The presence in the brain of the urea cycle intermediate citrulline in the absence of a complete urea cycle has never been adequately explained. In an attempt to clarify this problem, we developed antibodies to citrulline and determined the distribution of citrulline-immunoreactivity in fixed sections of rat brain using immunoperoxidase and indirect immunofluorescence techniques. Citrulline-positive neurons were found to have a re- stricted distribution within the brain. A few cells were present in the cortex and corpus callosum. A large population of strongly stained cells was diffusely scattered throughout the striatum, nucleus accumbens and olfactory tubercle. Less strongly stained cells were detected in the supraoptic and paraventricular nuclei of the hypothalamus, the dorsal raphe, and the laterodorsal and pedunculopontine tegmental nuclei of the pons. The citrulline-immunoreactive cells were similar to those previously shown to contain NADPH-diaphorase activity, and double staining experiments indicated that citrulline-immunoreactivity was present in a subpopulation of NADPH-diaphorase-positive neurons. We have recently identified NADPH-diaphorase as a nitric oxide synthase. Thus the presence of citrulline in these cells suggests that it is formed within the brain as a coproduct during nitric oxide formation from arginine.
Brain tissue is capable of synthesizing arginine from citrulline via argininosuccinate in a manner similar to that in liver [11, 12]. However, two essential urea cycle enzymes, ornithine transcarbamoylase and carbamoyl- phosphate synthetase 1, appear to be absent from brain [2, 6, 8]. Early researchers therefore proposed that the brain was incapable of citrulline biosynthesis and that the appreciable amounts of citrulline observed in brain [12, 13] were derived from the blood, since systemic citrulline loading resulted in increased brain levels of citrulline [2, 12]. However, recent evidence indicates that another possible origin for brain citrulline might be through the action of the enzyme nitric oxide synthase (NO synthase). NO synthase produces NO and citrulline from arginine in a Ca2+-calmodulin and NADPH-de- pendent fashion [1, 7]. Thus neurons possessing NO syn- thase may produce appreciable levels of citrulline.
In order to determine the localization of citruUine in the brain, we raised an antibody specific for citrulline and undertook immunohistochemical studies. We were struck by the similarity of the labelled neurons to those
Correspondence: S.R. Vincent, Department of Psychiatry, The Univer- sity of British Columbia, Vancouver, B.C., V6T 1W5, Canada.
previously described using the NADPH-diaphorase his- tochemical technique [16]. We have recently discovered that NADPH-diaphorase is a NO synthase [4]. Double labelling experiments were therefore undertaken to com- pare the distribution of citrulline-immunoreactivity with NADPH-diaphorase staining.
To produce the primary antibody, 1 ml of citrulline solution (10 mg/ml in H20) was mixed with 1 ml of bo- vine serum albumin (BSA) (30 mg/ml in 3 M sodium ace- tate buffer; pH 7.8). One ml of 5% glutaraldehyde was then added and allowed to react for 3 min before adding 1 ml of 10 mM sodium borohydride. This mixture was then dialyzed overnight against 0.9% NaC1, then centri- fuged for 30 min at 20,000 g. Rabbits were immunized with 1 ml of the BSA-citrulline conjugate mixed with 2 ml Freund's complete adjuvant. The animals received booster injections of antigen in 2 ml incomplete adjuvant after 3 and 6 months. Blood was collected and cen- trifuged for 30 min at 20,000 g. To the supernatant was added 0.02% sodium azide and aliquots were frozen at
_ 20oc.
The antibody specificity was tested using conjugates of various amino acids with whole brain macromolecules [10]. Briefly, whole rat brain was homogenized and dia- lyzed, and the resultant macromolecular suspension (10
156
mg protein/ml) was mixed with an amino acid (2.5 mmol/mg protein) in 0.1 M sodium phosphate buffer to- gether with 25 mmol/ml glutaraldehyde, at 22°C for 1 h. After dialyzing overnight', the conjugates were blotted onto nitrocellulose membranes. The membranes were then incubated in the primary antibody and stained with the same avidin-biotin complex immunoperoxidase tech- nique used for immunohistochemistry [5].
The distribution of citrulline-immunoreactivity was determined in brain sections from adult male Wistar rats (200-300g). Animals were deeply anaesthetized with chloral hydrate and perfused through the dorsal aorta with 50 ml 0.9 % NaCI containing heparin prior to fixa- tion with 300 ml of 4% paraformaldehyde in 0.1 M sodium phosphate buffer. Brains were removed and postfixed for 120 min, stored in 15 % sucrose for 2~48 h, then sectioned at 25 #m on a cryostat. Free-floating sec- tions were collected in 0.02 M phosphate-buffered saline (PBS) pH 7.4 and then incubated with the rabbit anti- citrulline antibody diluted to 1:20,000 in PBS with 0.3 % Triton X-100, 2% normal goat serum, and 0.01% sodium azide (PBST) for 48 h at 4°C. After rinsing for 3 x 20 min in PBS, the sections were incubated in biotinylated goat anti-rabbit IgG (1:200; Vector ABC Kit) for I h at room temperature, washed again and incubated in the avidin- biotinylated horseradish peroxidase complex for 1 h. The sections were then washed, and reacted for peroxidase using the glucose oxidase-diaminobenzidine-nickel tech- nique [ 14].
Other sections were incubated in rabbit anti-citrulline serum diluted to 1:2000 in PBST for 48 h at 4°C. After washing, sections were incubated in Texas red-conju- gated goat anti-rabbit IgG (Jackson Labs) diluted to 1:40 in PBST for 1 h at room temperature. Sections were
LYS
Q CIT
ARG
GLN
ASA
Fig. 1. Photograph of whole brain binding specificity conjugates after ABC-immunoperoxidase staining. The various amino acids fixed to whole brain homogenate by glutaraldehyde are: lysine (LYS), citrulline (CIT), arginine (ARG), glutamine (GLN), and argininosuccinate (ASA). The positive staining of the citrulline conjugate and the absence of cross-reaction with the other conjugates indicates the specific affi- nity of the antibody for aldehyde conjugates of citrulline and protein.
then examined and photographed with a Leitz fluores- cence microscope. After photography the sections were rinsed for 15 min in 50 mM Tris-C1 buffer pH 8.0 with 0.3% Triton X-100, then reacted for NADPH-diaphor- ase histochemistry by incubation in 10 ml Tris-C1 buffer containing Triton X-100, 10 mg fl-NADPH, and 1 mg Nitro blue tetrazolium at 37°C for 20 min. They were then re-examined and photographed under bright field illumination.
Immunoperoxidase staining of nitrocellulose mem- branes on which various amino acid-brain protein conju- gates had been blotted resulted in staining of only the citrulline conjugate (Fig. 1). The antibody did not cross- react with lysine, glutamine, arginine or argininosucci- nate conjugated to whole brain homogenate, nor with brain protein itself. The antibody also does not appear to recognize citrulline residues produced in proteins by posttranslational modification via peptidylarginine dei- minase, since the distribution of citrulline-immunoreac- tivity was completely distinct from that of this enzyme in brain (Vincent and Watanabe, unpublished observa-
a b
• '.'~ ~ :::-.
Fig. 2. Camera lucida mapping of citrulline-immunoreactive neurons in the rat brain at the level of the striatum (a), hypothalamus (b), mid- brain (c) and rostral pons (d). Positive neurons are indicated by black dots and are scattered in the caudate-putamen (cp), nucleus accumbens (na) and olfactory tubercle (or). Clusters of neurons are present in the paraventricular (pv) and supraoptic (so) hypothalamic nucleL the dor- solateral periaqueductal grey (pg), the interpeduncular nucleus (ip), the dorsal raphe (dr) and the laterodorsal tegmental nucleus (tld). Scat- tered cells are also seen in the piriform cortex, endopiriform nucleus,
and the superior and inferior colliculus (ic). gp, globus pallidus.
157
E
Fig. 3. Photographs of sections stained for citrulline-immunoreactivity by the ABC-immunoperoxidase method. Citrulline-positive cells exist in the striatum (A,B), the paraventricular nucleus of the hypothalamus (C,D), the supraoptic nucleus (E), the interpeduncular nucleus (F), inferior colliculus (G), dorsal raphe nucleus (DR) (H), and the laterodorsal tegmental nucleus (TLD) (H,I). Scale bars indicate 100 am (A,E,H,I) and 50 (B,C,D,F,G)
um.
158
tions). Preabsorption of the diluted antibody overnight with 10 nmol/ml of citruUine-BSA conjugate abolished
cell staining. Preabsorption with 100 nmol/ml free citrul-
line, or with BSA alone, did not block staining. Thus, the antiserum specifically recognizes aldehyde conjugates of citrulline and protein.
Fig. 4. Citrulline immunofluorescence in the striatum (A,C) and the pedunculopontine tegmental nucleus (E,G) followed by NADPH-diaphorase histochemistry (B,D,F,H). Note that all of the citrulline-immunoreactive neurons also contain NADPH-diaphorase, but many NADPH-diaphorase-
positive neurons do not display citrulline-immunoreactivity. Scale bars represent 100 am (A,B,E,F), 50 am (G,H), or 30/~m (C,D).
Immunoperoxidase staining with the highly diluted citrulline antibody (1:20,000) revealed numerous citrul- line-immunoreactive cell groups, particularly in the fore- brain. The distribution of citrulline-immunoreactive neurons was mapped by camera lucida at representative levels through the rat brain (Fig. 2). Strong citrulline- immunoreactivity was displayed by numerous cells scat- tered throughout the striatum (Fig. 3A,B), nucleus accumbens and olfactory tubercle. Occasionally, isolated cells embedded within the corpus callosum displayed citrulline-immunoreactivity. A small scattered popula- tion of positive neurons was also present in the piriform cortex and a few positive cells were found in the endopir- iform nucleus. Within the hypothalamus, well defined clusters of positive neurons were observed in the para- ventricular (Fig. 3C,D) and supraoptic (Fig. 3E) nuclei. All of these citrulline-immunoreactive forebrain neurons displayed intensely stained soma as well as staining of larger processes.
Several citrulline-immunoreactive cell groups were also observed in the midbrain. A few small, strongly stained cells were present in the interpeduncular nucleus (Fig. 3F). A small number of isolated cells was observed scattered in the external layers of the inferior (Fig. 3G) and superior colliculi. A moderately dense cluster was observed within the dorsal raphe nucleus (Fig. 3H), and additional groups of cells were present in the laterodor- sal (Fig. 3H,I) and pedunculopontine tegmental nuclei. Most of the positive cells in the hindbrain were noticea- bly less intensely stained than the forebrain neurons.
Since all the above cell groups have previously been shown to be stained selectively by NADPH-diaphorase histochemistry [16], double staining experiments were undertaken in which citrulline was detected using indi- rect immunofluorescence, the sections photographed and then restained for NADPH-diaphorase activity. All citrulline-positive cell detected were also NADPH-dia- phorase positive (Fig. 4). However, many more cells were labelled by NADPH-diaphorase histochemistry than by citrulline immunostaining. No cases of citrul- line-positive and NADPH-diaphorase-negative cells were observed.
These results provide the first report of the cellular localization of citrulline in the brain. Citrulline-immu- noreactivity was only detected in neurons possessing NADPH-diaphorase activity. NO synthase is known to be an NADPH-dependent enzyme [1, 7] and we have recently shown that neuronal NADPH-diaphorase is a NO synthase [4]. Thus citrulline would appear to be formed from arginine together with NO in these cells.
Not all neurons known to contain NADPH-diaphor- ase activity were observed to display citrulline-immuno- reactivity. For example, many NADPH-diaphorase-
159
positive cells are known to exist within the striatum [15, 16], however, only about 60 % of these display citrulline- immunoreactivity. The lack of detectable citrulline- immunoreactivity in many NADPH-diaphorase-positive cells may be due to a high turnover rate of citrulline in these cells, such that levels are below those detectable by the immunohistochemical techniques used.
What becomes of the citrulline produced in brain? Endothelial cells can maintain a constant level of argi- nine despite continuous release of nitric oxide [9] and, when grown in an arginine free medium, can generate arginine from citrulline [3]. This suggests that endothe- lial cells contain a modified urea cycle whereby citrulline formed from arginine by NO synthase is recycled to argi- nine via argininosuccinate [3]. We suggest that a similar cycle may exist in brain in those neurons possessing NO synthase. This would explain the occurrence of citrulline [12, 13] and the presence of the enzymes of the latter half of the urea cycle [11] in brain.
Supported by a grant from the Medical Research Council of Canada and the British Columbia Health Care Research Foundation. S.R.V. is an MRC Scientist. We thank Ms. E. Leung for her excellent technical assist- ance.
1 Bredt, D.S. and Snyder, S.H., Isolation of nitric oxide synthetase, a calmodulin-requiring enzyme, Proc. Natl. Acad. Sci. U.S.A., 87 (1990) 682-685.
2 Buniatian, H.Ch. and Davtian, M.A., Urea synthesis in brain, J. Neurochem., 13 (1966) 743-753.
3 Hecker, M., Sessa, W.C., Harris, H.J., Anggard, E.E. and Vane, J.R., The metabolism of L-arginine and its significance for the bio- synthesis of endothelium-derived relaxing factor: cultured endothe- lial cells recycle L-citrulline to L-arginine, Proc. Natl. Acad. Sci. U.S.A., 87 (1990) 8612-8616.
4 Hope, B.T., Michael, G.J., Knigge, K.M. and Vincent, S.R., Neur- onal NADPH-diaphorase is a nitric oxide synthase, Proc. Natl. Acad. Sci. U.S.A., 88 (1991) 2811-2814.
5 Hsu, S.-M., Rawo, L. and Fanger, H., Use of the avidin-biotin-per- oxidase complex (ABC) in immunoperoxidase technique, J. Histo- chem. Cytochem., 29 (1981) 577-580.
6 Jones, M.E., Anderson, A.D., Anderson, C. and Hodes, S., Citrul- line synthesis in rat tissues, Arch. Biochem. Biophys., 95 (1961) 499-507.
7 Knowles, R.G., Palacios, M., Palmer, R.M.J. and Moncada, S., Formation of nitric oxide from L-arginine in the central nervous system: a transduction mechanism for stimulation of the soluble guanylate cyclase, Proc. Natl. Acad. Sci. U.S.A., 86 (1989) 5159-- 5162.
8 Krebs, H.A. and Henseleit, K., Untersuchungen fiber die Harn- stoffbildung im Tierkfrper, Hoppe-Seyler Z. Physiol. Chem., 210 (1932) 33-66.
9 Mitchell, J.A., Hecker, M., .~nggard and Vane, J.R., Cultured endothelial cells maintain their L-arginine level despite the conti- nuous release of EDRF, Eur. J. Pharmacol., 182 (1990) 573-576.
10 Ottersen, O.P. and Storm-Mathisen, J., Glutamate- and GABA- containing neurons in the mouse and rat brain, as demonstrated
160
with a new immunocytochemical technique, J. Comp. Neurol., 229 (1984) 374-392.
11 Ratner, S., Morell, H. and Carvalho, E., Enzymes of arginine meta- bolism in brain, Arch. Biochem. Biophys., 91 (1960) 28(~289.
12 Sadasivudu, B. and Rao, T.I., Studies on functional and metabolic role of urea cycle intermediates in brain, J. Neurochem., 27 (1976) 785~94.
13 Shaw, R.K. and Heine, J.D., Ninhydrin positive substances present in different areas of normal rat brain, J. Neurochem., 12 (1965)
151-155,
14 Shu, S., Ju, G. and Fan, L., The glucose oxidase-DAB-nickel meth- od in peroxidase histochemistry of the nervous system, Neurosci. Lett., 85 (1988) 169 171.
15 Vincent, S.R., Johansson, O., H6kfelt, T., Skirboll, L., Elde, R.P., Terenius, L., Kimmel, J. and Goldstein, M., NADPH-diaphorase: a selective histochemical marker for striatal neurons containing both somatostatin- and avian pancreatic polypeptide (APP)-like immunoreactivities, J. Comp. Neurol., 217 (1983) 252-263.
16 Vincent, S.R., NADPH-diaphorase histochemistry and neuro- transmitter coexistence. In P. Panula, H. P~iiv~irinta and S. Soinila (Eds.), Neurohistochemistry: Modern Methods and Applications, Alan R. Liss, Inc., New York, 1986, pp. 375-396.
