The Normal Human Choroid Is Endowed with a Significant Number of Lymphatic Vessel Endothelial Hyaluronate Receptor 1 (LYVE-1)–Positive Macrophages

Abstract

purpose. Lymphatic vessel endothelial hyaluronic acid receptor (LYVE-1) is a newly discovered lymphatic endothelium-specific marker that is also expressed by a subpopulation of macrophages. To date, there is no report on its expression in the posterior human uvea. The purpose of this study was to investigate the expression of LYVE-1 in normal human choroids.

methods. Eyes of body/cornea donors (55–89 years of age; 4–9 hours postmortem) were obtained. Choroids were dissected and prepared for cryosections followed by immunohistochemistry with anti-human LYVE-1 antiserum and immunogold labeling. In addition, anti-human antibodies against macrophage markers (CD68, MHC class II) and lymphatic (podoplanin) and blood vascular endothelium (CD31, vWF) were used. For documentation, light-, fluorescence-, confocal laser scanning-, and electron-microscopy were used.

results. The normal human choroidal stroma contained 274 ± 86 LYVE-1 positive cells/mm². The cells displayed irregular shapes with a relatively uniform diameter of 32 μm. Costaining with CD68 and negativity for CD31, podoplanin, and melan-A/HMB45, as well as electron microscopic features, suggest these LYVE-1⁺ cells to be macrophages. Besides that, no classic LYVE-1⁺/podoplanin⁺ lymphatic vessels were detected within the normal adult human choroid.

conclusions. The normal adult human choroid does not contain typical lymph vessels, but is endowed with a significant number of LYVE-1 positive macrophages. These cells may be involved in choroidal hyaluronic acid metabolism or contribute to temporary formation of lymphatic channels under inflammatory conditions.

Lymphangiogenesis research has been hampered for several decades due to lack of specific markers for lymphatic vessels. ^{1 2 3 4} This lack of studies has changed only recently with the advent of novel specific markers for lymphatic vascular endothelium such as LYVE-1 and podoplanin. ^{1 2 3 4 5} Whereas the eye has conventionally been thought of as an organ devoid of lymphatic vessels, recent work using these novel antibodies has demonstrated that there are numerous lymphatic vessels in the conjunctiva up to the limbus and also in vascularized human corneas. ⁶ Indeed, the normally avascular cornea has become a very useful model system for the study of mechanisms of lymphangiogenesis. ^{7 8 9} In contrast, it is still unclear whether there are lymphatic vessels in the posterior pole of the normal human eye. Some animal studies suggest the existence of lymphatics in the choroid, ^{10 11 12 13 14} and there is at least a functional drainage of cells from the choroid into the regional lymph nodes, ¹⁵ but, besides electron-microscopic data from nonhuman primates ^{12 13 16} there have been no formal studies into the presence of lymphatic vessels in the normal human choroid. ¹⁷ Also, data from existing electron microscopic studies in human tissue have not provided any data about lymphatics. ^{16 18 19 20 21 22 23} Until recently, electron microscopy was the method of choice for the unambiguous detection of lymphatics; nevertheless, its reliability is not absolute, and it is definitely not suitable for clinical routine. ^{3 24}  

The marker most commonly used to study lymphangiogenesis, is the lymphatic vascular endothelium-specific hyaluronate receptor (LYVE)-1. Besides lymphatic vascular endothelium, it also reacts with macrophages involved in hyaluronate metabolism and angiogenesis. ^{1 6 25} Since macrophage dysfunctions have recently been implicated in the pathogenesis of age-related maculopathy, ^{26 27} and since LYVE-1⁺ macrophages also very recently have been shown to be able to transform into lymphatic vascular endothelial cells, ⁹ we were interested in studying the presence of typical lymphatic vessels expressing novel specific markers of lymphatic vascular endothelium in the normal human choroid (LYVE-1 and podoplanin) and the expression of LYVE-1⁺ macrophages in the normal human choroid. In this study, we demonstrated that the normal adult human choroid does not contain typical lymphatic vessels, by using LYVE-1 and podoplanin as markers, and, in contrast, that there are numerous LYVE-positive macrophages present in the choroid. 

Materials and Methods

Specimen

In accordance with the Declaration of Helsinki, eyes of body/cornea donors (55–89 years of age; 4–9 hours postmortem; n = 17) were obtained from either the Cornea Bank of the Department of Ophthalmology or the Department of Anatomy I (University of Erlangen-Nuremberg). Choroids were dissected and fixed by immersion in phosphate-buffered saline (PBS) containing 4% PFA for 4 hours at room temperature (RT), rinsed in PBS (four times 5 minutes), and transferred into PBS containing 15% sucrose (overnight at 4°C). The choroids were frozen face down at −80°C in liquid nitrogen-cooled methylbutane and stored at −20°C for further processing. 

Immunohistochemistry

Tissue was defrosted in PBS containing 15% sucrose, and 20-μm-thick sections were cut in a cryostat with a flatmount technique that enabled us to investigate sections of the whole choroid. ²⁸ In addition, serial cross sections from the optic nerve center to the ora serrata were performed to clarify the potential three-dimensional architecture of LYVE-1-positive cells. All sections were air-dried for 1 hour at room temperature (RT) on poly-l-lysine (Sigma-Aldrich, St. Louis, MO)–coated slides. After a 5-minute rinse in Tris-buffered saline (TBS; Roth, Karlsruhe, Germany) slides were incubated for 1 hour at RT in TBS containing 10% donkey or goat serum (depending on the secondary antibodies used; Sigma-Aldrich), 1% BSA (Sigma-Aldrich), and 0.5% Triton X-100 (Merck, Darmstadt, Germany). After a 5-minute rinse, the slides were incubated with antibodies (all raised against human epitopes) for single or double labeling of the markers listed in Table 1 . After a rinse in TBS (four times, 5 minutes each) binding sites of primary antibodies were visualized by corresponding Cy3- or Alexa488-tagged antisera (1:1000; Invitrogen, Karlsruhe, Germany) in TBS, containing 1% BSA and 0.5% Triton X-100 (1 hour at RT) followed by another rinse in TBS (four times, 5 minutes each). The slides were embedded in TBS-glycerol (1:1 at pH 8.6). Negative controls were performed by omission of the primary antibodies during incubation. In addition, before immunohistochemistry a preabsorption control for the LYVE-1 antiserum with the corresponding protein (Acris, Heidelberg, Germany) was performed. Both procedures resulted in no staining. Human liver specimens served as the then positive control for LYVE-1 and podoplanin and showed staining of sinusoid endothelial cells. 

Documentation

To document single- and double-label immunohistochemistry, a confocal laser scanning microscope was used (MRC 1000; Bio-Rad, Munich, Germany, attached to a Diaphot 300; Nikon, Düsseldorf, Germany, and equipped with a krypton-argon laser; ALC, Salt Lake City, UT; ×20 dry or ×40 and ×60 oil-immersion objective lenses, with numeric apertures of 0.75, 1.30, and 1.4, respectively; Nikon). Sections were imaged with the appropriate filter settings for Cy3 (568-nm excitation, filter 605DF32; channel 1, coded red) and Alexa488 (488 nm excitation, filter 522DF32; channel 2, coded green). Colocalization of the same structures in channels 1 and 2 resulted in a yellow color. For quantitative assessments, an epifluorescence microscope (Aristoplan; Leica, Bensheim, Germany; Filterblock N2.1 for viewing Cy 3, Filterblock I 3 for Alexa488) with ×25 or ×40 dry objective lenses and equipped with a digital camera (Spot RT; Visitron Systems, Munich, Germany) was used. For quantification of LYVE-1-positive cells, seven randomly chosen sections from three body donors were evaluated. 

Electron Microscopy

For electron microscopy, pre-embedding protocols were used to demonstrate LYVE-1 or CD-68 immunohistochemistry via a nanogold detection system. 

Choroids were fixed by immersion in PBS containing 4% paraformaldehyde and 0.1% glutaraldehyde for 1 hour at RT. After a rinse in TBS (three times, 10 minutes each) a preincubation in TBS containing 1% BSA, 0.05% thimerosal, 0.05% Triton X-100, and 10% goat normal serum was performed (2 hours at RT). Tissue was then incubated with the LYVE-1 (1:700) or CD-68 (1:50) antibody (in TBS containing 1% BSA, 0.05% thimerosal, 0.05% Triton X-100; 72 hours at 4°C). Followed by another rinse in TBS (24 hours at 4°C), binding sites were visualized by using a goat anti-rabbit (for LYVE-1; 1:200) or goat anti-mouse (for CD-68; 1:20) nanogold antibody (Biotrend, Cologne, Germany; in TBS containing 1% BSA, 0.05% thimerosal and 0.1% CWFS gelatin; 48 hours at 4°C). This incubation was followed by a rinse in TBS (containing 1% BSA and 0.1% CWFS gelatin; three times, 1 hour each at RT) and another rinse in TBS (48 hours at 4°C). Tissue received a postfixation in TBS containing 2.5% glutaraldehyde (15 minutes) and was further rinsed in TBS (24 hours), followed by a rinse in distilled water (eight times, 10 minutes each) and treatment with an SE-EM-silver enhancement-kit (ready to use; 4 hours; Biotrend). After a final rinse in distilled water, tissue was postfixed in 0.1% osmium tetroxide containing 1.5% potassium-hexacyanoferrate (Merck), dehydrated through graded alcohols, embedded in Epon, and mounted on Epon blocks. Silver-gray serial ultrathin sections received another silver enhancement on the Ni-grid (4 hours; SE-EM; Biotrend) and were lightly contrasted with lead citrate and examined in a transmission electron microscope (EM 906; Carl Zeiss Meditec, Oberkochen, Germany) and photographed for documentation. Micrographs were digitized (SnapScan e50; Agfa, Cologne, Germany) and slightly adapted in contrast and brightness (Photoshop 6.0; Adobe, San Jose, CA). 

Results

LYVE-1 Immunoreactivity in the Normal Adult Human Choroid

In this study, a commercially and a noncommercially available LYVE-1 antibody was used. Since both antibodies revealed the same results, only the results from the commercially available antibody will be reported. 

Numerous LYVE-1 immunoreactive cells were detected in choroidal sections. These cells showed a clearly discernible nucleus (with a mean diameter of 12 μm), and displayed an irregular shape of the cell surface (Fig. 1)as well as its processes. In randomly chosen sections (n = 7), 274 ± 86 of these cells per square millimeter were counted. Some of these LYVE-1-positive cells displayed processes arising from the soma that extended out to 100 μm (Fig. 1C) . LYVE-1-positive cells were randomly distributed throughout the choroid (Fig. 1D) . 

In serial cross-sections, LYVE-1-positive cells were detected in the perineurium of the optic nerve and rarely within the nerve, as well as in outer layers of the sclera. Here, processes of LYVE-1-positive cells were arranged tangential to the surface, but not perpendicular in direction to the choroid (Fig. 1E) . Likewise, no perpendicular LYVE-1-positive structures were found within the sclera or piercing it. In the retina, LYVE-1 immunoreactivity was absent. 

LYVE-1 and Lymphatic Vessels in Normal Human Choroid

To analyze whether the LYVE-1-positive cells were lymphatic vascular endothelial cells as part of lymphatic vessels, double immunolabeling experiments with podoplanin as another marker specific for lymphatic vascular endothelium and with CD31/PECAM1 as well as von Willebrand factor as pan-endothelial markers (Table 1)were performed. As shown in representative images in Figure 2 , no double labeling was detectable with these markers. Furthermore, there was no typical staining pattern of LYVE-1 immunoreactivity suggestive of classic lymphatic vessels (e.g., as known from the conjunctiva, skin, or the vascularized cornea; see the Discussion section). Taken together, these findings suggest the absence of classic lymphatic vessels from adult human choroids under nonpathologic conditions. 

LYVE-1 Positive Structures and Macrophages in Normal Human Choroid

To further characterize the LYVE-1-positive cells in the choroid, numerous double labeling experiments were performed using the markers outlined in Table 1 . In single optical sections of the confocal microscope, double immunostaining experiments revealed that almost all the LYVE-1-positive cells were also positive for the macrophage marker CD68 (Fig. 2A) , as well as for the major histocompatibility complex II (MHCII). In contrast, not all the MHCII-immunoreactive cells were positive for LYVE-1 (Fig. 2B) . Also, not all CD68-positive cells were LYVE-1 positive. LYVE-1 immunoreactive cells were not colocalized with the lymphatic vascular endothelial marker podoplanin (Fig. 2C)nor the pan-endothelial marker CD31/PECAM1 (Fig. 2D) , making this cell population clearly discernible from vascular endothelium and lymphatic endothelium, respectively. Occasionally, nonvascular podoplanin-positive cells were observed. 

Since LYVE-1-positive cells also resemble the typical size and shape of melanocytes, we tested for this possibility by using two well-established melanocyte markers: HMB 45 and melanA. Double immunostaining revealed that neither HMB 45- (Fig. 2E)nor melanA-positive cells (Fig. 2F)were colocalized with LYVE-1, and vice versa, suggesting their monocytic rather than melanocytic origin. 

Electron Microscopic and Immunogold Features of LYVE-1-Positive Choroidal Cells

LYVE-1 immunoreactivity was visualized with electron-dense nanogold particles. These particles were highly concentrated within the cells, but only a few particles were visible within the intercellular matrix, indicating high affinity for the detected epitope. 

LYVE-1-positive cells showed a smooth contoured nucleus of low electron density. The cytoplasm of LYVE-1-positive cells was packed with numerous electron-dense granules. The surface of these granules appeared not sharply contoured as in melanin granules but rather disintegrated, becoming denser toward the center of the granules. Different grades of electron density were discernible within these granules (Figs. 3A 3A ′). 

To determine whether the same type of granules was also detectable in macrophages, immunoelectron microscopy was performed with a CD68 antibody. In accordance with earlier results, nanogold particles showed an intracytoplasmic accumulation. ^{29 30} Electron-dense granules within the CD68-positive cells resemble the same as described in LYVE-1-positive cells, indicating the same cell population (Figs. 3B 3B ′). 

LYVE-1-positive cells were forming processes that appeared to span open spaces of up to 100 μm in diameter. Occasionally, processes of LYVE-1-positive cells were contacting each other (Fig. 4) . 

Discussion

The current studies allow two important conclusions to be drawn: (1) the normal human choroid does not contain typical lymphatic vessels, as assessed by immunohistochemistry with novel specific markers of lymphatic vascular endothelium (such as LYVE-1 and podoplanin). (2) The normal human choroid is endowed with significant numbers of LYVE-1⁺ macrophages. 

Although there is clinical and experimental evidence of drainage of fluid from the choroid to the posterior pole of the eye and the regional lymph nodes, ^{17 31} there has been no clear analysis of the presence or absence of classic lymphatic vessels in the human choroid so far. In contrast, several animal experiments suggest the existence of choroidal lymphatic vessels at least in some species. ^{7 11 12 13 14} Our findings suggest that there are, at least under normal conditions, no classic lymphatic vessels present in the adult human choroid. The classic concept of lymphatics refers to structures that represent blind ending capillaries building a netlike framework throughout the tissue, converging to larger lymphatic vessels and collector vessels, eventually entering lymph nodes. ³ However, blind ending capillaries or a netlike framework was not detectable in our investigation with the LYVE-1 antiserum. Although, by virtue of their tubular processes, some structures appeared lymphatic-like at a first glance, they neither converged nor formed larger compounds in the periphery. Because of the absence of podoplanin on the one hand and melanocyte markers on the other, and because of the presence of macrophage markers, we conclude that these LYVE-1-positive structures represent macrophages instead. In contrast, it was possible to identify lymphatic vessel-like structures with a clear vessel lumen in fetal human eyes (Cursiefen C, Schroedl F, unpublished observations, 2006). The presence of classic lymphatic vessels in fetal choroid and its absence in adult human tissue may suggest a regression of choroidal lymphatic vessels. That would be in contrast to the human cornea, for example, which is primarily devoid of both blood and lymphatic vessels, supporting the evolutionary importance of corneal angiogenic privilege. ³¹ However, tissue from younger persons was not available and therefore lack of these data is a limitation of the present study. Future work should be undertaken to test this hypothesis. 

Our findings are in line with those in a recent study examining murine eyes for classic lymphatic vessels, which could only be detected outside of the eye around the optic nerve and at the limbal arcade. ¹⁷ Together with reports on the presence of lymphatic vessels in the human orbit around the posterior pole of the eye, ² these data suggest that “true” lymphatic drainage from the eye only starts outside of it (e.g., at the limbus or at the posterior pole around the optic nerve). Our findings support the concept of a normally alymphatic eye, with lymphatic vessels physiologically in the adult starting only at the limbus and the outer sclera. ¹⁷  

Since the “pseudovessel” shown in our micrograph was lacking LYVE-1, the vessel may represent, if anything, a specialized form of lymphatics in the posterior uvea. For example, it is known that LYVE-1 is absent in collecting lymph vessels, and this specialization may also be in line with the observation that obviously no LYVE-1-positive lymphatics piercing the sclera were detectable. Alternatively, macrophages may be integrated in forming a drainage system, the function of which and especially the outflow is not understood yet (see below). This, however, also implies that a lymphatic outflow is routed more or less exclusively via anterior parts of the eye, as shown recently (Gupta N, et al. IOVS 2008;49:ARVO E-Abstract 2879). 

LYVE-1 has been extensively used as a lymphatic marker since it was first identified in 1999 by Banerji et al. ³² Besides lymphatic vessels, its expression was also reported on sinusoidal endothelial cells within the liver and placental syncytiotrophoblasts and Kupfer cells and, to a lesser extent, on cerebral cortex neurons, renal tubular cuboidal epithelium, and pancreatic exocrine cells/islet of Langerhans. ^{2 29} Recently, LYVE-1-expression was shown on CD45⁺, CD11b⁺, and CD31⁻ cells in the murine conjunctiva, indicating expression on cells of a monocytic lineage. ³³ Using the corneal inflammation model, this group also demonstrated downregulation of LYVE-1 expression in the conjunctiva during corneal inflammation while lymphangiogenesis occurred. ³⁴ The downregulation may be due to LYVE-1⁺ macrophages acting as a reservoir for lymphangiogenesis and cell recruitment under corneal inflammatory conditions. Indeed, macrophages have recently gained renewed interest for their novel and important role in (inflammatory) (lymph)angiogenesis. ^{34 35}  

Macrophages contribute to inflammatory lymphangiogenesis in two ways: First, they release a whole cocktail of hem- and lymphangiogenic growth factors such as VEGF-A, -C, and -D. ¹ Therefore, by locally depleting macrophages, it is possible to completely inhibit inflammatory corneal hem- and lymphangiogenesis. ¹ Second, CD11b⁺ macrophages have recently been shown to physically integrate to form new lymphatic vessels. ^{9 36} In diabetic skin, reduced wound healing and lymphangiogenesis was related to a reduced number of functionally active macrophages. ^{37 38} Taken together with the unexpectedly high number of LYVE-1-positive macrophages in the choroid, it may well be that these macrophages act as a reservoir for temporary formation of lymphatic vessels or lymphatic channels in the response to inflammatory stimuli, necessitating increased and localized drainage from the retina toward the sclera. Supporting that concept was the macrophage networks observed in normal human eyes that mimicked vascular lumina. It has long been known that the choroid contains a high number of resident macrophages. Most of the F4/80 positive cells in the murine choroid are macrophages. ^{39 40} In addition, as our results show, at least some of the cells in the light microscope identified as melanocytes apparently represent choroidal macrophages instead. 

Another potential role for the vast number of LYVE-1-positive macrophages in the normal human choroid may be their role in hyaluronic acid metabolism. LYVE-1 is an endocytic receptor for HA from the CD44 family. ^{29 31} In that context, it is interesting to note that macrophages seem to play a key role in the pathogenesis of age-related macular degeneration (AMD). Mice deficient in monocyte chemoattractant protein-1 (MCP-1) or its cognate C-C chemokine receptor-2 (Ccr-2) develop clinical features typical of AMD. ²⁶ Since normal choroidal macrophages degrade C5 and IgG in eye sections of Ccl2^−/− or Ccr2^−/− mice, impaired macrophage recruitment may allow accumulation of C5a and IgG, which induces vascular endothelial growth factor (VEGF) production by RPE, possibly mediating development of choroidal neovascularization (CNV). ²⁷ On the other hand, local depletion of macrophages also causes CNV, ^{27 41} indicating a fine-tuned mechanism for macrophages in the choroid. Since LYVE-1+ macrophages are known to play a role in angiogenesis and lymphangiogenesis, this macrophage subpopulation may contribute to the pathogenesis of AMD. Further research into that question is under way and will have to analyze the pattern of LYVE-1 positivity in inflamed and diabetic eyes as well as in eyes with different forms of age-related maculopathy. 

In summary, our data suggest that the normal adult human choroid is devoid of conventional lymph vessels but contains a significant number of LYVE-1⁺ macrophages. Subtypes of macrophages have to be identified in upcoming studies. The LYVE-1⁺ macrophages may be involved in de novo formation of lymph-channel–like structures under inflammatory conditions. 

 Supported by the Interdisciplinary Center for Clinical Research (IZKF) Erlangen (A9), and the ELAN Fonds University Erlangen (CC, FS).

 Submitted for publication January 10, 2008; revised June 6 and July 22, 2008; accepted October 2, 2008.

 Disclosure: F. Schroedl, None; A. Brehmer, None; W.L. Neuhuber, None; F.E. Kruse, None; C.A. May; None; C. Cursiefen, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Antibodies

Antibody against	Raised in	Distributor	Dilution
LYVE-1	Rabbit	Courtesy of David G. Jackson, Weatherall Institute of Molecular Medicine, Oxford, UK	1:700
LYVE-1	Rabbit	Acris, Heidelberg, Germany	1:700
CD31 (PECAM-1)	Mouse	Acris	1:50
CD68	Mouse	Dako, Hamburg, Germany	1:50
Melanosome clone HMB-45	Mouse	Dako	Ready to use
MelanA (MART-1)	Mouse	Acris	1:50
MHCII	Mouse	Acris	1:50
Podoplanin	Mouse	Acris	1:100

Figure 1.

 

LYVE-1-positive cells in the choroid. The normal human choroid is endowed with a significant number of LYVE-1-positive cells. Immunofluorescence showed numerous LYVE-1-labeled cells within the human choroid (A; photograph micrograph). Confocal microscopy (extended focus mode) reveals small cells with radially arranged processes (B) and small cells with long tubular-like processes (C). LYVE-1 immunostaining of human choroidal flatmount section (D) reveals an even distribution of LYVE-1-positive cells within the choroid. Furthermore, no staining pattern resembling classic lymphatic vessels can be detected (photograph micrograph). In cross sections (E), LYVE-1-positive cells (arrowheads) were found on the outer margin of the sclera (sc), but no LYVE-1 immunoreactivity was detected in the scleral stroma. ch, choroid; (☆) autofluorescent retinal pigment epithelium.

Figure 2.

 

LYVE-1 colocalization study. Double-immunolabeling studies reveal choroidal LYVE-1-positive cells to be CD68-positive macrophages. All figures: confocal microscopy, single optical sections. (A) Immunohistochemistry against LYVE-1 (red) and CD68 (green) shows almost complete colocalization of both markers (yellow). (B) Immunohistochemistry against LYVE-1 (red) and MHCII (green). LYVE-1-positive cells show colocalization with MHCII (yellow; arrows), whereas some cells (presumably LYVE-1-negative macrophages) show reactivity for MHCII only (arrowhead). (C) Immunohistochemistry of LYVE-1 (red) reveals no colocalization with podoplanin (green). Occasionally, podoplanin⁺/Lyve-1⁻ cells were observed (arrowhead). (D) Immunohistochemistry against LYVE-1 (red) and CD31/PECAM1 (green) revealed no colocalization of both markers (absence of yellow). Note the staining of the blood vessel endothelium (arrowheads). (E, F) Immunohistochemistry against LYVE-1 (red) and two melanocyte markers HMB45 (E, green) and melanA (F, green) revealed no colocalization (absence of yellow).

Figure 3.

 

Electron microscopy LYVE-1 versus CD68: Immunogold labeling with LYVE-1 (A, A′) and CD68 (B, B′) showed identical localization of the immunoprecipitation for the respective markers in the cytoplasm of cells packed with numerous electron-dense granules. The similar staining pattern of these markers suggests the LYVE-1-positive cells to be CD68-positive macrophages. (A′) Magnification of inset in (A). (B′) Magnification of (B).

Figure 4.

 

LYVE-1 pseudovessel. Electron microscopy of LYVE-1-positive cells using immunogold labeling. These cells sometimes form netlike structures with a pseudovessel-like appearance. Insets: predominantly intracellular LYVE-1 immunogold-reactivity associated with electron-dense granules. However, no typical lymphatic vessels were detected.

The authors thank Susanne Fickenscher, Carmen Rummelt, Jasmine Onderka, and Hedwig Symovsky for expert help with immunohistochemistry and Ursula Schlötzer-Schrehardt, Andrea Hilpert, and Inge Zimmermann for excellent support with the electron microscope.