The presence of characteristic epithelial swirls called Hassall bodies within the human thymic medulla has been used as an indicator of ongoing or recent thymopoiesis. We present a case where Hassall bodies were present in the absence of current or past thymopoiesis. The patient had been treated with corticosteroids for presumed asthma before his diagnosis of X-linked SCID. Two other cases of nonimmunodeficient patients treated with high-dose corticosteroids had markedly increased numbers of thymic Hassall bodies. To determine whether corticosteroid treatment induces thymic epithelial (TE) differentiation to form Hassall bodies, mAbs reactive with specific cytokeratins (CKs), filaggrin, and involucrin were used to define distinct stages of TE cell differentiation. Treatment of primary TE monolayers with hydrocortisone in vitro induced expression of involucrin and high-molecular-mass CKs that are characteristic of TE differentiation. Treatment of thymic organ cultures with hydrocortisone induced both medullary and subcapsular cortical TE cells to express CK6, a differentiation marker that is normally expressed only by Hassall bodies in vivo. These experimental studies combined with the case observations indicate that exogenous corticosteroids can regulate terminal differentiation of TE cells both in vitro and in vivo. Thus, the presence of Hassall bodies in thymus from corticosteroid-treated patients cannot be taken as an absolute indication of previous thymopoiesis. Because corticosteroids are also made within the thymus under normal physiologic conditions, these studies support the hypothesis that endogenous corticosteroids may play a role in normal TE differentiation and Hassall body formation in vivo.
The medulla of normal thymus with ongoing thymopoiesis contains characteristic swirled epithelial structures known as Hassall’s corpuscles or Hassall bodies. In addition to their distinctive histologic appearance, Hassall bodies react with mAbs that also react with the terminally differentiated upper layers of the epidermis (1, 2, 3, 4, 5). Thus, Hassall bodies are thought to be composed of terminally differentiated medullary thymic epithelial (TE)3 cells. The purpose of Hassall bodies and their contribution to thymic function is not known. Interestingly, although they are found in guinea pig thymus (6, 7, 8), they are not typically seen in murine thymus.
We previously studied a large panel of human thymus tissues from normal donors from birth to age 78 as well as patients with myasthenia gravis and a variety of primary and secondary immunodeficiencies (9, 10, 11, 12, 13, 14). We observed that the thymus from patients with molecular evidence of robust thymopoiesis contained noncystic, noncalcified Hassall bodies. If Hassall bodies were present in thymus from previously immunocompetent patients with minimal to no current thymopoiesis, they were typically cystically dilated and/or calcified. Thymus tissues from patients with primary immunodeficiencies who lacked thymopoiesis and T cell function lacked Hassall bodies entirely (15). Based in part on these observations, we proposed immunohistologic criteria for assessing thymopoiesis in thymic tissue sections. These criteria included the following: 1) presence of a light, lacy cytokeratin (CK)-positive TE network that allows close interaction between TE cells and thymocytes; 2) presence of CD1a+mib-1+ immature thymocytes; and 3) presence of noncystic, noncalcified Hassall bodies (9). Because Hassall bodies had never been observed in patients with primary failure of T cell differentiation, the presence of Hassall bodies within a thymus has been taken as an indicator of previous or current thymopoiesis (13, 16).
In this report, we describe a case where Hassall bodies were present in the thymus of a patient with genetically defined X-linked SCID (X-SCID) and primary failure of thymopoiesis. This patient had been treated with corticosteroids for presumed asthma before his diagnosis of immunodeficiency. We identified two other cases of nonimmunodeficient patients treated with high-dose corticosteroids who also had markedly increased numbers of thymic Hassall bodies. These case observations led us to hypothesize that corticosteroids may regulate the differentiation of TE cells to form Hassall bodies. We identified a panel of mAbs reactive with specific CKs, filaggrin, and involucrin that define distinct phenotypically recognizable stages of TE cell differentiation to form Hassall bodies. Treatment of primary TE cell monolayers and thymic organ cultures with hydrocortisone in vitro altered the expression of involucrin and CKs that are characteristically expressed by Hassall bodies and terminally differentiated epidermal keratinocytes in vivo. Taken together, these studies indicate that exogenous corticosteroids can regulate terminal differentiation of TE cells in vitro and drive Hassall body formation in vivo. Because corticosteroids are also made within the thymus under normal physiologic conditions (17, 18, 19, 20), these studies raise the possibility that endogenous corticosteroids may play a role in normal TE differentiation and Hassall body formation.
Materials and Methods
Tissue and cells
Normal human thymus tissue was obtained anonymously from children undergoing corrective cardiovascular surgery, according to a Duke Institutional Review Board-approved protocol for use of discarded tissue. Removal of this tissue was required to expose the operative site. No tissue was removed solely for research purposes. Thymus tissues were also obtained from immunodeficient and nonimmunodeficient patients at autopsy after consent of the next of kin. Detailed study of these tissues for research was also approved by the Institutional Review Board. Tissues were fixed in 10% neutral buffered formalin, and then processed into paraffin blocks using standard protocols. If available, additional tissue was snap-frozen in TissueTek OCT compound (Miles Laboratories, Elkhart, IN) and stored at −80°C until used for frozen sections. Some normal and immunodeficient thymus tissues were also used to derive primary TE cultures, as described below. Sections of normal skin were obtained anonymously from diagnostic blocks after pathologic diagnosis had been rendered or from biopsies taken with informed consent for Institutional Review Board-approved research.
The following primary Abs were used: IgG1 isotype control (Sigma-Aldrich, St. Louis, MO); anti-pan-CK (AE1 plus AE3; used as a mixture) (Boehringer Mannheim, Indianapolis, IN) (21); anti-CK5/8 (C-50), anti-CK6 (LHK6B), anti-CK10 (DE-K10), anti-CK14 (LL002), and anti-involucrin (SY5) (NeoMarkers, Fremont, CA); anti-filaggrin (FLG01) (Oncogene Research Products, San Diego, CA); CD3 rabbit polyclonal (DAKO, Carpinteria, CA); CD20 (L26) (DAKO); and CD1a (O10) (22).
Immunoperoxidase assays were performed using standard techniques on acetone-fixed frozen sections or chamber slides. Nonspecific protein binding was blocked by incubation with normal horse or goat serum, diluted 1/20 in PBS. Sections were sequentially incubated with primary Abs for 45 min at 37°C, horse anti-mouse or goat anti-rabbit biotinylated secondary Abs for 20 min at 37°C, and then avidin-biotin-HRP macromolecular complexes (Vector Laboratories, Burlingame, CA) for 20 min at 37°C, with intervening PBS washes. Sections were developed with 3,3′-diaminobenzidine plus H2O2, counterstained with hematoxylin, and permanently mounted. Formalin-fixed paraffin-embedded tissue sections were deparaffinized, rehydrated, incubated in 0.6% H2O2 in absolute methanol to block endogenous peroxidase, and subjected to immunoperoxidase staining as described above. Where necessary, Ag retrieval was performed using the microwave citrate method (Citra; Vector Laboratories) (600 W for 5 min times two) or digestion with 1/4 dilution of Digest-All 2A (trypsin) (Zymed Laboratories, South San Francisco, CA). For both paraffin and frozen sections, the optimum dilution of each Ab was first determined by serial dilutions on control normal thymus or skin sections.
TE monolayer and organ cultures
TE cells were obtained by an explant technique using TE medium containing 67% DMEM (Life Technologies, Grand Island, NY), 22% Ham’s F-12 (Life Technologies), 5% Fetal Clone II serum (HyClone, Logan, UT), 0.4 μg/ml hydrocortisone, 5 μg/ml insulin, 11 ng/ml recombinant human epidermal growth factor (Collaborative Biomedical, Bedford, MA), 0.18 μM adenine, 10−10 M cholera toxin (ICN Biomedicals, Aurora, OH), 0.25 μg/ml Fungizone, and 50 μg/ml gentamicin on mitomycin-C-treated NIH-3T3 fibroblast feeder layers as described (23, 24). Contaminating fibroblasts were removed from TE cultures by treatment with 0.02% EDTA in PBS followed by complement-mediated lysis with mAb 1B10, which binds to a cell surface Ag on fibroblasts (25). T cells were depleted by the culture process and extensive washing. Cultured TE cells in passages 1–3 (corresponding to 3–8 wk in culture) were stored frozen in DMEM, 20% FBS, and 7.5% DMSO until expansion in TE medium on 3T3 monolayers for use in experiments. Expanded TE cells were cultured as adherent monolayers either in six-well dishes or in eight-well glass chamber slides (Lab-Tek; Nunc, Naperville, IL) without a feeder layer for 4 days using TE medium lacking hydrocortisone (control) or in TE medium containing 0.4, 4, or 40 μg/ml hydrocortisone. Metyrapone (Sigma-Aldrich), an inhibitor of the steroidogenic enzyme 11-β-hydroxylase, was also added at 180 μg/ml to some cultures.
For thymic organ cultures, 0.5- to 1-mm-thick slices of thymus tissue from normal pediatric donors were prepared using a Stadie-Riggs hand microtome (Thomas Scientific, Swedesboro, NJ). Slices were cultured on filters placed on sterile gelfoam sponges in six-well culture dishes as described previously (26). Thymic organ culture medium consisted of 85.5% Ham’s F-12 medium, 25 mM HEPES, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.25 μg/ml amphotericin B, 10% FBS (all medium components were Life Technologies) with hydrocortisone concentrations of 0, 0.4, 4, or 40 μg/ml. Medium was changed daily. Thymic slices were harvested at 7 days and frozen for immunohistochemical analysis.
The hydrocortisone or native cortisol levels present in TE monolayer or thymic organ cultures were determined using a competitive RIA kit (ICN Biomedicals).
For Western blots, TE cells were solubilized in situ within six-well culture dishes using 250 μl of SDS-sample buffer. Equal volumes of lysate were analyzed by SDS-PAGE using 4–15% gradient gels (Bio-Rad, Hercules, CA) and transferred to nitrocellulose. Membranes were blocked with 10% nonfat milk in 0.01 M Tris, 0.15 M NaCl, and 0.05% Tween 20 (pH 8.0) (TBST), and then sequentially reacted with primary and then secondary Ab coupled to HRP, with intervening TBST washes. Bound Ab was detected by chemiluminescence (SuperSignal West Pico chemiluminescence substrate; Pierce, Rockford, IL). A lysate prepared from normal human epidermis was included on each blot as a positive control and was reactive with bands of the appropriate molecular mass for each mAb tested. Reactivity with mAb specific for poly(A) binding protein, a ubiquitously expressed housekeeping gene (27), was used to verify equal loading of samples.
Report of three cases
The patient was an 8-mo-old male with X-SCID. He initially presented to his local pediatrician with recurrent respiratory tract infections and wheezing. Treatment with inhaled budesonide (Pulmicort respules; 0.25 mg twice daily) was begun at age 3 mo for presumed asthma. The patient developed pneumonia requiring hospitalization at 6 mo of age. Inhaled budesonide was discontinued and i.v. methylprednisolone was substituted at 3.3 mg/kg/day for 7 days. Further investigation revealed a low absolute CD3+ T cell count (75/mm3; normal, 2280–6450/mm3) with moderate numbers of B cells (1741/mm3; normal, 430-3350/mm3), severe hypogammaglobulinemia (95 mg/dl; normal, 286-1680 mg/dl), and lack of proliferation to PHA and Con A mitogens. Further investigation into the health of a previous male infant given up for adoption some years earlier revealed that that child died from infection in infancy, with autopsy findings consistent with primary immunodeficiency. Taken together, these findings were consistent with a diagnosis of X-SCID for the current patient. A taper of steroid dosage was begun. At the time of his transfer to this hospital at 7 mo of age, his methylprednisolone dosage was 2.2 mg/kg/day. Documented infections included Streptococcus pneumoniae bacteremia/pneumonia, Pneumocystis carinii pneumonia, parainfluenza, and respiratory syncytial virus infections. He was treated for these infections, but developed disseminated adenoviral infection. He received a haploidentical T cell-depleted bone marrow transplant from his mother on day 10 after transfer. His methylprednisolone dosage was tapered to 0.07 mg/kg/day by day 16 after transfer. On day 18 after transfer, he received 0.9 mg/kg methylprednisolone plus 81 mg/m2 hydrocortisone. He subsequently was treated with hydrocortisone, tapering from 122 mg/m2/day on day 19 to 41 mg/m2/day on days 26–29 after transfer (Fig. 1). The patient died from pulmonary hemorrhage associated with adenoviral infection on day 29 after transfer. An autopsy was performed. Postmortem genetic analysis confirmed the diagnosis of X-SCID by identifying a R222C mutation in the IL2RG gene that codes for the γ-chain common to IL-2 and other cytokine receptors.
The patient was a 610-g female infant delivered by cesarean section at 28 wk for severe intrauterine growth retardation. The mother was treated with dexamethasone before delivery to facilitate fetal lung maturation. The infant was immediately begun on i.v. hydrocortisone (100 mg/m2/day) due to a concern about adrenal dysfunction. This dosage was continued for 6 days, discontinued for 1 day, and then replaced with dexamethasone (0.5 mg/kg/day times 3 days, then 0.3 mg/kg/day times 2 days) (Fig. 1). The patient died on day 12 of life from refractory hypotension and multiorgan failure. A full autopsy demonstrated periventricular leukomalacia of the brain, pulmonary immaturity, and acute myocardial infarction with hemorrhage, as well as congenital islet cell dysplasia and bilateral adrenal hypoplasia.
This male infant was admitted at 5 mo of age for failure to thrive, recurrent fevers, and persistent left lower lobe pulmonary consolidation. Cultures, bronchoalveolar lavage, and lung biopsy were negative for infectious agents. Serum Ig and complement levels were normal. The patient received 4 mg/kg/day methylprednisolone times 2 days for adult respiratory distress syndrome before death from respiratory failure at 6 mo of age (Fig. 1). Autopsy findings and the patient’s clinical course were consistent with a diagnosis of familial erythrophagocytic lymphohistiocytosis or immune-associated hemophagocytosis syndrome related to infection with EBV or CMV.
Hassall bodies are present in thymus from a corticosteroid-treated patient with X-SCID
The thymus from patient 1 was markedly hypoplastic and located high in the anterior mediastinum. It measured ∼2.2 × 2.5 × 0.7 cm and weighed 0.7 g (normal for age, 10 ± 4 g) (28). Microscopically, the thymus had a compact lobular structure consistent with primary failure to expand during fetal development rather than postnatal atrophy (Fig. 2,A). It consisted predominately of islands of epithelial cells, surrounded by adipose tissue with prominent connective tissue septae (Fig. 2,A). Occasional foci of CD3+ T cells (Fig. 2,E) and scattered CD20+ B cells were present (F); however, many of the epithelial islands completely lacked T cells (not shown). No CD1a+ cells were present (Fig. 2,G). Proliferating cells were rare, as indicated by minimal reactivity with mib-1 Ab against the Ki-67 nuclear proliferation Ag (Fig. 2,H). The lack of CD1a+mib-1+ small lymphocytes indicates lack of ongoing thymopoiesis (9), consistent with the documented mutation in IL2RG. Despite the lack of thymopoiesis, scattered cystically dilated Hassall bodies were observed in the thymus (Fig. 2,B). Insufficient tissue was available to react with the entire panel of differentiation-specific mAbs described below; however, these Hassall bodies and the cells surrounding them were reactive with pan-CK mAbs AE1/AE3 (Fig. 2 C) and with Abs against involucrin (D), similar to Hassall bodies seen in normal thymus. Thymus sections from six other patients with X-SCID who had not been treated with corticosteroids were also examined. Four of these patients had one or more unsuccessful bone marrow transplants or died of infection before engraftment, similar to patient 1. Each thymus showed a compact lobular structure and lacked thymopoiesis similar to patient 1, but no Hassall bodies were present (data not shown; Ref. 15).
Hassall bodies are increased in thymus from corticosteroid-treated patients without primary immunodeficiency
As expected from numerous previous reports on the sensitivity of human thymocytes to corticosteroids (29), thymus tissues from patients 2 and 3 demonstrated decreased cellularity, particularly of the cortical region, with loss of corticomedullary distinction. In addition, the medulla of each thymus contained a markedly increased number of Hassall bodies compared with normal thymus (Fig. 2, I–L). These Hassall bodies varied in size and amount of internal acellular material. All were strongly reactive with Abs against CK5/8, -6, -10, and involucrin (not shown), similar to Hassall bodies seen in normal thymus (see below).
Identification of distinct phenotypic stages of TE cell differentiation to form Hassall bodies
TE cells have previously been described to react with mAbs that also react with epidermal keratinocytes (3). Although whole thymus contains CK characteristic of terminally differentiated epidermal keratinocytes by two-dimensional gel analysis (30), the Abs required to localize expression of each of these specific CK in situ have only recently become available. To study the differentiation of TE cells to become Hassall bodies in greater depth, we determined the reactivity of TE cells in normal thymus with a panel of mAbs reactive with markers of epithelial differentiation and compared this reactivity to that of epidermis.
Pan-CK Abs AE1/AE3 used as a mixture recognize CK1/2, -5, -6, -7, -8, -10, -13, -14, -15, and -16 (21). Immunoperoxidase staining of normal thymus with AE1/AE3 demonstrated a light lacy network of TE cells in the thymic cortex as well as a more condensed TE network in the medulla, as described previously (9). TE cells in the subcapsular cortex and all portions of Hassall bodies were strongly reactive with AE1/AE3 (Fig. 3,B). AE1/AE3 was also reactive with all layers of epidermis; however, staining was stronger and more granular in the stratum basale (Fig. 3,J). CK14-specific mAb reacted strongly with TE of the subcapsular cortex and medulla and with the outer portion of Hassall bodies, but reacted either very weakly or not at all with the majority of cortical TE (Fig. 3,C). CK14 mAb did not react with the center of Hassall bodies. In epidermis, CK14 reactivity was seen in the stratum basale, stratum spinosum, and stratum granulosum, but not in the more differentiated stratum corneum layer (Fig. 3,K). mAb specific for CK5 and -8 reacted with cortical and medullary TE cells similar to AE1/AE3, but also failed to react with the central portion of Hassall bodies (Fig. 3,D). CK5/8 reactivity in epidermis was limited to the stratum basale and the stratum spinosum, and was absent from both the stratum granulosum and the stratum corneum (Fig. 3 L).
In contrast, CK10-specific mAb reacted only with Hassall bodies in thymus (Fig. 3,E). All layers of the Hassall bodies were stained including the anuclear material present in their centers. In skin, CK10-specific mAb failed to react with the stratum basale, but reacted with the three upper layers of epidermis (Fig. 3,M). CK6-specific mAb reacted only with the outer layers of Hassall bodies in thymus (Fig. 3,F). Its reactivity with epidermis was minimal (Fig. 3 N), although eccrine ducts stained strongly (not shown).
Expression of filaggrin and involucrin is associated with terminal differentiation of epidermal keratinocytes (31). Therefore, we also determined the reactivity of thymus sections with mAbs specific for these proteins. Involucrin-specific mAb reacted with Hassall bodies as well as the medullary TE immediately surrounding them, but did not react with the remainder of the thymus (Fig. 3,G). Anti-involucrin mAb predominately reacted with the stratum granulosum of skin, with minimal staining of the stratum corneum (Fig. 3,O). Anti-filaggrin mAb reacted only with the anuclear debris in the center of the Hassall bodies (Fig. 3,H). This mAb also appeared to recognize portions of the stratum granulosum in epidermis, although fewer cells were recognized than with anti-involucrin mAb. Anti-filaggrin mAb had variable weak-to-absent reactivity with the stratum corneum of epidermis (Fig. 3 P).
Taken together, this panel of mAbs can be used to define distinct phenotypic changes that occur in medullary TE cells as they differentiate to form Hassall bodies. Medullary TE cells express CKs reactive with AE1 and AE3 throughout their ontogeny. Medullary TE also express CK14. As these TE begin to terminally differentiate and become the outermost cells of a Hassall body, they acquire expression of involucrin. The cells in the outer layers of Hassall bodies also acquire expression of CK6 and CK10. As their differentiation continues and they become anuclear squames analogous to the cells present in the stratum corneum of the epidermis, they lose reactivity with mAbs specific for CK5/8 and -6 and acquire expression of filaggrin. Thus, the outer cellular portions of Hassall bodies are defined by coexpression of CK5/8, -6, -10 and involucrin. The anuclear central portion of the Hassall body is defined by coexpression of involucrin, CK10, and filaggrin, and absence of CK5/8 and -6.
Hydrocortisone treatment of TE monolayers in vitro induces expression of markers characteristic of terminal differentiation
Hassall bodies have been reported to form only in thymus with ongoing thymopoiesis (9, 13, 16), suggesting that developing thymocytes induce factors that regulate TE differentiation. Steroid-producing enzymes were shown to be present in thymus with active thymopoiesis but were absent in thymus that has been depleted of thymocytes by irradiation (20). These previous studies combined with our case observations led us to develop the hypothesis that corticosteroids may regulate the terminal differentiation of TE cells that results in Hassall body formation in vivo. To minimize potential contributions from endogenous thymic corticosteroids, TE cells were derived from two different patients with X-SCID who lacked thymopoiesis in vivo. Analysis of spent medium from these cultures by competitive RIA showed no detectable endogenous production of cortisol. The effect of 0–40 μg/ml hydrocortisone on TE differentiation in vitro was determined by in situ immunostaining of TE grown as monolayers on chamber slides. Expression of involucrin, the first specific marker of TE differentiation toward Hassall body formation in vivo (see above), was used as an indicator of TE differentiation. X-SCID TE cultured in the absence of hydrocortisone contained distinct foci of involucrin-positive cells (Fig. 4, A and B) that increased in size and number with hydrocortisone treatment (C and D). Increased differentiation of these X-SCID TE with hydrocortisone treatment was also reflected by changes in morphology from small and cobblestone-like (Fig. 4 B) to large and squamous (D), similar to what has previously been reported for differentiating epidermal keratinocytes (32, 33).
We next determined the effect of hydrocortisone on differentiation of TE monolayers obtained from three normal donors. Most normal TE demonstrated low-level reactivity with anti-involucrin mAb, both in the absence (0 μg/ml) and in the presence of 0.4, 4, or 40 μg/ml hydrocortisone. Increased reactivity with anti-involucrin mAb was seen on larger cells with squamous, differentiated morphology that were present in all cultures (not shown). Thus, normal TE cells appear to undergo at least the earliest stage of differentiation, as indicated by production of involucrin, spontaneously in adherent monolayer culture regardless of exposure to exogenous hydrocortisone. However, increased differentiation of normal TE with hydrocortisone treatment was reflected by increased expression of high-molecular-mass CK species on Western blot (Fig. 5). The expression of high-molecular-mass CKs (CK1/2, at 65–67 kDa) that are recognized by both mAbs AE1 and AE3 have been previously described as markers for terminal epithelial differentiation in both skin and thymus (21, 30). Expression of these high-molecular-mass CKs represents a later stage of TE differentiation toward Hassall body formation compared with induction of involucrin (see above). The spontaneous differentiation seen in these monolayer cultures of normal TE was not due to endogenous corticosteroid production. No cortisol was detectable in spent medium from cultures without added hydrocortisone, whereas cortisol levels of 0.3 ± 0.1 μg/ml were measured in spent medium to which 0.4 μg/ml hydrocortisone had been originally added. Also, addition of 180 μg/ml metyrapone, an inhibitor of the 11-β-hydroxylase enzyme needed for steroid biosynthesis, had no effect on TE differentiation. Specific bands corresponding to CK10, a differentiation marker that occurs late in terminal differentiation to form Hassall bodies in vivo, were not seen in Western blots of either normal or X-SCID TE, with or without hydrocortisone treatment (not shown). Thus, although hydrocortisone treatment stimulated differentiation of both X-SCID and normal TE adherent monolayers in vitro, these TE did not develop the full range of differentiation marker expression that is characteristic of terminally differentiated TE cells in Hassall bodies in vivo over the time period studied.
Hydrocortisone treatment induces terminal differentiation of TE in thymic organ culture
The microenvironment of TE cells in adherent monolayer cultures in vitro is likely very different from what these same TE cells experience in vivo. To more closely model the in vivo microenvironment, organ cultures of normal thymus were performed using hydrocortisone concentrations ranging from 0 to 40 μg/ml and analyzed immunohistologically. H&E staining showed that, although thymocyte cellularity decreased during culture, significant numbers of lymphocytes were still present within thymic slices cultured for 7 days, especially within cortical regions (Fig. 4, E and F). Corticomedullary distinction and Hassall body formation were also maintained during organ culture. The thickness of the subcapsular cortical TE layer increased with culture, and a variable amount of tissue necrosis was observed (not shown).
Increased numbers of small Hassall bodies were present in thymic slices cultured with hydrocortisone (Fig. 4,F) compared with slices cultured in the absence of hydrocortisone (E). The pattern of reactivity with mAbs specific for CK5/8, CK10, CK14, filaggrin, and involucrin did not change with hydrocortisone exposure in vitro (not shown). In thymus slices from two of three normal donors that were cultured in vitro without hydrocortisone, CK6 reactivity was limited solely to Hassall bodies as seen in normal thymus in vivo (Fig. 4,E). However, TE cells adjacent to Hassall bodies and many TE cells in the subcapsular cortex also strongly reacted with anti-CK6 mAb (Fig. 4 F) in thymic slices cultured with 0.4, 4, or 40 μg/ml hydrocortisone. In the remaining thymus, only Hassall bodies were reactive with anti-CK6 mAb at time zero. However, all slices from this thymus (cultured with 0, 0.4, 4, or 40 μg/ml hydrocortisone) had at least focal anti-CK6 reactivity of subcapsular cortical TE as well as CK6+ Hassall bodies after culture for 7 days. Analysis of the spent medium showed that no endogenously produced cortisol was present in these organ cultures. These studies show that exposure of intact thymus to hydrocortisone induces formation of Hassall bodies and induces expression of CK6, an early marker of TE differentiation, by medullary and subcapsular cortical TE cells. The thymic organ culture studies complement the studies of TE monolayer cultures, and provide further evidence that exposure to increased levels of corticosteroids regulates the terminal differentiation of TE cells in vitro. However, the spontaneous differentiation seen in at least one culture in the absence of added hydrocortisone indicates that additional signals besides corticosteroids can also regulate TE differentiation.
We report the presence of Hassall bodies in thymus from a patient with a genetically defined immunodeficiency, primary failure of thymopoiesis, and deficient T cell function. This patient had previously been treated with corticosteroids. Enhanced Hassall body formation was observed in two additional patients with pre-existing thymopoiesis who were also treated with corticosteroids. In addition, we have identified a panel of mAbs that define phenotypically distinct stages in the terminal differentiation of TE cells in tissue sections and in vitro. Hydrocortisone treatment also induced differentiation of TE cells in vitro in both adherent monolayer and thymic organ cultures. Taken together, these studies indicate that corticosteroids regulate TE differentiation in vitro and support the hypothesis that endogenous corticosteroids may drive Hassall body formation in the thymus in vivo.
TE cells physically contact thymocytes and are exposed to soluble factors produced by thymocytes undergoing T cell selection and differentiation within the thymus (34, 35). That Hassall bodies are not typically observed in the absence of maturing thymocytes suggests that contact with or substances produced by maturing thymocytes drive the differentiation of TE cells in normal thymus to form Hassall bodies. However, our data indicates that the need for developing thymocytes to induce Hassall body formation may not be absolute. The thymic histopathology in cases 1, 2, and 3, and our thymic organ cultures demonstrates that corticosteroid treatment can drive formation of Hassall bodies, irrespective of the presence of thymopoiesis.
It is possible that partial function of the R222C mutant IL2RG expressed by patient 1 may have facilitated differentiation of his TE in vivo, with or without thymopoiesis. Sharfe et al. (36) previously described a patient with a R222C mutation in IL2RG who, although susceptible to opportunistic infections and unable to reject an allogeneic skin graft, had a normal-sized thymus with abundant thymocytes and Hassall bodies. Further studies demonstrated that this R222C mutant IL2RG was expressed on the cell surface, but had a reduced ability to both bind IL-2 and to stimulate Jak3 activation (36). However, unlike the patient described by Sharfe et al. (36), patient 1 in our study had a vestigial thymus without evidence of prior thymic function. In particular, the changes in size and increased adipose tissue that accompany atrophy of a previously functional thymus were notably absent. It is also unlikely that the Hassall bodies observed in thymus from case 1 were due to the bone marrow transplant the patient received 16 days before death. This patient showed no clinical evidence of engraftment at any time posttransplant, and the thymus was totally devoid of developing thymocytes at autopsy. In humans, de novo generation of T cell function after bone marrow transplantation for SCID typically requires 3–4 mo (37, 38). Examination of thymus from six other patients with X-SCID and failed bone marrow transplants or who died before engraftment could be achieved did not demonstrate Hassall bodies (15). However, the time required to create a Hassall body or the length of time that it remains in the thymus once it is formed is not known for humans. A 7-year-old human patient with acute leukemia and transfusion-associated graft-vs-host disease had thymic involution with loss of Hassall bodies at autopsy 6 wk after receiving the first of several nonirradiated blood transfusions (39). Unfortunately, the onset of this patient’s thymic depletion cannot be known with certainty, because the patient’s cachexia (18 kg for 108-cm height) and previous chemotherapy with prednisone, cyclophosphamide, vincristine, and methotrexate (last treatment, 1 wk before transfusion) also probably contributed to the severity of thymic involution. A study in the guinea pig showed that acute thymic involution induced by cyclophosphamide was associated with marked cystic enlargement of Hassall bodies, which could still be detected up to 3 wk after treatment (8). Abundant cystically dilated Hassall bodies were present in the thymus of a 14-mo male infant who died without evidence of current thymopoiesis 4 wk after chemotherapy (busulfan, melphalan, and anti-thymocyte globulin) and cord blood transplantation for treatment of mucopolysaccharidosis I (Hurler’s syndrome) (L. P. Hale, unpublished data). Thus, given the potential longevity of Hassall bodies, we cannot rule out the possibility that the Hassall bodies in the thymus of case 1 were induced by mutant T cells as described by Sharfe et al. (36) before corticosteroid treatment or by transient effects of the bone marrow transplant and persisted short term, despite the absence of ongoing thymopoiesis. However, the evidence provided by the other cases and the in vitro studies make corticosteroid treatment a much more likely explanation for the Hassall bodies found in this patient.
The effect of exogenous and stress-induced corticosteroids in inducing rapid apoptosis of immature thymocytes has been well studied in both humans and animals (29). Most of the normal thymus tissues available for study are obtained from nonimmunodeficient human infants and children undergoing corrective cardiovascular surgery. These children typically receive a dose of 10 mg/kg methylprednisolone at 6–12 h and again 1–6 h before surgery. Thymus tissues derived from these patients commonly have a starry-sky appearance, characteristic of mild stress involution. Apoptotic thymocytes are phagocytosed by tingible body macrophages, leaving relatively open areas that appear as stars in the dark sky of closely packed basophilic thymocytes within the cortex. Thus, although normal thymus donors receive a relatively large dose of corticosteroids before surgery, this dose is insufficient to induce thymic depletion or formation of large numbers of Hassall bodies in the short time before thymus excision. However, variations in timing and total dose of preoperative corticosteroids may affect differentiation of TE during in vitro thymic organ culture. This may have contributed to the spontaneous differentiation observed in one donor thymus culture in the absence of added hydrocortisone. It is interesting that, although adrenal and exogenous glucocorticoids promote thymocyte apoptosis, endogenous thymic glucocorticoids have also been shown to promote the survival of thymocytes after TCR engagement (17). In vivo glucocorticoid blockade results in selection of T cells with lower avidity for self Ag/MHC than would otherwise occur in the presence of corticosteroids (40). Endogenous thymic corticosteroids thus play an important role in regulating both positive and negative thymic selection processes in vivo.
We found that normal TE cells differentiated spontaneously in monolayer cultures in vitro, as indicated by cellular morphology and expression of involucrin and high-molecular-mass CKs, in the presence and absence of hydrocortisone. However, TE derived from X-SCID patients had less spontaneous differentiation. Increased differentiation of hydrocortisone-treated X-SCID cultures could be documented by cellular morphology and by expression of involucrin. The pattern of markers of terminal differentiation acquired by hydrocortisone-treated normal and X-SCID TE cell monolayers in vitro agrees with what was established using thymus sections, i.e., that TE acquire expression of CK14, then involucrin, and then CK6 and CK10 as they undergo terminal differentiation. Thymic organ cultures showed expanded reactivity of anti-CK6 mAb with subcapsular cortical TE and cells adjacent to Hassall bodies in hydrocortisone-treated cultures. Because CK6 mAbs react only with the terminally differentiated TE present in Hassall bodies in normal thymus in vivo, these studies provide further evidence that corticosteroids regulate terminal differentiation of TE cells in vitro. Reactivity with anti-CK10 Abs was seen for TE located within Hassall bodies, further confirming that expression of CK10 is limited to the final stages of TE terminal differentiation both in vitro and in vivo.
The work presented in this report is consistent with the hypothesis that maturation and function of the stromal component of the thymic microenvironment to form Hassall bodies normally depends on soluble factors liberated by or induced by direct contact with developing T cells. Several studies have shown that TE cells themselves can produce corticosteroids (18, 19, 20). Active thymocyte development may be required for TE expression of the 11-β-hydroxylase enzyme required for corticosteroid synthesis in vivo, because this enzyme is detected in normal thymus with ongoing thymopoiesis but not in irradiated, thymocyte-depleted murine thymus (20) or in our TE monolayer or thymic organ cultures. TE differentiation to form Hassall bodies is also typically observed only in thymus with ongoing thymopoiesis. However, the data presented here demonstrates that TE differentiation can also be driven by exogenous corticosteroids, in the presence or absence of developing T cells. Because corticosteroids are produced within the thymus when developing thymocytes are present and appear to be necessary for normal thymic selection and maturation, these studies raise the possibility that endogenous thymic corticosteroids may regulate TE differentiation and Hassall body formation in vivo.
We thank Ms. Leona Whichard and Dr. Barton F. Haynes for providing TE cells. We acknowledge the expert technical assistance of Jie Li with immunostaining and Paula Greer for culture and analysis of TE cells and thymic organ cultures.
This work was supported by funds from Department of Pathology, Duke University Medical Center, and by National Institutes of Health Grant R01-AI 47040 to M.L.M.
Abbreviations used in this paper: TE, thymic epithelial; X-SCID, X-linked SCID; CK, cytokeratin.