Reprints Available Directly from the Publisher Photocopying Permitted by License Only a Comparative Analysis of the Murine Thymic Microenvironment in Normal, Autoimmune, and Immunodeficiency States

It is widely accepted that the thymic microenvironment regulates normal thymopoiesis through a highly coordinated and complex series of cellular and cytokine interactions. A direct corollary of this is that abnormalities within the microenvironment could be of etio-logic significance in T-cell-based diseases. Our laboratory has developed a large panel of monoclonal antibodies (mAbs) that react specifically with epithelial or nonepithelial markers in the thymus. We have taken advantage of these reagents to characterize the thymic microenvironment of several genetic strains of mice, including BALB/cJ, ALY/NscJcl-aly/aly mice, and littermate control animals. We report herein that control mice, including strains of several backgrounds, have a very consistent phenotypic profile with this panel of monoclonal antibodies, including reactivity with thymic epithelial cells in the cortex, the medulla and the corticomedullary junction, and the extracellular matrix. In contrast, the disease-prone strains studied have unique, abnormal staining of thymic cortex and medulla at both the structural and cellular levels. These phenotypic data suggest that abnormalities in interactions between developing thymocytes and stromal cells characterize disease-prone mice.


t, the specia
ized nature of which de- rives from both the diverse cellular content and the complex structural organization (Boyd et al., 1993).Recently, interest in the nature and function of the thymic microenvironment has intensified with the real- ization that defects in it may underlie abnormal T-cell development and hence diseases such as autoimmunity and immunodeficiency.The most important component of the microenvironment is the epithelial cells that de- fine three major areas: the subcapsule, cortex, and medulla.Epitheli l cells in the subcapsular region do not express MHC class II, but are believed to initiate thymopoiesis (Van de Wijngert et al., 1983; Boyd et al.,  1992, 1993).Thymocytes presumably then contact un- derlying cortical epithelial cells including thymic nurse cells that express both MHC class and II molecules and most likely have an important function in positive selection (Berg et al., 1989; Ashton-Rickardt et al.,   1993).Phenotypically mature single-positive T cells (CD4 + CD8-or CD4-CD8 + are concentrated in the medulla, suggesting that this region completes thymocyte-positive selection (Ritter and Boyd, 1993); however, it may have been initiated earlier but was sim- ply manifested in the medulla.In fact, medullary thymic epithelial cells induce tolerance toward class Irestricted self-peptides presented on their own surface (Oukka et al., 1996).

Clearly, the thymic microenvironment, in particular epithelial cells, are indispensable for T-cell maturation.The logical corollary to this is that defects in the cellu- lar content or spatial organization of the thymic stroma may be of etiologic significance in so e diseases.In this context, we previously described thymic stromal abnormalities in several murine models of human systemic lupus erythematosus (SLE), including New Zealand Black (NZB), MRL/MP-Faslpr, C3H/HeJ-Fas- gta, and BXSB/MpJ.Yaa mice (Watanabe et al., 1993;   Takeoka et al., 1995a, 1995b).In addition, there are several other genetically determined murine models for immunodeficiency and autoimmune diseases.These include SM/J, NOD, ALY/Nsc Jcl-aly/aly, and mothe- aten (Hpchme/Hpchme) mice.SM/J mice have high lev- els of natural thymocytotoxic autoantibodies (NTA) found commonly in murine lupus strains (Mittal et al.,  1970; Terasaki et al., 1970; Shirai and Mellors, 1971;   Huang et al., 1973; Goldblum et al., 1975; Eisenberg et  al., 1979) yet surprisingly do not manifest severe path- ogenic changes.NOD mice are a well-known model of human insulin-dependent diabetes mellitus (IDDM), an autoimmune disease targeting the insulin-secreting cells in pancreatic islets (Tochino, 1987).The "mothe- aten" autosomal recessive mutation disrupts the struc- tural gene for a cytoplasmic protein tyrosine phos- phatase called hematopoetic cell phosphatase (Shultz et  al., 1993).Mice homozygous for motheaten (npchme/npchme) manifest immunodeficiency' and au- toimmunity and exhibit severe defects in the development and function of both T and B lymphocytes Green and Shultz, 1975).Finally, the "alymphoplasia" mouse- termed ALY/Nsc Jcl-aly/aly from the gene symbol "aly" is a spontaneous autosomal recessive mutant that causes a systemic absence of lymph nodes and Peyer's patches, and reduced levels of IgM and severely de- pressed levels of IgG and IgA (Miyawaki et al., 1994).Because we have shown in earlier studies that lupusprone mice have characteristic defects in the thymic microenvironment (Watanabe et al., 1993; Takeoka et  al., 1995a, 1995b), the present study was undertaken to determine whether such abnormalities are generic to Tcell disorders in general.


RESULTS


Classification of the mAbs

Fourteen mAbs used in this study have been classed into three categories, based on their reactivity pattern with normal thymic tissue: (1) those reactive with thymic epithelial cells, (2) those reactive with both thymicepithe lia

nd me
ullary thymocytes by MTS37.The extracellular matrix stained with MTS 16 was decreased in NOD mice, again similar to NZB mice (Fig. 5).Finally, staining with MTS12, Mac-1, or MTS28 was normal.

Thymic Architecture of NOD-scid Mice NOD-scid mice had a severe involution of the thymus, as expected, with no clearly defined cortex, only rudi- mentary medullary pockets and no distinct corti- comedullary regions.Furthermore, both the cortical (MTS44-positive) and medullary (MTS 10-positive) ep- ithelial cells where present were irregularly shaped but NOD-scid, (f) Hpch /Hpchm', or (g) aly/aly mice; M: medulla, C: cortex, S: subcapsule.(a) C57BL/6 mice show normal expression.(b) NZB mice were similar to normal mice.However, (c) SM/J and (f) ttpchm'7ttpch mice had elevated MTS35-positive thymocytes in their corticomedullary and medullary regions, and aly/aly mice had elevated MTS35-positive thymocytes in their cor icomedullary regions.Arrows indicate MTS35-positive thymocytes in the medulla or corticomedullary regions.The bar indicates 250 lam.100) glGURE 5 MTS16 staining of thymus from (a) C57BL/6, (b) NZB, (c) SM/J, (d) NOD, (e) NOD-scid, (f) Hpch /Hpch or (g) aly/aly mice.Extracellular matrix, as recognized by MTS16, was markedly increased in (e) NOD-scid or (g) aly/aly mice.The extracellular matrix was aggregated and expressed large irregular-shaped clusters in (g) aly/aly mice, but (e) NOD-scid mice exhibited small clusters.The thymus of (b) NZB or (d) NOD mice had a mild reduction of MTS16 expression.Arrows indicate abnormalities of MTS16 stain- ing in the thymus ofaly/aly or NOD-scid mice.The bar indicates 250 !am.


100)

quite different from NZB mice (Figs. 2 and 3).These staining patterns were confirmed with the pan-epithelium markers MTS1 and MTS5.Consistent with their immature status, the thymocytes were MTS33-and MTS35-positive (Fig. 4), and although very infrequent, there were isolated MTS33-and MTS35-positive epithelial cells in the medullary pockets.The density of extracellular matrix, recognized by MTS16, was in- creased (Fig. 5).Although less extensive because of the small size of the thymus, the staining of macrophage-like cells by Mac-1 or MTS28, or vascular endothelium by

S12 was sim
lar to control NOD-+ /+, C57BL/6, or BALB/c mice.

Thymic Architecture of Motheaten me me

(Hcph /Hcph Mice

The thymuses of Hcph 7Hcph mice were smaller than the aged-matched controls, the moderate involu- tion including both the cortex and medulla the epithelial networks (MTS44and MTS 10-positive, respectively), however, was similar to the control strains (Figs. 2 and 3).Hcphme/Hcph me mice also demonstrated elevated MTS35-positive thymocytes in the corticomedullary and medullary regions (Fig. 4).

Staining with MTS16 (Fig. 5), MTS12, Mac-l, or MTS28 was normal as compared with +/Hcphme-het erozygous, C57BL/6, or BALB/c mice.

Thymic Architecture of aly/aly Mice Mice homozygous for the alymphoid (aly) mutations (aly/aly) exhibited marked alterations in their thymic architecture.The most dramatic and distinguishing fea- ture was the presence of epithelial cysts identified by hematoxylin and eosin staining, MTS 1, or anti-keratin antibody; these were at the corticomedullary junction or medulla.The cortical epithelial network was severe- ly disrupted and irregular when stained with MTS1, MTS5, or MTS44 (Fig 2), with many areas actually lacking MTS44-positive epithelial cells.In addition, isolated clusters of MTS 10 staining were found in the cortex.The medullary epithelium or aly/aly mice was also severely disrupted, as shown by staining with MTS1 or MTS10 (Fig. 3), with fewer of the clusters positive for MTS33, MTS20, or MTS24 (Fig. 3).The aly/aly mice had elevated MTS35-positive thymocytes around the corticomedullary junction (Fig. 4), whereas they are normally in the outer cortex.However, staining of thymocytes with MTS33 or MTS37 was similar to +/aly-heterozygous, C57BL/6, or BALB/c mice.

There was marked increase in expression of the extra- cellular matrix recognized by MTS16, the staining being more aggregated and irregularly shaped.Thymic staining with MTS 12, Mac-1, or MTS28 was essential- ly normal.


DISCUSSION

One of the striking features of the thymus is the phylogenetically conserved cellular content and structural or- ganization of the stromal components.These constitute the specific microenvironment of this organ and or- chestrate the induction and regulation of thymopoiesis in such a manner that pathological states rarely arise.The corollary of this is that defects in the microenvi- ronment at the structural and/or cellular levels could predispose the individual to T-cell-based disorder such as autoimmunity or immunodeficiency.In this regard, we have previously shown that the lupus-prone NZB mice and related strains are characterized by abnormal- ities in the thymic microenvironment manifest as ec- topic expression of medullary epithelial antigens in the cortex and the marked expansion of lymphocyte-rich areas lacking epithelial cells.The abnormalities have been already occurring since month before disease and become worse with aging (Watanabe et al., 1993).

The present study was undertaken to determine if these thymic microenvironment abnormalities are generic to other autoimmune and also immunodeficient mice.

Collectively, the data generated in this study are con- sistent with defects in thymic architecture in several of the murine strains evaluated.Because these are manifest before overt signs of disease, they do not develop as a consequence of disease and may thus be of etiologic significance.The changes were specifi for the disease- prone mice and were never observed in a panel of nor- mal mice strains (e.g., BALB/c, CBA, and C57BL/6).SM/J (thymotoxic autoantibodies) and NOD mice had extensive epithelial-cell-free regions and disrupted cor- tical epithelial networks similar to those of NZB and other models of murine lupus (Watanabe et al., 1993;  Takeoka et al., 1995a; 1995b).These were also ob- served in one of the immunodeficient models, aly/aly mice.However, the thymic abnormalities of NOD-scid mice--sever thymic involution, poorly defined cortex, and rudimentary medulla increased extracellular matrix are more typical of aged mice such as (Takeoka et al.,  1996) than murine lupus.These changes are unlikely to be related to any disease processes, but rather a conse- quence of the lack of T cells, particularly mature T cells, which are required for epithelial-cell maturation and ex- pansion.The thymus in motheaten (Opchme/opchme) mice also showed some atrophy, but the stromal archi- tecture was relatively normal.

Given the importance of cortical epithelium in thymocyte differentiation and positive and negative selection (Berg et al., 1989; Ashton-Rickardt et al., 1993; Boyd et  al., 1993), any abnormalities in this region in particular are very likely to be translated into defective T-cell de- velopment and hence predisposition to disease.The role of the medullary epithelium is less clear, but it presumably has a role in the final maturation of thymocytes, in- cluding their capacity to migrate to the periphery; it may also induce tolerance through energy.A possibility in many of the models (e.g.; NZB, SM/J, and NOD), there- fore, is that the presence of medullary epithelium in the cortex delivers premature maturation signals to sur- rounding lymphoc tes, expediting their progression to functionally competent cells, thereby escaping normal tolerance induction mechanisms.Because these alter- ations were not found in npchme/Hpchme, they are generic to all autoimmunity-prone mice.

The significance of the ECF is still unclear.By defi- nition, they lack epithelial cells, but they have a minor population of supporting stromal cells and are filled with thymocytes of multiple phenotypes, including di- viding cells (Boyd et al., unpublished observations).Our working hypothesis is that these thymocytes accu- mulate through lack of normal deletion mechanisms, creating space through the elasticity of the stromal ar- chitecture.They could, therefore, harbor potentially au- toreactive cells.

The extracellular matrix, which is composed of a mixture of collagen, reticulin fibers, glycosaminogly- cans, and glycoproteins surr,ounding the epithelial cells, is also important to thymic integrity and thymocyte de- velopment, including mediating cell migration and binding of soluble cytokines (Lannes-Vieira et al.,  1991; Boyd et al., 1993; Savino et al., 1993).It has been reported that 12-month-old normal mice exhibit a denser and irregular-shaped extracellular matrix pattern than young adult normal mice (Lannes-Vieira et al.,  1991; Takeoka et al., in press).Our data demonstrate that NOD-scid or aly/aly mice exhibited increased number and density of extracellular matrix.Further, the extracellular matrix was aggregated and expressed a large irregular shape in the thymus of aly/aly mice.The abnormalities of the extracellular matrix in thymus of NOD-scid or aly/aly mice were similar to age-related degenerative changes in normal mice, and may be a factor in immunosenescence due to involution of thymus.On the other

atrix
was decreased in NZB and NOD mice as compared to C57BL/6, BALB/c, SM/J, or heterozygous controls.This may again disrupt T normal thymopoiesis through reduced cell migration and binding of soluble cytokines.

SM/J mice possess some thymic epithelial abnormal- ities common to murine SLE, but.one important differ- ence was found by staining with MTS35.This mAb recognizes the thymic shared antigen-1 (TSA-1), a marker of immature thymocytes having an inverse rela- tionship with CD3 expression (Randle et al., 1993), and isolated thymic stromal cells, which include dendritic cells and a medullary epithelial-cell subset.It is, how- ever, upregulated on activated mature T cells (Kosugi et al., 1994).This accumulation of TSA-l-positive thy- mocytes may be related to abnormalities in the thymic epithelium and could either be due to disruption of the cortical network, allowing release of immature thymocytes, or activation of a select subset in the medulla.In fact, SM/J mice have a high level of naturall

occurring thymocytotxic autoa
tibody (NTA) that may contribute to either of these possibilities (Mittal et al., 1970).Motheaten (Hpchme/Hpchme) mice did not have marked abnormalities in the stroma, but did have the elevated MTS35-positive cells at the corticomedullary junction.The relationship between this and the high levels of IgM, which are 25 to 50 times greater than that of con- trol mice (Shultz and Green, 1976), is unknown.

Collectively, these data clearly demonstrate that pathological st

es such as auto
mmunity and immun- odeficienc are associated with marked abnormalities in the thymic microenvironment.Although a direct causal link between these two has yet to be verified, this study underlies the importance of understanding the nature and function of the thymic microenviron- ment as a prelude to determining the etiology of sever- al T-cell-based disorders.We are currently pursuing the identification and gene cloning of the antigens invest - gated in this study.


MATERIAL AND METHODS


Mice

Female NZB/BlnJ, C57BL/6J, BALB/cJ, SM/J, or NOD/Ltz-+/+ mice, and male NOD/Ltz-scid/sz (herein referred to as NZB, C57BL/6, BALB/c, SM/J, NOD, or NOD-scid) mice were obtained from Jackson Laboratory (Bar Harbor, ME), and were maintained by the Animal Resource Services of the University of California at Davis.Female or male motheaten C57BL/6J-Hcphme/Hcph me and heterozygous control C57BL/6J-+ /Hcph me (herein referred to as Hcphme/Hcph me and +/Hcphme-heterozygous) mice were maintained by Jackson Laboratory.Female or male ALY/Nsc Jcl-aly/aly mice and heterozygous con- trol ALY/Nsc Jcl-+/aly mice (herein referred to as aly/aly and +/aly-heterozygous) were obtained from CLEA Japan (Osaka), and were maintained by the In- stitute of Bio-Active Science, Nippon Zoki Pharmaceu- tical (Hyogo).Groups of up to ten mice were individu- ally studied, and 4-10-week-old mice were used.Mice of both sexes were studied.Because no sex differences were observed, the data from males and females mice were combined.

Monocional Antibodies (mAbs) A panel of 14 mAb with previously defined specificity for mouse thymic stromal (MTS) elements and thymo- cytes were used in this study.The MTS mAbs were prepared from the fusion of P3-NS-1-Ag4-1 (NS-1) cells with spleen cells or popliteal lymph node cells from LOU/M r ts immunized with enriched mouse thymic stromal-cell suspensions (Godfrey et al., 1990).The characteristics of each of these mAb were described be- fore, summarized in Table I, and illustrated in Fig. 1.


Histochemistry

Thymi were removed from mice, snap-frozen in dry, ice-cold 2-methyl butane, or dry, ice-cold ethanol, and embedded in TISSUE-T