In this report, introducing in rats a method of combining electrophysiological recordings followed by microiontophoretic injections of CTb through the same micropipette, we obtained evidence for specific afferents to the rat LC. We confirmed many of these afferents with control retrograde and anterograde tracing experiments. Afferents to the LC that contained a substantial number of retrogradely labeled neurons after injections that were apparently restricted to the LC, and that were confirmed with substantial to large fiber innervation of the LC using anterograde tract-tracing with PHAL or CTb, included the lateral paragigantocellular and Kölliker-Fuse nuclei, the preoptic area, and the ventrolateral periaqueductal gray. We also found a substantial number of retrogradely labeled neurons in posterior hypothalamic areas after injections that were apparently restricted to the LC but we observed only a moderate to small fiber innervation of the LC using anterograde tract-tracing with PHAL or CTb. The dorsomedial rostral medulla also contained many cells retrogradely labeled from the LC; anterograde tracing was not conducted as it was reported in a previous study 6
The sensitivity of tract tracing molecules, and the methods for their use, have advanced considerably since the first studies were performed to examine inputs to the LC 10, 13. Moreover, the "micro-anatomy" of the LC region, with marked differences in inputs to closely neighboring small areas surrounding the LC vs the LC proper, has become apparent only recently with the advent of more sensitive tracers allowing tracing from small deposits and with the development of sensitive anterograde tract-tracers (e.g., PHAL) 24. Our results differ from previous tract-tracing studies of afferents to LC in two major ways: (i) we found that several previously reported inputs to the LC project instead to areas surrounding the LC, and (ii) we observed afferents to the LC that were not reported in previous studies. Both of these differences between our results and those in previous studies can be attributed to increased sensitivity of tracers, and the increased awareness of the "micro-anatomy" of the dorsolateral pontine tegmentum.
Injections of unconjugated HRP yielded labeling in several structures that were not substantially retrogradely labeled in our study (e.g., amygdala, dorsal spinal horn, nucleus of the solitary tract) 10, 13. Our results and those of Aston-Jones et al. 6, 7 using anterograde tracing indicate that these inputs are directed to surrounding nuclei but not to the core LC nucleus (see detailed discussion below). Thus, it appears that injections of unconjugated HRP may have diffused out of the LC proper to label afferents not only to the LC, but also inputs that selectively innervate areas and nuclei around the LC but not the LC proper. This is a potential difficulty of retrograde studies of afferents to small brain nuclei, and indicates the importance of making focal injections of retrograde tracers restricted to the nuclear core of the LC. These findings also underscore the importance of confirming suspected afferents with anterograde tract tracing.
While the above analysis indicates the importance of making small injections of retrograde tracers restricted to the nuclear core of the LC, a potential problem with such small tracer deposits is that the sensitivity of the tracer may be too low to reveal all inputs to the area injected. Indeed, the present study as well as recent retrograde tracing experiments using pressure injections of WGA conjugated to inactivated (apo) HRP and coupled to colloidal gold (WGA-apoHRP-Au) in the LC (Aston-Jones and Zhu, unpublished observations) reveals certain inputs that were not apparent in previous experiments that used WGA-HRP or Fluoro-Gold as retrograde tracers. This analysis indicates that CTb (and WGA-apoHRP-Au) may be more sensitive tract-tracers than WGA-HRP and Fluoro-Gold when employed in very small deposits.
There is another advantage of the smaller injections made possible by the increased sensitivity of CTb and WGA-apoHRP-Au compared to previous tracers: It becomes more feasible to analyze possible inputs from neurons in the areas surrounding the injection site. For example, previous studies with WGA-HRP injections into the LC were inconclusive regarding possible inputs from the periaqueductal gray due (at least in part) to the close proximity of this area to the LC injection site 6.
It is noteworthy that CTb may have its own limitations as a retrograde tracer as well. For example, although we and others found no evidence for trans-synaptic transport in the central and peripheral nervous system 41, 54, detailed studies remain to be done to completely exclude such a possibility. In addition, the size of the injection site in the first hours after CTb injections is unknown. It is possible that the site in these early hours is larger than after 1-7 days, and that apparently small injections in animals with long survival times may actually have been considerably larger during the earlier periods of uptake and transport. In addition, it is possible that CTb (or any other tracer) is selectively taken up and transported by certain fibers and not by others. If this is true, there may be additional inputs to the LC or the surrounding area that are not revealed in the present set of experiments. Of interest in this regard, we observed only a limited number of labeled cells in the LDT and the periLC area after CTb injections in the LC. In contrast, after small WGA-apoHRP-AU injections into the LC, a considerable number of labeled neurons were seen in certain nuclei immediately surrounding LC, e.g., Bar (Aston-Jones et al., unpublished observations). One explanation for such a difference in local labeling obtained with these two tracers might be that WGA-apoHRP-Au, but not CTb, is taken up and transported by dendrites of neurons that may invade the LC. Another possibility is that CTb excludes axonal inputs to the LC from surrounding regions based upon some other selective uptake or transport property of this compound. These considerations underscore the importance of confirming results of retrograde labeling with anterograde tracing.
Although we found that CTb is a good anterograde tracer, it also has some limitations (see anterograde section in Results), the most serious being that this compound strongly retrogradely labels neurons and their processes. Retrogradely labeled neuronal processes could be mistaken for anterogradely labeled fiber inputs. Therefore, injections into areas that are efferent targets of the LC may lead to erroneous conclusions regarding sources of anterograde fiber labeling in the LC. This is a particularly common problem with studies of afferents to the LC because of the uniquely widespread projections of LC neurons. Therefore, while negative results with anterograde labeling of CTb may be very informative (assuming effective transport from the cells of interest is demonstrated), positive fiber labeling in the LC after injections of CTb into most CNS areas should be taken as only tentative evidence of afferent innervation. For this reason, we confirmed many of our results by an alternative method that do not generate robust retrograde labeling (e.g., PHAL).
Neurons and dendrites of the peri-coerulear region
Anatomical studies have shown that the region surrounding the LC is complex and contains neurons as well as neuropil. Locus coeruleus neurons are multipolar and have long dendrites that extend into surrounding regions 26, 60 ,66. Recent studies 22, 61 have confirmed that LC dendrites extend for considerable distances into peri-LC regions, but are polarized being particularly densely located in two peri-LC zones, the rostromedial and caudal juxtaependymal zones. In our study, after PHAL or CTb injections in some of these areas (preoptic area, posterior hypothalmic areas, periaqueductal gray, Kolliker-Fuse), we observed generally more anterogradely labeled fibers in these periLC areas than in the nuclear core of the LC. These results suggest that LC noradrenergic cells might receive considerable inputs on their dendrites outside the nuclear core of the LC. PeriLC areas also receive a great deal of input from regions that do not, or only minimally, innervate the LC (e.g., nucleus of the solitary tract, prefrontal and infralimbic cortex, amygdala, dorsal raphe), raising the possibility that LC dendrites may receive innervation from areas that do not innervate the LC proper. One way to examine this possibility is use of electrophysiological methods to measure the functional influence of afferents to such peri-LC regions on LC impulse activity. This has been done for some possible afferents, including the central nucleus of the amygdala 6, nucleus of the solitary tract 19 and prefrontal cortex 12. In each case, little or no effect was found in LC activity of stimulating the test afferent with single electrical pulses. However, similar tests need to be conducted for other possible afferents as well. Moreover, it is possible that stimulation paradigms may not reveal modulatory effects of inputs to distal dendrites of LC neurons. Therefore, ultrastructural studies of the peri-LC region employing anterograde tracing and dopamine-beta-hydroxylase staining are also needed to evaluate the possible status of peri-LC afferents as inputs to LC dendrites. It is also noteworthy that LC dendrites do not extend laterally into the 5Me, nucleus parabrachialis or similar areas, nor caudally into the vestibular nucleus, so that afferents to these regions (e.g., frontal cortex, amygdala, spinal cord) are unlikely to contact LC neurons or dendrites.
Conversely, examination of the LC region using NADPH or cholinacetyltransferase staining has revealed that cholinergic neurons of the LDT region possess extensive dendrites, some of which innervate the LC nucleus or periLC dendritic zone 33 ,53. This raises issues of interest to the study of afferents to the LC: (i) do dendrites of these LDT neurons act presynaptically on LC neurons via a cholinergic mechanism, and (ii) do dendrites of LDT neurons located within the LC receive inputs from neurons that project into the LC. The latter possibility raises especially important issues that have not been addressed in studies to date, e.g., do some inputs to the LC preferentially innervate non-LC elements located within the LC? This possibility seems most reasonable for those projections that (i) are not particularly dense within the LC proper and (ii) densely innervate the area of source neurons in the region around LC that give rise to the non-LC dendrites located in the LC nucleus (e.g., dorsal raphe, lateral hypothalamus, preoptic area, periaqueductal gray and nucleus of the solitary tract, among others). These various data illustrate the complexity of connectivity studies in small deep brain nuclei, and point out the need for functional (electrophysiological) and ultrastructural studies to test possible inputs.
After LC injections, Cedarbaum and Aghajanian 10 and Clavier 13 observed some labeled cells in the insular cortex 10, 13. In agreement with anterograde tracing studies with PHAL or WGA-HRP showing that the insular cortex projects densely to areas corresponding to the Peri-5Me and the 5Me 44, 72 rather than to the LC, we observed labeled cells in the insular cortex only after 5Me or Peri-5Me injections.
We observed a few labeled cells in the frontal and infralimbic cortex and the claustrum after LC injections. However, the frontal and infralimbic cortex contained many more cells after a specific MVe injection or Peri-5Me injections. Moreover, after PHAL injections in the frontal or infralimbic cortex, we found only occasional anterogradely labeled fibers in the LC and many more fibers in the MVe and the Peri-5Me. These results are in agreement with previous anterograde tracing studies with PHAL or WGA-HRP showing that the infralimbic cortex projects to an area corresponding to the Peri-5Me rather than to the LC proper 12 ,32, 68.
Together, these results suggest that the infralimbic, insular and frontal cortices and the claustrum provide only a very limited input to the nuclear core of the LC and rather strongly project to the periLC areas and the surrounding nuclei. As noted above, such limited fiber labeling in the LC could reflect inputs to LC neurons or to non-LC elements located within the LC. Also, it remains to be determined whether cortical projections to periLC areas that densely contain LC dendrites (ventromedial and caudodorsal to the LC 42, 65) are directed to the dendrites of noradrenergic neurons or to non-noradrenergic cells and dendrites also present in these areas.
After LC injections, previous reports showed either a limited number 10 ,13 or only occasional labeled cells in some animals 6 in the preoptic area. In contrast, after LC injections we observed a substantial number of cells in the preoptic area dorsal to the supraoptic nucleus. By means of anterograde tracing with PHAL and CTb, we confirmed that the preoptic area dorsal to the supraoptic nucleus substantially innervates the nuclear core of the LC itself. It is noteworthy that, as found for several inputs to the LC (described above), anterograde tracing revealed a denser innervation from the preoptic area in periLC areas than in the LC proper. It remains to be determined with electronmicroscopic observations whether this projection to the periLC is directed to the dendrites of noradrenergic neurons or to non-noradrenergic cells and dendrites also present in these areas.
After injections of CTb in the Bar, we observed a large number of cells in the same area but also in the medial preoptic nucleus and the lateral preoptic area. By means of anterograde tracing with PHAL and CTb, we confirmed that the preoptic area dorsal to the supraoptic nucleus very densely innervates the Bar. Recent studies 53 using WGA-apoHRP-Au and PHAL have also confirmed that the preoptic area sends direct projections to the LC, although projections were much denser to areas surrounding the LC like the Bar. In agreement with these results, Swanson 67 using autoradiography demonstrated a projection from the lateral preoptic area to an area corresponding to the area of the Bar and rostral LC. Conrad and Pfaff 15 using autoradiography, and Simerly and Swanson 63 with PHAL injections in the medial preoptic nucleus or area, also demonstrated anterogradely labeled fibers in a region corresponding to Bar.
We also saw a large number of labeled cells in the medial part of the bed nucleus of the stria terminalis after Bar injections. Fewer cells were found after LC+periLC, 5Me or Peri-5Me injections. Moga et al. 44 reported anterogradely labeled fibers in the Bar after WGA-HRP injections in the medial part of the bed nucleus of the stria terminalis confirming that this nucleus is mainly projecting to the Bar.
Cedarbaum and Aghajanian 10 observed a relatively large number of cells in the lateral part of the bed nucleus of the stria terminalis after HRP injections into the LC. We saw a large number of cells in this area only after 5Me or Peri-5Me injections. In agreement with our results, Moga et al. 44 saw anterogradely labeled fibers in the peri-5Me and Me5 after WGA-HRP injections in the lateral part of the bed nucleus of the stria terminalis. These data indicate that the lateral part of the bed nucleus of the stria terminalis projects to the 5Me and peri-5Me but not to the LC.
We and other previous authors 6, 10 excepting Clavier 13 found a small number of retrogradely labeled neurons in the dorsal cap of the hypothalamic paraventricular nucleus after LC injections. Such a weak projection has also been observed in anterograde tracing experiments using tritiated amino-acid and PHAL 15, 38. We further found that the surrounding nuclei, particularly the Peri-5Me and the Bar, also received a projection from the paraventricular nucleus but from the medioventral parvocellular aspect.
At the same level, Cedarbaum and Aghajanian 10 and Clavier 13 described a large number of retrogradely labeled cells in the central nucleus of the amygdala after HRP injections into the LC. In contrast, we and Aston-Jones et al. 6 saw no or only occasional labeled cells in this nucleus. We observed a large number of labeled cells in the central nucleus of the amygdala only after CTb injections in the 5Me or Peri-5Me. After injections of the anterograde tracer WGA-HRP in the central nucleus of the amygdala, no fibers were observed in the LC while the 5Me, Peri-5Me and the median parabrachial nucleus contained a great number of anterogradely labeled fibers 6, 44. Stimulation of the central nucleus of the amygdala yielded little or no response in LC but markedly excited nearby parabrachial neurons 6. Taken together, these results indicate that the central nucleus of the amygdala does not project to the LC but rather to the adjacent 5Me, Peri-5Me and median parabrachial nucleus.
After LC injections, Aston-Jones et al. 6 saw occasional cells in some animals in the posterior hypothalamus. Cedarbaum and Aghajanian 10 and Clavier 13, after large LC injections very likely encroaching upon neighboring areas, reported the presence of cells in posterior hypothalamic areas. We observed a substantial number of small and medium-sized retrogradely labeled cells in the lateral and dorsal hypothalamic areas and the rostral perifornical nucleus after LC injections. We determined that these regions also project with a topographic organization for each nucleus to the Bar, 5Me, Peri-5Me and MVe. After Bar injections, labeled neurons were clustered 1) in the caudal part of the dorsal hypothalamic area and 2) around and ventromedial to the fornix. After 5Me and Peri-5Me injections, they were mostly located in the lateral hypothalamic area just medial to the internal capsule and the subthalamic nucleus. Finally after MVe injections, medium-sized cells were diffusely distributed over the posterior hypothalamic areas and the perifornical nucleus. We confirmed these results for peri-LC injections with anterograde tracing with CTb and PHAL; note however that the LC only contained a small number of fibers following such anterograde tracing. Studies using anterograde tracers such as tritiated amino-acids, WGA-HRP and PHAL have reported a projection from the posterior hypothalamic areas to the region of the LC 30, 69, 70. However, for most of them, the distribution of the terminals was not precise enough to draw a comparison with our study. Only Allen and Cechetto 1 recently reported a strong specific projection from the perifornical region to the Bar and from the lateral hypothalamic area to a region corresponding to the Peri-5Me and 5Me. On the other hand, Shirokawa and Nakamura 62 reported antidromic activation of dorsal hypothalamic area neurons from the area of the LC. These data taken together indicate that the dorsal and lateral posterior hypothalamic areas and the perifornical nucleus may provide inputs to the LC and surrounding nuclei with a very specific topographic organization. It is noteworthy that, as found for several of the inputs to the LC, anterograde tracing revealed that these hypothalamic regions innervate the peri-LC region more strongly than the LC itself (discussed above in Neurons and Dendrites in Peri-LC). Further experiments are necessary to determine the neurotransmitter content of these pathways as well as their physiological roles.
We also observed occasional cells in the hypothalamic ventromedial nucleus after LC+periLC injections. Although sparse anterograde labeling has been reported in the LC after tritiated amino-acids and PHAL injections in this hypothalamic nucleus 36 ,38, 58, more precise studies seem to benecessary to define the exact terminal field of this projection.
In contrast with previous studies, we consistently found a few retrogradely labeled cells in the tuberomammillary nucleus after LC but also Bar, Peri-5Me, 5Me and MVe injections. This nucleus contains only histamine neurons 48. A significant number of histamine immunostained fibers has been observed in the area of the LC 47. These data indicate that histamine neurons of the tuberomamillary nucleus may provide a weak diffuse innervation of the LC and surrounding nuclei. Additional studies with anterograde transport from the tuberomammillary nucleus or histamine immunohistochemistry are needed to test this possibility.
Cedarbaum and Aghajanian 10 and Clavier 13 reported large numbers of retrogradely labeled neurons in the ventrolateral periaqueductal gray following injections of HRP into the LC. However, as discussed above, injections in these studies very likely encroached upon neighboring areas, so that these retrograde labeling may have reflected inputs to peri-LC areas. Aston-Jones et al. 6 found that periaqueductal gray cells were not consistently labeled following injections of WGA-HRP into the LC; periaqueductal gray cells were often labeled near the LC injection site in that study, but as they were near the halo of the injection it was concluded that additional studies would be necessary to determine the status of the periaqueductal gray as a possible afferent to the LC. Retrograde and anterograde labeling in our study confirm that the ventrolateral part of the periaqueductal gray directly innervates the LC. These findings are also consistent with recent studies using retrograde transport of WGA-apoHRP-Au (which allows very small injections but is nonetheless very sensitive)(Aston-Jones et al., unpublished results). However, as noted above for other afferents (and discussed above under Neurons and Dendrites in the Peri-LC Area), the innervation of the peri-LC region by the periaqueductal gray was much greater than to the LC proper. In general, our results agree with those of a more recent study of periaqueductal gray inputs to the LC region 21. These authors found only a modest number of cells retrogradely labeled in the ventrolateral periaqueductal gray after apparently restricted injections of WGA-HRP into the LC. Moreover, PHAL anterograde labeling in that study revealed strong inputs to the Bar and other peri-LC regions, but more moderate input to the LC proper. These anatomical results were consistent with physiological studies in that paper, which indicated a modest connection (via antidromic activation) and synaptic influence from the periaqueductal gray on LC neurons but a much greater functional connection from the periaqueductal gray to nearby peri-LC regions.
No anterograde data is available for the projection from the dorsal part of the periaqueductal gray to the LC. Suggesting the specificity of this afferent, we found only a few retrogradely labeled cells in the dorsal periaqueductal gray after control injections in the nuclei surrounding the LC. However, neurons were not observed in the dorsal part of the periaqueductal gray in previous reports 6, 10, 13.
We and Cedarbaum and Aghajanian 10 found a large number of neurons in the mesencephalic reticular formation after large LC+periLC injections. We observed a substantial number of cells even after LC injections apparently not involving peri-LC areas. The cells were localized lateral to the median raphe nucleus and in a region inside and just dorsal to the medial lemniscus, in the area of the B9 serotonin cell group 16. This area does not correspond to the ventral tegmental area previously reported to project to the LC using anterograde tracing methods 28 ,64. Indeed, Aston-Jones et al. 6, using WGA-HRP as a anterograde tracer, have demonstrated that the ventral tegmental area projects to the median parabrachial nucleus and a region located rostral to the Peri-5Me but not to the LC itself. On the contrary, after injections of PHAL in the B9 region, we observed a few anterogradely labeled fibers in the nuclear core of the LC. Together, these data suggest that the LC receives a small projection from the B9 region and no input from the ventral tegmental area. Further anterograde studies are necessary to confirm projections to the LC from the other parts of the mesencephalic reticular formation.
In agreement with Cedarbaum and Aghajanian 10 and Clavier 13, we observed a substantial number of rather small retrogradely labeled cells in the nucleus raphe dorsalis particularly its rostral and dorsolateral extensions after LC+periLC injections. We and Cedarbaum and Aghajanian 10 observed also fewer cells in the median raphe nucleus after such injections. After LC or Bar injections, we observed fewer cells in these raphe nuclei. Aston-Jones et al. 6 saw a few weakly labeled cells in some animals in the dorsal and median raphe nuclei with LC injections of WGA-HRP. After CTb injections in the Peri-5Me, 5Me and MVe, we observed mainly medium-sized retrogradely labeled neurons in the medio-ventral part of the nucleus raphe dorsalis and occasional cells in the median raphe nucleus. In agreement with these results, after tritiated proline injections in the dorsal and median raphe nuclei, weak projections to the area of the LC have been reported 14. Furthermore, after CTb injections in the medioventral part of the nucleus raphe dorsalis, we found a substantial number of fibers in the nuclei surrounding the LC and only a few in the LC proper. This is consistent with previous results obtained with PHAL 7.
Using radioautographic detection of serotonin, Leger et al. 37 found only a partial decrease in the number of serotonin terminals in the LC after lesions of the dorsalis, median or pontis raphe nuclei. In addition, recent studies by Pieribone et al. 52 found no discernable decrease in the density of 5HT-immunoreactive fibers in the LC after lesions of the nucleus raphe dorsalis. Furthermore, in our preliminary experiments combining CTb retrograde staining with 5HT immunohistochemistry after even large LC injections of CTb, we found only a few double-stained neurons in the nucleus raphe dorsalis. In contrast, many 5HT- and CTb- immunoreactive cells were observed in the median raphe nucleus and the B9 serotoninergic cell group 9. Together, these results suggest that the serotonin innervation of the LC arises primarily from the median raphe nucleus and the B9 serotonin cell group and only to a minor extent from the nucleus raphe dorsalis. Additional double-labeling and anterograde tracing experiments are necessary to confirm these preliminary results.
In the pons, after LC or LC+periLC injections, a large number of retrogradely labeled neurons was present bilaterally in the Kölliker-Fuse nucleus. Few or only occasional retrogradely labeled cells were localized in this nucleus after control injections in the Bar, the MVe, peri-5Me or 5Me. Clavier 13 also described a contralateral projection from the Kölliker-Fuse nucleus to the LC. In contrast, Cedarbaum and Aghajanian 10 and Aston-Jones et al. 6 reported a few or no cells in this nucleus. However, we confirmed the strong bilateral input from the Kölliker-Fuse nucleus to the LC using PHAL and CTb as anterograde tracers and this projection has also been observed recently with retrograde transport of WGA-apoHRP-Au from small pressure injections into the LC (Aston-Jones and Zhu, unpublished observations). Such discrepancies between ours and previous results might be due to the weaker sensitivity of the tracers previously used compared with that of CTb and WGA-apoHRP-Au (as discussed above; see Technical Considerations).
We observed a large number of cells in the lateral parabrachial nucleus after LC+periLC injections. A small number of cells was also present in this nucleus after LC, Bar or 5Me injections. Cedarbaum and Aghajanian 10 and Clavier 13 reported cells in this region only contralaterally. Aston-Jones et al. 6 often saw cells located within the halo of their LC injection sites in this region. Further anterograde experiments are necessary to determine whether the retrograde labeling in the present study reflects afferents to areas neighboring the LC, or to the LC itself.
After LC or LC+periLC injections, we consistently observed a small number of cells lying dorsal and lateral to the superior olive, medial to the exciting fibers of the facial nerve at the location of the A5 noradrenergic cell group 16. We observed only a few or no cells in this area after control injections in surrounding nuclei suggesting the specificity of this input to the LC. Cedarbaum and Aghajanian 10 and Clavier 13 also observed labeled cells in this area, but Aston-Jones et al. 6 did not consistently obtain such labeling with focal WGA-HRP injections in the LC. Anterograde and immunohistochemical experiments are necessary to test this possible projection and to determine if it consists of noradrenergic A5 neurons.
After LC, Bar, peri-5Me, 5Me or MVe injections, we observed only a few retrogradely labeled cells in the pedonculopontine and laterodorsal tegmental nuclei where most of the pontine cholinergic cells are located. We also saw only few anterogradely labeled fibers in the LC after CTb injections in the laterodorsal tegmental nucleus. In agreement with these results, only a few scattered cholinergic axons were observed in the LC after choline-acetyltransferase immunohistochemistry 33, 55.
After LC+periLC injections, we, Cedarbaum and Aghajanian 10 and Clavier 13 observed a few labeled cells in the contralateral LC. However, Aston-Jones et al. 6 observed no contralateral labeling after WGA-HRP injections into the LC. Consistent with this result, we saw no cells after CTb injections apparently restricted to the LC or with control injections in surrounding nuclei. After LC+periLC injections, we also observed a few anterogradely labeled fibers in the contralateral LC but not in adjacent nuclei. Taken together, these results suggest only a weak, if any, contralateral projection of the LC.
After LC injections, Cedarbaum and Aghajanian 10, Clavier 13 and Fung et al. 23 reported a large number of labeled cells in the vestibular nuclei. Aston-Jones et al. 6 noted that the vestibular nuclei often contained labeled cells but only within the halo of their injection sites. We saw labeled cells in the vestibular nuclei only after LC+periLC injections involving the adjacent MVe. After a control injection restricted to the MVe area located just lateral to the LC, we observed a large number of cells and anterogradely labeled fibers bilaterally in the vestibular nuclei. These results indicate that the MVe region close to the LC, but not the LC itself, may be reciprocally innervated by the other vestibular nuclei.
Cedarbaum and Aghajanian 10 and Clavier 13 observed a large number of cells in the ventrolateral part of the medullary reticular formation after LC injections, which they ascribed to the lateral reticular nucleus. Aston-Jones et al. 6 described a large number of cells specifically in the lateral paragigantocellular reticular nucleus (PGi) localized in the rostral ventrolateral medulla, but not extending more caudally into the lateral reticular nucleus. After either LC or LC+periLC injections, we also saw a large number of cells specifically in the PGi. We found fewer retrogradely labeled cells in the PGi after Bar, 5Me, Peri-5Me or MVe injections. Confirming these data, we and Aston-Jones et al. 6 found a large number of anterogradely labeled fibers in the LC after PHAL, CTb or WGA-HRP injections in the PGi.
Cedarbaum and Aghajanian 10 and Clavier 13 reported a few and no cells, respectively, in the dorsomedial rostral medulla (ventromedial nucleus prepositus hypoglossi and dorsal paragigantocellular nucleus) after LC injections. Aston-Jones et al. 6 found a large number of retrogradely labeled neurons bilaterally in the dorsomedial rostral medulla after WGA-HRP injections in the LC. We also found bilaterally a large number of cells in this region after LC or LC+periLC injections of CTb. Note that the counts given in the Results here were for the ipsilateral dorsomedial rostral medulla only; this projection is strongly bilateral, so that many contralateral cells also innervate the LC. Also, the main collection of LC-projecting neurons in the dorsomedial rostral medulla is a rostrocaudally oriented narrow column of neurons, so that counts in one frontal section may not fully represent the labeled neurons. It is noteworthy that in Aston-Jones et al. 6 study and the present investigation, the neurons labeled retrogradely from the LC in the nucleus prepositus hypoglossi were specifically located in its ventromedial aspect; the bulk of the nucleus did not contain labeled cells.
We also observed a large number of cells in the dorsomedial rostral medulla with a similar distribution but with a strong contralateral predominance after a control injection in the adjacent MVe area just lateral to the LC. Like for LC injections, the neurons in the nucleus prepositus hypoglossi were specifically located in its ventromedial aspect; the bulk of the nucleus did not contain labeled cells.
Only a few cells were localized in the dorsomedial medulla after CTb injections in the Bar, 5Me and Peri-5Me.
After WGA-HRP injections in the dorsomedial rostral medulla, Aston-Jones et al. 6 further reported the presence of a large number of anterogradely labeled fibers in the LC. Also, Ennis and Aston-Jones 20 obtained frequent antidromic activation of neurons in the dorsomdial rostral medulla after focal LC stimulation, and potent synaptically mediated inhibition of LC activity after stimulation of the dorsomedial rostral medulla. Taken together, these results suggest that the dorsomedial rostral medulla strongly projects to the LC and the adjacent MVe area.
Cedarbaum and Aghajanian 10 and Clavier 13 found a large number of cells in all subdivisions of the nucleus of the solitary tract. In contrast, after small LC injections, we and Aston-Jones et al. 6, respectively found a few and no cells in this nucleus. We observed numerous cells in the rostral or caudal parts of the nucleus of the solitary tract only after 5Me or Peri-5Me injections, respectively. After WGA-HRP or PHAL injections in the nucleus of the solitary tract, Aston-Jones et al. 6 and Herbert et al. 27 reported sparse scattered fibers in the LC and Bar and a large number of fibers in the peri-5Me and Me5. These and additional electrophysiological data 19 suggest that the nucleus of the solitary tract strongly projects to the peri-5Me and Me5 and provides only a weak, if any, input to the LC itself.
In agreement with Cedarbaum and Aghajanian 10, we observed a small number of cells in the caudoventrolateral medullary reticular formation in the region of the A1 noradrenergic cell group 16 after LC or LC+periLC injections. In support of the specificity of this afferent, we found only a few cells in this area after our control injections in nuclei surrounding the LC. Nevertheless, additional anterograde experiments are necessary to confirm these results.
Inactivation of LC noradrenergic neurons during paradoxical sleep
Although it is well accepted that the noradrenergic neurons of the LC cease firing during paradoxal sleep (PS-off cells) 4, 29, the mechanisms responsible for this inactivation have not been elucidated. Nevertheless, as first proposed by Aston-Jones and Bloom 4, evidence suggests that these cells might be tonically inhibited by extrinsic afferents rather than defacilitated during PS. These investigators reported spontaneous and sensory-evoked field potentials without unit activity in the LC during paradoxal sleep 4. These potentials were proposed to reflect concerted excitatory postsynaptic responses in the presence of strong tonic inhibition preventing discharge. Consistent with this view, LC cells display a tonic pacemaker activity in brain slices 71, in the absence of possible facilitatory inputs. The neurons responsible for such tonic inhibition of the LC might be located in the brainstem. Indeed, pontile cats with sections at the level of the superior colliculus display periods of PS alternating with wakefulness 34 suggesting that the neurons responsible for the inhibition of the monoaminergic neurons including the locus coeruleus neurons are localized in the brainstem. In its reciprocal interaction model, Sakai 56 has proposed that during PS, the PS-off noradrenergic cells of the LC might be under a monosynaptic inhibition provided by the PS executive neurons (PS-on cells) located in the medial or mediodorsal pontine reticular formation. Our results do not support the existence of a monosynaptic projection from such neurons. Indeed, we found no or only a weak projection from the medial or mediodorsal pontine reticular formation to the LC. The only strong pontine input to LC identified in the present study was from the Kölliker-Fuse nucleus localized in the lateral part of the pontine reticular formation. Of interest regarding this result, Data et al. 17 recently reported that injection of carbachol in the peribrachial area just caudal to the Kölliker-Fuse nucleus induces a long-term enhancement of PS. Additional experiments are necessary to determine whether the Kölliker-Fuse nucleus might participate in this effect.
It has also been proposed that the GABAergic neurons recently observed in the region of the pedunculopontine nucleus and the nuclei surrounding the LC such as the Bar the peri-5Me and the laterodorsal tegmental nucleus 33 might be, in the reciprocal interaction model, the missing inhibitory link between the presumed cholinergic PS-on cells in these nuclei and the noradrenergic PS-off cells of the LC. Although we cannot rule out possible dendro-dendritic interactions not revealed using CTb as a retrograde tracer (see above "neurons and dendrites of the peri-coerulear region"), our results do not support the existence of short inhibitory projections to the LC from interneurons located in the vicinity of the LC nor from the pedunculopontine nucleus. Indeed, we observed only a few CTb labeled cells in the pedunculopontine nucleus and the nuclei surrounding the LC after CTb injections in the LC.
Together, these data suggest that the neurons inhibiting the LC during PS might be rather located in one of the three strong brainstem afferents we report here namely the Kölliker-Fuse nucleus, the PGi or the dorsomedial rostral medulla. Among these, the projection from the dorsomedial rostral medulla to the LC seems to be a good candidate for such inhibition. Indeed, Ennis and Aston-Jones 19, 20 have shown that this pathway is inhibitory and uses GABA as a neurotransmitter.
We have recently demonstrated that in addition to GABA, the LC is strongly inhibited by glycine 40. It has also been shown that during PS the motoneurones are tonically hyperpolarized by glycine 11. In view of these results, we made the hypothesis that during PS, the LC noradrenergic cells are tonically inhibited by the glycinergic neurons responsible for the inhibition of the somatic motoneurons. Locus coeruleus cells might also be hyperpolarized by other types of neurotransmitters found to inhibit LC cells such as noradrenaline or enkephalin 7. In this regard it is also noteworthy that Aston-Jones et al. 3 have recently found potent alpha2 adrenoceptor-mediated inhibtion of LC neurons from the PGi, presumably reflecting the strong input from C1 neurons located within the PGi 2, 35, 50 ,51. Moreover, the PGi and the dorsomedial medulla provide a dense enkephalinergic input to the LC 18, and enkephalin strongly inhibits LC neurons 46. Thus, its is also possible that inhibition of LC during PS reflects PGi input to these neurons.
Also of great interest regarding the sleep-waking cycle, we found inputs from the preoptic area dorsal to the supraoptic nucleus and the posterior hypothalamic areas. Indeed, a large body of evidence indicate that these two regions play crucial roles respectively in the onset of sleep and in wakefulness. In this regard, it would be of great interest to determine if these projections participate in the regulation of the activity of the noradrenergic cells of the LC accross the sleep-waking cycle. In conclusion, our results indicate that a large number of structures project to the LC and might therefore play a role in the modulation of the activity of the noradrenergic neurons accross the sleep-waking cycle. Further experiments are now necessary to determine the specific role of each of these afferents.