the total power).

Animal preparation for recording of sleep-wake states

MRS measurements

Recording of sleep-wake states

Vehicle and CAP administrations at light onset

Administration of saline produced no changes over the 12 h of the light period (Table 1, Fig. 4). This stability was also reflected in the episode number and duration (Table 1) as well as in the spectral components of the EEG (Fig. 5). CAP, at a dose of 400 mg/kg (i.p.), induced a significant increase in the waking state duration [+51%, F(31,27) = 36.52, p < 0.01, Table 1, Fig. 4] during the first 6 h of the light period when compared with saline-injected animals. This effect was attributable to a decrease in the number of episodes [−30%, F(31,27) = 2.82, p < 0.01, Table 1] and a marked increase in their duration [+183%, F(31,27) = 9.40, p < 0.01, Table 1]. Total duration of SWS sleep was slightly but significantly decreased [–18%, F(31,27) = 32.24, p < 0.01, Table 1, Fig. 4] over the first 6 h of the light period when compared with saline-injected animals. This effect was attributable to a decrease in the number of episodes [−36%, F(31,27) = 4.34, p < 0.01, Table 1] and an increase in their duration [+29%, F(31,27) = 6.76, p < 0.01, Table 1]. Finally, the duration of REM sleep markedly decreased during the first 6 h of the light period [−55%, F(31,27) = 16.25, p < 0.01, Table 1, Fig. 4] as compared to saline-injected animals. This effect was attributable to a consistent decrease in the number of episodes [−54%, F(31,27) = 15.53, p < 0.01, Table 1].

Table 1.   Sleep-wake changes induced by chloramphenicol (CAP, 400 mg/kg) and thioamphenicol (TAP, 400 mg/kg) administered intraperitoneally at the light-phase onset (values are mean ± SEM)

	Amount (min)			Episode number			Episode duration (min)
	Waking	SWS	REM sleep	Waking	SWS	REM sleep	Waking	SWS	REM sleep
REM sleep, rapid-eye movement sleep; SWS, slow-wave sleep; n = number of injections. *Indicate significance in a least significant difference post hoc test following anova,*p < 0.05 vs. vehicle.
Vehicle (n = 7)
10 am–4 pm	105.57 ± 6.33	229.93 ± 5.82	24.50 ± 3.62	45.43 ± 6.15	51.86 ± 4.84	15.14 ± 2.04	2.58 ± 0.37	4.70 ± 0.50	1.64 ± 0.20
10 am–10 pm	209.50 ± 12.30	444.50 ± 11.66	66.00 ± 5.31	95.43 ± 11.49	110.00 ± 9.69	39.86 ± 3.79	2.50 ± 0.52	4.25 ± 0.41	1.68 ± 0.08
CAP (n = 9)
10 am–4 pm	159.94 ± 12.63*	188.89 ± 11.68*	11.17 ± 2.09*	32.00 ± 4.90	33.44 ± 4.77*	7.11 ± 1.65*	7.32 ± 2.20*	6.65 ± 0.99*	1.74 ± 0.14
10 am–10 pm	251.78 ± 15.43	414.22 ± 14.50	54.00 ± 2.78	75.56 ± 8.21	86.67 ± 7.20	30.11 ± 1.88*	3.97 ± 0.79*	5.04 ± 0.43*	1.83 ± 0.11*
Vehicle (n = 10)
10 am–4 pm	114.15 ± 9.80	223.80 ± 7.76	22.05 ± 3.20	50.60 ± 5.37	54.10 ± 4.87	15.44 ± 1.20	2.36 ± 0.18	4.49 ± 0.46	1.42 ± 0.18
10 am–10 pm	208.95 ± 14.25	450.85 ± 13.61	60.20 ± 4.62	103.70 ± 10.43	114.90 ± 9.74	37.10 ± 1.45	2.11 ± 0.15	4.21 ± 0.39	1.60 ± 0.08
TAP (n = 10)
10 am–4 pm	129.85 ± 8.51	209.35 ± 8.15	20.80 ± 2.82	44.30 ± 2.31	49.70 ± 1.99	15.90 ± 2.12	3.01 ± 0.27	4.26 ± 0.22	1.31 ± 0.07
10 am–10 pm	222.85 ± 11.79	435.55 ± 12.29	61.60 ± 4.46	96.10 ± 7.54	108.90 ± 7.31	41.10 ± 3.47	2.40 ± 0.19	4.15 ± 0.27	1.52 ± 0.07

In after-CAP injection versus saline-injected animals, the powers in the θ and σ bands of the EEG in the waking state were enhanced (third, fourth, and fifth hour; p < 0.05, Fig. 5) and decreased (second, third, and fifth hour; p < 0.05, Fig. 5), while the beta-1 and beta-2 powers remained unchanged. For SWS, the power in the delta band was increased (third, fourth, and fifth hour hours, p < 0.05, Fig. 5), while the powers in the θ (third and fourth hours, p < 0.05, Fig. 5), α (first, third, fourth, and fifth hour, p < 0.05, Fig. 5), σ (second, third, fourth hour, p < 0.05, Fig. 5), and beta-1 (fourth hour, p < 0.05, data not shown) bands were decreased. Finally, for REM sleep, there were no significant variations in the spectral power within any of the bands (Fig. 5).

Vehicle and TAP injections at light onset

After injection of the vehicle (2/3 DMSO + 1/3 saline), no changes were observed in peak amplitudes of PCr, ATPα, ATPβ, and ATPγ, (Fig. 3a–d). After TAP injection, contrary to CAP, the ³¹P metabolites peak amplitudes remained constant over time (Fig. 3a–d). After vehicle administration (DMSO + saline), as for saline, there were no changes in sleep-wake states over the 12 h of the light period (Table 1, Fig. 4). TAP at a dose of 400 mg/kg did not induce significant changes in waking states, SWS, and REM sleep versus vehicle (Table 1, Fig. 4). The spectral components of the EEG remained unchanged after TAP administration when compared with vehicle (data not shown).

³¹P spectroscopy

The data obtained when using ³¹P spectroscopy in anesthetic-free animals indicate that CAP and TAP structural analogs have different effects on the cerebral energy production. Indeed, while CAP administration elicits successive and significant deficits in PCr (−30%) and ATP (−40%) components, TAP has no effects under the same experimental conditions.

Our results indicate that the sensitivity of the MRS method used (Ackerman et al. 1984; Naressi et al. 2001) was sufficient to detect significant variations in a low magnetic field strength. Using a surface resonator positioned on the rat skull in combination with dedicated software, we were able to acquire NMR data of signal-to-noise ratios ≥15. Moreover, the reproducibility of the measurements between animals also supports the validity of the entire set-up.

The effects exerted by CAP on ATP components were remarkable, with the magnitudes and kinetics of the CAP-induced declines occurring in a narrow range for all the components. This antibiotic appears to be an efficient blocker of complex-1 of the respiratory chain, avoiding its redox processes and the energy transfer necessary for ATP synthesis (Freeman and Haldar 1970; Abou-Khalil et al. 1980, 1980; Bories and Cravedi 1994). This result has also been suggested from data obtained using 355-nm laser stimulation of the rat brain via an optic fibre, capable of detecting the evoked fluorescence at 460 nm related to NADH (Mottin et al. 1997, 2003). Our present results confirm that CAP decreases ATP synthesis. The specificity of the changes reported is also reinforced by the fact that TAP, a structural analog of CAP, has no effect on ATP components (Abou-Khalil et al. 1980; Bories and Cravedi 1994; Freeman and Haldar 1970; Abou-Khalil et al. 1980). As already suggested, the differences observed between CAP and TAP are likely to be associated with the substitution of the p-NO₂ group (CAP) by a –CH₃–SO₂– group (TAP) (Freeman and Haldar 1970; Abou-Khalil et al. 1980, 1980; Yunis 1988; Bories and Cravedi 1994). Our results also indicate that the PCr component decreases after CAP treatment, this effect preceding the declines exhibited by ATP components. These particularities might suggest that PCr sensitivity to CAP is related to the ability of this antibiotic to inhibit protein synthesis (Stoner et al. 1964; Freeman and Haldar 1970; Abou-Khalil et al. 1980; Bories and Cravedi 1994). It is now clearly documented that synthesis of PCr molecule results from a cascade of events including the synthesis of creatine itself (Ellington et al. 1998; Ellington 2001). While creatine is a non-protein amino acid, its synthesis requires glycine, arginine, and methionine, its conversion to PCr being catalysed by the creatine kinase isoenzyme from creatine and ATP (Ellington 2001). The hypothesis of a CAP influence on protein synthesis, at least on the enzymes involved in the above cascade of events, is unlikely as TAP which is also capable of inhibiting protein synthesis (Cannon et al. 1990; Holt et al. 1993), remains inefficient on the PCr component. The PCr decline observed is thus dependent on a deficit in the aerobic production of ATP, most likely the ATP fraction required by creatine kinase isoenzyme processing. The shorter delay between CAP injection and the beginning of the PCr decline, when compared with ATP components, may reflect a faster diffusion of PCr (Speer et al. 2004). This also supports the notion that the creatine kinase system plays a key role in the energy buffering, i.e., ATP generation and consumption (Sauer and Schlattner 2004).

Estimation of the free ADP under control and CAP has been performed from pH knowledge at 2 T and repeated in an other field value (9.4 T). Considering that total creatine is 10.3 μmol/g wet weight (Fitzpatrick et al. 1989) in both experimental situations, free ADP was found close to 20 μM before CAP, growing to 26 μM under CAP. It is well known that the relationship between pH and [PCr] is reflected in the ADP, which decreases when the pH falls and increases when [PCr] falls. The decrease seen after induction of CAP may explain the increase in ADP but not the slight decrease seen in pH values after CAP administration. The increase in [ADP] after CAP induction reflects clearly an impaired mitochondrial function indicating that this antibiotic (CAP) is an efficient blocker of complex-1 of the mitochondria respiratory chain.

Sleep-wake cycle

The results obtained for the sleep-wake cycle are in good agreement with those obtained using spectroscopy. We found that CAP significantly reduces REM sleep, slightly decreases SWS, and significantly increases the waking state during the first 6 h of the light period. These changes are also accompanied by specific modifications of the EEG band powers. TAP has no effect on sleep-wake states or spectral components.

A set-up restraining the animals was used for MRS measurements, while the rats were allowed to move freely during polygraphic recordings. In keeping with ethical rules on animal experimentation, this protocol difference was accepted to limit animal use out of their normal housing. Irrespective of whether the animals were restrained or unrestrained, a pharmacological effect was obtained with CAP but not with TAP. Effects of CAP and TAP on sleep have also been observed previously in different animal species, and discussed mainly in relation with their ability to inhibit the general synthesis of proteins (Drucker-Colin et al. 1979; Petitjean et al. 1979; Chastrette et al. 1990). The ability of CAP to inhibit ATP synthesis and suppress REM sleep, as hypothesized previously (Jouvet 1994; Cespuglio et al. 1999, 2005), was not demonstrated when using direct in vivo measurements. Our finding that the CAP-induced declines in the ATP components correlated well with the inhibition of REM sleep supports the notion that this state of sleep is energy gated and extremely sensitive to ATP decrements in the brain (Jouvet 1994; Cespuglio et al. 1999, 2005). CAP also moderately reduced the occurrence of SWS while increasing the waking state. This SWS decrease might question the specificity of the CAP effect observed on REM sleep, but this is not supported by the changes occurring in the spectral power of the EEG bands. Indeed, CAP enhanced the power in the delta band while reducing those in the α and σ bands – such changes indicate a reinforcement of deep sleep (increase in delta band) at the expense of light SWS (decreases in σ and α bands). These observations might indicate compensatory homeostatic processes triggered in conditions of energy deficiency to save energy (Benington and Heller 1995; Cespuglio et al. 2005). This was not encountered with REM sleep, as besides its marked drop, the infrequent episodes were not associated with changes in the spectral band powers. It thus appears likely to limit the ATP production by blockage of complex-1 of the respiratory chain, thereby primarily affecting the occurrence of REM sleep. Finally, the main EEG modifications characterizing the waking state after CAP treatment consisted of a parallel increase and decrease in the θ and σ band powers, respectively, while the beta-1 and beta-2 band powers remained unchanged. These results indicating primarily, a good maintenance of the spectral components related to the waking state, were surprising as the deficiency in energy metabolites accompanying CAP treatment was expected to affect beta-1 and beta-2 band powers, which are well known for being associated with active waking (Benca 2000). The full maintenance of the components characterizing the EEG of the waking state may reflect the existence of a functional basic state protected against sudden variations in the energy transaction. Such a persistent functionality may rely on the poikilotherm energy status, these species being capable of maintaining a basic interaction with the environment in poorer conditions of energy supply (Karmanova 1982; Guppy et al. 1987).

In conclusion, the data reported here indicate that blocking of the respiratory chain by CAP elicits significant deficits in PCr and ATP production. Such deficits occur together with a marked decrease in REM sleep, an intensification of SWS and maintenance of the waking state. By directly evaluating the brain production of ATP, our study provides new evidence for a strong dependence of REM sleep on energy production.