Mice heterozygous for both A₁ and A_2A adenosine receptor genes show similarities to mice given long-term caffeine

Abstract

Caffeine is believed to exert its stimulant effects by blocking A_2A and A₁ adenosine receptors (A_2AR and A₁R). Although a genetic knockout of A_2AR eliminates effects of caffeine, the phenotype of the knockout animal does not resemble that of caffeine treatment. In this study we explored the possibility that a mere reduction of the number of A₁Rs and A_2ARs, achieved by deleting one of the two copies of the A₁R and A_2AR genes, would mimic some aspects of long-term caffeine ingestion. The A₁R and A_2AR double heterozygous (A₁R-A_2AR dHz) mice indeed had approximately one-half the number of A₁R and A_2AR, and there were little compensatory changes in A_2B or A₃ adenosine receptor (A_2BR or A₃R) expression. The ability of a stable adenosine analog to activate receptors was shifted to the right by caffeine and in A₁R-A_2AR dHz tissue. Caffeine (0.3 g/l in drinking water for 7–10 days) and A₁R-A_2AR dHz genotype increased locomotor activity (LA) and decreased heart rate without significantly influencing body temperature. The acute stimulatory effect of a single injection of caffeine was reduced in A₁R-A_2AR dHz mice and in mice treated long term with oral caffeine. Thus at least some aspects of long-term caffeine use can be mimicked by genetic manipulation of the A₁R and A_2AR.

caffeine is the most widely consumed psychoactive drug (15). It is generally believed that most of its actions in habitually consumed doses can be explained by its being an antagonist of adenosine receptors (12, 15). Since competitive antagonists are active if the receptors are occupied by an agonist, only adenosine receptors that are activated by endogenous adenosine under physiological conditions are likely targets for the effects of caffeine. Therefore, the focus has been on adenosine A₁ and A_2A receptors (A₁R and A_2AR) (15), because these receptors where abundantly expressed are activated by endogenous adenosine levels. Indeed, it has been shown that some of the effects of caffeine are eliminated by knocking out the A_2AR (10, 20). However, the phenotype of the A_2AR knockout (KO) mouse is different from that of a caffeine-treated mouse. In doses commonly consumed by humans, caffeine (and its breakdown products) will bind to 25–50% of the A₁R and A_2AR. Such blockade if continued for a long time may have functional consequences similar to a reduction in receptor number. Our group (2, 22) has previously shown that mice with only a single copy of the adenosine A₁ or A_2A gene have approximately one-half the number of receptors. Therefore, we hypothesize that mice that have a reduced number (rather than a complete elimination) of both A₁Rs and A_2ARs may show features of mice that are treated for a long time with caffeine, and subsequently, the A₁R and A_2AR double heterozygous mouse could be used as a genetic animal model to mimic the effect of long-term caffeine consumption.

In the present experiments we have tested some of the underlying assumptions and also some corollaries. We have examined whether mice that are heterozygous for the A₁R and the A_2AR have reduced receptor numbers and show reduced responses to adenosine. We also have investigated whether there are major compensatory changes in other receptors. We have in particular examined whether these mice show increased locomotor activity such as mice given caffeine for a longer term do and whether they show reduced responses to caffeine. The results indicate that these mice show at least some features resembling long-term caffeine consumption.

MATERIALS AND METHODS

Animals.

The study was performed in accordance with the “Guide for the Care and Use of Laboratory Animals” published by the National Institutes of Health and the ethical guidelines of the European Union and Sweden, and it was approved by the Animal Ethics Committee of Northern Stockholm.

The adenosine A₁R KO mice were generated by inactivating the second protein-coding exon of the mouse adenosine A₁ receptor gene as previously described (22). The A_2AR KO mice were generated by inactivating the exon 2 of mouse adenosine A_2A receptor gene as described previously (6). The mice were backcrossed repeatedly against C57Bl/6 ( initially by 6 generations according to the Jackson Laboratory “speed congenic procedure” for A₁R KO mice and more than 10 generations for A_2AR KO mice, followed by 2–4 additional backcrosses during generation and maintenance of double heterozygote breeding) to ensure that they were practically congenic. The A₁R and A_2AR double-knockout (dKO) mice were generated by cross-breeding heterozygous A₁R KO and A_2AR KO mice and then using resulting double heterozygous (dHz) mice in further breeding. Genotyping relied on polymerase chain reaction (PCR). Mice were kept in individual cages under normal conditions with a 12:12-h light-dark cycle during the experiments. All animals were given free access to both food and tap water as described in more detail below.

Materials.

Bovine serum albumin fraction V (BSA), l-norepinephrine hydrochloride, 2-chloroadenosine, adenosine 5′-triphosphate (ATP), β-nicotinamide adenine nucleotide (NAD), and adenosine 3′,5′-cyclic monophosphate (cAMP) were obtained from Sigma (St. Louis, MO). 1,3-Dipropyl-8-cyclopentylxanthine (DPCPX) was obtained from Research Biochemicals International (Natick, MA). Collagenase (type 1; CLS) was obtained from Worthington (Lakewood, NJ), and adenosine deaminase, glycerol kinase, and glycerol-3-phosphate dehydrogenase (GAPDH) were obtained from Boehringer (Mannheim, Germany). [2,8-³H]adenosine 3′,5′-cyclic phosphate was obtained from PerkinElmer (Boston, MA).

Telemetry study.

The telemetry system (Data Sciences, St. Paul, MN) consists of implantable transmitters (TA10ETA-F20), telemetry receivers (DSI PhysioTel Receivers, RPC-1 model), and eight universal adapters (UA 10 PC). The computer program (Data quest A.R.T gold acquisition) sampled calibrated values of ECG and body temperature as well as noncalibrated locomotion activity (LA) counts.

The protocol of the telemetry study is shown in Fig. 3C. After a 7-day recovery from transportation, the transmitters were implanted in mice. As previously described (37), under anesthesia induced and maintained by isoflurane, the transmitter for telemetry was implanted into the peritoneal cavity and the leads were sutured in a lead II position. Analgesia was ensured by local application of 5% Xylocain salve (AstraZeneca, Södertälje, Sweden) and subcutaneous injection of Temgesic (0.3 mg/kg; Schering-Plough Europe, Brussels, Belgium) after the implantation. Each individual mouse was allowed at least 7 days of recovery before the start of registration. After 5 days of baseline registration (control), the mice received saline, followed by 7.5 mg/kg and then 30 mg/kg caffeine (Sigma; dissolved in saline) intraperitoneally at 2-h intervals. Two days later, the drinking water was replaced by tap water with 0.3 g/l caffeine or tap water only followed by a second period with tap water (caffeine or water). Heart rate (HR), body temperature, and LA were recorded on the 5th, 6th, and 7th days after the start of the second period experiment. The fluid consumption during caffeine treatment was, as reported earlier (33), similar between groups and on the order of 50–60 ml/wk.

Behavioral studies in an open field.

In some experiments activity was analyzed by recording the LA in a square open field arena (500 × 500 × 225 mm), enclosed in a solid and sound-attenuating box (Kungsbacka Mät och Reglerteknik, Fjärås, Sweden) (5). The locomotor box with the open field arena was equipped with two rows of photocells sensitive to infrared light, each having 16 photocells per side. The space between the photocells was 31 mm, and the outermost was placed 17.5 mm from the wall. The number of photocell interruptions was collected by a computer, and the following variables were recorded and analyzed: horizontal activity (total number of beam breakings), locomotion (interruptions of photocells in the lower rows when there is a new beam broken; i.e., the animal has made an actual transfer), rearing activity (all interruptions of photocells in the upper rows), and forward locomotion (interruptions of 2 or more consecutive photocell beams). This equipment does not allow us to record small movements, e.g., tremor, reflexes, and tail movements. The data were subjected to a square root transformation before statistical analysis to ensure a more normal distribution.

Before the recording, all animals were allowed a period of 30–45 min in a behavioral testing room. Adult mice were analyzed with an experimental protocol that spanned over 2 days. Two habituations (45 min each) were performed on day 1, separated by 2 h. On day 2, the third habituation (45 min) was followed by 90 (occasionally 240)-min recording after stimulation with caffeine (15 mg/kg). All measurements were made in the dark and performed during the light cycle between 8:00 AM and 4:00 PM.

Measurement of the concentrations of caffeine and its active metabolites in brain.

The concentrations of caffeine and its active metabolites in brain and plasma were measured 90 min after acute caffeine injection and at midnight after 10 days of caffeine drinking. When caffeine is dissolved in the drinking water, its plasma concentration in mice stays rather constant during the dark period but is very low during the light period (21). We used this method of administration to mimic the concentration profile of a human who consumes caffeine regularly. Mice were anesthetized with carbon dioxide and decapitated. The brain was rapidly dissected out and frozen on dry ice. Preparation of brain samples for HPLC analysis and the HPLC assay was done as described previously (16, 21). Cerebellum, known to be representative of brain as a whole (21), was weighed and homogenized in 20 vol of 0.4 M perchloric acid. After centrifugation at 12,000 g, the supernatant was neutralized with 1:10 (vol/vol) 4 M KOH and 1:20 (vol/vol) 1 M Tris. Samples were stored at −20°C until HPLC. The column was a LiChrocart RP-18. The mobile phase was 20 mM sodium phosphate, pH 3.5, containing 10% acetonitrile, and the wavelength of the detector was 254 nM.

Receptor autoradiography.

Receptor density was determined using receptor autoradiography with the adenosine A₁ receptor antagonist [³H]DPCPX (0.5 nM) (11), and the adenosine A_2A receptor antagonist [³H]SCH-58261 (0.2 nM) (17). Sections (14 μm) were preincubated in 170 mM Tris·HCl buffer containing 1 mM EDTA and 2 U/ml adenosine deaminase at 37°C for 30 min. Sections were then washed twice for 10 min at 23°C in 170 mM Tris·HCl buffer. Incubations were performed for 2 h at 23°C in 170 mM Tris·HCl buffer containing DPCPX (120 Ci/mmol, 0.2–20 nM) or SCH-58261 (77 Ci/mmol; 0.1–10 nM) and 2 U/ml adenosine deaminase. In the experiments with DPCPX, 1 mM MgCl₂ was added to preincubation and incubation buffers. The incubation with DPCPX was made in the presence of 100 μM GTP to convert all the receptors to the low-affinity state for agonists and thereby remove all endogenous adenosine (11). Sections were then washed twice for 5 min each in ice-cold Tris·HCl, dipped three times in ice-cold distilled water, and dried at 4°C over a strong fan. Slides were exposed to ³H-labeled film with ³H-labeled microscales for 4–8 wk. Analysis of receptor expression and binding was performed using a computerized image analysis system (MCID, InterFocus Imaging, Cambridge, UK). Relative optic density of expression or binding was measured in autoradiograms, and amounts of receptor-bound radioactivity of the specific brain regions were determined using ³H-labeled microscale standards. Specific binding was calculated by subtraction of the optical density values in sections where nonspecific binding was determined.

cDNA preparation and real-time RT-PCR.

The brains were dissected out and rapidly frozen. They were homogenized in Qiazol lysis reagents (Qiagen), and RNA was isolated from the brain with the Qiagen RNeasy kit according to the manufacturer's protocol (Qiagen). The cDNA synthesis was carried out with the High-Capacity cDNA reverse transcription kit with random primers and multiscribe reverse transcriptase enzyme according to the manufacturer's instructions (Applied Biosystems, Foster City, CA). The Applied Biosystem 2720 terminal cycler was used for the reverse transcription, and the incubation conditions were set to 25°C for 10 min, followed by 37°C for 2 h.

Detection of A_2BR and A₃R mRNA was performed using real-time RT-PCR (7). The real-time RT-PCR reactions were run in an ABI Prism 7500 sequence detector system (Applied Biosystems). Each run consisted of 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 10 s and 60°C for 1 min. Primers (900 nM), probes (200 nM), and TaqMan Universal PCR No AmpErase UNG master mix were used in each reaction. All the reactions were performed in triplicates. GAPDH was used as an endogenous control. Values are expressed as 2^−ΔCT (where CT = cycle at threshold) using GAPDH as a reference (ΔCT = CT_{adenosine receptor} − CT_GAPDH).

Isolation of adipocytes and lipolysis experiments.

Mouse adipocytes were isolated as described elsewhere (23) from epididymal fat pads of male mice older than 5 mo treated with collagenase (CLS type 1, 1 mg/ml) for 45–60 min. After the collagenase treatment, adipocytes were filtered through a nylon mesh, washed, counted, and diluted. Glycerol release was measured as an index of lipolysis (26). Aliquots of the cell suspension (∼150,000 adipocytes/ml) were placed into separate tubes (1.9 ml/tube) and incubated with or without adenosine deaminase (0.1 U/ml) for 15 min in a shaking water bath at 37°C. 2-Chloroadenosine (0.003–3 μM) was added to the adipocytes, and after 5 min of incubation, norepinephrine (1–300 nM) was added. Incubation was stopped after another 60 min, and the glycerol release was measured.

Statistical analysis.

Graphs and statistical analyses were done using GraphPad Prism 4 or 5 (GraphPad Software, San Diego, CA). Sigmoidal dose-response curves were calculated using nonlinear regression. Two-way ANOVA or Student's t-test was employed to evaluate the differences between groups. Data are means ± SE. Statistical significance was defined as P < 0.05.

RESULTS

Knocking out one or both copies of A₁R and/or A_2AR genes causes little adaptive changes in other adenosine receptors.

To examine whether the genetically modified mice exhibited major adaptive changes that would obscure the interpretation of results, we first investigated the expression of A₁R and A_2AR in wild-type (WT), A₁R KO, A_2AR KO, A₁R heterozygous (A₁R Hz), A_2AR heterozygous (A_2AR Hz), and A₁R-A_2AR dHz mice by using receptor autoradiography. In agreement with our previous results (22), we found that heterozygous mice with only one copy of the receptor gene had a reduced number of receptors. As shown in Fig. 1A, the number of A₁ adenosine receptors was reduced to about one-half (∼57% in this set of experiments) in heterozygote mice. The A₁ adenosine receptor binding was abolished in A₁RKO mice (not shown). It was furthermore shown that A₁R binding in cortex was unaltered in A_2ARKO mice and A_2AR Hz mice, whereas they were the same in single and A₁R-A_2AR dHz mice. Similar results were found in other brain regions, including hippocampus. An analogous situation was found for A_2AR, which were studied in the basal ganglia (Fig. 1B). The data shown are from caudate-putamen, and similar results were seen in nucleus accumbens (not shown).Thus elimination of one or both copies of the A₁R gene did not affect A_2AR binding, whereas A_2AR Hz mice had about two-thirds the number of receptors irrespective of the number of A₁R copies (Fig. 1B). The results also indicate that A_2ARs are not up- or downregulated when the number of A₁Rs is genetically altered, and A₁Rs are not up- or downregulated when the number of A_2ARs is altered.

Fig. 1.

Effect of A₁-A_2A double heterozygosity on the number of adenosine receptors and on the response to adenosine. A and B: autoradiographic quantitation of A₁ (A₁R) and A_2A adenosine receptors (A_2AR) in mice with different genotypes. Graph in A shows a binding isotherm for A₁R binding in cortex using 1,3-dipropyl-8-cyclopentylxanthine (DPCPX) as the radioligand. Binding was not detectable in A₁R knockout (KO) mice. K_d values (mean and 95% confidence intervals) for mice were as follows: wild-type (WT), 4.5 (3.1–5.9); A₁R heterozygous (Hz), ±4.0 (3.0–5.0); A_2AR KO, 3.6 (2.5–4.7); A_2AR Hz, 3.7 (2.6–4.7); and A₁R-A_2AR double heterozygous (dHz), 4.0 (2.8–2.3). Maximum binding (B_max) values were as follows: WT, 148 (130–165); A₁R Hz, 86 (77–94); A_2AR KO, 135 (119–150); A_2AR Hz, 138 (124–153); and A₁R-A_2AR dHz, 85 (75–95). Two to three sections were analyzed at each concentration of radioligand. Graph in B shows a binding isotherm for A_2AR binding in caudate-putamen using SCH-58261 as the radioligand. Binding was not detectable in A_2AR KO mice. K_d values (mean and 95% confidence intervals) were as follows: WT, 0.7 (0.5–0.9); A₁R Hz, 0.7 (0.4–0.9); A₁R KO, 0.6 (0.4–0.8); A_2AR Hz, 0.8 (0.4–1.1); and A₁R-A_2AR dHz, 0.9 (0.6–1.2). B_max values were as follows: WT, 159 (144–174); A₁R Hz, 162 (141–184); A₁R KO, 171 (157–185); A_2AR Hz, 115 (101–130); and A₁R-A_2AR dHz, 113 (100–126). Three sections were analyzed at each concentration of radioligand. Graph in C shows that adenosine responses were shifted to the right in mice with one-half the receptor number, approximately to the same extent as by caffeine in commonly used doses. Adenosine responses are shown as the inhibition of lipolysis by a stable adenosine analog, 2-chloroadenosine (2-CADO), in adipocytes from WT (•) or A₁R-A_2AR dHz mice (○). The cells were incubated in the presence of adenosine deaminase and 10 nM norepinephrine. In some experiments, 30 (WT caff 30) or 100 μM caffeine (WT caff 100) was added in addition. Increasing concentrations of 2-CADO were then added, and glycerol release was measured after the end of incubation. Results are from triplicate determinations from 3 (WT) or 2 (dHz) batches of cells prepared from 4 mice.

The next set of experiments examined whether there were adaptive changes in A_2BRs or A₃Rs. Since these receptors are less abundant and superb radioligands or antibodies for their study in mouse tissues are not readily available, we used real-time RT-PCR to determine the amount of mRNA. As shown in Table 1, mRNA abundance is close to identical (completely overlapping 95% confidence intervals) between WT mice and A₁R-A_2AR dHz mice. Single or double knockouts also were not different from wild types (and in knockouts as well; results not shown).

Table 1.

Expression of A_2B or A₃ adenosine receptor mRNA

	WT	A1R-A2AR dHz
n	11	6
A2BR	5.19 (4.20-6.03) × 10−4	5.82 (4.67-7.02) × 10−4
A3R	1.50 (1.05-1.82) × 10−4	1.20 (1.03-1.38) × 10−4

Results are expressed as the 2^−ΔCT using GAPDH mRNA as reference data are shown as means and 95% confidence intervals: 2^{−(ΔCT + SE × T0.05)} − 2^{−(ΔCT − SE × T0.05)}; n = no. of mice per group. WT, wild-type mice; A₁R-A_2AR dHz, A₁ and A_2A adenosine receptor double heterozygous mice; A_2BR, A_2B adenosine receptor; A₃R, A₃ adenosine receptor.

Xanthine levels in mouse brain after caffeine administration.

Administration of caffeine by injection or by addition of caffeine to the drinking water increased brain caffeine levels, as well as the levels of caffeine metabolites theophylline and paraxanthine, regardless of adenosine receptor genotype (Table 2). The peak level of caffeine during voluntary drinking of water containing 0.3 g/l caffeine was less than one-half the peak level of caffeine after a single injection of 15 mg/kg ip. The minor differences between groups of animals were not significant. Levels of caffeine at the time points studied were much higher than the levels of theophylline and paraxanthine, as described earlier (21, 25).

Table 2.

Xanthine levels in mouse cerebellum after caffeine administration

	Caffeine Injection					Oral Caffeine
	WT	A1R KO	A2AR KO	A1R-A2AR dKO	A1R-A2AR dHz
n	10	12	7	10	12	12
Caffeine, nmol/g	32.3±6.8	37.5±4.73	40.85±10.27	40.9±7.33	24.17±2.72	12.07±1.2
Theophylline + paraxanthine, nmol/g	8.9±1.29	7.33±0.64	8.86±2.05	6.8±1.28	6.5±0.77	2.84±0.30

Caffeine injection results indicate the xanthine level in mouse cerebellum 90 min after intraperitoneal injection of 15 mg/kg caffeine (means ± SE). Oral caffeine results indicate the xanthine level in WT mouse cerebellum after 10 days of ingestion of water containing 0.3 g/l caffeine. Values are means ± SE; n = no. of mice per group.

A₁R-A_2AR dHz mice mimic the antilipolytic effect induced by caffeine.

Using the above-described data, we next compared the influence of heterozygosity and caffeine administration on the ability of a stable adenosine analog to activate receptors. The results of such experiments are shown in Fig. 1C. We utilized the antilipolytic effect of the metabolically rather stable but receptor subtype nonselective adenosine analog 2-chloroadenosine to assess changes in the potency of the natural ligand (23). With the use of wild-type adipocytes, the IC₅₀ value was 13.7 (9.4–20) (mean and 95% confidence intervals). In the wild-type cells, the addition of caffeine caused a shift to the right in IC₅₀ value: with 30 μM caffeine to 44 (26–73), and with 100 μM caffeine to 112 (73–172). The IC₅₀ value in adipocytes from dHz mice was 36 (23–57). Thus we found that heterozygosity produces an effect qualitatively similar to the administration of caffeine.

A₁R-A_2AR dHz mimic effect of long-term caffeine ingestion on HR and LA.

After completing these initial experiments, which supported our underlying assumptions, we could start to compare the phenotype of mice treated with caffeine or genetically modified to have only a single copy of the relevant genes. We first examined body temperature and found no differences compared with WT mice (results not shown), in contrast to the rather substantial effects of completely eliminating these receptors (Ref. 37; Yang JN, Chen JF, Fredholm BB, unpublished observations).

By contrast, as shown in Fig. 2, A₁R-A_2AR dHz mice had lower HR than WT mice, especially in females. After long-term caffeine ingestion, the HR in WT mice was decreased (again, this was most obvious in female mice), but this was not seen in A₁R-A_2AR dHz mice. After caffeine ingestion, WT and A₁R-A_2AR dHz mice had similar HR in both sexes. Figure 2 shows the time course of HR over the 24-h period in the different groups and also the cumulated HR over the entire 24-h period. A clear effect is shown in female, but not male, mice. An analysis of different time periods showed significant effects of both long-term caffeine and the dHz state during the daytime, with less consistent effects on peak or nighttime HR. These data suggest that oral caffeine ingestion and double heterozygosity have similar minor effects on HR in female mice and that these are not additive.

Fig. 2.

Effect of caffeine ingestion on heart rate (HR). HR was recorded using telemetry for consecutive 10-min periods in WT and A₁R-A_2AR dHz mice. In the top, and middle panels the dark period (shaded) is from 7:00 PM to 7:00 AM. Basal HR data are presented as 5-day averages, shown as WT control (n = 8 females, n = 8 males) and A₁R-A_2AR dHz control (n = 8 females, n = 9 males). HR data during caffeine ingestion are presented as 3-day averages (the 5th, 6th, and 7th days after start of the period), shown as WT caffeine and A₁R-A_2AR dHz caffeine. Bottom panel: summary of the data presented in top and middle. Values show the summarized HR from 12:00 AM to 12:00 PM (beats/day). Data are means (top and middle) or means ± SE (bottom). *P < 0.05 compared with WT control mice. #P < 0.05 compared with WT caffeine.

In the awake mice subjected to telemetric analysis of LA, a clear diurnal locomotor pattern is shown (Fig. 3). Activity was low in the daytime and increased markedly at night, starting slightly before the expected gradual dimming of the light. To examine the effect of caffeine on the LA, we performed two periods of examination. One period started when the animals had accustomed themselves to the home cage, and the average for 5 days was calculated. The second period was started, as described in materials and methods (see Fig. 3C), when the mice had received oral caffeine in the drinking water or only tap water for 7 consecutive days. We then calculated the average for days 5 to 7 when the mice continued the caffeine or tap water ingestion. As shown in Fig. 3, chronic caffeine treatment did increase LA in both sexes (P < 0.05). Interestingly, compared with the first period (control), LA was decreased in both female and male mice with water ingestion during the second period (P < 0.05), and it was also showing a decreasing tendency in female mice. With the use of this method of recording locomotion, there was no clear effect of deleting one of the copies of the A₁R and A_2AR genes on daytime activity in awake, habituated mice (data not shown).

Fig. 3.

Increased activity in mice receiving oral caffeine ingestion. The locomotor activity (LA) of WT female (A) and male mice (B) mice was recorded using telemetry. LA data before caffeine injection are presented as 5-day averages in 30-min intervals, shown as control (n = 13). LA data during the second period are presented as 3-day averages (the 5th, 6th, and 7th days after start of the period); mice drank water containing 0.3 g/l caffeine (caffeine, n = 8) or drank only tap water (water, n = 5), as described in materials and methods. The dark period (shaded) is from 7:00 PM to 7:00 AM. Values are means ± SE. *P < 0.05 compared with control. #P < 0.05 compared with water. C: the protocol of telemetry recording: 5 days were recorded for baseline HR, body temperature (Temp), and LA (control); 7 days were recorded during mouse ingestion of caffeine or tap water, for which the average of the 5th, 6th, and 7th days is shown as second period data (caffeine or water).

Instead, when the locomotor box was used, an effect of the genotype could be detected. There was a small, but significant, increase in the activity of dHz mice placed in the locomotor box for the first time (Fig. 4). This increase in activity counts was approximately equal in magnitude to that seen in A₁R KO mice and reported elsewhere (5). Mice that had received oral caffeine (Fig. 4) also showed an increased activity in this situation. There were no significant differences in the amount of time the mice spent in the periphery or the center of the arena depending on treatment or genotype (not shown). Although the effects of genotype and drug treatment appeared larger in the females than in the males (Fig. 4, A and B), an essential similarity was observed between sexes.

Fig. 4.

LA measured in a locomotor box. A and B: the LA of tap water-treated WT (WT H₂O) (n = 15 females, n = 16 males), long-term caffeine-treated WT (WT chr caff; n = 7 females, n = 10 males), A₁R-A_2AR dHz (n = 11 females, n = 11 males), A₁R-A_2AR double knockout (A₁R-A_2AR dKO; n = 7 females, n = 8 males), A₁R KO (n = 16 females, n = 15 males), and A_2AR KO mice (n = 6 females, n = 12 males) was measured in an open field as described in materials and methods. LA was increased in WT chr caff, A₁R-A_2AR dHz, and A₁R KO mice (*P < 0.05) but not in A₁R-A_2AR dKO and A_2AR KO mice, including both sexes.

Long-term caffeine ingestion and A₁R-A_2AR double heterozygosity reduce responses to acute caffeine.

It is well known that long-term caffeine ingestion can lead to a tolerance to the acute effects of the drug. We wanted to examine whether this could be observed under our experimental conditions and whether dHz mice also would show a lower response to caffeine injection. Effects of the chosen doses of caffeine on HR were small, in agreement with our previous results (37), and we therefore focused on locomotor behavior. As shown in Fig. 3, the activity of mice during the daytime was low (2–5 counts/min) in the telemetry studies. However, when they were taken up and injected intraperitoneally, the activity was transiently quite elevated in male mice (Fig. 5). When caffeine was later injected at 7.5 or 30 mg/kg in the same mice, the increase in activity was slightly higher and more prolonged (Fig. 5). It clearly can be seen that the effect of acute caffeine was much less in dHz mice. A similar effect was seen in female mice. In particular, the locomotor stimulation after 30 mg/kg was much more pronounced in WT than in dHz mice.

Fig. 5.

The stimulatory effect of caffeine injection is reduced in A₁R-A_2AR dHz male mice. Saline, 7.5 mg/kg caffeine, and then 30 mg/kg caffeine were administrated at 2-h intervals to WT (•) and A₁R-A_2AR dHz (○) male mice (n = 8). The LA was measured at 30-min intervals. Values are means ± SE. *P < 0.05, WT vs. dHz mice.

Essentially similar data were observed in the locomotor boxes. The mice were placed in the box, and this caused a marked increase in activity despite two prior habituation episodes. In this setting of the experiment, intraperitoneal injection of saline after 45 min had little effect and the activity soon returned toward baseline (data not shown). Mice given long-term caffeine treatment had a higher baseline, and activity did not return during the observation period. Injection of caffeine increased activity in both groups of mice over that seen in the untreated WT mice. The result of the caffeine (15 mg/kg) injection in several genotypes is shown in Fig. 6. It is clear that the caffeine effect is very much reduced (or even abolished, since the remaining activity could be basal) in mice with their A_2AR deleted. By contrast, mice lacking A₁R had normal responses to caffeine, whereas the dHz mice showed a significantly reduced response, as also did the mice that had received long-term caffeine (at least this is true for the male mice).

Fig. 6.

Responses to caffeine in genetically modified mice. Reduced response to caffeine injection in A₁R, A_2AR KO, A₁R-A_2AR dKO, and A₁R-A_2AR dHz mice. After acute administration of 15 mg/kg caffeine, the accumulated LA was measured between 46 and 135 min in WT H₂O (n = 10 females, n = 4 males), WT chr caff (n = 4 females, n = 5 males), A₁R-A_2AR dHz (n = 7 females, n = 8 males), A₁R-A_2AR dKO (n = 6 females, n = 6 males), A₁R KO (n = 6 females, n = 6 males), and A_2AR KO mice (n = 4 females, n = 6 males) using a locomotor box as described in materials and methods. *P < 0.05 compared with WT H₂O.

DISCUSSION

In this study we showed that deletion of one of the copies of the A₁Rs and A_2ARs led to a reduction of each of those receptors to about one-half the original number. We also found that deleting both copies of the A₁R gene did not lead to any compensatory upregulation of A_2ARs, or conversely. In addition, we did not see any compensatory adaptive changes in A_2BR or A₃R mRNA expression. These results are consistent with several previous studies (22, 27, 31, 37) indicating there are no compensatory adaptive changes in adenosine receptors when one or both copies of the gene for one of them is deleted. Together with all the previous information, the present results clearly suggest that signaling via adenosine receptors is not a process that is homeostatically regulated by altering receptor expression. If and when adaptive changes occur, they involve other genes and proteins.

We next confirmed and extended the previous observation that a reduction in receptor number leads to a parallel shift of the dose-response curve for adenosine (22, 23) by showing that this also was true for dHz mice (Fig. 1). In the present experiments, a 2.5-fold shift to the right was observed. A larger shift was observed with 30 and 100 μM caffeine, amounting to three- and eightfold shifts. These data implicate a pA₂ value of ∼10 μM, which agrees well with the previously determined K_i value from binding experiment 20 μM at A₁R and 9 μM at A_2AR (33). It can be noted that this is almost exactly the concentration of caffeine in brain at night during periods of consumption of water with 0.3 g/l caffeine added (21, 25, 33). Although concentrations are lower in daytime (21), the prediction is upheld that a state of double heterozygosity for A₁R and A_2AR would produce a shift in the activation of these receptors similar to that seen after long-term caffeine.

Caffeine is a competitive antagonist of adenosine receptors. A competitive antagonist is expected to exert a biological effect in vivo provided that the receptors are occupied by endogenous adenosine. The dose of caffeine used for long-term treatment (0.3 g/l) in this study mimicked the dose commonly consumed by humans, which mainly influences A₁R and A_2AR (12, 14, 15). It is well known that caffeine also can act on other molecular targets (such as phosphodiesterases, GABA receptors, or ryanodine receptors), but the concentrations needed to do this are generally believed to be higher than those used in this study (15). In a separate study (Yang JN, Chen JF, and Fredholm BB, unpublished data), our group has shown that the effects of caffeine (7.5 and 15 mg by injection or 0.3 g/l in drinking water) are lost in mice that lack A₁Rs and A_2ARs. The present data also show that levels of caffeine (and its major metabolites) are lower after the oral caffeine regimen than after injection of 15 mg/kg. Hence, we are reasonably certain that the caffeine by ingestion indeed acts via these adenosine receptors to produce the effects recorded in this study. Since caffeine does have biological effects, the implication is that adenosine is tonically acting on some adenosine receptors, and therefore we ought to be able to observe some effects of altering receptor number given the relationship between receptor number and effect.

We found that neither caffeine treatment nor the double heterozygosity for A₁R and A_2AR caused any major change in body temperature (data not shown). This lack of effect of caffeine on body temperature is consistent with the previous information as recently reviewed by Armstrong et al. (1). By contrast, there were clear effects on heart rate, measured via telemetry, and locomotor activity, measured in two locomotion assays.

Long-term caffeine treatment decreased heart rate in female WT mice, which is consistent with previous findings in humans (9, 36) and is apparently due to the blockade of 25–50% of A₁Rs and A_2ARs, leading to a reduced sympathetic tone (29). Endogenous adenosine, via A₁Rs and A_2ARs, regulates firing of cell groups in the central nervous system and directly influences cardiac control centers (3, 4). A₁R-A_2AR dHz mice also had a heart rate that was lower than that of WT mice, thereby mimicking the effect of long-term caffeine treatment. Interestingly, long-term caffeine treatment did not affect heart rate in A₁R-A_2AR dHz mice, which could be due to the tolerance to long-term caffeine effect (15, 30) or to a maximal effect being achieved by either caffeine treatment or the dHz state. We did not observe any clear effect of acute caffeine treatment on heart rate; this could indicate that tolerance is a less likely explanation. Even more importantly, the fact that acute caffeine administration does not produce any fall in heart rate even in female mice (37), where long-term caffeine ingestion clearly does, suggests that the effect of long-term caffeine is not due to the acute blockade of adenosine receptors, but instead to a consequence of a longer-term blockade. Indeed, adaptive changes in sympathetic regulation that develop with time during caffeine administration have been reported (30).

Acute as well as long-term caffeine intake did cause an increased locomotor activity, as expected. The brain concentration of caffeine after 15 mg/kg caffeine injection was about three times higher than that seen after long-term intake of water containing 0.3 g/l caffeine. Since peak concentration of caffeine in blood and tissues is essentially linearly related to dose, we also expect that 7.5 mg/kg caffeine would give a caffeine concentration at least as high as that observed during long-term caffeine intake. It was earlier estimated that oral consumption of water containing 0.3 g/l caffeine would be similar to habitual human consumption of caffeine and that 7.5 mg/kg injection might correspond to the consumption of a few cups of coffee at a single occasion (15). It is therefore not surprising that the effects on locomotion are modest.

The present data also emphasize the importance of sex when examining the phenotype of mice. The marked effect of sex on basal heart rate and locomotion was earlier noted (5, 37). We do not think that this is entirely due to estrus cycle-dependent stimulation by estrogen because we averaged locomotor activity over 5 days, and had the effect been due to a particular phase of the female estrus cycle, we would expect very much larger variability between days in locomotor activity in females than in males, but that was in fact not the case. In this study we note that the effects of caffeine (and the dHz genotype) were sex dependent. Basal heart rate is higher in females than males, and this could perhaps explain why it is easier to detect a decrease such as that seen in dHz mice and in mice treated with caffeine for several days. Basal locomotion was higher in female than in male mice. Since caffeine produces an increase this might be more difficult to observe in females, and this was indeed the case. Deleting one of the copies of the A₁R and A_2AR caused a small increase in locomotor activity, which was most readily seen when tested in the locomotor boxes. By contrast, no such increase was seen in the double knockout mice. Instead there was a decrease, probably mainly due to the elimination of the A_2A receptor (Fig. 6).

After successive administration of a drug, a decreasing effect, often denoting tolerance, is often noted and has been repeatedly shown for caffeine (15, 24, 30, 34). In the present study, after long-term caffeine treatment, the motor-stimulating response to caffeine injection was reduced, as shown in two different locomotion assays (Figs. 5 and 6). A₁R-A_2AR dHz mice could, at least partly, mimic this aspect of long-term caffeine treatment effects. The motor-stimulating effect of caffeine injection was also smaller in A₁R-A_2AR dHz mice than in WT control mice. Thus the tolerance effect of long-term caffeine treatment is likely at least to some extent dependent on the adenosine receptors, in line with previous suggestions (e.g., Ref. 33).

It is still debated which mechanisms underlie development of tolerance to caffeine development. It often has been suggested that tolerance is due to increased metabolism of caffeine, but it appears that this can at most explain a minor part of the phenomenon (15, 34). There are some data suggesting increased sensitivity of adenosine in the absence of changes in receptor number (19, 35), but it is not obvious how this would be brought about. Conlay et al. (8) reported long-term treatment of caffeine (1 mg/ml in drinking water for 2 wk) dramatically increased plasma adenosine 10-fold in rats. This finding has not been replicated, and it is not clear how dramatically increased levels of adenosine could explain both a stimulatory effect of long-term caffeine and a reduced response to a single caffeine injection.

It also was reported that chronic methylxanthine (such as caffeine and theophylline) treatment could increase the density of adenosine binding sites in variant sites of brain, including cerebral cortex (11, 13, 18), brain stem, and cerebellum (28), suggesting that the upregulation of adenosine receptors was involved in caffeine tolerance (32). However, it has been convincingly argued that upregulation of adenosine receptors is not an adequate explanation for tolerance development (15, 19, 33). Perhaps several mechanisms are involved, and hopefully, the present data indicating that genetic reduction of receptor number can also induce a phenomenon resembling tolerance will help provide clues to the mechanism(s) involved.

Together, our results show that A₁R-A_2AR dHz mice indeed show a phenotype that resembles an aspect of animals treated repeatedly with caffeine. These dHz mice thus show somewhat increased locomotion and a slightly reduced response to acute caffeine treatment. Thus at least some aspects of long-term caffeine use can be mimicked by genetic manipulation of the A₁R and A_2AR. Furthermore, the results demonstrate that heterozygous mice can exhibit very a different phenotype from that of double knockout mice and suggest that such mice may be quite useful to delineate effects of long-term incomplete blockade of the receptors.

GRANTS

This study was supported by Swedish Science Council Grant 2553, National Institute of Neurological Disorders and Stroke Grant R01 NS048995, the Swedish Heart and Lung Foundation, Karolinska Institutet, and the European Commission (STREP “Enough sleep”), National Institutes of Health Grant ES10804, US Department of Defence Grant W81XWH-04-1-0881.

DISCLAIMER

The contents of this report are the responsibility of the authors, and the supporting bodies take no responsibility.