Abstract CD38 is a type II transmembrane glycoproteininvolved in signaling and adhesion which is expressedmainly by immature hematopoietic cells and activatedlymphoid cells. Central lymphatic channels of humansmall intestinal villi, the so-called lacteals, were coinci-dentally found to express CD38. Gastric and large intes-tinal mucosae, pancreas, liver, lung, nasal mucosa, kid-ney, thymus, palatine tonsil, Peyer’s patches, appendix,and mesenteric lymph nodes, and rodent intestinal muco-sa were subsequently examined for lymphatic expressionof CD38. Cryosections prepared from biopsy or surgicalresection specimens were immunostained with four dif-ferent antibodies to CD38 combined with antibodies tovon Willebrand factor and CD31 to differentiate lym-phatics from blood vessels, or with antibody to lysoso-mal protein. Sections were evaluated by ordinary andconfocal immunofluorescence microscopy. Jejunal cryo-sections were subjected to in situ hybridization forCD38. All CD38 antibodies decorated human lacteals,and some of these were positive for CD38 mRNA. Lym-phatics draining Peyer’s patches and appendix as well asafferent lymphatics of mesenteric lymph nodes ex-pressed CD38 weakly. CD38 was not detected on lym-phatics in other organs or in rodent lacteals. We proposethat CD38 is a novel marker of human small intestinallymphatic vessels.

Keywords CD38 · Lymphatic vessels · Humans · Rodents

Introduction

CD38 is a multilineage type II transmembrane glycopro-tein primarily expressed by lymphoid cells [31]. It is abifunctional ectoenzyme with ADP-ribosyl cyclase andhydrolase activities [23] and has signaling properties[15, 29]. CD38 participates in adhesion of circulatingcells to vascular endothelium via its ligand CD31 whichis abundantly present on endothelial cells [7]. Further-more, CD38 exists as a soluble molecule with retainedenzymatic activity, occurring in both normal and patho-logical body fluids [16].

The CD38 gene contains eight exons, and two poly-morphic variants have been identified [14]. Increasingfocus on CD38 in human diseases has revealed that inchronic lymphatic leukemia, cellular expression of CD38appears to indicate poor prognosis [6], as does CD38 ex-pression on circulating CD8+ cells in HIV+ patients [36].The CD38-cyclic ADP-ribose signaling system seems tobe involved in insulin secretion [34], and autoantibodiesto CD38 have been linked to the development of bothtype 1 and type 2 diabetes [35]. Finally, CD38 seems todefine a unique signaling pathway in CD8+ intestinallamina propria lymphocytes [9].

Rodents harbor molecules similar to CD38, but theirtissue distribution differs from that in humans [19, 30].Murine CD38 is expressed mainly by B lymphocytes andstem cells but to a lesser extent by other leukocyte sub-sets and not at all by plasma cells [30]. In rats CD38 hasbeen detected at the RNA or protein level in lung, liver,intestinal, and neural tissues [25, 26, 27].

Lymphatic vessels draining the gut were describedseveral hundred years ago and were called lacteals be-cause of their whitish (“milky”) appearance due to lipidsabsorbed from the intestinal lumen [3]. The central lym-phatic channels of small intestinal villi were also calledlacteals and are found in all vertebrates studied so far, in-cluding humans, rats, mice, rabbits, and dogs [1, 33, 37,40]. Lacteals merge to form a plexus near the muscularismucosae and in the submucosa. In organized gut-associ-ated lymphoid tissue (GALT) such as Peyer’s patches

I.N. Farstad (✉) · G. Haraldsen · P. BrandtzaegLaboratory for Immunohistochemistry and Immunopathology, Rikshospitalet, University of Oslo, 0027 Oslo, Norwaye-mail: i.n.farstad@labmed.uio.noTel.: +47-2-3071454, Fax: +47-2-3071511

F. MalavasiLaboratory of Transplant Immunology, Department of Genetics,Biology and Biochemistry, University of Torino, Turin, Italy

H.S. HuitfeldtSection for Toxicopathology, Institute of Pathology, Rikshospitalet, University of Oslo, 0027 Oslo, Norway

Virchows Arch (2002) 441:605–613DOI 10.1007/s00428-002-0679-9

O R I G I N A L A RT I C L E

Inger Nina Farstad · Fabio MalavasiGuttorm Haraldsen · Henrik Sverre HuitfeldtPer Brandtzaeg

CD38 is a marker of human lacteals

Received: 6 November 2001 / Accepted: 5 May 2002 / Published online: 31 July 2002© Springer-Verlag 2002

(PPs), draining lymphatics in the interfollicular zonesmerge to form perifollicular sinuses that eventually draininto submucosal plexuses and become mixed with lymphfrom the lacteals. Lymph that drains GALT structurescontains approximately ten times more lymphoid cellsthan that from the intestinal lamina propria [37].

We previously described efferent lymphatics in PPs[13] and found them to contain much more lymphoidcells than lacteals in adjacent villi, in agreement with da-ta from rats. More detailed phenotyping of lymphatics inthe human gut unexpectedly showed that lacteals ex-pressed CD38 in addition to CD31. We therefore includ-ed several human tissues and rodent intestinal mucosaein an extended study to substantiate lymphatic expres-sion of CD38.

Materials and methods

Tissue specimens

The human tissue material was obtained in accordance with Nor-wegian laws on the use of human tissues. We used biopsy or surgi-cal resection specimens of antrum and body gastric mucosa (n=4;two had Helicobacter pylori gastritis), small intestinal mucosa(n=14; two from treated celiac mucosae, one from a patient withtotal IgA deficiency), ileum with PPs (n=6), appendix (n=2), largeintestinal mucosa (n=4), mesenteric lymph nodes (MLN; n=3), na-sal mucosa (n=3), lung (n=3), liver (n=3), pancreas (n=4), thymus(n=3), palatine tonsil (n=4), spleen (n=1), and kidney (n=6). Nasalmucosa was control specimens obtained from healthy volunteersduring a study of nasal allergy, and tonsils were from patients un-dergoing resection due to recurrent tonsillitis. The PPs, MLNs,spleen, appendiceal and intestinal tissues were obtained eitherfrom organ donors or from biopsy specimens taken for diagnosticpurposes. Thymic tissue was obtained from children under 5 yearsof age undergoing cardiac surgery. Pancreatic specimens werefrom Whipple’s resections due to pancreatic carcinomas and lungtissue from lobectomies due to malignant tumours; in all casessampling was at least 1 cm away from the actual tumor. Kidneytissue was similarly obtained from two resections for renal carci-noma, and additionally from diagnostic biopsy specimens of fourtransplanted kidneys. Liver tissue was from biopsy specimens ofone transplanted patient with minimal rejection, one with slightunspecific hepatitis, and one with normal liver morphology. Un-less otherwise indicated, all samples revealed normal histology.

Tissue blocks (2–5 mm3) prepared from the fresh surgical re-section specimens and whole-biopsy specimens were within15 min oriented on a slice of carrot embedded in OCT compound(Tissue-Tek, Miles Laboratories, Elkhart, Ind., USA), snap-frozenin liquid nitrogen, and stored at –70° C until cryosectioning at4–6 µm. Sections were then fixed for 10 min in acetone at roomtemperature, wrapped in aluminum foil, and stored at –20° C. Inparallel, five of the human small intestinal biopsy specimens werefixed 1 h in 1% PLP (per iodate lysine paraformaldehyde) prior toOCT embedding, freezing in liquid nitrogen, and sectioning toprovide immunostaining with improved morphology. Animal tis-sue specimens consisting of the distal small intestine with PPsfrom conventionally reared BALB/c mice and from germ-free aswell as conventionally reared AGUS rats [21] were treated asabove. The animals were kept in accordance with the principles oflaboratory animal care (NIH publication no. 85-23, revised 1985).

Immunostaining, microscopical evaluation, and photography

Multicolor immunostaining was performed in three steps with dif-ferent anti-CD38 monoclonal antibodies (mAbs) combined withanti-CD31 or anti- lysosome-associated membrane protein(LAMP) 1 for 1 h (Table 1; all diluted in phosphate-buffered sa-line, PBS, with 1.25% w/v bovine serum albumin), mixtures ofsecondary reagents (pretitrated Cy3 or fluorescein isothiocyanate(FITC) conjugated goat anti-mouse IgG1, IgG2a, or IgG2b(Southern Biotechnology, Birmingham, Ala., USA) and rabbit an-ti-human von Willebrand factor, vWf; Table 1) for 1.5 h, and preti-trated amino-coumarine (AMCA) conjugated goat anti-rabbit IgG(Vector Laboratories, Burlingame, Calif., USA) for 30 min. Con-trols were irrelevant isotype-specific primary antibodies andFITC-conjugated goat IgG used at concentrations comparable tothose of the specific antibodies. For animal studies a rat anti-mouse CD38 or a polyclonal goat antibody recognizing both mu-rine and rat CD38 (Table 1) was combined with a rabbit anti-CD3recognizing the ε chain of the CD3/TCR complex in several ani-mals (Table 1), or rabbit anti-vWf as described above for 1 h, fol-lowed by AMCA-conjugated goat anti-rabbit IgG combined withpretitrated ALEXA 488-conjugated donkey anti-goat IgG (Molec-ular Probes, Eugene, Ore.,USA) for 1.5 h. Controls for animalstudies were sections incubated with 1.25% bovine serum albuminin PBS instead of primary antibody, followed by relevant second-ary reagents.

The specimens were examined in a Leitz DMRXE microscope(Leica, Wetzlar, Germany) equipped with filter blocks for observa-tion of red (Cy3), green (FITC and Alexa 488), blue (AMCA), andcombined red/green emissions. Pictures were obtained in a NikonEcLipse 800 fluorescence microscope equipped with a 3518charge-coupled device video camera and captured by Foto-Station

606

Table 1 Primary antibodies used for immunohistochemistry

Clone, designation Specificitya Isotypeb Working conc. Source(µg/ml or dilution)

A0452 CD3 Rabbit IgG Purified Ig; 1/20 Dako, Glostrup, DenmarkMoon-1 CD31 IgG1 Purified Ig; 1 F. MalavasiHB-7 CD38 IgG1 Purified Ig; 0.25 Becton-DickinsonIB4 CD38 IgG2a Purified Ig; 1 F. MalavasiOKT10 CD38 IgG1 Purified Ig; 1 ATCCIB6 CD38 IgG2b Purified Ig; 1 F. MalavasiH4A3 LAMP-1 IgG1 Ascitic fluid; 1/800 Developmental Studies Hybridoma

Bank, Iowa, USAA0082 Von Willebrand factor Rabbit IgG Purified Ig; 1/1600 DakoGoat anti-CD38 mouse/rat Mouse and rat CD38 Goat IgG Purified Ig; 20 Research Diagnostics, Flanders, N.J.,

USAClone 90 Mouse CD38 Rat IgG2a Purified Ig; 10 PharMingen, San Diego, Calif., USA

a Human specificity when not otherwise indicatedb Murine origin when not otherwise indicated

software (FotoWare, Høvik, Norway). Sections selected for confo-cal microscopy were examined in a Leica TCS confocal micro-scope (Leica Microsystems, Heidelberg, Germany) equipped withone argon and two HeNe lasers. A PL Apo 100X oil objective wasused. Pinhole was set to 2, and emissions from each fluorochromewere sequentially acquired. The image representing each colorwas an accumulation of eight scans.

From all specimens, parallel hematoxylin and eosin (H&E)stained sections were examined for morphological orientation.PLP-fixed specimens from human ileum were subjected to im-munoenzyme staining for anti-CD38 (HB-7) followed by the alka-line phosphatase–anti-alkaline phosphatase method [32].

In situ hybridization for CD38 mRNA

In situ detection of mRNA for CD38 was performed in four jeju-nal specimens. The full-length cDNA for CD38 (1.3 kb) [24] wassubcloned into pBS (Stratagene) in both orientations, linearized,and used as template for antisense and sense riboprobe synthesis.The probes were digoxigenin-labeled with the digoxigenin RNA-labeling kit according to the manufacturer’s directions (Boehrin-ger-Mannheim, Indianapolis, Ind., USA). All steps were per-formed at room temperature unless otherwise noted. Frozen sec-tions (8 µm) of OCT-embedded tissue were fixed in 4% parafor-maldehyde/diethylpyrocarbonate (DEPC) treated PBS (15 min)and subsequently washed twice (15 min each) in PBS containing0.1% active DEPC (Sigma). After 15 min-equilibration in 5× sodi-um saline citrate (SSC), sections were prehybridized (2 h, 59°C)in a solution of 50% formamide, 5× SSC, 50 µg/ml yeast tRNA,100 µg/ml heparin, 1× Denhardt solution, 0.1% Tween 20, 0.1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate, and5 mM EDTA (Sigma). The sections were subsequently hybridizedovernight at 59°C with 500 ng/ml riboprobe in hybridization solu-tion. A high-stringency wash was performed in the following se-quence: 2× SSC (30 min), 2× SSC (1 h, 65°C), and 0.1× SSC (1 h,65°C). For signal amplification, a horseradish peroxidase rabbitanti-digoxigenin antibody (Dako, Carpinteria, Calif., USA) wasused to catalyze the deposition of biotin-tyramide (GenPoint kit,Dako). Further amplification was achieved by adding horseradishperoxidase rabbit anti-biotin (Dako), biotin-tyramide, and then al-kaline-phosphatase rabbit anti-biotin (Dako). Signal was detectedwith the alkaline phosphatase substrate Fast Red TR/Napthol AS-MX (Sigma, St. Louis, Mo., USA), and sections were counter-stained with hematoxylin.

Results

General interpretation of lymphatics

With routine H&E staining all human tissues except forthe thymus showed thin-walled vessels deemed to belymphatics that contained occasional mononuclear cellsbut no erythrocytes. They were found adjacent to arteriesand veins in connective tissue septae except in the gas-trointestinal tract, organized lymphoid tissue, and kid-ney, in which they were interspersed between other tis-sue elements. Parallel immunostained sections were ex-amined for expression of CD38, vWf, and CD31, orCD38 and LAMP-1 (Fig. 1 and 2). The results obtainedare listed in Table 2. All organs were tested with the IB4and HB-7 mAbs to CD38 (Fig. 1), and positivity on lym-phatics was confirmed with mAbs OKT10 and IB6 (datanot shown). In all organs in which lymphatic vesselswere identified by H&E staining such vessels expressedCD31 albeit at a lower level than blood vessels which

were strongly positive for vWf (Fig. 2d, f, l). The lym-phatics were virtually negative for vWf.

Gastrointestinal tract

Immunofluorescence microscopy showed that all smallintestinal specimens contained vessels reacting with alltested CD38 mAbs, corresponding to the lacteals(Figs. 1a-c, 2b, c). In triple immunofluorescence stain-ings for CD38, CD31, and vWf, vessels correspondinganatomically to the location of lacteals and lymphaticplexuses in the muscularis mucosae and submucosa ex-pressed CD38 and CD31 but no vWF (Figs. 1a–c, 2b–d).The latter was always apparent on adjacent blood vesselsthat expressed no CD38 but stronger CD31 than the lym-phatics. CD38+ vessels showed a similar distribution asrecently described for LYVE-expressing lymphatics insmall intestinal mucosa [5]. Submucosal and muscularismucosae-related vessels expressed CD38 more stronglythan villous lacteals (Fig. 1c). The intensity of lactealCD38 was variable but approximately similar to that ofintraepithelial lymphocytes and weaker than that of adja-cent plasma cells. There was no apparent difference inCD38 staining intensity of lymphatics in coeliac or normalmucosa, or in the case with IgA deficiency (Fig. 2b-c).

In PPs and appendix, draining lymphatics showed faintstaining for CD38 compared to that of lacteals (Fig. 1c).No lymphatics in normal gastric mucosa, H. pylori gastri-tis, or normal large intestinal mucosa expressed CD38.

607

Fig. 1a–g Immunohistochemistry, confocal microscopy and insitu hybridization of CD38-expressing lacteals in human small in-testinal biopsy specimens. Examined markers are indicated in theirrespective colors on each photomicrograph. a, b Villi of normaljejunal mucosa immunostained with two different mAbs to CD38to visualize lacteal CD38 expression (horizontal arrows) as op-posed to adjacent vWf-positive blood vessels (vertical arrows).The scattered CD38-expressing cells in the lamina propria repres-ent mainly plasma cells. Original magnification ×400. c Alkalinephosphatase–anti-alkaline phosphatase staining for CD38 in PLP-fixed ileal biopsy specimen with adjacent Peyers’ patch (PP) folli-cle. Horizontal arrows Lacteals positive for CD38, also traversingthe muscularis mucosae (mm); vertical arrow lymphatic thatdrains PP shows weaker CD38 expression. Note that the brushborder of villus epithelium is positive because of endogeneous al-kaline phosphatase. Original magnification ×250. d, e Confocalmicroscopy of small intestinal lymphatic endothelial cells (L lu-men of lacteal). d Costaining for CD38 and CD31 in part of villuslacteal containing five endothelial cells shows coexpression (yel-low, horizontal arrow) of markers on the apical surface, althoughin some areas CD38 is expressed in the absence of CD31 (verticalarrow). Original magnification ×100, zoom ×4. e Costaining forCD38 and the lysosome-associated membrane protein (LAMP) re-veals that relatively few lysosomes are present within the lymphat-ic endothelial cells because they are mostly green. Horizontal ar-rows Small areas of CD38 and LAMP coexpression (yellow) intwo lymphatic endothelial cells; vertical arrow lysosomal com-partment in adjacent cell (probably a macrophage) without CD38.Original magnification ×100, zoom ×2. f In situ hybridization forCD38 mRNA shows two positive endothelial cells of submucosallymphatic (arrows); lower parts of epithelial crypts are seen at theright aspect. Asterisks Positive lymphoid cells. g Adjacent sectionhybridized with sense probe shows no positive cells. Originalmagnification ×400

▲

608

Fig. 1a–g Legend see page 607

609

Fig. 2 Legend see page 610

Tonsils, thymus, spleen, and MLNs

Lymphatics in tonsils did not express CD38. In the thy-mus, lymphatics were not identified by H&E staining,and no vessels with the CD38+CD31+vWf– phenotypewere observed. Splenic tissue did not contain CD31+

vWf– vessels and no vessel-like structures expressedCD38. In contrast, afferent lymphatics of MLNs often

expressed CD38 (Fig. 2e, h), approximately similar tothe staining intensity of GALT lymphatics.

Nasal mucosa, lung, liver, pancreas, and kidney

Subepithelial lymphatics in the nasal mucosa were nega-tive for CD38, as were lymphatics in connective tissueseptae of the lung and pancreas and in portal tracts of theliver (shown for the pancreas in Fig. 2j–k). In contrast,liver sinusoidal endothelial cells expressed low levels ofCD38 but were negative for vWf and only weakly posi-tive or negative for CD31; stronger sinusoidal CD38 ex-pression was seen in the hepatitis specimen. The kidneysamples contained only few lymphatics localized in themedulla, and they were negative for CD38.

Rodent intestinal mucosa

Frozen sections of murine and rat ileum with PPs wereexamined for CD38 expression (Table 1). In the mouseCD38 was evident on B lymphocytes of PP follicles butnot on lamina propria cells (data not shown). Lactealsand lymphatics draining PPs, as identified by vesselscontaining occasional CD3+ cells in adjacent sectionsand by lack of vWf staining that was otherwise evidentin vascular endothelium, were negative for CD38 (Ta-ble 2). This staining pattern was found both with themonoclonal and the polyclonal antibodies.

The distribution of CD38 in rats was different fromthat in the mouse. In addition to positive cells corre-sponding to PP follicles, also intraepithelial lymphocytesand some lamina propria cells expressed CD38 (data notshown). Lacteals and lymphatics draining PPs were iden-

610

Fig. 2 Histological and immunohistochemical demonstration oflymphatics in surgical resection samples of small intestine frompatient with total IgA deficiency (a–d), mesenteric lymph node(e–h), and pancreas (i–l). Color codes for investigated markers(lower left) as well as symbols (lower right) are indicated. a H&Estaining reveals villous mucosa with mononuclear cells in the epi-thelium and lamina propria. Thin-walled lymphatics (lacteals)containing occasional mononuclear cells are indicated (L). b Tri-ple immunofluorescence staining for CD38, CD31, and vWf. LLacteals (yellow to green) expressing CD38 and weak CD31 butno visible vWf. In contrast, blood vessels are pink because theyexpress both vWf and CD31 but no CD38. c, d Single exposuresof lacteals and blood vessels (B) shown in b. Note the differencein CD31 expression intensity by lacteals (weak) and blood vessels(strong). e, f Single exposures of lymph node cortex area contain-ing afferent lymphatics and blood vessels as shown by H&E stain-ing in g. The lymph vessels show very faint CD38 staining andrelatively weak CD31 staining; it also contains CD38+ mononu-clear cells. h Triple immunofluorescence staining of the same areaas in g shows, again, that lymph vessels (red to orange) express novWf as opposed to blood vessels that express vWf and CD31 butno CD38 and therefore appear pink. Asterisks Subcapsular sinus ing, h. i H&E staining of pancreas, showing blood vessels and lym-phatics. j Triple color exposure showing CD31 expression on bothvessels types; blood vessels are also distinctly positive for vWfwhereas this protein is hardly detectable on lymphatics. k, l Singleexposures of CD38 and CD31; the lymph vessels are negative forCD38 whereas the walls of blood vessels containing smooth mus-cle and connective tissue shows background greenish fluorescencealso seen in j. Original magnifications a–h ×250; i–l ×400

▲

Table 2 Distribution of lym-phatic endothelium positive forCD38, von Willebrand factor(vWF), or CD31 (n.d. not deter-mined)

Tissue material No. of samples Lymphatic vessel CD38 vWF CD31identified by H&E staining

HumanGastric mucosa 4 +a – –b +Small intestinal mucosa 20 + + – +Large intestinal mucosa 4 +a – – +Peyer’s patches 6 + +c – +

Appendix 2 + +c – +Mesenteric lymph nodes 3 + +c – +Tonsil 4 + – – +Thymus 3 –Liver 3 +d –e – +Pancreas 4 + – – +Kidney 6 +a – – +Nasal mucosa 3 + – – +Lung 3 + – – +Spleen 1 –

RodentBALB/c mouse ileum 1 + – – n. d.with Peyer’s patchesAGUS rat ileum 2 + – – n. d.with Peyer’s patches

a Lymphatics difficult to identi-fy; only in submucosa of gas-tric and colonic samples; onlyvery few in kidney medullab Lymphatics in all organs vir-tually negative; occasional ves-sels showing very faint stainingc Staining for CD38 weakerthan in lactealsd Lymphatics in portal tractseLiver sinusoidal endotheliumpositive only

tified as described for the mouse and were likewise neg-ative for CD38 (Table 2). No difference in CD38 expres-sion patterns was detected between the germ-free andconventionally reared animals.

Confocal microscopy

Combined staining for CD38 and CD31 revealed coex-pression of the molecules mainly at the periphery of theendothelial cells (Fig. 1d). With costaining for CD38 anda protein associated with lysosomes (LAMP-1; Table 1),lacteal endothelial cells were found to contain relativelyfew lysosomes compared with adjacent cells such asmacrophages and enterocytes, and there was little colo-calization of the two molecules (Fig. 1e).

In situ hybridization

Examination of CD38 mRNA was performed on cryo-sections from two jejunal biopsy specimens and twoWhipple resection specimens. A protocol yielding virtu-ally no background resulted in occasional distinctly posi-tive signals in some submucosal lymphatics (Fig. 1f) butno convincingly positive villous lacteals (not shown).Lymphoid cells found in the lumina of these vesselswere sometimes positive for CD38 mRNA (data notshown). Adjacent plasma cells that are known to expressCD38 strongly [12], served as a positive control, andalso occasional intraepithelial lymphocytes revealedCD38 mRNA (not shown).

Discussion

Here we show that CD38 is a novel marker of humanlacteals and GALT-draining lymphatics. Furthermore,CD38 expression appeared to be unique for these lym-phatic vessels. Our immunohistochemical result wasbased on mAbs recognizing different epitopes of CD38[4, 22]. This finding represents a new contribution to thebiology of CD38, which, in addition to its reported ex-pression on leukocyte subsets, neural cells, prostate epi-thelial cells, pancreatic β-cells, osteoclasts, and retinaland muscle cells [8], apparently also can be expressed byendothelial cells. In contrast, rodent lacteals were nega-tive for CD38. In the rat both airway lining epitheliumand hepatocytes have been reported to express CD38[25, 26], whereas in the human liver we found CD38 on-ly in sinusoidal endothelial lining cells. Intestinal distri-bution of CD38 in mice and rats is not well known, butthe present study confirms that differences exist betweenhumans and rodents.

In our hands, the identified lymphatics were virtuallynegative for factor VIII related antigen/vWf. This was inagreement with the results of Banerji et al. [5] who re-cently described the lymph-specific hyaluronan receptorLYVE-1 and found the LYVE-1 positive vessels to be

negative for vWf. Lymphatics have been reported to ex-press this marker [18, 28], but we have found it to bevery weak and most likely negative by our in situ immu-nohistochemical detection. Other recently identified mol-ecules such as podoplanin and vascular endothelialgrowth factor receptor 3 apparently recognize most lym-phatic endothelia [28] although their distrubution in thesmall intestine has not been described.

It is theoretically possible that CD38 is released bythe numerous CD38-expressing cells present in the lam-ina propria, especially plasma cells, and then endocyto-sed by the lymphatic endothelium. CD38 can exist as asoluble molecule, probably released from cell-boundCD38 by proteolytic cleavage [16], and lymphocyteshave been shown to endocytose such CD38 [17]. Thispossibility seems unlikely because lacteal endothelialcells contained relatively few lysosomes as judged byexpression of the lysosomal protein LAMP-1, and therewas little colocalization with CD38 (Fig. 1e). LAMP-1is expressed by late endosomes and lysosomes [2], andcoexpression with CD38 might have suggested that lym-phatic CD38 resulted from endocytosis of solubleCD38. In preliminary experiments we were able to iso-late CD38-expressing endothelial-like cells from sam-ples of human small intestinal mucosa, using a collage-nase-dispase mixture primarily designed to isolate mi-crovascular intestinal endothelial cells [20], followed byparamagnetic beads armed with the non-activatingCD38 mAb IB6 (Farstad et al., unpublished observa-tions). This result indicated that CD38 is expressed onthe surface of the intestinal lymphatic endothelial cells,in agreement with our observations based on confocalmicroscopy (Fig. 1d). However, attempts to expandthese putative lacteal-derived endothelial cells in culturehave failed.

Based on the protein expression pattern of CD38 inthe small intestinal mucosa, it was surprising that lac-teals revealed less CD38 mRNA than adjacent plasmacells and intraepithelial lymphocytes. The basis for thisdiscrepancy remains elusive. It is possible that lactealCD38 is not identical to that expressed by lymphoidcells, or that the level of CD38 synthesis is lower in lac-teals. However, this result was in agreement with immu-nohistochemical data that CD38 seemed to be morestrongly expressed by lymphatics near the muscularismucosae and submucosa, although we have no explana-tion to this finding.

Because lymphatics in no other investigated tissuesexpressed CD38, the lacteal positivity might be relatedto the absorptive function of the small intestine. Interest-ingly, vitamin D3 has been shown to upregulate CD38 onlymphocytes [38], and it is well known that a metaboliteof vitamin A, all-trans-retinoic acid, induces CD38 ex-pression in promyelocytic leukemia [11]. These fat-solu-ble vitamins are absorbed from the gut lumen and coex-ist in chylomicrons within the lacteals where they mightinfluence lacteal CD38 expression. By immunofluores-cence microscopy, vitamin D receptor was detected inenterocyte nuclei but not in those of lacteal endothelial

611

cells (Farstad et al., unpublished observations); however,this finding does not exclude an effect of fat-soluble vi-tamins on lacteals.

The function of CD38 on lacteals is an open question.Three of the mAbs (IB4, HB-7, OKT10) used in thisstudy detect the carboxy-terminal end of CD38 whichconfers ecto-NADase activity [22], and it is thereforepossible that lacteal CD38 has enzymatic activity. IB4and OKT10 also mediate cell signaling [4], but thisseems to depend on the association of CD38 with othersurface molecules, such as the T-cell receptor/CD3 com-plex on T cells and the B-cell receptor complex on Bcells [8]. Whether lacteals harbor molecules that couldserve this function, remains to be shown. The coexpres-sion of CD38 and its ligand CD31 by lacteals (Fig. 2b–d)is a feature shared with other lymphoid cells [39], but thefunctional implication of this is unknown.

In conclusion, we have shown that four different anti-bodies to CD38 react with human but not rodent lacteals.CD38 can thus be used to identify these structures in hu-man small intestinal mucosa.

Acknowledgements Audun Berstad, Frode Jahnsen, and LarsHelgeland, Laboratory for Immunohistochemistry and Immunopa-thology (LIIPAT), are gratefully acknowledged for providing gas-tric and nasal mucosae, and rat intestinal mucosa, respectively.Ståle Sund, Department of Pathology, is gratefully thanked forproviding kidney biopsies. David G. Jackson, Oxford, UK, isgratefully thanked for providing the cDNA for CD38, and Finn-Eirik Johansen, LIIPAT, for generating the riboprobes. We thankthe Departments of Surgery and Medicine, Rikshospitalet, for pro-viding other tissue samples.

References

1. Abbas B, Hayes TL, Wilson DJ, Carr KE (1989) Internalstructure of the intestinal villus: morphological and morpho-metric observations at different levels of the mouse villus.J Anat 162:263–273

2. Akasaki K, Michihara A, Mibuka K, Fujiwara Y, Tsuji H(1995) Biosynthetic transport of a major lysosomal membraneglycoprotein, lamp-1: convergence of biosynthetic and en-docytic pathways occurs at three distinctive points. Exp CellRes 220:464–473

3. Aselli, G (1969) The lacteals. JAMA 209:7674. Ausiello CM, Urbani F, Lande R, la Sala A, Di Carlo B,

Baj G, Surico N, Hilgers J, Deaglio S, Funaro A, Malavasi F(2000) Functional topography of discrete domains of humanCD38. Tissue Antigens 56:539–547

5. Banerji S, Ni J, Wang S-X, Clasper S, Su J, Tammi R,Jones M, Jackson DG (1999) LYVE-1, a new homologue ofthe CD44 glycoprotein, is a lymph-specific receptor for hyal-uronan. J Cell Biol 144:789–801

6. Damle RN, Wasil T, Fais F, Ghiotto F, Valetto A, Allen SL,Buchbinder A, Budman D, Dittmar K, Kolitz J, Lichtman SM,Schulman P, Vinciguerra VP, Rai KR, Ferrarini M, ChiorazziN (1999) Ig V gene mutation status and CD38 expression asnovel prognostic indicators in chronic lymphocytic leukemia.Blood 94:1840–1847

7. Deaglio S, Morra M, Mallone R, Ausiello CM, Prager E,Garbarino G, Dianzani U, Stockinger H, Malavasi F (1998)Human CD38 (ADP-ribosyl-cyclase) is a counter-receptor forCD31, an Ig superfamily member. J Immunol 160:305–402

8. Deaglio S, Mehta K, Malavasi F (2001) Human CD38: a(r)evolutionary story of enzymes and receptors. Leuk Res25:1–12

9. Deaglio S, Mallone R, Baj G, Donati D, Giraudo G, Corno F,Bruzzone S, Geuna M, Ausiello C, Malavasi F (2001) HumanCD38 and its ligand CD31 define a unique lamina propria Tlymphocyte signaling pathway. FASEB J 15:1–4

10. Dianzani U, Funaro A, DiFranco D, Garbarino G, BragardoM, Redoglia V, Buonfiglio D, De Monte LB, Pileri A, MalavasiF (1994) Interaction between endothelium and CD4+CD45RA+lymphocytes. Role of the human CD38 molecule. J Immunol153:952–959

11. Drach J, Zhao S, Malavasi F, Mehta K (1993) Rapid inductionof CD38 antigen on myeloid leukemia cells by all-trans-reti-noic acid. Biochem Biophys Res Commun 195:545–550

12. Farstad IN, Halstensen TS, Lazarovits AI, Norstein J, FausaO, Brandtzaeg P (1995) Human intestinal B-cell blasts andplasma cells express the mucosal receptor integrin α4β7.Scand J Immunol 42:662–672

13. Farstad IN, Norstein J, Brandtzaeg P (1997) Phenotypes of Band T cells in human intestinal and mesenteric lymph. Gastro-enterology 112:163–173

14. Ferrero E, Saccucci F, Malavasi F (1999) The human CD38gene: polymorphism, CpG island, and linkage to the CD157(BST-1) gene. Immunogenetics 49:597–604

15. Funaro A, Alessio M, Roggero S, Funaro A, De Monte LB,Peruzzi L, Geuna M, Malavasi F (1990) CD38 molecule:structural and biochemical analysis on human T lymphocytes,thymocytes, and plasma cells. J Immunol 145:878–884

16. Funaro A, Horenstein AL, Calosso L, Morra M, Tarocco RP,Franco L, De Flora A, Malavasi F (1996) Identification and characterization of an active soluble form of human CD38in normal and pathological fluids. Int Immunol 8:1643–1650

17. Funaro A, Reinis M, Trubiani O, Santi S, Di Primio R, Malavasi F (1998) CD38 functions are regulated through aninternalization step. J Immunol 160:2238–2247

18. Gnepp DR (1987) Vascular endothelial markers of the humanthoracic duct and lacteal. Lymphology 20:36–43

19. Harada N, Santos-Argumedo L, Chang R, Grimaldi JC,Lund FE, Brannan CI, Copeland NG, Jenkins NA, Heath AW,Parkhouse RME, Howard M (1993) Expression cloning of acDNA encoding a novel murine B cell activation marker. Ho-mology to human CD38. J Immunol 151:3111–3118

20. Haraldsen G, Rugtveit J, Kvale D, Scholz T, Muller WA, Hovig T, Brandtzaeg P (1995) Isolation and longterm cultureof human intestinal microvascular endothelial cells. Gut37:225–234

21. Helgeland L, Vaage JT, Rolstad B, Midtvedt T, Brandtzaeg P(1996) Microbial colonization influences composition and T-cell receptor Vβ repertoire of intraepithelial lymphocytes inrat intestine. Immunology 89:494–501

22. Hoshino S, Kukimoto I, Kontani K, Inoue S, Kanda Y, Malavasi F, Katada T (1997) Mapping of the catalytic and epi-topic sites of human CD38/NAD+ glycohydrolase to a func-tional domain in the carboxyl terminus. J Immunol 158:741–747

23. Howard M, Grimaldi JC, Bazan JF, Lund FE, Santos-Argumedo L, Parkhouse RM, Walseth TF, Lee HC (1993) For-mation and hydrolysis of cyclic ADP-ribose catalyzed by lym-phocyte antigen CD38. Science 262:1056–1059

24. Jackson DG, Bell JI (1990) Isolation of a cDNA encoding thehuman CD38 (T10) molecule, a cell surface glycoprotein withan unusual discontinuous pattern of expression during lym-phocyte differentiation. J Immunol 144:2811–2815

25. Khoo KM, Chang CF (1998) Purification and characterizationof CD38/ADP-ribosyl cyclase from rat lung. Biochem MolBiol Int 44:841–850

26. Khoo KM, Chang CF (2000) Localization of plasma mem-brane CD38 is domain specific in rat hepatocyte. Arch Bio-chem Biophys 373:35–43

27. Koguma T, Takasawa S, Tohgo A, Karasawa T, Furuya Y,Yonekura H, Okamoto (1994) Cloning and characterization ofcDNA encoding rat ADP-ribosyl cyclase/cyclic ADP-ribosehydrolase (homologue to human CD38) from islets of Langer-hans. Biochim Biophys Acta 1223:160–162

612

28. Kriehuber E, Breiteneder-Geleff S, Groeger M, Soleiman A,Schoppmann SF, Stingl G, Kerjaschki D, Maurer D (2001)Isolation and characterization of dermal lymphatic and bloodendothelial cells reveal stable and functionally expressed lin-eages. J Exp Med 194:797–808

29. Lund EF, Yu N, Kim KM, Reth M, Howard MC (1996) Sig-naling through CD38 augments B cell antigen receptor (BCR)responses and is dependent on BCR expression. J Immunol157:1455–1467

30. Lund EF, Cockayne DA, Randall TD, Solvason N, Schuber F,Howard MC (1998) CD38: a new paradigm in lymphocyte ac-tivation and signal transduction. Immunol Rev 161:79–93

31. Malavasi F, Funaro A, Roggero S, Horenstein A, Calosso L,Mehta K (1994) Human CD38: a glycoprotein in search of afunction. Immunol Today 15:95–97

32. Mason DY (1985) Immunocytochemical labeling of monoclo-nal antibodies by the APAAP immunoalkaline phosphatasetechnique. In: Bullock GR, Petrusz P (eds) Techniques in im-munocytochemistry, vol 3. Academic, London, pp 25–42

33. Ohtani O, Ohtsuka A (1985) Three-dimensional organizationof lymphatics and their relationship to blood vessels in rabbitsmall intestine. A scanning electron microscopic study of cor-rosion casts. Arch Histol Jpn 48:255–268

34. Okamoto H (1999) The CD38-cyclic ADP-ribose signalingsystem in insulin secretion. Mol Cell Biochem 193:115–118

35. Pupilli C, Giannini S, Marchetti P, Lupi R, Antonelli A, Malavasi F, Takasawa S, Okamoto H, Ferrannini E (1999) Au-toantibodies to CD38 (ADP-ribosyl cyclase/cyclic ADP-ribosehydrolase) in Caucasian patients with diabetes: effects on insu-lin release from human islets. Diabetes 48:2309–2315

36. Savarino A, Bottarel F, Malavasi F, Dianzani U (2000) Role ofCD38 in HIV-1 infection: an epiphenomenon of T-cell activa-tion or an active player in virus/host interactions? AIDS14:1079–1089

37. Steer HW (1980) An analysis of the lymphocyte content of ratlacteals. J Immunol 4:1845–1858

38. Stoeckler JD, Stoeckler HA, Kouttab N, Maizel AL (1996)1α,25-Dihydroxyvitamin D3 modulates CD38 expression onhuman lymphocytes. J Immunol 157:4908–4917

39. Vallario A, Chilosi M, Adami F, Montagna L, Deaglio S, Malavasi F, Caligaris-Cappio F (1999) Human myeloma cellsexpress the CD38 ligand CD31. Br J Haematol 105:441–444

40. Yamanaka Y, Araki K, Ogata T (1995) Three-dimensional or-ganization of lymphatics in the dog small intestine: a scanningelectron microscopic study on corrosion casts. Arch Histol Cy-tol 58:465–474

613