The authors have declared that no competing interests exist.
Conceived and designed the experiments: CC VM RG. Performed the experiments: CC VM HL JD AR GO ST CT. Analyzed the data: CC VM HL JD AR ST. Contributed reagents/materials/analysis tools: AR ST CT RG. Wrote the paper: CC VM RG.
C57BL/6 mice are the most widely used strain of laboratory mice. Using
In the last three decades, mouse models have contributed to enormous advances in characterizing the molecular underpinnings of disease through the use of transgenesis and other approaches. The C57BL/6 strain is one of the most widely used for both transgenesis and environmental exposure experiments
Our facility (CIBM) is a regional imaging center for
1H MRS spectra displaying high Gln and low Ins are typically attributed to liver failure or to portosystemic (PS) shunting
All animal experiments were conducted according to federal and local ethical guidelines, and the protocols were approved by the local regulatory body of the Canton Vaud, Switzerland (EXPANIM (Expérience sur animaux) – SCAV, Département de la sécurité et de l’environnement, Service de la consommation et des affaires vétérinaires).
All our animals (wild-type and genetically modified) were provided by official suppliers or different academic or commercial animal facilities nationwide. For the purpose of the present study we are presenting only some of the “High Gln” mice identified in our laboratory (total of n = 23 mice). All the mice were C57BL/6J or back-crossed with this strain.
Metabolite concentrations were estimated using LCModel (
Plasma ammonia and glutamine concentrations were measured from all animals using an Analox GM7 analyzer (Analox Instruments, London, UK) as previous published
One high glutamine (Gln) female and male was used to obtain two litters of offspring. All offspring (n = 9) were furthermore subject to only
Five “High Gln” (n = 3 at 4 months and n = 2 at 12 months, males) and five “Normal Gln” (n = 3 at 4 months, males and n = 2 at 12 months, females) mice underwent 1H MRS measurements prior to portal angiography, to confirm the abnormal neurochemical pattern. Mice were anesthetized using isoflurane (3% for induction and 2% thereafter, 1L/min O2 flow). Following a midline incision, the mesentery was overturned to the left side of the abdomen in order to best expose the inferior mesenteric vein. The vein was isolated and encircled with two 7–0 silk ties, then a heparin-flushed 26-French catheter was introduced into the vein and secured with the silks. The tip of the catheter was placed between the gastro-splenic and the duodeno-pancreatic veins.
Portal angiographies were performed on a digital subtraction angiography system equipped with a 40 × 48 cm flat panel detector (Allura Xper FD 20; Philips Medical System, Best The Netherlands). The mice were placed in the supine position on the angiography table. The 26-F catheter in the portal vein was attached through a 26-F connecting tube to an infusion pump (Module DPS Orchestra; Frenesius Vial, Brezins, France). Standard postero-anterior angiograms were obtained during continuous injection of iodinated contrast media (350 mg iodine/ml of iohexol, Accupaque; GE Healthcare, Oepfikon, Switzerland), Cork, Ireland) with the following image parameters: tube voltage, 120 kV; modulated tube current; inherent filtration, 2.7 mm aluminium; pulse width, 6 ms; system dose, 0.36 µGy per pulse; focal spot, 0.7 mm; tube detector distance, 0.4 m; matrix size, 1024×1024; zoom factor, 5x; field of view, 7×4 cm; acquisition time, 15 s; delay time, 2 s. Then, a 3D rotational computed tomography (CT) was acquired with the following parameters: a total acquisition angle of 222°, projection increment of 0.8°. The CT images were reconstructed and analyzed on a dedicated workstation (Xper CT; Philips Medical System, Best The Netherlands). In order to optimize contrast media injection, we performed successive portal angiographies with different flow rate (75 mL/h; 150 mL/h; 300 mL/h; 600 mL/h) on two test animals. The optimal flow rate of 150 mL/h (corresponding to 0.042 mL/s) was used in the “Normal Gln” group (n = 5) and in the “High Gln” group (n = 5).
Livers were harvested following portal angiography (n = 5 for “High Gln” mice and n = 5 for “Normal Gln” mice). To control for injection artifacts (i.e. modifications of the liver parenchyma restricted to the lobules, which did not impact assessment of portal tract and centrilobular vein), we also harvested livers from two 4 months old “High Gln” female mice and two 4 months old “Normal Gln” female mice (previously confirmed by 1H MRS) having not undergone portal angiography.
All tissues were fixed in 10% buffered formalin overnight at room temperature, then paraffin-embedded. For routine histological examination, 3 µm-thick sections were stained with haematoxylin-eosin (H&E), Masson’s trichrome, and a reticulin stain. Immunohistochemistry was performed on paraffin-embedded sections mounted on positively charged slides, using an automated DAKO immunostainer (DakoCytomation, Glostrup, Denmark). The following primary antibodies were applied: CD31 (PECAM-1, M-20, sc-1506, goat polyclonal, Santa Cruz Biotechnology, Inc., Heidelberg, Germany), and D2–40 (Podoplanin, goat polyclonal, R&D Systems, Abingdon, UK). First, for antigen retrieval, deparaffinized and rehydrated sections were treated using an electric pressure cooker for CD31 (Pascal Tris EDTA buffer, pH7.0, for 30 seconds) and a microwave for D2–40 (citrate buffer, pH6.0, for 10 minutes). Then the sections were mounted in the DAKO autostainer, covered with blocking peroxidase for 5 minutes. Slides were incubated for 60 minutes with the diluted antibody (CD31 1:200; D2–40 1:1000). This step was followed by applying the labelled rabbit anti-goat HRP method (DAKO; 1∶20, 30 minutes). DAB (DAKO) was used as a chromogen.
Apart from routine liver architecture, portal tract, centrilobular vein, and hepatocyte evaluation, the following were also assessed: relative sizes of portal veins and lymphatics, total numbers of portal tracts and lymphatic vessels, and mean lymphatic profiles per portal tracts.
All the results are presented as mean ± SD of n mice, unless otherwise indicated. A two-way ANOVA followed by the Bonferroni’s multi-comparison post-test was used to compare the concentration of neurochemicals. ANOVA during development was performed for each neurochemical in 2 groups×5 time points (ages from P10 to P90), whereas in adult mice was performed for each neurochemical at each age in 2 groups×3 brain areas. Statistical analyses were performed using Prism 5.03 (GraphPad, La Jolla CA USA).
The ultra-short echo time sequence combined with the availability of ultra-high magnetic field (14.1T) yielded 1H spectra with excellent signal-to-noise ratio (SNR) and with a clear separation of the Gln and glutamate (Glu) peaks (
The increase of Gln and decrease of Ins in the cortex of the “High Gln” mouse is visually apparent. Only those metabolites displaying a concentration change are labeled (i.e. Gln and Ins) for the “High Gln” spectrum.
A) Evolution of brain metabolite concentrations during mouse development at P 10, 20, 30, 60 and 90; open squares indicate “Normal Gln” mice and triangles indicate “High Gln” mice; and B) metabolite concentrations in the striatum (str), hippocampus (hip) and cortex (ctx) of “High Gln” and “Normal Gln” C57BL/6 mice at 4 and 12 months of age. Two-way ANOVA was performed at 5 developmental time-points (ages P10 to P90) comparing the “High Gln” mice to “Normal Gln” mice for each metabolite. Statistically significant differences for Gln, Tau and Ins between “High Gln” and “Normal Gln” mice (df = 1, F value between 23.6 and 625) are marked *(p<0.05), **(p<0.01) and ***(p<0.001). The age comparison showed statistical differences for all plotted metabolites (p<0.0001, df = 4, F value between 19.2 and 139) (not shown). We observed statistically significant age-dependent differences between groups for Ins and Gln (p = 0.002 and p<0.0001 respectively, df = 4, F value 4.47 and 66.7, respectively) (not shown). In adult mice two-way ANOVA was performed for each neurochemical at each age (4 and 12 months) in 2 groups (“High Gln” vs “Normal Gln” mice)×3 brain areas. Statistically significant differences for Gln, Glu, Tau, Ins and tCr between “High Gln” and “Normal Gln” mice (df = 1, F value between 14 and 519) are marked *(p<0.05), **(p<0.01) and ***(p<0.001). Additionally, the brain regions comparison showed statistical differences for some of the plotted metabolites (Glu, NAA, Tau, Ins p<0.001, df = 2, F value between 10 and 87) (not shown). Lac: lactate, GABA: γ-aminobutyrate, NAAG: N-acetylaspartylglutamate, NAA: N-acetylaspartate, Gln: glutamine, Glu: glutamate, Asp: aspartate, Cr: creatine, PCr: phosphocreatine, PE: phosphoethanolamine, PCho: phosphocholine, GPC: glycerophosphocholine, Tau: taurine, Ins: myo-inositol, Gly: glycine, GSH: glutathione, Asc: ascorbate.
To determine whether the neurochemical changes in Gln and Ins could be ascribed to developmental fluctuations, we performed serial
To determine whether the increase of Gln was region specific, we performed
To determine the presence of a systemic phenotype, we performed serial weight and biochemical measurements. There was no significant difference in mean weight between the “High Gln” and “Normal Gln” mice at any time point: “High Gln” at 4 months weighed 30±3 g and “Normal Gln” weighed 29±3 g. No statistical differences were observed for plasma ammonia (270±40 µM for “High Gln” mice and 270±60 µM for “Normal Gln” mice, p = 0.9) and plasma glutamine concentrations (0.90±0.06 mM for “High Gln” mice and 0.88±0.12 mM for “Normal Gln” mice, p = 0.8). As the aforementioned changes are a hallmark of hepatic encephalopathy in humans
To characterize the heritability pattern of this trait, a “High Gln” male and “High Gln” female were crossed, yielding two litters (n = 9). None of the pups showed the characteristic abnormal neurochemical profile (data not shown) by 1H MRS, and therefore no further measurements were performed on these animals.
To assess the presence of portosystemic shunts, portal angiograms were obtained in all animals (
Normal filling of the portal tree is visible in B. In A, injection in the superior mesenteric vein leads to immediate filling of the inferior vena cava. Inferior vena cava (IVC), portal vein (PV), shunt (Sh).
PS shunts are associated with subtly abnormal liver histology
Original magnification 100x. A and C: CD31 immunostaining highlights endothelial cells. In the “High Gln” mouse (A), the hepatic artery branch is of normal size, similar to that of the interlobular bile duct; the portal vein is small and hypoplastic, and the inlet venules are dilated. In the “Normal Gln” mouse (C), the portal vein is large, with a normal size ratio to the interlobular bile duct (the hepatic artery branch is not seen in this section). The inlet venule is thin. B and D: D2–40 expression confirms the lymphatic nature of the dilated channels at the periphery of the portal tracts in the “High Gln” mouse (B), being selectively reactive in lymphatic endothelial cells, contrary to arterial and venous endothelial cells. Of note, D2–40 (podoplanin) reactivity is also seen in bile duct epithelium. Abbreviations: A = hepatic artery; B = interlobular bile duct; CLV = centrilobular vein; IV = inlet venule; L = lymphatic vessel; V = portal vein.
This study shows that a high fraction of C57BL/6J mice present an abnormal neurochemical profile consisting of elevated cerebral Gln concentrations and decreased cerebral Ins (“High Gln” mice), and that these changes are associated with the presence of congenital portosystemic shunts.
To the best of our knowledge, this is the first report describing an abnormal neurochemical profile identified using
Note the excellent agreement with previous studies of regional and developmental changes (
The abnormal neurochemical profile consisted of increased cerebral Gln concentration by an average of 3 fold and 50% decreased cerebral Ins affecting all brain regions. It has to be emphasized that a 3 fold increase in Gln is well in excess of any change considered statistically significant in a given mouse for all of the studies performed at our center.
Therefore, this unique phenotype is far from a variation of the norm. This characteristic profile is typical of hepatic encephalopathy associated with chronic liver disease and portosystemic shunting
The commonly accepted pathophysiology of central nervous changes in liver disease and portosystemic shunting is that portal blood bypasses the liver, thereby reaching the systemic circulation without undergoing the necessary transformation in the liver. Thus, exogenous and endogenous substances such as ammonia enter the systemic circulation directly. In the central nervous system, ammonia is detoxified into Gln via the astrocyte-specific enzyme glutamine synthethase
High cerebral Gln concentration is detectable
Congenital PS shunts are characterized by similar pathophysiology and histology, whether intra- or extrahepatic. The shunts identified in our animals resemble incomplete ductus venosus (DV) closure, typical for intrahepatic shunts
It is interesting to note that the same abnormal neurochemical profile identified using
We observed that as many as 25% (or even higher) of C57BL/6J animals enrolled in a given study conducted in our facility displayed an abnormal neurochemical profile in all regions of the brain, and that this finding existed in animals with congenital PS shunts. The frequency of this characteristic neurochemical profile, typically attributed to PS shunting or chronic liver disease, raises the question of the reliability of this model for neurobiology in the absence of appropriate tools to screen the animals.
The implications of the current findings are likely far-reaching: in humans, PS shunting is associated with many systemic changes, namely cardiac, renal, hepatic and pulmonary
We conclude that, although congenital anomalies associated with the inbreeding of C57BL/6 have already been identified (ocular defects, genetic polymorphisms among substrains, behavioral differences among substrains
The authors acknowledge the personal communication of Profs. Ivan Tkac and Gulin Oz (Department of Radiology, University of Minnesota) regarding the “High Gln” mice identified in their center. We thank Dr. Sylvain Lengacher (Laboratory of Neuroenergetics and Cellular Dynamics, EPFL) and Prof. Kim Quang Do (Center for Psychiatric Neuroscience, University Hospital Lausanne, Switzerland) for providing some of the mice for this study. The authors thank Drs. Vladimir Mlynarik, Lijing Xin, Bernard Lanz and Gregory Lodygenski, as well as Olivier Brina and Dario Sessa for their help and support.