Tyrosine Hydroxylase in Vasopressinergic Axons of the Pituitary Posterior Lobe of Rats Under Salt-Loading as a Manifestation of Neurochemical Plasticity

In this.study, we attempted to test whether tyrosine hydroxylase (TH), the first ratelimiting enzyme of catecholamine synthesis, is confined to the perikarya of activated magnocellular vasopressinergic (VPergic) neurons or is also present in their distal axons in the pituitary posterior lobe (PL). In addition, we evaluated the possible correlation between TH and VP turnover in the axons of rats drinking 2% NaCl for 1, 2, and 3 weeks. To this aim, we examined the large swellings of VPergic axons, the so-called Herring bodies, using the doubleimmunofluorescent technique and the avidinbiotin technique, combined with image analysis. Here we have demonstrated for the first time a colocalization of TH and VP in Herring bodies, which is a strong argument in favor of TH transport from the perikarya of VPergic neurons via axons toward their terminals. THimmunoreactive (IR) and VP-IR materials were distributed in Herring bodies with seeming zonality. The number of VP-IR Herring bodies decreased by a factor of four over the first week of osmotic stimulation, remaining at almost the same low level until the end of the experiment. Conversely, the content of the VP-IR material within the individual Herring bodies fell gradually during the three weeks of saltloading. The results suggest that VP depletion from Herring bodies prevails in its transport into these structures during the whole period of osmotic stimulation. In contrast to VP-IR Herring bodies, the number of TH-IR Herring bodies and the content of TH-IR material within the individual Herring bodies increased progressively during the entire experiment. The synchronization of the VP depletion and TH accumulation in Herring bodies during longterm osmotic stimulation raised the question about a possible functional interaction between both substances.

In this study, we attempted to test whether tyrosine hydroxylase (TH), the first ratelimiting enzyme of catecholamine synthesis, is confined to the perikarya of activated magnocellular vasopressinergic (VPergic) neurons or is also present in their distal axons in the pituitary posterior lobe (PL). In addition, we evaluated the possible correlation between TH and VP turnover in the axons of rats drinking 2% NaC! for 1, 2, and 3 weeks. To this aim, we examined the large swellings of VPergic axons, the so-called Herring bodies, using the doubleimmunofluorescent technique and the avidinbiotin technique, combined with image analysis.
Here we have demonstrated for the first time a colocalization of TH and VP in Herring bodies, which is a strong argument in favor of TH transport from the perikarya of VPergic neurons via axons toward their terminals. THimmunoreactive (IR) and VP-IR materials were tCorresponding author: Tel: +7 095 135 88 42 Fax: +7 095 135 80 12 e-mail: michael-ugrumov@mail.ru distributed in Herring bodies with seeming zonality. The number of VP-IR Herring bodies decreased by a factor of four over the first week of osmotic stimulation, remaining at almost the same low level until the end of the experiment.
Conversely, the content of the VP-IR material within the individual Herring bodies fell gradually during the three weeks of saltloading. The results suggest that VP depletion from Herring bodies prevails in its transport into these structures during the whole period of osmotic stimulation. In contrast to VP-IR Herring bodies, the number of TH-IR Herring bodies and the content of TH-IR material within the individual Herring bodies increased progressively during the entire experiment. The synchronization of the VP depletion and TH accumulation in Herring bodies during longterm osmotic stimulation raised the question about a possible functional interaction between both substances.
Taking into account that TH is a cytosolic nonsecreted protein, one might expect its accumulation in the perikarya of VPergic neurons during long-term osmotic stimulation, followed by its possible transport toward the axon terminals.
However, a previous attempt to detect differences in TH-immunoreactivity (TH-IR) in the PL of intact and salt-loaded animals was unsuccessful (Meister, 1990b (Saavedra et al., 1975, Saavedra, 1985 belonging to the neurons of the periventricular (Kawano & Daikoku, 1987) nucleus, as well as by noradrenergic fibers issued from the superior cervical ganglia (Saavedra, 1985;Vanhatalo et al., 1995).
The difficulties in detecting both substances in VPergic axons most probably could be overcome by studying Herring bodies, the large axon swellings that are generally considered a specific attribute of VPergic axons, but not of catecholaminergic axons (Dellmann & Rodriguez, 1970;Polenov & Garlov, 1971). In contrast to the relatively small swellings of the VPergic axons, Herring bodies partly retain VP under osmotic stimulation (Ugrumov & Belenkii, 1974), namely, after the onset of TH synthesis in VPergic neurons (Kiss & Mezey, 1986). The present study attempted to test whether TH is an attribute not only of the cell bodies but also of the axons of magnocellular VPergic neurons, subsequently evaluating a possible correlation between TH and VP dynamic in Herring bodies during long-term salt-loading of the rats.

Animals and Processing of Materials
Adult male Wistar rats, weighing 200 to 300 g each, were used in this study. The animals were maintained at 21C to 23C in a light-dark cycle (light period between 8:00 and 20:00). Control (intact) rats had free access to food and tap water (nine animals), whereas the experimental rats were given 2% NaCI instead of tap water for either week (nine animals), 2 weeks (three animals), or 3 weeks (three animals). The rats were anesthetized with Pentobarbital (50 mg/kg body wt) and perfused via the heart--first with saline for 10 min at 37C, and then with 4% paraformaldehyde in 0.1 M sodium phosphate buffer (pH 7.3) for 15 min at 4C. Thereafter, the pituitaries were postfixed by immersion in the same fixative for 2 h at 4C, rinsed for h in 0.02 M phosphate buffer with 0.9% NaCI (PBS, pH 7.2 to 7.4) at 4C, immersed overnight in 15% sucrose PBS at 4C, and then frozen in isopentane at -45C.

Immunohistochemistry and Quantification
In the first series of immunocytochemical studies, the pituitaries of intact rats and of rats that wer salt-loaded for 7 d (3 rats in each group) were used for TH and VP double-immunolabeling on the same cryostat sections, 14 Ixm or 30 ktm thick.
The sections were incubated overnight at 4C in PBS containing a mixture of either (a)mouse monoclonal antibodies to TH (diluted 1:1000) (Sigma, USA) and rabbit polyclonal antibodies to VP (diluted 1:4000) (Chemicon, USA) or (b) mouse monoclonal anti-bodies to VP (diluted 1:20000) (Robert et al., 1985) and rabbit polyclonal antibodies to TH (diluted 1:2000), 1% normal goat serum, and 0.1% Triton X-100. The sections were then incubated for 2 h at room temperature with a mixture of the secondary antibodies, rhodamine-conjugated goat anti-rabbit (diluted 1:40) (DAKO, Denmark) and fluoresceinconjugated goat anti-mouse (diluted 1" 100) (Pasteur, France). The sections were rinsed twice in PBS, embedded in a mounting medium for fluorescence (Vectorlab, USA), coverslipped, and examined using either a fluorescence microscope (Leitz Diaplan, Germany) or a confocal microscope (Leica TCS 4D, Germany). The fluorescence microscope was equipped with filters for fluorescein-induced fluorescence (450-490 nm excitation filter, 520-560 nm barrier filter, and 520-550 nm extra barrier filter) and for rhodamine-induced fluorescence (546/10 nm excitation filter and 610 nm barrier filter). The confocal microscope was equipped with an argon/krypton laser 25 mW adjusted at 488 nm and 568 nm wavelengths for the respective excitations, as well as with filter sets for the selective detection of fluorescein (488 nm, green) and rhodamine (568 nm, red) fluorescence. The 30 lam-thick sections were x/y scanned along the z-axis with 1-1.5 Ixm step. Each section was scanned twice" at first with oil immersion 63x objectives for mapping Herring body distribution and then with an oil immersion 63x with zoom 2x objective for evaluating the fine structure of the Herring bodies. Vertical scanning planes were within the whole depth of the section.
The specificity of the antiserum to TH was determined previously (Arluison et al., 1984). Additional controls for the specificity of antibodies to VP and TH consisted of either omitting the primary antibodies or preadsorbing them with antigens. The possible cross-reactivity between primary and secondary antibodies was excluded by incubating for each antigen the corresponding mouse primary antibodies with goat anti-rabbit secondary antibodies or, conversely, rabbit primary antibodies with goat anti-mouse secondary antibodies, which resulted in no labeling. The intermediate lobe of the pituitary, being free of VP-immunorective (VP-IR structures, served as an additional physiological control. In the second series of the immunocytochemical studies, the pituitaries of intact rats, as well as those of rats that were salt-loaded for 1, 2, or 3 weeks (three animals in each group), were cut with a cryostat microtome for serial sections, 14 Ixm thick. Adjacent sections were mounted on two sets of slides and processed separately for either VP or TH immunocytochemistry. Each slide displayed the sections of an intact animal, as well as those of animals at any period of salt-loading (six sections per animal). The sections were incubated successively with the following: (a) goat normal serum (1%) for 30 min at room temperature; (b) rabbit polyclonal antibodies to TH (diluted 1:2000)  and salt-loaded rats. According to this approach, the relative concentrations of VP and TH could be accurately assessed, and the results obtained at different periods of osmotic stimulation could easily be compared in-between and with those in control animals. Indeed, all preparative procedures of the sections were performed under standardized conditions that allowed us to apply meaningful quantitative and semi-quantitative analysis. Each section in the entire series of the PL was reviewed under a light microscope using a 25x objective, followed by transferring the image via video camera (CCD) to a computer monitor (IBM PC Pentium 60) and recording. The relative concentrations of VP-IR and TH-IR materials were measured as the 'gray level' (GL), related to the optical density (OD) of the specimen, as follows (Conn, 1990): The GL of the background was measured in the vicinity of the PL, namely, in the anterior lobe with no specific immunostaining. The following average parameters of Herring bodies were estimated in intact animals, as well as in each experimental group: (a) OD ofVP-IR or TH-IR material in Herring bodies; (b) number of VP-IR or TH-IR Herring bodies per section, (c) area of the individual VP-IR or TH-IR Herring bodies, (d) relative amount of VP-IR or TH-IR material per Herring body (area x OD) as a percentage of its level in the control (intact) animals (100%). The profiles of individual Herring bodies were outlined with a light pen to obtain a morphometric mask.
Statistical analysis was carried out using descriptive statistics and the non-parametric Wilcoxon matched pair test.

Intact Rats
The PL of intact rats contained an abundant network of VP-IR fibers having transient and terminal swellings. A minor portion of the swellings was represented by Herring bodies, more than 6 gm in diameter (Fig. a). The Herring bodies were either scattered throughout the PL or concentrated in its proximal portion, as well as along the border with the intermediate lobe. In the individual Herring bodies, VP-IR material was distributed either homogeneously or mainly at the periphery, thereby showing weak immunostaining of the central portion.
The PL of intact animals was also occupied by a network of TH-IR fibers (Fig. b), being particularly dense along the intermediate lobe and in the most proximal region of the PL. In contrast to VP-IR fibers, TH-IR fibers possessed swellings that were usually no more than 2 lam in size. Nevertheless, occasional, relatively large and intensely immunostained TH-IR swellings were also observed (Figs. 2 to 4). No Herring bodies with colocalization of VP-IR and TH-IR materials were detected in the PL of intact animals when using immunofluorescence double-labeling.

Rats Salt-Loaded for 1 Week
Herring bodies, immunoreactive for both VP and TH, were distributed in the same fashion as in the PL of intact rats (Fig. c, d). They were usually seen in close vicinity of the intermediate lobe and the hypophysial stalk. According to the avidin-biotin immunocytochemical study, the VP-IR Herring bodies decreased dramatically in number (by fourfold) (Fig. 2) and slightly but significantly in size (Fig. 3). Although no change in the OD of VP-IR Herring bodies was observed (Fig. 4), the relative content of VP-IR material per Herring body slightly decreased (Fig. 5).  In contrast to the VP-IR Herring bodies, the number (Fig. 2) and the average area (Fig. 3) of the TH-IR Herring bodies increased considerably, which was accompanied by an increase of the amount of the TH-IR material per Herring body (Fig. 5).
The double-immunofluorescence study showed that numerous VP-IR Herring bodies contained TH-IR material (Fig. 6a, b), although monolabeled VP-IR (Fig. 6c), and TH-IR were also observed. The distribution of VP-IR and TH-IR materials in some Herring bodies was not homogeneous (Fig. 6d). The VP-IR material was often distributed at the periphery of Herring bodies. In addition to Herring bodies, double-immunolabeled thin fibers with small swellings (<1.5 lam) were regularly seen in a confoeal microscope (Fig. 6e).

Rats Salt-Loaded for Two Weeks
Over the second week of osmotic stimulation, no change in the number and size of the VP-IR Herring bodies was observed (Figs. 2, 3), whereas their OD and the content of the VP-IR material per Herring body slightly decreased (Figs. 4,5). In contrast to the VP-IR Herring bodies, the number (Fig. 2) and OD (Fig. 4) of TH-IR Herring bodies, as well as the content of the TH-IR material per Herring body (Fig. 5), increased significantly. The average area of the TH-IR Herring bodies did not change during the second week of osmotic stimulation (Fig. 3).

Rats Salt-Loaded for Three Weeks
During the third week of osmotic stimulation, the number, size, and OD of the VP-IR Herring bodies did not change (Figs. 2 to 4), whereas the content of the VP-IR material per Herring body slightly decreased. The same was true for the average area of the TH-IR Herring bodies (Fig. 3), whereas their numbers (Fig. 2), OD (Fig. 4), and content of TH-IR material in the individual Herring bodies increased considerably.

Technical Remarks
In this study, a double-immunofluorescence technique was applied to determine whether TH is an attribute not only of the perikarya but also of the distal axons of stimulated magnocellular VPergic neurons. Moreover, an image analysis made it possible to specify VP-IR and TH-IR Herring bodies in rats, under normal laboratory conditions and following 1, 2, or 3 weeks of saltloading, by measuring the number of Herring bodies, their individual area, and the OD of IR material. We have emphasized (see Experimental Methods) that the PL of intact rats and of those at any stage of salt-loading were processed under standardized conditions (fixation, cutting, immunostaining of the sections mounted on the same slides) for quantitative studies and particularly for the semi-quantitative analysis of TH-IR and VP-IR materials. The validity of this approach was proved in earlier technical studies (Conn, 1990) and then confirmed by concrete applications (Beltramo et al., 1994(Beltramo et al., , 1997. Still, it should be taken into account that we could compare only the relative intracellular amounts of the IR materials in the animals of different experimental groups, but separately for TH-IR and VP-IR materials. In other words, this technique failed to evaluate the absolute content of the IR material and to compare the OD or the relative amounts of TH-IR and VP-IR materials in Herring bodies. This is the first study attempting to compare the dynamics of the TH and VP turnover in the axons of VPergic neurons of the supraoptic nucleus during long-term osmotic stimulation by using avidin-biotin immunoeytochemistry and image analysis. When evaluating the intra-axonal dynamics of TH and VP, it should be taken into account that these substances differ considerably in some specific properties. Thus, TH is a nonreleasable eytosolic protein that is degraded by cytosolic peptidases (Fernandez & Craviso, 1999). Conversely, VP is a secreted peptide that is stored in secretory granules and protected from enzymatic degradation by the granule membrane (Castel et al., 1984). From the above comments follows that the relative average concentration (OD) or the content (ODxarea of a Herring body) of TH-IR material per Herring body is proportional to the difference between the TH transport from the cell bodies of VPergie neurons and the enzymatic degradation of TH in Herring bodies. In turn, the relative average concentration (OD) or oment (ODxarea of a Herring body) of the VP-IR material per Herring body is proportional to the difference between VP transport from the cell bodies of VPergie neurons and VP depletion from these structures. The relative contents of polypeptides (IR materials) per Herring body appeared to be a more informative index for their turnover than their relative concentrations were because the Herring body size changed during the course of the experiment. For instance, the relative concentration of the VP-IR material (OD) in Herring bodies did not change for the first week of osmotic stimulation, whereas the relative content of the VP-IR material per Herring body decreased over the same period (see Results).

Colocalization of TH and VP in Herring Bodies
In this study, we have demonstrated for the first time a colocalization of TH and VP in Herring bodies in the PL by using an immunofluorescence double-labeling technique. In addition to Herring bodies, small double-labeled axon swellings were seen regularly in the PL of the rats that were saltloaded for a week when using eonfocal microsocopy. The data strongly suggest that TH is transported from the cell bodies located in the supraoptic nucleus toward the axon terminals in the PL. Although TH mRNA has been detected before in the VPergic axons in the median eminence on sections (Skutella et al., 1994) and in the homogenate of the 'neurointermediate' (PL + intermediate lobe) lobe (Melia et al., 1994), the authors are skeptical about the possible translation of TH mRNA in the axon terminals because of the absence of translational machinery in this neuronal compartment.
Although both intact rats and those that were salt-loaded for a week were tested for eolocalization of TH and VP, double-labeled Herring bodies were detected only in the salt-loaded animals. Apparently, this result is accounted for by the extremely low frequency of TH-eontaining Herring bodies in intact rats, whereas their number significantly increased for the first week of osmotic stimulation (see Results). The same was also true for VPergic cell bodies expressing TH gene and synthesis in rats during the first week of salt-loading (Abramova et al., unpublished).
Surprisingly, some double-labeled Herring bodies did not show full superposition of the TH-IR and VP-IR profiles. Moreover, VP-immuno-staining of some Herring bodies was stronger at the periphery than in the center. These findings appear to agree with earlier electron microscopic observations of the zonality in the distribution of subeellular structures in Herring bodiesthe VP-containing secretory granules occupied the periphery, whereas the central area was filled with microtubules (Dellmann & Rodriguez, 1970;Polenov & Garlov, 1971;Ugrumov & Belenkii, 1974). It is noteworthy that VP-IR Herring bodies decreased in size for the first week of osmotic stimulation, while TH-IR Herring bodies increased in size during the same period, becoming as large as VP-IR Herring bodies. After that, the Herring bodies immunostained for TH or VP did not change in size until the end of the experiment. These observations might be a result of two synchronous processes: (1) a sharp depletion of the peripheral area of Herring bodies from the VP-containing secretory granules that perhaps was followed by the shrinkage of Herring bodies, and (2) a spread of the TH-IR material from the center to the periphery of Herring bodies, due to its diffusion either through axoplasm or via the hypertrophied microtubular system. Indeed, the concentration of intra-axonal microtubules increases under functional stimulation of the VP magnocellular system (Knowles & Vollrath, 1974).

TH and VP Turnover in Herring Bodies
In the present study, peculiar attention was paid to VP and TH turnover in Herring bodies during the long-term osmotic stimulation of VPergic neurons. The number of VP-IR Herring bodies decreased fourfold over the first week of osmotic stimulation, remaining almost at the same low level until the end of the experiment. In other words, the depletion of VP from Herring bodies appeared to occur at a higher rate when compared with the depletion of the VP axons in the PL as a whole (Abramova et al., unpublished), which contradicts earlier qualitative studies (Ugrumov & Belenkii, 1974). The relative content of VP-IR material within individual Herring bodies fell gradually over the entire period of osmotic stimulation studied here, whereas its concentration decreased only for the second and the third week. This result suggests that the VP depletion from Herring bodies prevailed during its transport from the cell bodies for the whole experiment. In contrast to VP-IR Herring bodies, the number of TH-IR Herring bodies progressively increased during the 3 weeks of salt-loading. This phenomenon is apparently a consequence of the gradually increasing number of VPergic neurons co-expressing TH (Abramova et al., unpublished) and finally, of the transport of TH in these neurons from the perikarya toward axon terminals.
The functional significance of TH in magnocellular VPergic neurons remains unclear. The VPergic cell bodies in the supraoptic nucleus lacked the specific markers of TH enzymatic activity: guanosine triphosphate-cyclohydrolase, the enzyme cathalizing the TH co-factor production (BI-I4), and L-DOPA, the immediate precursor of dopamine (Marsais & Calas, 1999). Moreover, one cannot exclude the superfluous gene expression of TH as that of tryptophan hydroxylase in the neurons of the zona incerta (Vanhatalo & Soinila, 1996). Still, the synchronization of gradual VP depletion and TH accumulation in Herring bodies during the entire period of long-term osmotic stimulation studied here raises the question of the functional interaction between VP and TH in axon terminals.
Thus, the VPergic axons of osmotically stimulated neurons contained TH. The VP release from Herring bodies gradually increased during longterm osmotic stimulation and was synchronized with TH accumulation, at least for the second and third weeks. This correlation suggests functional interactions between VP and TH in axon terminals.

ACKNOWLEGMENTS
The present study was supported by the following grants: PICS (98-04-22018), NATO