SINCE their introduction in 1978 Lucifer yellow CH and Lucifer yellow VS have been used with considerable success as intracel-lular markers in a wide variety of biological systems. Both dyes contain the same sulphonated 4-aminonaphthalimide moiety, but they differ in the substituent on the imide nitrogen (Fig. 1). Their syntheses, to be reported elsewhere, are based on the commercial wool dye brilliant sulphoflavine (CI56205). The dyes have similar spectral properties: absorption maxima at 280 nm and 430 nm, corrected emission maxima near 540 nm, and quantum yields of about 0.25. The high quantum yields make possible the detection of the dyes at low concentration, and the wide separation between the absorption and emission maxima facilitates excitation at one wavelength and observation of fluorescence at another. The dyes differ in their chemical properties, but both can be bound to tissue. Lucifer yellow VS is a vinyl sulphone that reacts rapidly with amino and sulphydryl groups but is extremely stable in water; even in weak base (pH 9.0) its half life is about one week at room temperature. The dye combines rapidly and covalently with proteins and presumably with other tissue constituents containing sulphydryl or amino groups. Lucifer yellow CH, on the other hand, has a free hydrazido group and reacts with aliphatic aldehydes at room temperature. Aldehyde-containing fixatives bind the CH dye to tissue, presumably through the hydrazido group. Aqueous solutions of this dye appear to be chemically stable for at least several months at room temperature.So far the CH dye has been used more often than the VS dye, perhaps because the latter is not yet commercially available; which one is best suited to a particular application can only be determined empirically. Fig. 1 Structural formula of Lucifer dyes. Both dyes are 3,6-disulphonated 4-aminonaphthalimides. For the lithium salt of Lucifer yellow CH (often referred to simply as Lucifer yellow CH), R is -NH-CO-NH-NH2; for the lithium salt of Lucifer yellow VS, R is m-Phenyl-SO2-CH=CH2. About x24,000,000.The most frequent application of Lucifer yellow CH has been to reveal the shape of individual neurones1-5. It has been used to determine the branching pattern and course of regenerating neurones6-7. Because of the low toxicity of the dye, direct in vivo observations on the regeneration of dye-filled neurones may be possible. Lucifer yellow CH has been used in the study of developing systems: in the development of molluscs8, for example, and in the exciting wbrk by Goodman and Spitzer on the development of the grasshopper nervous system9.I describe here other applications of these fluorescent tracers in the hope that their many potential uses can be rapidly exploited. Dye-coupling revealed by intracellular injection of Lucifer yellowNeurones and other cells can be marked with Lucifer yellow CH by injecting the dye through a micropipette using pressure or iontophoresis. The dye spreads quickly throughout the body and processes of the injected cell, but does not cross the cell membrane1. The low molecular weight (457.2) and intense fluorescence of the dye offer several advantages over other tracers. (1) A cell can be stained in its entirety with a relatively small amount of dye. Illumination sufficient for photomicrography of live cells does not noticeably alter their physiology. (2) In preparations which are sufficiently transparent one can determine whether the distribution of dye found after fixation is the same as that seen in the living state. (3) Most importantly, the dye frequently spreads in vivd from the injected cell to certain nearby cells. This movement of dye from cell to cell has been termed dye-coupling1. In certain cases, dye-coupled cells are known to be electrically coupled to each other. Two examples are horizontal cells in the turtle retina1 and the septate axon of the crayfish (K. Futamachi and W.S., unpublished). In the embryo of the mollusc Patella, however, dye-coupling between macromeres was detected without knowing whether the cells were electrically coupled and was used to study the development of the embryo8. Presumably in many (but not necessarily all) instances dye-coupling and electrical coupling have a common basis. In any case, dye-coupling offers a method of recognizing certain functional connections between cells by morphological means. A pair of neurones called Leydig cells occurs in each seg-mental ganglion of the leech Hirudo medicinalis. These cells are known to be electrically coupled30, and Fig. 2 shows that they are dye-coupled as well. Two neurones in the ganglion contain dye: the Leydig cell that was injected and the contralateral Leydig cell into which dye spread. The transfer of dye presumably occurred at the sites of contact between the arborizations of the two neurones. Unexpectedly, dye also spread from the injected neurone to processes in the ipsilateral axon bundles leaving the ganglion (Fig. fyb). These are probably axons from Leydig cells in adjacent ganglia, as Kent Keyser has independently demonstrated in greater detail (personal communication). As both adjacent ganglia had been removed before injection, the sites of coupling to the two dye-coupled processes must have been in the remaining ganglion or in the short lengths of attached nerve. Here and in similar situations Lucifer yellow CH is useful for gross localization of the sites of intercellular communication.Fig. 2 Leech ganglion containing a Leydig cell injected intracel-lularly with the lithium salt of Lucifer yellow CH. a, The lower cell was pressure-injected with a 5% solution of dye. The tissue was allowed to stand for 90 min at room temperature and was then fixed at 4C for 12 h in 4% formaldehyde in 0.15M sodium phosphate buffer, pH 7.4. The ganglion was embedded in glycol methacrylate as previously described1. The fainter cell body (top) is the contralateral Leydig cell, which is dye-coupled to the injected cell. X107. b, Detail from a. The ipsilateral nerve contains two fluorescent processes-one from the injected neurone, the other from a dye-coupled neurone in the adjacent ganglion (see text). xl80. The sensitivity of Lucifer yellow CH was compared semiquantitatively with that of two other tracers, Procion yellow and horseradish peroxidase. Horseradish peroxidase is often used for intracellular staining18,19 and is detectable by both light and electron microscopy. When 5% solutions of Lucifer yellow CH and horseradish peroxidase were injected by pressure into leech neurones, the two tracers appeared roughly equal in their ability to reveal morphological detail as judged visually with the light microscope (W.S. and K. Muller, unpublished). Horseradish peroxidase, however, cannot be seen in living tissue and is thought not to pass between dye-coupled cells20. Procion yellow, a fluorescent tracer previously used for intracellular injection21, has about 1% of the fluorescence intensity of Lucifer yellow and in comparably marked cells reveals much less morphological detail1. Backfilling with Lucifer yellow CHNeurones can also be filled with dye through their processes rather than by injection of dye into their somata. Backfilling, introduced by lies and Mulloney22, is carried out by simply immersing the cut end of a nerve in a pool of tracer. The tracer moves up the axons of the nerve, either by diffusion or in response to an externally applied electric field, and fills the cell bodies of the neurones with processes in that nerve. This technique has been used with Procion yellow22, with cobalt chloride23,24, and with horseradish peroxidase25. Lucifer yellow CH is at least as suitable for backfilling as these tracers. Moreover, since neither the dye nor the process of backfilling appears to alter physiological properties noticeably, and since the dye can usually be seen in vivo, one can first identify a backfilled cell by its fluorescence, then impale it with a micro-pipette and investigate its electrophysiology. Figure 3a is a fluorescence photomicrograph of the buccal ganglia of the freshwater planorbid snail Biomphalaria glabrata. The preparation was backfilled with the lithium salt of Lucifer yellow CH through the nerve innervating the left salivary gland. The two largest cells containing dye can usually be identified by other criteria such as size, position, and branching pattern. These cells appear to be salivary effector neurones and may be homologous to the cells described by Kater in Helisoma26 and Benjamin in Lymnaea27. The fluorescent cells were impaled after they had been backfilled, and their physiology appeared normal. In particular, action potentials in the left cell produced one-for-one depolarizing potentials in the salivary cells (Fig. 3b, c), and depolarizing and hyperpolarizing current passed from either cell to the other, suggesting that they were electrically coupled (Fig. 3d, e). These results are similar to those obtained from preparations which had not been backfilled. After fluorescence photomicrography of the ganglia (Fig. 3a), the physiology of both cells was still normal, and they remained electrically coupled (Fig. 3f, g). In this system, therefore, the amount of light needed for fluorescence photomicrography is much less than that which causes photosensitized damage or killing of cells that contain dye. In the method of cell killing devised by Miller and Selverston, the illumination level which killed Lucifer-filled cells in 5 min was 1,000 times the level required for visualization of processes in the neuropil28.Lucifer dyes as tracers in electron microscopy Since the Lucifer dyes described here do not contain a heavy atom, they have no useful electron density. Their intense fluorescence, however, allows them to be used as tracers for electron microscopy by an indirect procedure. An ultrathin section mounted on a grid i^ examined with a fluorescence microscope to identify cells or processes that contain dye, and then the same section is examined with the electron microscope. Figure 4 shows light and electrpn micrographs of the same areas of a neurone in the right parietal ganglion of the snail B. glabrata; this neurone can often be identified by its size and position. To make this type of comparison, tissue is conventionally processed for electron microscopy except that osmication is omitted, since even brief exposure to osmium tetroxide greatly reduces Lucifer fluorescence.Fig. 3 Live neurones stained by backfilling with the lithium salt of Lucifer yellow CH, and electrical responses recorded after staining. The cut end of the left salivary nerve of a snail (B. glabrata) was drawn into a small drop of a 3% solution of dye on a square of polyethylene about 1 mm on each side. For 1 h a potential was imposed so that a positive, constant current of 30 nA flowed from the bath to the drop of dye. The buccal ganglia were rinsed in a Ringer's solution, and the stained cells were identified with fluorescence optics. The large dye-filled cell in the left buccal ganglion was impaled with a microelectrode (b and c, lower trace). A salivary cell was impaled with a second electrode (b and c, upper trace). The microelectrode was then withdrawn from the salivary cell and used to impale the large cell on the lateral margin of the right buccal ganglion (d and e, records for the right cell are on top). The current traces show when current (about 0.25 nA) was injected, using an active bridge circuit (not accurately balanced in all records), first into one cell and then the other. The live prepartion, a, was then photographed using epifluorescence with a Zeiss Ultraphot. The conditions of photography were: 6.3X Neofluar lens (0.20 numerical aperture), HBO 200 W mercury arc lamp, heat filter and BG-12 excitation filter, FL-500 dichroic splitter, 500 nm barrier filter, 30 s exposure. x66. After photography, both lateral cells were impaled again, producing the records shown in f and g. Vertical calibration indicates 50 mV for all records; horizontal calibration indicates 1 s for c and 2 s for all other records. Fig. 4 A cell in the right parietal ganglion of the freshwater snail Biomphalaria glabrata was injedted with Lucifer yellow CH, fixed for 4 h at 23 C in a pH 7.4 12.5 mM Na+/H+ HEPES buffer containing 2% glutaraldehyde, 1% formaldehyde, and 25 mosM balanced salts, dehydrated in graded alcohols, and transferred to anhydrous acetone at 4 C. The ganglion was stained for 1 h at 4 C in 1% hafnium tetrachloride in acetone, then rinsed in acetone and embedded in Spurr's resin42. (The fixative and rinse solutions also contained 1% HfCl4 in this experiment, but not in most.) a, Silver section on a hexagonal grid (Pdlaron) was illuminated through a Zeiss 63X OD lens (0.90 numerical aperture) with an Osram 50 W HBO lamp and photographed with a 15 s exposure on Ektachrome 400 film pushed to ASA 800. x640. b, The same section shown in a, photographed with the electron microscope after uranium and lead staining. x640. c, Detail from a. x1,500. d and e, Detail from b. x1,500 and x3,700. Photographed at 80 kV.Fig. 5 Isthmo-optic nucleus of chick labelled by retrograde axonal transport of Lucifer yellow VS. The vitreous humour of the contralateral eye was injected with 25 uJ of 5% Lucifer yellow VS; the injection damaged a portion of the retina. After 21 h the chick was anaesthetized and fixed by perfusion with buffered 4% formaldehyde. The other eye was not injected, and its contralateral isthmo-optic nucleus showed almost no fluorescence. x200. Post-fixation staining can be omitted entirely, but tissue processed in this way generally lacks sufficient contrast and electron density to give useful images. Adequate contrast is generated by staining tissue en bloc in 1% hafnium tetrachloride in anhydrous acetone at 4 C, a procedure introduced by Morris Karnovsky (personal communication). The tissue is then rinsed in acetone and embedded in Spurr's resin. When a thin section (grey or silver) on an ordinary electron microscope grid is illuminated using epifluorescence with an intense arc lamp, the fluorescence of the marked cell is bright enough to be photographed on a fast colour film with a 15-s exposure. The same thin section is then conventionally stained with uranium and lead; the areas that had been found fluorescent are located and photographed with the electron microscope. Surprisingly small areas can be identified with confidence when the light and electron micrographs are compared. In Fig. 4c, d, for example, there are dye-free areas only 500 nm in diameter that are easily recognized in both the light and electron micrographs. The areas from which dye is excluded lie within the perimeter of the cytoplasm of the injected cell and may be penetrating processes of other cells29. Dye-filled processes 500 nm in diameter can usually be found with ease, while 200 nm processes are locatablewith greater difficulty. The necessity of avoiding osmication to preserve fluorescence entails some disadvantages. Membranes either cannot be seen or, more often, appear as thin grey lines (Fig. 4e) instead of the dark black bilayer commonly seen with conventional fixation. The poor visualization of membranes makes the method described here unsuitable for some applications. For example, synapses in the leech central nervous system are difficult to recognize because membranes and associated synaptic specializations lack the contrast they normally have with osmium post-fixation (K. Muller, personal communication).Since Lucifer yellow CH has no significant electron density of its own, it was possible to test whether either the dye or the process of injection had an effect on the ultrastructure of an injected neurone. The paired Retzius cells of the leech30 were convenient test objects. In each of three segmental ganglia, one Retzius cell was injected with dye and one was not. The tissue was immersed in a cacodylate-buffered fixative20, then processed as described in the legend to Fig. 4. When thin sections were examined with the electron microscope, both cells were recognizable by their size jand position. (The injected cell had previously been identified [by its fluorescence.) The ultrastruc-tural morphology of the j two cells was virtually .the same, suggesting that neither thd dye nor the process of injection had caused significant morphological changes (K. Keyset and W.S., unpublished). This experihient illustrates an advantage of an ultrastructural tracer that cjoes not rely on electron density for its detection: the appearance Of the marked tissue is not necessarily changed. Retrograde axonal transportA useful technique for studies of vertebrate neuroanatomy relies on the phenomenon of retrograde axonal transport, which causes certain exogenous proteins to be carried from axon terminals to the cell body31. Kristensson, Olsson, and Sjostrand showed in 1971 that horseradish peroxidase appeared in the cell bodies of neurones near whose terminals a solution of horseradish peroxidase had been introduced32. This marker has since been used both to trace pathways in the central nervous system and to study the mechanism of transport itself. The La Vails and others have shown that peroxidase moves within axons at a rate of about 100mm per day (ref. 33). Some exogenous proteins seem not to be subject to retrograde axonal transport34, but the reasons for this are not understood. Kuypers and his colleagues have recently shown that certain other fluorescent dyes are also subject to retrograde movement35,36, so it is not surprising that Lucifer yellow VS serves as an effective marker for retrograde axonal transport from the chick eye to the cell bodies of neurones which have their axon terminals within the retina33 (Fig. 5). Perhaps the dye is transported as a protein conjugate formed in situ. It is not known whether Lucifer yellow VS will prove useful for tracing connections in other parts of the nervous system. Fig. 6 Indirect immunofluorescence staining of actin filaments with Lucifer-labelled antibodies. Cultured human lung cells, W138, were fixed and treated with rabbit anti-actin antibody, then with Lucifer-labelled goat IgG anti-(rabbit IgG); the procedures were those described by Lazarides43. The Lucifer-labelled antibody was prepared as follows. To 50 mg of Miles-Yeda goat IgG anti-(rabbit IgG) dissolved in 1.8 ml 0.5 M pH 9.0 carbonate buffer was added 0.52 mg Lucifer yellow VS in 1.0 ml carbonate buffer. These weights correspond to a nominal 3 mol of Lucifer per mol of protein, but since Lucifer yellow VS is hygroscopic, less dye was actually present. The mixture was allowed to stand for 3 h at room temperature, then unreacted dye was removed by chromatog-raphy with phosphate-buffered saline on Sephadex G-25. The optical densities at 280 nm and 430 nm indicated a dye to protein molar labelling ratio of approximately 2.3. About x500.Recently wheat germ agglutinin labelled covalently with Lucifer yellow VS (see below) has been found to provide good staining of motoneurones in the spinal cord of chick embryo after a small amount (0.5 ul of 0.5% agglutinin) was injected into leg muscles (M. McPheeters and L. M. Okun, personal communication). After retrograde transport had occurred, it was possible to dissociate the spinal cord and isolate the labelled neurones by means of a fluorescence-activated cell sorter37. Covalent labelling of macromolecules with Lucifer yellow VSLucifer yellow VS labels proteins rapidly and covalently under mild conditions. In 0.1 M NaHCO3 the reaction is complete in 2 h at room temperature. The conjugates generally have quantum yields between 0.1 and 0.2 (R. Chen and W.S., in preparation), so Lucifer-conjugated antibodies should be suitable for immunofluorescence staining. Figure 6 shows actin filaments of cultured fibroblasts stained with Lucifer-labelled antibodies. Lucifer yellow VS is similar to fluorescein isothiocyanate in its ease of use and fluorescence intensity and has the advantage that its emission peak at 540 nm is considerably to the red of fluorescein's at 515 nm, so that its fluorescence contrasts more strongly with tissue autofluorescence. But since Lucifer has no other clear-cut advantage and since less is known about the chemistry and the stability of the bond that is formed, it is likely that fluorescein isothiocyanate will remain the dye of choice for most immunofluorescence applications. When fluorescence is needed at pHs below 6, however, Lucifer yellow VS may be useful. The fluorescence intensity of Lucifer VS is unchanged from pH 2 to pH 10 (ref. 1), an unusual property for a water-soluble dye. This pH-independence makes the dye suitable for monitoring pH-induced conformational changes in macromolecules. Other uses of Lucifer yellow dyesAmong published examples of the possible uses of Lucifer yellow dyes it is worth noting that intracellular staining with Lucifer has been followed by immunocytochemical staining with rhodamine-labelled antibodies38; the goal of this approach is to characterize the immune reactivity as well as the electrophysiology of the same cell. This dye also has been used to follow changes in junctional permeability caused by moderate trans-junctional polarization39. Another interesting application is suggested by experiments in which an individual dye-filled neurone or even a part of a neurone was selectively damaged by means of an intense, focused spot of light28. This phenomenon is sometimes referred to as "cell killing," but as the originators of the technique have pointed out40, it is not known whether the affected cells are killed or simply inactivated temporarily. Also, it is not yet known what margin of safety there is for cells that do not contain dye. Finally, there is an interesting recent report that when turtle retina is bathed in a calcium-free Ringer's solution containing Lucifer, the bipolar cells, but not other cell types, take up dye from the medium41. Uptake is not influenced by light, but is blocked by metabolic inhibitors. It will be of interest to see whether similar phenoma occur in other neuronal systems. Dye-coupled systems of cells have been discovered through the use of Lucifer yellow CH10,11. It has been used to monitor dye-coupling in regenerating neurones12,14, and it greatly facilitated the discovery that an electrical connection between two leech neurones was mediated by a pair of small interneurones15 and was not direct, as had previously been thought. The dye has been used to obtain information on the position of the micro-electrode tip, either in neurones16 or, in one case, in giant mitochondria17. But because of its rapid spread, presumably by diffusion, this dye generally gives only limited information on tip position, and so it may not be the best marker for this purpose.In addition to the two sulphonated 4-aminonaphthalimides that have been used until now, more than 40 compounds in this group have recently been prepared by a convenient one-step synthesis44. These compounds have a wide range of substituents on the imide nitrogen, and most are water-soluble and have an intense yellow-green fluorescence. Because of the ease with which functional groups can be incorporated into sulphonated 4-aminonaphthalimides, these compounds will probably find new applications as biological tracers.The experiment shown in Fig. 2 was performed at the 1978 Leech Course at Cold Spring Harbor Laboratory. I thank Elias Lazarides for carrying out the immunofluorescence staining and Kent Keyser, Enid Applegate, and Jennifer LaVail for collaboration in performing the experiments described in the legends to Figs 3, 4 and 5 respectively.