Evidence for the ectopic synthesis of melanin in human adipose tissue

Abstract

Melanin is a common pigment in animals. In humans, melanin is produced in melanocytes, in retinal pigment epithelium (RPE) cells, in the inner ear, and in the central nervous system. Previously, we noted that human adipose tissue expresses several melanogenesis-related genes. In the current study, we confirmed the expression of melanogenesis-related mRNAs and proteins in human adipose tissue using real-time polymerase chain reaction and immunohistochemical staining. TYR mRNA signals were also detected by in situ hybridization in visceral adipocytes. The presence of melanin in human adipose tissue was revealed both by Fontana-Masson staining and by permanganate degradation of melanin coupled with liquid chromatography/ultraviolet/mass spectrometry determination of the pyrrole-2,3,5-tricarboxylic acid (PTCA) derivative of melanin. We also compared melanogenic activities in adipose tissues and in other human tissues using the L-[U-¹⁴C] tyrosine assay. A marked heterogeneity in the melanogenic activities of individual adipose tissue extracts was noted. We hypothesize that the ectopic synthesis of melanin in obese adipose may serve as a compensatory mechanism that uses its anti-inflammatory and its oxidative damage-absorbing properties. In conclusion, our study demonstrates for the first time that the melanin biosynthesis pathway is functional in adipose tissue.—Randhawa, M., Huff, T., Valencia, J. C., Younossi, Z., Chandhoke, V., Hearing, V. J., Baranova, A. Evidence for the ectopic synthesis of melanin in human adipose tissue.

Melanin is a common pigment in animals. In humans, melanin is present in two major forms, eumelanin and pheomelanin (1). Both forms of melanins are produced in melanocytes that are derived from the neural crest and in retinal pigment epithelium (RPE) cells that originate from the optic cup (2, 3). In melanocytes and in the RPE, melanin is produced and stored in specialized lysosome-related organelles called melanosomes. In melanocytes, those organelles are transported out into the dendritic processes before being transferred to keratinocytes, where melanin protects the skin from UV-induced DNA damage. This is a major biological function of melanin in humans, along with its role as a powerful antioxidant capable of scavenging reactive oxygen species (ROS), such as singlet oxygen, hydroxyl radicals, and superoxide anions (4, 5). The pathway for melanin biosynthesis includes a number of melanocyte-specific proteins. The key role in this biosynthetic pathway is played by tyrosinase (TYR), a multifunctional copper-containing glycoenzyme.

In addition to melanocytes and RPE cells, the production of melanin occurs in other locations of the body, such as the inner ear and the central nervous system, particularly in the substantia nigra and in the locus coeruleus, where it normally accumulates with age and possibly plays a role in the pathogenesis of Parkinson disease (6). Among nonmammalian vertebrates, the biosynthesis of the melanin has been found in pigmented macrophages of the spleen and the liver of amphibians (7) and in some pathogenic microorganisms, where its production correlates with an increased virulence (8).

In our previous study of visceral adipose tissue of morbidly obese patients, we detected a statistically significant overexpression of melanogenic genes encoding tyrosinase related protein 1 (TYRP1), dopachrome tautomerase (DCT/TYRP2), melanosome transport protein RAB27a, and Melan-A (MLANA) (9). Those findings prompted us to hypothesize that the melanin biosynthesis pathway is functional in adipose tissue. Our current study demonstrates for the first time that the biosynthesis of the melanin, indeed, takes place in visceral adipose tissue of morbidly obese subjects and may have some physiologically relevant functions.

MATERIALS AND METHODS

Tissue samples, preparation of tissue sections, and fixation

The visceral adipose (n=8) and gastric (n=2) tissue samples used for this study had been collected with informed consent during bariatric surgery; 2 of the patients had biopsy-confirmed nonalcoholic steatohepatitis, 5 had simple steatosis of the liver, and 1 had normal liver biopsy. Adipose sample from the latter patient had been collected at a time of reversal surgery when the patient’s body mass index (BMI) decreased to 26. Adipose samples from the nonobese kidney donors (n=2) and liver specimens from patients undergoing liver resections for liver mass with no underlying liver disease (n=2) were used as controls. Samples were collected after obtaining informed consent as a part of an Epidemiology of Nonalcoholic Fatty Liver Disease (EPI-NAFLD) study according to the protocol approved by Institutional Review Board of Inova Fairfax Hospital (protocol 01.050). Specimens were snap-frozen in liquid nitrogen immediately after collection and were stored at −80°C until the time of assay. Skin samples (10) were used as a positive control for Fontana-Masson staining. Frozen adipose tissue sections (10 μm) were cut in the cryostat chamber at a temperature of −18 to −20°C and stored at −80°C until further processing. Fixation of tissue sections in 4% paraformaldehyde was followed by dehydration with 70, 90, and 100% methanol for 5 min each at 4°C. Paraffinized tissue sections were deparaffinized with xylene and dehydrated with ethanol, followed by treatment with cold methanol and boiling in 10 mM citrate buffer (pH 6.0) for antigen retrieval. Human promyelocytic leukemia HL-60 cells were maintained as recommended by the supplier (American Type Culture Collection, Manassas, VA, USA).

Fontana-Masson staining

Fontana-Masson staining (AMTS Inc., Lodi, CA, USA) was performed according to the manufacturer’s recommendations with modifications. After fixation, the tissue sections were rinsed with distilled water three times, followed by a quick rinse with running tap water followed by another quick rinse with distilled water. Fontana silver solution was prepared by adding ammonium hydroxide solution drop by drop to 10% silver nitrate solution followed by continuous stirring until a faint opalescence appears. The slides were placed in ammoniacal silver solution and incubated in a 60°C water bath for 35 min. After incubation, the slides were rinsed with distilled water 3 times and were placed in 0.1% gold chloride for 1 min. The slides were then rinsed with distilled water 2 times and placed in 5% sodium thiosulfate for 2 min. The slides were rinsed twice in running tap water for 3 min and placed in Nuclear Fast Red Stain for 5 min, followed by two rinses in tap water and dehydration with 3 changes of fresh absolute alcohol. The slides were cleared with 3 changes of fresh xylene. Finally, each slide was covered with a coverslip using a permanent mounting medium, and pigment was observed using a light microscope.

Extraction of melanin from hair

The acid/base method to extract melanin from human hair was performed according to Bolt’s procedure (11) with some modifications. Hairs were washed with a detergent, then with water and acetone to remove dirt and lipid. The air-dried hair was minced into 2- to 5-mm pieces and digested in 1 N NaOH overnight. Concentrated HCl was added to the whole mixture to precipitate a brown gum sediment that was dissolved in 1 N NaOH, precipitated by adding HCl, and centrifuged. Base solubilization and acid precipitation were repeated 15 times until the supernatant of the acid wash was almost colorless. The melanoprotein concentrate obtained through the above procedure was stirred in 1 M HCl for 4 h and was centrifuged. This step was repeated 10 times until the supernatant was colorless. The resultant melanoprotein concentrate was washed with distilled water 6 times, with ethanol twice, and finally with ether, and then was dessicated in a SpeedVac and lyophilized.

Extraction of melanin from adipose tissues

Adipose tissue samples (1 g) were homogenized by sonication in 4 ml of RIPA buffer until the tissue was converted to liquid, then centrifuged at 1200 g. The fat was removed, and the liquid phase was stored at −80°C. Pellets consisting of cell debris and black pigment were washed twice with PBS and were centrifuged again at 1200 g. The pellet was dissolved in 1 M NaOH to dissolve the pigment, which was subsequently analyzed by liquid chromatography and mass spectrometry (LC-MS) as described below.

Permanganate oxidation of synthetic and extracted melanin

Oxidation experiments were performed using 1 mg of synthetic melanin, 1 mg of lyophilized hair melanin, and melanin extracts obtained from 1 g of sonicated tissue samples as described above. To achieve complete solubilization of the melanin, the mixtures were sonicated for 24 h in 1 ml of 1 M NaOH. Each sample underwent permanganate oxidation essentially as described previously (12). Each sample was assayed in duplicate. Aqueous suspensions were prepared by sonicating 100 μl of respective homogenates in separate test tubes with 100 μl of BSA (20 mg/ml) and 800 μl of 1 M H₂SO₄ at room temperature overnight. This step is required for the breakdown of melanin to pyrrole-2,3,5-tricarboxylic acid (PTCA), which serves as a specific indicator of eumelanin that can be quantified by HPLC. To this, 3% KMnO₄ was added in portions of 20 μl while mixing until the purple color persisted. At 10 min after the first addition, the residual KMnO₄ and newly formed MnO₂ were decomposed by the addition of 100 μl of 10% Na₂SO₃. The resulting solution was evaporated overnight in a SpeedVac. Samples were evaporated to dryness using a Centrivap sample concentrator (Labconco, Kansas City, MO, USA). Samples were reconstituted in 1 ml of 18 MΩ ultrapure water and filtered into 12 × 32 mm deactivated amber autosampler vials (National Scientific Co., Rockwood, TN, USA) using Millex hydrophilic polytetrafluoroethylene syringe filters with 0.45-μm pores (Millipore Corp., Billerica, MA, USA).

LC-UV-MS analysis of PTCA

The detectors were configured serially with the nondestructive photodiode array (PDA) upstream to the mass spectrometer. The PDA monitored UV spectra from 190 to 400 nm and recorded a chromatogram at 270 nm for additional confirmation of PTCA. The MS electrospray ionization (ESI) probe was operated in negative ionization mode with a capillary voltage of 3.5 kV and an extractor voltage of 5 V. The source temperature was 150°C, and the desolvation temperature was 350°C. The nitrogen desolvation gas flow was 250 L/h, and the cone flow was 50 L/h. The mass spectrometer was operated in selected ion recording (SIR) mode, monitoring ions 198, 154, and 110 m/z, with a dwell time of 0.5 s for each ion. The cone voltage was alternated between 14 and 25 V during each 1.5-s MS scan. The 198 m/z parent ion was produced at 14 V. The higher cone voltage of 25 V increased the kinetic energy of the PTCA ion, resulting in collision-induced fragmentation in the source to produce the 154 and 110 m/z secondary ions.

Authentic PTCA was obtained as a gift from Prof. Shosuke Ito (Fujita Health University, Toyoake, Japan). Synthetic melanin (Sigma, St. Louis, MO, USA) and melanin extracted from hair were used as positive controls, whereas HeLa cell extracts were used as a negative control. Quantitative analysis was accomplished using MassLynx ver. 4.0 software (Waters Corp., Milford, MA, USA). Five calibration standards were made in 18 MΩ ultrapure water at concentrations of 100, 250, 500, 750, and 1000 ng/ml. A peak from an unknown sample component partially coeluted with the primary ion 198; therefore, the area response of secondary ion 154 m/z, which was free from interference, was used for external standard quantitation, and the ions 198 and 110 m/z were used for identification confirmation (Fig. 2). A 5-point linear calibration curve with forced 0,0 axis generated a coefficient of determination value of 0.9997 for ion 154 m/z (not shown).

Figure 1.

Fontana-Masson stain of human adipose tissue demonstrates melanin pigment (black staining) mainly in the periphery of the adipocytes. A, B) Multiple conglomerates of melanin granules are present at the periphery of the adipocytes in adipose tissue from morbidly obese subjects (×20). C, D) Melanin granules are scarce in the adipocytes of adipose tissue from nonobese subjects (×20). E) No melanin granules were observed in the microvessels located in the adipose tissue (×20). F) Melanin staining in skin tissue used as a positive control (×10).

Figure 2.

Presence of melanin in extracts of adipose tissue as revealed by LC-UV-MS. A) Homogenized samples of adipose tissue separated into three phases: supernatant (fat), aqueous, and sediment. Cell debris sediments from adipose tissue samples from morbidly obese subjects contain visible amounts of black pigment. A, B) Adipose sediments of nonobese subjects. C, D) Adipose sediments from obese subjects. B) CAD of the PTCA precursor ion at mass-to-charge ratio (m/z) 198 produces abundant product ions at m/z 154 and 110 peaks. C) LC-MS multi-ion SIM chromatogram of PTCA peak at a retention time of 6 min. D) Negative ESI mass spectra of PTCA peak at 6 min. E) HPLC-UV/VIS chromatograms at 270 nm.

Melanogenic assays

The radioactive substrates L-[U-¹⁴C] tyrosine and L-[3,5-³H] tyrosine were obtained both from Amersham/Searle (Arlington Heights, IL, USA) and from Schwarz/Mann (Orangeburg, NY, USA). Serum albumin, chloramphenicol, cycloheximide, and penicillin G were purchased from Sigma Chemical; Celite 545 and Norit A were purchased from Fisher Scientific Co. (Fair Lawn, NJ, USA).

The ¹⁴C-tyrosine melanogenic activity was measured by a modification of the original technique described before (13). The protein extract obtained from human melanocytes were used as a positive control. Negative controls (blanks) were incomplete reactions with omitted protein extract. The assay was performed in a total volume of 50 μl, including 30 μl of sample (equal to 50 μg of protein extract), 10 μl of ¹⁴C-tyrosine (25 μCi/ml, sp radioactivity 100 μCi/mmol), and 10 μl of buffer solution (1 M KPO₄, pH 7.2; 1 mg/ml chloramphenicol; 1 mg/ml cycloheximide; 0.1 mg/ml serum albumin; 1000 U/ml penicillin B; and 0.25 mM DOPA cofactor). The assays were set up on ice in triplicate in 96-well round-bottom microtiter plates and were incubated for 16 h at 37°C. After incubation, the plates were put on ice to stop the reaction, then 40 μl of sample from each well was placed on Whatman 3MM filter discs. The filters were washed as follows: one 15-min wash with 0.1 N HCl (1 L/50 filters) containing unlabeled tyrosine (1 g/l), two 15-min washes with 0.1 N HCl, two 5-min washes with 95% (v/v) ethanol (200 ml/50 filters), and one 5-min wash with acetone (200 ml/50 filters). The filters were allowed to dry in air and were then counted for radioactivity in 0.4% (w/v) 2,5-diphenyloxazole in toluene (with 70% efficiency). Each sample was measured in triplicate. Recorded values of blanks (no protein extract added) were 234 ± 37 cpm. Averaged values of blanks were subtracted from mean of three replicates of each sample.

In situ hybridization of adipose tissue specimens

Full-length human TYR cDNA clone MGC9191 (id: 3923096; American Type Culture Collection) was sequenced from both ends to confirm orientation and identity of the insert. Target polymerase chain reaction (PCR) sites were selected after BLAST analysis to prevent any evidence of nonspecific annealing to human transcripts. Oligonucleotide primers specific for human TYR were designed with PRIMER3 software (http://frodo.wi.mit.edu/). Best results were obtained with the following probes: TYR-sense, 5′-CATACGATTTAGGTGACACTATAG-gggcaggcggaggcagagga-3′; TYR-antisense, and 5′-GCGCGTAATACGACTCACTATAGGG-gctgctcagctcgccgatgtcc-3′. The probes were converted to RNA and 3′ tailed with digoxigenin-11-dUTP using a DIG RNA labeling kit (Roche, Basel, Switzerland), according to the manufacturer’s recommendations. Nonradioactive in situ hybridizations using DIG-labeled cRNA probes were carried out as described previously, with minor modifications (14). Briefly, slides with fixed tissue were washed in glycine solution (2 mg/ml in PBS) for 10 min, then washed twice in PBS, and placed in 200 ml of acetylation buffer (0.1 M triethylamine, pH 8.0, containing 0.25% acetic anhydride) for 15 min. After washing in 4× saline sodium citrate (SSC) for 10 min, samples were incubated in prehybridization solution (2× SSC, 50% deionized formamide) for 1 h at 47°C overnight. After that, samples were placed in hybridization solution containing 10 μl of purified DIG-labeled antisense riboprobe. Samples then were incubated in 10 mM Tris-HCl, 0.5 M NaCl, and 0.25 mM EDTA (TNE) buffer, treated with RNaseA for 30 min, and returned to TNE buffer for 3 min, all at 37°C. After washing in 0.1× SSC for 15 min at 47°C, samples were blocked for 30 min and incubated with anti-DIG/HRP conjugate (Dako, Carpinteria, CA, USA) for 40 min at room temperature. For detection, we used the tyramide signal amplification system (GenePoint kit; Dako) and VIP solution (Vector Laboratories, Burlingame, CA, USA) according to the manufacturer’s instructions. For each in situ experiment, signals for cTYR (T7) and control (SP6) probes were compared for neighboring sections of the adipose tissue sample simultaneously. Samples were observed and photographed in a Leica DMRB microscope (Leica Microsystems, Bannockburn, IL, USA).

RNA extraction and analysis

RNAs were extracted from tissues obtained from morbidly obese and from nonobese subjects. Total RNAs were extracted by following the procedure provided with the RNeasy Lipid Tissue Midi Kit (Qiagen, Valencia, CA, USA) using 200 mg of tissue. RNA was quantified by absorbance at 260 nm in a spectrophotometer. Integrity of RNA was assessed after electrophoresis in 1% agarose gels stained with ethidium bromide.

Real-time PCR

First-strand reverse transcription cDNA synthesis was performed on 2 μg of RNA samples using a first-strand cDNA synthesis kit with Superscript II (Invitrogen, Carlsbad, CA, USA) and 250 ng random hexamers (Invitrogen) according to manufacturer’s instructions. Real-time PCR reactions were incubated at 94°C for 10 min, for 1 cycle, then 94°C for 20 s, 60°C for 20 s, and 72°C for 30 s for a total of 40 cycles. This set of cycles was followed by an additional extension step at 72°C for 5 min. All PCR reactions were validated by the presence of a single peak in the melt curve analysis, and amplification of a single specific product was further confirmed by electrophoresis on 2% agarose gels. Primers were designed for the amplicon size in a 100- to 150-bp size range. Real-time PCR results were calculated using the previously described method (9). All reactions were performed in triplicate. For normalization purposes, the level of 18S ribosomal RNA was tested in parallel with the genes of interest as described previously (9).

Immunohistochemistry for human tissue slides

Immunohistochemistry was performed on frozen adipose tissues slides obtained from obese and from nonobese subjects as explained above, and skin sections were used as a positive control. After fixation, the slides were washed with PBS twice for 3 min each. The tissue sections were marked with a Dako pen. The slides were incubated in 20% goat serum (Vector Laboratories) for 1 h at room temperature and then were washed with PBS 3 times for 3 min and incubated with the antibody overnight at 4°C in a humidified chamber. Protein extracts (20 μg) were separated on 8–16% gradient SDS-polyacrylamide gels (Invitrogen). After electrophoresis, proteins were transferred electrophoretically from the gels to Invitrolon polyvinylidene difluoride (PVDF) transfer membranes (Invitrogen). The rabbit polyclonal antibodies to LPL and CD31 were purchased at Santa Cruz Biotechnology (Santa Cruz, CA, USA) and Abcam (Cambridge, MA, USA), respectively, and were used at 1:50 dilution. The rabbit polyclonal antibodies against the carboxyl termini of TYR (αPEP7h), TYRP1 (αPEP1h), and TYRP2 (αPEP8h) (15) were diluted in PBS containing 2% goat serum up to 1:700, 1:700, and 1:7500, respectively. Adipose sections were incubated overnight. Next day, the slides were given 3 washes with PBS containing 0.05% Tween-20 for 5 min each and incubated with Alexa Fluor 594 goat anti-rabbit IgG (Molecular Probes, Eugene, OR, USA) for 1 h at room temperature at 1:200 dilution in the presence of 2% goat serum. Red florescence signal was observed and analyzed with a Leica DMRB/DMLD laser microscope (Leica Microsystems) and ScionImage software (Scion, Frederick, MD, USA).

RESULTS

Presence of melanin granules in adipose tissue is revealed by Fontana-Masson staining

Adipose tissue is mainly composed of adipocytes, fibroblasts, macrophages, and endothelial cells. Fontana-Mason staining of human adipose samples showed an accumulation of black pigment in the periphery of adipocytes in morbidly obese patients. Few or no black granules were detected in the adipose tissue of nonobese subjects. The aforementioned staining was specific to adipocytes, as microvessels and stromal fibroblasts were negative. In adipocytes, the staining pattern reflected the cytoplasmic location of melanin granules, as the central vacuoles filled with lipids remained unstained (Fig. 1).

Presence of melanin is revealed by LC-UV-MS in extracts of adipose tissue

To determine whether the black pigment present in adipose tissue is eumelanin rather than lipofuscin or some other type of nonmelanin pigment, we developed an LC-UV-MS assay that detects PTCA, the most characteristic degradation product of eumelanin. Synthetic melanin and melanin extracted from human hair were used as positive controls, and HL-60 human leukemia cells were used as a negative control. Samples of adipose tissue were liquefied by sonication, and the resulting pellets were dissolved in NaOH, then subjected to permanganate oxidation and used for PTCA quantification by LC-UV-MS. Even during the stage of homogenization, differences in pigmentation between adipose tissue extracts of nonobese and obese adipose samples were noted (Fig. 2A).

LC-UV-MS analysis positively identified PTCA and quantified its secondary ions in visceral adipose samples (Fig. 2B–D). The collisionally activated dissociation of the PTCA precursor ion at a mass-to-charge ratio (m/z) 198 produces abundant product ions at m/z 154 and 110 (16). In adipose samples, PTCA ion 193 m/z coelutes with an interfering component of the extract, while ion 154 m/z does not. Therefore, ion 154 m/z was used as the quantitative ion for all samples.

Quantification of the PTCA ion 154 m/z in the profiled visceral adipose sediments of three morbidly obese subjects revealed the presence of the ion in concentrations ranging from 0.19 to 0.12 ng/μl, while its concentrations in oxidized adipose sediment from two nonobese subjects were 0.05 ng/μl (abdominal visceral adipose) and 0.0009 ng/μl (perirenal fat). PTCA ion 154 m/z was not detected in sediments of the gastric tissue samples or in HL-60 cells used as negative controls.

Melanin biosynthetic activity is revealed in adipose tissue by the L-[U-¹⁴C] tyrosine assay

The total output of the melanogenic pathway was quantitatively evaluated by incorporation of labeled L-[U-¹⁴C] tyrosine into its final product, acid-insoluble melanin. Protein extracts of human gastric and liver tissues were used as negative controls, while an extract of highly pigmented human melanoma MNT1 cells served as a positive control. Melanogenic activities in the liver and the gastric controls were similar to the blank negative control with no protein extract added, while activities in the adipose tissue samples of all 7 obese subjects were much higher (Fig. 3) and were characterized by marked heterogeneity. Activity in the adipose tissue sample of the nonobese individual was less than half that of the obese subjects (obese subjects, 0.17 ± 0.03 pmol product/μg/h; lean subject, 0.05 pmol product/μg/h). (Fig. 3).

Figure 3.

Results of the L-[U-¹⁴C] tyrosine assay. OA1-7, adipose samples from morbidly obese individuals; NOA, adipose sample from nonobese subject; Ga, gastric sample; Liv, liver sample; MNT1, highly pigmented human melanoma cells. Recorded values of melanogenic pathway activity for blanks (no protein extract added) and for gastric and liver side controls were similar (Blanks, 234±37 cpm; Ga, 237±15 cpm; Liv, 238±23 cpm). Recorded values for obese adipose samples were different from those of nonadipose samples (OA1-7, 845±99; Ga&Liv, 237±1). Recorded value for nonobese sample was 428 ± 25. For representation purposes, the recorded count per minute values were adjusted by subtraction of blank values and were transformed to picomole product per microgram per hour values (left-hand scale). Activity of melanogenic pathway per microgram of protein extract of MNT1 cells was ∼20 times higher that that of the adipose samples (right-hand scale).

Expression of melanogenesis-related genes in adipose tissue

In our previous microarray experiments, we detected the expression of melanogenesis-related genes in adipose tissue samples collected from morbidly obese subjects undergoing bariatric surgery (9). To confirm those findings, we extracted mRNA from adipose tissue samples from 6 morbidly obese subjects and from 2 nonobese subjects, then performed real-time PCR quantification of expression levels for TYR, TYRP1, TYRP2, and MC1R genes (Fig. 4). In adipose tissue, samples of obese subjects, TYR, TYRP1, and MCR1 genes were expressed at relatively higher levels than in nonobese individuals. Attempts to detect mRNAs for these genes in gastric samples yielded negative results.

Figure 4.

Relative abundance of MC1R, TYR, TYRP1, and TYRP2 transcripts in adipose samples collected from morbidly obese (OA) and nonobese (NOA) subjects after normalization against 18S RNA. Height of each square corresponds to the level of gene expression in the given sample. MCR1, dark blue; TYR, purple; TYRP1, yellow; TYRP2, light blue. Mann-Whitney P values: P < 0.05 for TYR and MC1R; nonsignificant for TYRP1 and TYRP2.

To confirm the adipocyte-specific expression of tyrosinase, we performed an in situ hybridization of a human TYR probe on sections of the visceral adipose tissue from morbidly obese and from nonobese subjects (Fig. 5). The staining predominantly reflected a cytoplasmic location of TYR mRNA in adipocytes.

Figure 5.

In situ hybridization of a human TYR RNA probe on sections of visceral adipose tissues from morbidly obese and nonobese subjects demonstrated cytoplasmic and membrane-associated staining pattern (×20). A, B) T7 (cTYR) probe, visceral adipose tissue from a morbidly obese subject. C) T7 (cTYR) probe, visceral adipose tissue from a lean subject. D) SP6 (control) probe, visceral adipose tissue from a morbidly obese subject.

TYR, TYRP1, and TYRP2 proteins are present in visceral adipose tissue of morbidly obese subjects

To confirm that enzymes participating in melanin biosynthesis are synthesized in the adipose tissue, we performed immunohistochemical staining of cryosliced adipose tissue samples with αPEP7h, αPEP1h, and αPEP8h antibodies, which are specific for TYR, TYRP1, and TYRP2, respectively.

The staining patterns were the same as Fontana-Masson staining, showing that the expression of melanogenesis-related enzymes highlights the structure of adipose tissue (Fig. 6). Counterstaining of TYR with adipocyte-specific markers LPL (Fig. 7A–C) and adipsin (not shown) revealed almost complete overlap of the uniformly stained comb-like patterns, while staining for endothelial marker CD31 revealed only isolated cells (Fig. 7D–F). The cytoplasmic distribution of TYR, TYRP1, and TYRP2 was mainly restricted to the periphery of the cells. Expression of all three melanogenesis-related proteins was higher in the visceral adipose tissue from morbidly obese subjects compared to nonobese subjects. TYR staining was higher than that of TYRP1 and TYRP2.

Figure 6.

Immunohistochemical staining of visceral adipose tissue sections from morbidly obese and from nonobese subjects for human TYR, TYRP1, and TYRP2 proteins (×20). Red: TYR, TYRP1 or TYRP2 staining. Blue: DAPI (nuclei). A, C, E, G) Visceral adipose tissue from a morbidly obese subject. B, D, F, H) Visceral adipose tissue from a nonobese subject. A, B) Tyrosinase. C, D) Tyrosinase-related protein Tyrp2 (dopachrome tautomerase). E, F) Tyrosinase-related protein Tyrp1. G, H) Negative control (secondary antibodies) and DAPI.

Figure 7.

Immunohistochemical staining of visceral adipose tissue sections from morbidly obese subjects for human TYR, adipocyte-specific markers LPL, and endothelial marker CD31. Red denotes TYR staining. Green denotes LPL or CD31 staining. A, D) Tyrosinase. B) LPL. C) Tyrosinase and LPL counterstaining. E) CD31. F) Tyrosinase and CD31 counterstaining.

DISCUSSION

While studying the transcriptome of visceral adipose tissue of morbidly obese individuals, we noted that human adipose tissue expresses several melanogenesis-related genes (9). Moreover, when the visceral fat of morbidly obese individuals was compared to that of lean subjects, these genes were found to be expressed at higher levels. In the current study, we confirmed the expression of melanogenesis-related mRNAs and proteins, namely TYR (tyrosinase), DCT (Tyrp2), TYRP1, and MITF (microphthalmia transcription factor) in human adipose tissue using real-time PCR and immunohistochemical staining. Additionally, TYR mRNA signals were detected by in situ hybridization in visceral adipocytes. The presence of melanin in human adipose tissue was revealed both by Fontana-Masson staining (Fig. 1) and by permanganate degradation of melanin coupled with LC-UV-MS determination of the PTCA derivative of melanin.

To exclude the possibility that certain levels of melanogenic activity are a common characteristic of many human tissues, we compared melanogenic activities in adipose tissues and in other human tissues using the specific and sensitive L-[U-14C] tyrosine assay. Tyrosinase activities in the adipose tissue samples of obese subjects were much higher than in the samples from nonobese subjects (Fig. 2), while protein extracts of human gastric and liver tissues were at background levels. A marked heterogeneity in the melanogenic activities of individual adipose tissue extracts was noted.

A connection between the extracutaneous accumulation of melanin and morbid obesity have never been described before, although one type of hyperpigmentation of the skin known as acanthosis nigricans has been associated with either clinical or subclinical insulin resistance, a component of the metabolic syndrome. What would be a plausible explanation for the fact that adipose tissue of morbidly obese patients produces higher levels of melanin compared to lean subjects? We hypothesize that the ectopic synthesis of melanin may serve as a compensatory mechanism that utilizes its anti-inflammatory (17) and its oxidative damage-absorbing (18, 19) properties. With the progression of obesity and an increase of the cellular fat deposition, adipocytes become more exposed to endogenous apoptotic signals, especially ROS. To counteract proapoptotic ROS effects, the adipocytes, in turn, may ectopically activate the genetic program of melanogenesis, thus neutralizing excessive ROS. Adipocytic melanin would also suppress the secretion of proinflammatory molecules (17), thereby decreasing the proinflammatory background in obese subjects and alleviating the metabolic syndrome. High levels of polymorphisms in human genes regulating melanin biosynthesis may account for the highly individual melanogenic responses of adipocytes and for the differences in an individual’s propensity to develop secondary complications of obesity.

Currently, it is unknown whether the extent of melanin biosynthesis in adipose tissue could be negatively correlated with patients’ morbidity. Most of the adipose tissue donors enrolled in our study had an underlying disease of the nonalcoholic fatty liver disease (NAFLD) spectrum. Interestingly, morbidly obese subjects with the highest content of the melanin by LC/UV/MS assay (0.42 ng/μl) had normal liver biopsy. Respective sample of fat was taken at a time of bariatric surgery reversal, when patient had almost normal BMI of 26. This finding may indicate that the melanin accumulated in obese individuals remains in the adipose despite a loss of weight. Both patients diagnosed with NASH had substantially lower melanin contents in their adipose, 0.16 ng/μl and 0.18 ng/μl, respectively). Interestingly, limited clinical information available for 5 out of the 8 obese patients sampled for visceral adipose tissue revealed a positive correlation between fasting glucose levels and total outputs of the melanogenic pathway in adipose tissues (r=0.9685, P≤0.007). This observation might indicate a connection between adipocytic melanogenesis and insulin resistance. Larger study involving extended cohorts of the morbidly obese and lean patients is necessary to confirm a link of the melanin biosynthesis to the morbid obesity and the possible protective role of this compound.

One possible explanation for the activation of the melanogenic pathway in adipose tissue may be related to the significantly elevated levels of an endogenous melanogenic peptide α-MSH in the serum of obese subjects (20). Since human adipocytes express the melanocortin receptor MC1-R, which triggers α-MSH-induced melanogenesis in melanocytes (21), it is entirely possible that ectopic melanogenesis in human adipose tissue results from excessive exposure of human adipocytes to melanogenic signals. It is also possible that the described phenomenon is more complex and may be connected to other, not yet uncovered, facets of human obesity.

Interestingly, our attempts to detect melanogenesis in human and mouse adipocytes differentiated from primary cultured preadipocytes repeatedly failed. Despite substantial expression levels of TYR, TYRP1, DCT, and MITF mRNAs (not shown), we could not detect either an enzymatic activity of tyrosinase or the production of melanin in vitro. It is likely that cultured adipocytes are unable to support adequate post-translational modifications or the correct folding of tyrosinase and, therefore, are deficient in this sense. This notion supports previous evidence that the molecular networks activated during adipocyte development in vivo and in vitro, although overlapping, are in many respects quite different (22).

We realize that our study raises more questions than it answers. Both the causes and the consequences of melanin production in adipose tissue remain unknown. One obvious candidate for the regulatory molecule is α-MSH, which is intimately connected both to melanogenesis and to energy homeostasis. Our future research will be aimed at developing an appropriate cellular system to allow the study of adipocytic melanogenesis in vitro and to explore the connection between melanogenesis and metabolic syndrome.

In conclusion, our study demonstrates for the first time that the biosynthesis of melanin takes place in visceral adipose tissue of morbidly obese subjects. Thus, we have uncovered an entirely new phenomenon never reported before. Further research into this area is warranted.