Double Immunolabeling of Neuropeptides in the Human Hypothalamus as Analyzed by Confocal Laser Scanning Fluorescence Microscopy

Abstract

The main goal of this study was to develop a better light microscopic procedure for quantitative study of the cellular co-localization of neuropeptides in adult human brain tissue. To reach this goal, we opted for a method (proved to be optimal on rat brain) in which sections were double immunolabeled with two different fluorophore-conjugated secondary antibodies and analyzed with a confocal laser scanning fluorescence microscope. One of our main problems faced was a strong autofluorescence of the sections, mainly caused by lipofuscin granules normally present in adult human brain tissue, which made any analysis of specific fluorescence impossible. This problem could be solved by staining the sections after immunolabeling with the dye Sudan Black B, which completely blocked this autofluorescence. The complete optimized procedure that we eventually developed can be summarized as follows. After a relatively short fixation time (6-14 days) in 4% freshly depolymerized paraformaldehyde, the resected brain tissue can best be stored in a 30% sucrose solution supplemented with 0.05% NaN₃ at 4C. Stored under these conditions, cryosections from the tissue still reveal good histology and allow successful immunocytochemical staining after a period of 6 months. Double immunolabeling is done by incubating cryo- or paraffin sections in a mixture of two primary antibodies directed against the targeted antigens, followed by incubation with two different fluorophore-conjugated secondary antibodies. Amplification with a biotinylated secondary antibody followed by fluorophore-conjugated streptavidin is possible. Finally, the sections are stained with Sudan Black B, mounted in plain 80% Tris-buffered glycerol, and studied by confocal laser scanning fluorescence microscopy. Sections processed in this way are well suited for qualitative and quantitative analyses of co-localized neuropeptides in human brain tissue.

THE MAIN GOAL of this study was to develop a better light microscopic method for quantitative study of cellular co-localization of neuropeptides in adult human brain tissue than is presently available. One current method consists of mutually comparing serial thin cryoor plastic sections or the matching surfaces of vibratome sections after they have been alternately immunostained for two peptides (e.g., Okamura et al. 1986). Another method is double immunolabeling of sections by use of two different contrasting chromogens as secondary antibodies, e.g., diaminobenzidine (brown) with Fast Blue (e.g., Raadsheer et al. 1993). Although both methods are well suited for qualitative studies, various inherent suboptimal features render them unsuitable for reliable quantitative research. Because previous studies in our laboratory on the rat suprachiasmatic nucleus (SCN) had shown that double immunolabeling with fluorophore-conjugated secondary antibodies combined with confocal laser scanning microscopy (CLSM) is an ideal method in this respect (Romijn et al. 1998), we chose this method as a point of departure for the present experiments. The use of confocal fluorescence microscopy to analyze double immunolabeled sections has several advantages over conventional methods, i.e., little fading of fluorophores, high optical resolution, particularly in the Z-direction, computer-controlled acquisition of images and their permanent storage after they have been digitized, the possibility of retrieving two corresponding fluorescent images pertaining to one optical section and of portraying them in two contrasting artificial colors (e.g., green and red), and the availability of dedicated software tools to support quantitative analysis. Such an experimental approach promised to provide a reliable quantitative method for assessing the co-localization of two neuropeptides in neuronal cell bodies and nerve endings in sections of human brain tissue. In addition to optimizing the fixation and subsequent long-term storage of brain tissue, we had to overcome a terrible autofluorescence that was mainly emitted by an abundance of lipofuscin granules normally present everywhere in adult human brain tissue. This autofluorescence, which completely masks all specific immunofluorescence, is one of the main reasons why immunofluorescence microscopy on human brain tissue has largely been unsuccessful thus far. We succeeded in completely blocking this autofluorescence by staining the sections, after they had been double immunolabeled with fluorophore-conjugated secondary antibodies, with the lipid dye Sudan Black B.

The necessary experiments were done on the human hypothalamus, with emphasis on the paraventricular (PVN), supraoptic (SON), and suprachiasmatic nuclei (SCN), because these structures are the regions of interest of our research group. Arginine vasopressin (AVP), vasoactive intestinal peptide (VIP), gastrin-releasing pep-tide (GRP), and glutamic acid decarboxylase (GAD₆₅) were the peptides and protein of investigation.

Materials and Methods

Materials

Human hypothalami were obtained from the Netherlands Brain Bank (coordinator Dr. R. Ravid) and the Faculty of Medicine of the University of Craiova, Rumania (Dr. I. Radu). The hypothalami were resected after brain autopsy on adult male and female subjects. The postmortem delay was 3.4-15 hr.

Fixation

Resected human hypothalami were fixed in three different fixatives: (a) 4% formaldehyde in 0.1 M phosphate buffer (pH 7.5) made from freshly depolymerized paraformaldehyde; (b) commercially obtained, ready-made 4% formaldehyde solution neutralized with phosphate buffer at a pH of 6.8-7.2 (J.T. Baker; Deventer, The Netherlands); or (c) 3.6% formaldehyde solution [made by dilution with tapwater from a commercially obtained 36% formaldehyde stock solution (pH 3.2) stabilized with 12% methanol] (Lommerse-Pharma; Oss, The Netherlands). Different fixation times were studied, ranging from 3 days to several months. During the first week, fixation took place on a rocking device and the fixative was refreshed after the first or second day.

Long-term Storage and Sectioning

To perform a reliable, comparative immunocytochemical study on a number of hypothalami of (control) patients, fixed hypothalami (or their sections) have to be collected and stored for some weeks or months up to the time when the necessary number of samples has been obtained to be studied simultaneously. We therefore tried to determine a long-term storage condition that would affect histology and immunoreactivity as little as possible. To that end, hypothalami fixed in 4% freshly depolymerized paraformaldehyde in phosphate buffer for 6-14 days (the best option for fixation; see Results) were thoroughly rinsed in PBS, pH 7.5, on a rocking device for a couple of hours and then cut into right and left halves. These halves were used to test seven different storage conditions. First, some hypothalami were immersed overnight in 15% sucrose in PBS, followed by immersion in 30% sucrose in PBS, or in a mixture of 20% glycerol with 2% dimethylsulfoxide (DMSO) in Tris-buffered saline (TBS). Immersion was always performed on a rocking device at 4C (with the lid ajar to allow formaldehyde vapor to evaporate) and the solutions were refreshed each day up to the time when the hypothalami had sunk to the bottom of the jar (about 3 days). Thereafter, one part of the sucrose- and one part of the glycerol/DMSO-immersed hypothalami were stored at −80C, whereas the remaining hypothalami were split up again into two groups. From each group, one sucrose- and one glycerol/DMSO-immersed hypothalamus was directly used for cutting 25-μm-thick cryosections, which were then stored in their corresponding cryoprotectant at −80C. The remaining hypothalami were kept stored in their original immersion fluid, i.e., either in the 30% sucrose or in the glycerol/DMSO mixture, at 4C. The hypothalami immersed in the 30% sucrose solution at 4C still received sodium azide (NaN₃) to a final concentration of 0.05% to obtain extra protection against the growth of aerobic bacteria. This procedure thus resulted in six storage conditions. After a storage period of some weeks or months, 25-μm-thick cryosections of these hypothalami were cut and, together with the cryosections already cut at the beginning of the experiment, were stained with thionin (to judge the histology) or processed for double-label immunofluorescence. As a seventh manner of long-term storage, some hypothalami were routinely embedded in paraffin. From these hypothalami, 6-8-μm-thick sections were cut and also stained with thionin or antibodies.

Double Immunolabeling

In most cases, the immunoreactivity of cryo- and paraffin sections containing the PVN, the SON, and/or the SCN was tested by double immunolabeling for the presence of AVP and VIP. In a few cases, we also analyzed these nuclei for the combined presence of AVP with GRP or AVP with one of the GABA-synthesizing isoenzymes, i.e., the glutamic acid decarboxylase with an MW of 65,000 (GAD₆₅). The following primary antibodies and dilutions were used: a monoclonal mouse antibody against AVP-associated neurophysin (NP-AVP, PS41) diluted 1:200, which was a generous gift of Drs. H. Gainer and S.H. Key (NIH; Bethesda, MD), who also demonstrated its specificity for NP-AVP (Ben-Barak et al. 1985; see also Whitnall et al. 1985); a polyclonal antibody against VIP and one against AVP (both 1:2000), which were raised in rabbits and subsequently tested for their specificity at our Institute (see for refs. Romijn et al. 1998); a monoclonal mouse antibody against GRP obtained from Charles River (Southbridge, MA, USA) (1:2000); and a monoclonal mouse antibody against GAD₆₅ (GAD-6) from the Developmental Studies Hybridoma Bank (Department of Biology, University of Iowa, Ames, IA) (1:4). Secondary and tertiary antibodies, purchased from Jackson ImmunoResearch Laboratories (West Grove, PA), were a donkey anti-mouse and a donkey anti-rabbit antibody multilabeled with FITC (1:100) to demonstrate mouse anti-NP-AVP and rabbit anti-AVP, respectively, and a donkey anti-mouse antibody multilabeled with Cy3 (1:400) to demonstrate mouse anti-GRP and mouse anti-GAD₆₅. A biotinylated donkey anti-rabbit (1:400) antibody followed by a Cy3-labeled streptavidin antibody (1:1000) appeared to be necessary to sufficiently visualize rabbit anti-VIP. The vehicle for all incubations was so-called supermix (0.25 g gelatin and 0.5 ml Triton X-100 in 100 ml TBS, pH 7.5). After the sections had been thoroughly rinsed for 1-2 hr in TBS to remove the cryoprotectant, incubation took place first in plain supermix for 10 min followed by a mixture of two primary antibodies for 1 hr at room temperature (RT) and an overnight period at 4C. The next day, after rinsing, the corresponding mixture of fluorophore-conjugated antibodies was applied at RT for 1 hr, preceded by incubation with biotinylated donkey anti-rabbit when VIP was one of the targeted antigens. All intermediate rinsing steps, which must be done thoroughly and repeatedly (six to eight times for 30 min), were done in TBS. After immunolabeling, the sections were rinsed in TBS, put in a Petri dish with TBS supplemented with a drop of supermix to minimize the surface tension, mounted on gelatin-coated microscopic slides, allowed to air-dry and to fix in a slightly tilted position for a few minutes, and routinely coverslipped with a drop of plain 80% Tris-buffered glycerol at a pH of 7.0-7.6. The sections were stored at 4C and the coverslips sealed with nailpolish the next day. However, on analysis by confocal laser scanning fluorescence microscopy, the sections showed substantial autofluorescence which definitely required some further adaptation of the protocol (see below).

Confocal Laser Scanning Fluorescence Microscopy

A Zeiss 410 confocal laser scanning microscope (CLSM) was used with two lasers, i.e., an argon ion laser emitting at 488 nm and an HeNe laser emitting at 543 nm to excite FITC and Cy3, respectively. With the aid of two photomultipliers with appropriate filter settings (see below), the double immunolabeled sections were studied in two ways: (a) routinely by simultaneous excitation with both lasers on, which resulted in a combined two-color FITC-Cy3 image of both stained antigens (e.g., two neuropeptides) but with a little “bleeding through” of the right tail-piece of the emission spectrum of the FITC signal in the Cy3 image, and (b) in case of doubt and on behalf of photographic documentation, by sequential excitation with each laser separately on, which, after superposition of both individual images, gave more or less the same combined image as that obtained by simultaneous excitation, but without any “cross-talk” now. The FITC signal was displayed on the monitor in the pseudo-color green and the Cy3 signal in red. The 543-nm laser was always used with a smaller detection pinhole than the 488-nm laser to obtain the same thickness of the optical sections. During simultaneous scanning with both lasers on, we used a dichroic beam splitter (DCB) around the 560-nm, together with the following emission filters in front of both photomultiplier tubes: a bandpass (BP) of 510-525 nm for the FITC signal and a BP of 575-640 nm for the Cy3 signal. During separate scanning, we used a DCB of approximately 630 nm and as emission filters either a BP 515-540 nm when the argon ion laser (FITC signal) was used or a BP 575-640 nm when the HeNe laser (Cy3 signal) was used. A x40 objective (Zeiss plan-neofluar NA 1.3 oil) was used for routine studies. Because the scanned area in the sections was 320 × 320 μm and both sample frequency and pixel format were 1024 × 1024, a lateral resolution of about 0.4 μm was obtained (taking the Nyquist criterion into account). The corresponding optical section thickness was about 1.7 μm.

Blocking of Autofluorescence with Sudan Black B

At excitation with the 488- and 543-nm laser, the double immunolabeled sections showed considerable autofluorescence, which made analysis of specific immunofluorescence impossible. This autofluorescence was mainly caused by fatty lipofuscin granules abundantly present in nerve cell bodies and glial cells. Apart from these granules, the neuropil itself also exhibited some autofluorescence. The problem of autofluorescence was expected to be solved by staining the sections with the lipid dye Sudan Black B (SBB). SBB is known to stain all kinds of lipids, including lipofuscin granules. Commercial SBB is mostly composed of two isomeres, SBB-I (20%) and SBB-II (60%), whereas the remaining fraction contains all kinds of impurities (Lansink 1968; Pfüller et al. 1977; Subramaniam and Chaubal 1990). SBB has been suggested not to be a true dye (Kiernan 1981) forming a stable chemical bond with lipids but to stain fat because it is more soluble in fat than in the solvent ethanol from which it is to be applied. We tested two protocols to obtain a good SBB solution. One, given by Romeis (1948), prescribes heating a 0.1% SBB in 70% ethanol to boiling point to dissolve the dye, followed by cooling and filtration. The other one, according to Kiernan (1981), prescribes dissolution of the dye by stirring 0.3% SBB in 70% ethanol at RT in the dark for 2 hr and leaving it standing overnight, followed by filtering. Both solutions were applied before and after the sections had been immunolabeled, and various incubation times of 1-60 min were studied. Finally, three brands of SBB were mutually compared, i.e., Merck (Darmstadt, Germany), Aldrich (Zwijndrecht, The Netherlands), and BDH (Poole, UK).

Results

Fixation

Our fixation experiments demonstrated that hypothalami fixed for a period of 6-14 days in 4% freshly depolymerized paraformaldehyde in 0.1 M phosphate buffer (pH 7.5) gave the best histological and immunocytochemical results. A shorter or much longer fixation time, as well as a fixation in the ready-made fixative of Baker and LommersePharma, proved to be suboptimal.

Long-term Storage and Sectioning

Thionin-stained cryosections (25 μm thick) cut from hypothalami stored for 6 months in 30% sucrose at 4C still showed fairly good histology, which was slightly better than that of hypothalami stored for 3 months in 30% sucrose at −80C or hypothalami stored for 3 months in the glycerol/DMSO mixture at either 4C or −80C (not shown). In particular, the extra freezing step and subsequent thawing always introduced some freeze damage, however carefully this step was performed. Immunolabeling of cryosections derived from hypothalami stored in 30% sucrose at 4C for 6 months also showed (after autofluorescence had been blocked; see below) good immunofluorescence of neuronal cell bodies and nerve endings (boutons) for AVP, VIP, GRP, and GAD₆₅ in the PVN, SON, and SCN (Figures 1B–1E). The immunoreactivity for VIP was invariably a little lower than that for the other ligands despite the fact that, for the immunolabeling for VIP, a biotinylated secondary antibody was included to amplify the signal. Subsequent experiments with two different VIP antibodies made it very plausible that this lower immunoreactivity was not caused by the VIP antibody but by too low a VIP antigenicity in the tissue, the cause of which is unknown (data not shown). Remarkable, finally, was a conspicuously good immunolabeling of GAD₆₅-positive nerve endings (Figure 1C).

Cryosections (cut immediately after immersion in sucrose or glycerol/DMSO) and stored in one of both cryoprotectants for 3 months at −80C also showed some damage caused by the extra freezing and thawing step, but this was usually less than was seen in cryosections from hypothalami frozen in toto (not shown).

The embedding of hypothalami in paraffin showed, as could be expected, a substantial shrinkage of the tissue, and the paraffin sections (6-8 μm thick) always showed lower immunoreactivity than cryosections. Nevertheless, the strength of the immunofluorescent signal was, for the present antigens studied, sufficient to count cell profiles and to estimate co-localization with the CLSM (Figure 1F). An obvious advantage of paraffin-embedded tissue is, of course, its unlimited storage time and the possibility of obtaining three times more sections per hypothalamus than by cryosectioning.

Blocking of Autofluorescence with Sudan Black B

Figure 1A demonstrates the immense autofluorescence seen in a nonimmunolabeled frontal cryosection through the human PVN after excitation with the 488- and 543-nm laser of the CLSM and after superimposition of both images. Because the autofluorescence pattern was equally strong after excitation with either laser, superimposition of both images in the pseudo-colors green and red, respectively, resulted in the present yellowish color. This autofluorescence is mainly caused by lipofuscin granules which are present in most, if not all, neuronal cell bodies and glial cells; the neuropil itself also exhibits some autofluorescence (Figure 1A). The latter may be caused by lipofuscin-like substances and/or by scattered autofluorescent light derived from the granules. Because Figure 1A was taken with approximately about the same settings of lasers, filters, and photomultipliers as those of the double immunolabeled sections after their autofluorescence had been blocked with SBB (Figures 1B–1E), it is obvious that such a strong autofluorescence, if not blocked, will completely obscure the analysis of any specific immunolabeling.

Figure 1

Confocal images of neuronal cell body profiles and nerve endings in the SCN, PVN, and SON as observed in some frontal cryosections (A-E) and a frontal paraffin section (F) of the human hypothalamus stored, after fixation in formaldehyde, for more than 6 months in 30% sucrose at 4C. All pictures were obtained by superimposing two images of the same optical section, i.e., one image obtained after excitation with the 488-nm laser and displayed in the pseudo-color green and one image obtained after excitation with the 543-nm laser and displayed in the pseudo-color red. (A) Autofluorescence in an unlabeled frontal cryosection through the PVN. Most if not all neurons exhibit autofluorescence because of the presence of many lipofuscin granules, and the neuropil also exhibits some background autofluorescence (approximately the same settings of lasers, filters, and photomultipliers were used as those during analysis of the immunolabeled sections). The overall yellowish color is caused by the fact that each of the two superimposed images shows about the same pattern of autofluorescence. It is obvious that such strong autofluorescence will completely obscure the analysis of any specific immunolabeling. (B-F). Confocal images of sections were double immunolabeled for the presence of AVP with VIP in the PVN and SON, AVP with GRP in the SCN, and AVP with GAD₆₅ in the PVN, and finally incubated in a Sudan Black B solution to block all autofluorescence. Yellowish- and orange-stained cell profiles point to the coexistence of two neuropeptides. (C) Observe the presence of many red GAD₆₅-IR axonal terminal and/or nonterminal varicosities, some of them apposed to a green AVP-IR neuronal cell body profile and its proximal dendrite. Such apposing varicosities suggest the presence of classical synaptic contacts. A micrograph of an unlabeled section stained only with Sudan Black B to prove the complete blocking of all autofluorescence is not given because such a micrograph only shows a black field. Bars: A,B,D = 25 μm; C,E,F = 10 μm.

Each of the SBB solutions (Romeis 1948; Kiernan 1981) was able to fully block all autofluorescence when sections were incubated for a 10-min period at RT after they had been immunolabeled. When done beforehand, the SBB is extracted again for the most part by the Triton X-100 present in the immuno-incubation media. A comparable effect was found when some Triton X-100 or ethanol was added to any of the final rinsing steps. Further comparison of both SBB solutions revealed that the solution of Kiernan was superior to that of Romeis because it yielded less nonspecific precipitation of SBB on the sections during the first subsequent rinsing step in Tris-buffered saline (TBS). Although this nonspecific precipitation was minimal, it could be completely prevented if SBB of BDH (one of the three different SBB brands tested) was used and subsequent rinsing in TBS was done rapidly and repeatedly (10 times). The SBB solution can be stored in the dark at 4C for at least 2 months.

Unfortunately, SBB itself was also found to exhibit some autofluorescence, starting at about 630 nm and gradually increasing at higher wavelengths (not shown). This excludes the use of a 633-nm laser.

Worth mentioning, finally, is that the ready-made glycerol-based antifading mounting medium Vectashield (Vector Laboratories; Burlingame, CA) and the antifading agents PPD (p-phenylene-diamine; Sigma, St. Louis, MO) or DABC0 (1,4-diazabicyclo [2,2,2]-octane; Sigma) dissolved in glycerol were found to react to such an extent with SBB that a substantial reddish background autofluorescence was generated during analysis with the 543-nm laser, regardless of the pH of the solution or the SBB brand used. Therefore, plain 80% Tris-buffered glycerol at a pH of 7.0-7.6 (not higher) is recommended as the best mounting medium for fluorophore-labeled and SBB-stained sections if analyzed by confocal laser scanning microscopy. If the intensity of the laser excitation is kept at a moderate level, fading of the fluorophores FITC and Cy3 is minimal and background autofluorescence is negligible, while the fluorescence signal is generally sufficient for qualitative and quantitative analysis (Figures 1B–1F). For more details about antifading agents, the reader is referred to Florijn et al. (1995).

Discussion

The main outcome of this study is the development of a procedure for reliable quantitative research on co-localization of neuropeptides in adult human brain tissue. The optimal condition for fixation proved to be in 4% freshly depolymerized paraformaldehyde in phosphate buffer at 4C for 6-14 days. The under and upper time limits of 6-14 days, respectively, were chosen for the following reasons. Formaldehyde, after its initially rapid but reversible binding to proteins, starts to irreversibly crosslink proteins, which requires 1-2 weeks for completion at RT (Kiernan 1981). Such crosslinking is needed to some extent to retain antigens in the tissue and to solidify the tissue a little and thus to enhance resistance against long-term storage in a solution at 4C and subsequent cryosectioning. Much longer fixation times, e.g., several weeks or months, were found to seriously hamper antibody penetration and to reduce antigenicity so that retrieval procedures (e.g., a microwave treatment) were required (Shi et al. 1991; Evers and Uylings 1994; Login and Dvorak 1994; Zhou et al. 1996). A fixation time of 6-14 days therefore proved to be optimal. The fact that ready-made commercial formaldehyde solutions always gave worse immunocytochemical results than a freshly depolymerized paraformaldehyde solution may be caused by some deterioration of commercial formaldehyde in the course of a few weeks, by the addition of an unwanted stabilizer such as methanol to the solution to retard this deterioration, and/or by a suboptimal pH (Kiernan 1981).

As an optimal condition for the long-term storage of brain tissue in toto, a 30% sucrose solution in PBS (supplemented with 0.05% NaN₃) kept at 4C is recommended. We observed that after 6 months under such a conditions hypothalami still showed good histology and immunoreactivity. Cryosections cut immediately after immersion either in 30% sucrose in PBS or in a mixture of 20% glycerol and 2% DMSO in TBS and stored in the same solutions at −80C are a second choice because some extra, although minimal, freeze artifacts are induced by the extra freezing and thawing step. In theory, these freeze artifacts can be expected to counterbalance at a certain moment the slowly but steadily proceeding decay of tissue stored in sucrose at 4C if longer storage times than 6 months are required.

The main problem to be overcome was a substantial autofluorescence during excitation with both the 488- and the 543-nm laser, which completely masked any specific immunofluorescence. This autofluorescence was mainly caused by the wear-and-tear pigment lipofuscin aggregated in granules, which are abundantly present in neuronal cell bodies and glial cells in almost all brain regions (Figure 1A). Apart from these granules, the neuropil itself also showed some autofluorescence, which could be caused by lipofuscin-like substances and/or by scattered autofluorescence light derived from the granules. By covering these lipofuscins with Sudan Black B after the sections had been double immunolabeled, this autofluorescence could be fully blocked while specific autofluorescence decreased only minimally.

The omission of an antifading agent in the mounting medium is necessary to prevent background autofluorescence emitted by a reaction product formed between the antifading agent with one of the Sudan Black B components. Little fading of the fluorophores FITC and Cy3 is seen, however, if the intensity of the excitation light of both lasers is kept at a normal moderate level.

Our decision to use the relatively new fluorophore Cy3 instead of its conventional sister compounds Texas Red or TRITC was made because Cy3 is much brighter and photostable and yields less background fluorescence. As a result of some overlap between the emission spectrum of FITC and that of Cy3, however, a little bleeding through of the FITC signal into the Cy3 signal was always seen if double-immunolabeled sections were simultaneously scanned with both lasers on. This bleeding through, which cannot be fully filtered away without substantial loss of Cy3 signal intensity, made it sometimes necessary to reanalyze a particular structure by sequential excitation with each laser separately on. In theory, it should be possible to further minimize this cross-talk by replacing the usual emission filter BP 575-640 nm by a BP 595-640 nm in combination with the fluorophore Texas Red instead of Cy3. The emission spectrum of Texas Red extends over a longer wavelength region than that of Cy3, which enables a much better separation from FITC. We did not test this option, however, because such a filter was not available or for sale when these experiments were performed. Worth mentioning in this respect is that an emission filter passing light with a longer wavelength than 640 nm results in a substantial decrease of the signal/noise ratio due to autofluorescence of Sudan Black B itself.

Finally, the pesent protocol also gives good results when applied to sections derived from other brain regions than the hypothalamus, such as the cerebral cortex, which is also characterized by an abundance of autofluorescence emitted by lipofuscin granules.

A quantitative study based on this optimized procedure is to be published elsewhere.