Arrest of In Vitro T Cell Differentiation of Normal Bone Marrow-Derived CD34+ Stem Cells with Thymic Epithelial Fragments from Children with AIDS



A novel approach is presented to assess the ability of thymic tissues obtained from children with end stage AIDS to attract normal bone marrow (BM)-derived CD34+ (lineage negative) stem cells (SCs) and support lymphopoiesis in vitro. Chemokinesis of BM-derived CD34+ SCs was analyzed by time-lapse videomicroscopy to ascertain whether an alteration in SC motility could contribute to abnormal thymopoiesis under conditions of HIV infection. The migration of SCs derived from an HIV+ donor into thymic tissue was not significantly altered compared to normal controls, as were normal SCs migrating toward thymic epithelial cell monolayers derived from an HIV+ patient. Thymic tissue obtained from children with AIDS contained nests of CD34+ SCs identified by immunofluorescence, indicating SC homing to the thymus is apparently supported in HIV infection. The ability of HIV-affected thymic epithelial fragments to support lymphopoiesis was determined by examining the initial thymocyte populations present, compared to thymocytes produced de novo in T cell-depleted thymic fragments, following a single pulse of lineage negative CD34+ CD38 SCs. In comparison to normal controls, thymocytes derived from the HIV-affected thymic epithelial fragment coculture had an increased percentage of triple negative thymocytes (28% of lymphocytes from HIV-affected tissue versus 1.5% in controls, p < 0.01) and a decreased percentage of double and single positive CD4+ thymocytes. However, CD3+CD8+ TCRαβ+ expression was comparable to control cultured thymic epithelial fragments indicating that HIV-affected thymic epithelia were capable of supporting the development of the CD8+ lineage. In an effort to extend the information obtained to date from the histological examination of HIV-affected thymic tissue, select patient thymic tissues were maintained in culture to evaluate the capacity of undifferentiated thymic epithelial cell guirlandes to differentiate in vitro. A partial regeneration of certain subpopulations of the thymic epithelium defined by TE-4 monoclonal antibody (mAb) and CDR2 mAbs occurred during the in vitro culture. The epithelial and mesenchymal components of thymic tissues were distinguished by immunostaining for keratins (indicative of epithelium) and vimentin (a mesenchymal marker). Further evaluation of the modulation of HIV thymus, with respect to the testing of new therapeutic strategies on SCs, will be possible with this in vitro model.


AIDS is manifested by recurrent microbial infections and malignancies as the result of a progressive cellular and humoral dysfunction caused by HIV infection [1]. In particular, the T cell immune deficiency is characterized by the dysfunction and eventual depletion of CD4+ T helper cells. Several mechanisms have been presented to explain the depletion of peripheral T cell pools including direct cytolysis, disruption of cytokine circuits, syncytia formation, apoptosis, autoimmune cytolysis, gp120-mediated innocent bystander cytolysis and the immunosuppressive effects of viral proteins [1]. In addition to this direct depletion, the regeneration de novo of lymphocytes by thymopoiesis is compromised [2].

In HIV+-infected individuals, there is a steady state maintained between virus production and immune clearance. It has been reported that the kinetics of viral replication ([viral RNA]/mm3 plasma) and viral burden ([proviral DNA]/106 peripheral blood mononuclear cells) suggest a viral turnover rate of 9.7 × 109 virions per day, and the turnover of CD4+ lymphocytes to be ∼1.8 × 109 cells per day [3]. The demands of clearing a persistent high titer virus along with intervening infection will progressively challenge and deplete the peripheral T cell pool. The capacity for regeneration of these T cells during HIV infection is not clear. Mature resident CD4+ lymphocytes can be expanded in the peripheral T cell pool, but this population has a limited proliferation potential defined by telomere length [4]. During the later stages of HIV infection, the peripheral T cell pool is depleted and dysfunctional. There is a notable disruption of the peripheral lymph node architecture with cellular depletion and collapse of the follicular dendritic cell (FDC) network coincident with increasing viremia. Furthermore, the bone marrow (BM) and the thymic microenvironments (ME) become incapable of regenerating the lymphoid populations destroyed by the virus. The contribution of thymopoiesis to the maintenance of the adult human peripheral T cell pool is controversial, since thymopoiesis is maximal in newborns and young children.

An intact thymic ME is an absolute requirement for T cell differentiation to occur. Thymic stromal dysgenesis, damage by drugs, irradiation or infection will not support lymphopoiesis until the stromal environment is repaired. The thymus is dependent on a continual supply of BM-derived SCs to maintain both lymphopoiesis and the integrity of the thymic stromal ME. One fundamental issue that needs to be resolved regarding thymopoiesis in HIV is whether the virus induces a primary acquired thymic stromal defect that compromises lymphopoiesis, or whether there is a secondary and contingent effect on the thymic stroma associated with the SCs populating the thymus that have become HIV-infected.

The CD34+ SC population of the BM is a heterogeneous mixture of progenitors, in which only a minor percentage represents the prelymphoid subset. There is an even smaller percentage (0.01%) of CD34+ SCs constituting the totipotent SC population that is responsible for maintaining the hematopoietic steady state. How HIV affects these subpopulations of cells in terms of direct infection, viral reservoirs, trafficking of infected cells, cell growth/function and the cytokine regulation of HIV expression, remains to be determined. The capacity for productive HIV infection of BM CD34+ SCs has been variably reported as occurring in only a subset of HIV seropositive individuals [5], or in CD34+ SCs coexpressing CD4+ [6], or without evidence of productive HIV infection [7-9]. Monocyte differentiation has been noted to play an essential role in viral production in the BM [10]. How this may relate to the different subpopulations of CD34+ SCs in their respective ME is not known. The growth of CD34+ SCs in vitro is inhibited by HIV or its products, and this HIV-induced suppression of progenitor cells may adversely affect the BM capacity to support lymphopoiesis in vivo [11-14].

The thymus ME is adversely affected by HIV. Infection of resident T cells, thymocytes, CD34+CD4+ SCs and thymic epithelial cells (TEC) specifically affects the generation of naive CD4+ T cells by thymopoiesis. The thymus manifests a loss of thymocytes due to direct infection [14-17]. TECs can be infected with HIV resulting in their dysfunction and destruction [18, 19]. The thymic stromal architecture becomes dysplastic, displaying an effacement of the cortical thymic epithelium, a loss of corticomedullary differentiation, a loss of Hassall's corpuscles, thymic epithelial necrosis or calcification, and hyaline changes of the perivascular spaces [20-26]. Morphologically, the epithelium may be represented as sheets of undifferentiated cells called guirlandes [22, 24].

A novel approach to cellular immune reconstitution involving the transplantation of cultured thymic epithelial fragments (CTEFs) has been attempted prior to the advent of retroviral medication [27-29]. In some cases, the CTEF transplants provided transient increases in T cell number and function (despite a long-term graft failure). The reasons for graft failure were not clear, although theoretically could have been due to rejection, inadequate numbers of pre-T SCs capable of homing or entering the thymus, HIV infection of newly differentiating thymocytes, or HIV infection of TEC or dendritic cells (DCs) resulting in their dysfunction or destruction. Thymic biopsy studies obtained by two groups revealed the absence of thymic epithelial tissue with HIV-infected T cells at the graft site, or with thymic epithelial tissue present but involved in a multinucleated giant cell inflammatory response [27, 28]. These biopsy results suggest that the thymic epithelial tissue was being damaged by an ongoing T cell-mediated inflammatory response due to HIV infection rather than being involved in rejection. Supporting this contention are in vitro studies which demonstrate that HIV-infected thymocytes and T cells induce TEC injury [14, 18-20]. More recently, evidence suggesting primary thymic epithelial failure in lentiviral infection is supported by the following example of competitive reconstitution [30]. In a study of the efficacy of human thymus transplantation in a simian immunodeficiency virus (SIV)-infected rhesus, one animal which had previously received a T cell-depleted human cultured thymic epithelial fragment (TCD-CTEF) transplant succumbed to a Mycobacterium avium intracellulare (MAI) infection. At autopsy, the monkey thymus demonstrated many hyalinized lobules, appeared atrophic and devoid of lymphocytes, but did not appear morphologically stressed. The transplanted human CTEF was replenished with monkey lymphocytes, thus illustrating that SIV-infected host BM SCs were capable of repopulating a transplanted (human donor) thymus. Similarly, the host thymus capacity for attracting or retaining SCs could have been functionally altered by the lentiviral infection.

In the present study, thymic tissue obtained from children with end stage HIV was assessed for the ability of the thymic epithelium to attract normal allogeneic BM-derived CD34+ SCs and to support lymphopoiesis in vitro. In an effort to extend the information obtained to date from the histological examination of HIV-affected thymic tissue, thymic tissue obtained from children with end stage HIV were grown in culture, utilizing a novel coculture system recently developed in our laboratory [31]. The capacity of HIV-affected thymic epithelial fragments to support lymphopoiesis was evaluated by examination of the initial thymocyte populations present and thymocytes produced in coculture following a single pulse of BM-derived CD34+CD38 (lineage negative) SCs. The epithelial guirlandes present in the cultured fragments were evaluated utilizing laser scanning confocal microscopy (LSCM) and three-dimensional reconstruction to assess the potential for thymic stromal regeneration.

Materials and Methods


Thymus glands were obtained at autopsy from three children with end-stage AIDS. The relevant clinical data on these patients are as follows. Case 1 represents a 35-month-old female with congenitally acquired HIV dying of respiratory failure secondary to aspiration/endogenous lipoid pneumonia. Case 2 represents a neonatally acquired HIV infection dying at 13 years of age, of overwhelming fungal (Aspergillus) sepsis and disseminated MAI. Case 3 represents a neonatally acquired HIV infection dying at 15 years of age of respiratory failure secondary to disseminated pulmonary and mediastinal B cell lymphoma. Normal control thymus tissues were obtained from children undergoing corrective cardiac surgery who had portions of their thymus removed as part of the surgical process. This study was approved by the Saint Louis University Health Sciences Center Institutional Review Board, and informed consent was obtained from the subjects' parents. CTEF cultures were established by an explant technique and subcultured as previously described [31-33]. The thymic capsule was removed by blunt dissection and the remaining tissue minced into 1 mm3 fragments which were agitated gently in Dulbecco's modified Eagle's medium (DMEM) (GIBCO BRL; Long Island, NY) supplemented with 5% fetal calf serum (FCS) to wash out as many thymocytes as possible. Thymocytes and other hematopoietic cells including DCs were depleted from the fragments by incubation in 1.35 mM 2′-deoxyguanosine (Sigma; St. Louis, MO). The CTEFs were incubated on sterile gelfoam tissue rafts (1 cm × 3 cm) set in 6 cm Petri dishes. The tissue rafts were covered with sterile filters and were partially immersed in Ham's F-12 medium supplemented with 10% FCS, 1.35 mM 2′-deoxyguanosine, 25 mM HEPES, 2 mM glutamine, 50 U/ml penicillin, 1 μg/ml streptomycin (complete medium) at 37°C in a 5% CO2 atmosphere for 14 days for the normal thymic tissue and five to six days for the HIV thymic tissue. Subsequently the CTEFs were transferred to 24-well transwell culture plates (Nunc; Naperville, IL) and incubated for three to seven days in Iscove's/Ham's medium at a 1:1 ratio supplemented with 5% FCS, 0.4 μg/ml hydrocortisone (Calbiochem Behring; LaJolla, CA), 11 ng/ml epidermal growth factor (Collaborative Research; Bedford, MA), 1 × 10–10 M cholera enterotoxin (Sigma), 5 μg/ml insulin (Sigma), 1.8 × 10–4 g/ml sodium pyruvate, 50 μg/ml gentamicin, and 0.2 μg/ml fungizone prior to being seeded with CD34+CD38 enriched SCs. A portion of each thymic tissue was reserved for tissue analysis by LSCM and placed in the coculture media described below.

Isolation of CD34+ SCs

After completion of routine BM cell harvests, marrow infusion filters were collected as discarded specimens. The filters were removed from the cell pack under sterile conditions and the remnant marrow tissue pieces and spicules were washed from the filters by flushing with phosphate-buffered saline (PBS). The BM cells were diluted in normal saline, and nucleated cells obtained by Ficoll-Hypaque density centrifugation. Soybean agglutinin (SBA) positive cells were removed by adherence on SBA-coated flasks (Applied Immune Sciences, Inc.; Menlo Park, CA). The CD34+ SCs were positively selected by panning, using CD34+ monoclonal antibody (mAb)-coated flasks (Applied Immune Sciences, Inc.) [34]. Remaining CD38+ cells were depleted from the CD34+ population by immunomagnetic microbead depletion. One mg magnetic microbeads (Immunotech; Cedex, France) were precoated with anti-CD38 mAb (Immunotech) at a ratio of 5 μg per 1 mg microbeads for one h at 23°C, and then washed. For the negative depletion of CD38+ cells, 1 to 10 × 106 CD34+ cells were incubated with 1 mg of anti-CD38-coated magnetic microbeads for 30 min with gentle rotational mixing at 23°C. The CD38+ cells were removed using a magnetic separation apparatus (Dynal; Oslo, Norway). Anti-CD38 depletion was repeated to further remove any residual CD38+ cells. The purity of the fractionated population was monitored using flow cytometry. CD38+ cells comprised <5% of the CD34+CD38 population. T cell contamination of the CD34+CD38 population was <1% CD3+ cells.

Coculture of CTEF and CD34+ SCs

The CTEF cultures were set up in parallel from both normal thymic tissue and HIV thymic tissue in 24-well transwell plates. Negative control cultures with CTEF alone or with CD34+ SCs alone were included. The CTEFs were seeded with 5 × 104 CD34+ SCs/well. The cocultures were placed in medium containing Iscove's/Ham's medium at a 1:1 ratio supplemented with 3% FCS, 10% serum-free media substitute HL-1 (Ventrex; Walkersville, MD), 5 μg/ml insulin, 2 mM glutamine, 11 ng/ml epidermal growth factor (Collaborative Research), 50 U/ml penicillin, 50 μg/ml streptomycin, and 25 U/ml of interleukin 2 (IL-2) at 37° C in a 5% CO2 atmosphere. The medium was replaced with fresh medium every four days.

Detection of T Cell Phenotypes

T cell surface phenotypes of thymocytes freshly isolated from the processed tissue, and later from thymocytes harvested after the SC coculture, were determined by staining with a panel of mAbs conjugated with different fluorochromes and analyzed by flow cytometry. Cells were aspirated from the culture wells, washed, resuspended in RPMI containing 5% FCS, then subjected to Ficoll-Hypaque density centrifugation to remove dead cells. The mononuclear cells isolated from the interface were washed, resuspended in PBS containing 10% FCS and mixed with optimal concentrations of mAb. Ten thousand cells were stained with mAb and 5000 events recorded for analysis. mAbs used included: CD45, CD14, CD34, CD3, CD4, CD8, CD16, CD38, CD56, CD20, T cell receptor for antigen (TCR)αβ and TCRγδ obtained from Becton Dickinson (San Jose, CA) (Table 1). Negative controls (murine IgG1 fluorescein isothiocyanate/phycoerythrin/peridinin chlorophyll protein [FITC/PE/PerCP]) (Becton Dickinson) emitted only a small percent of positive fluorescence and were used to set the positive gate. The lymphocyte region was selected using CD45 gating. Combinations of mAbs were used to identify immature thymocytes, such as double positive (CD3CD4+CD8+) and triple positive (CD3+CD4+CD8+) thymocytes, and mature CD3+CD4+CD8 or CD3+CD4CD8+ T cells. FITC, PE and PerCP fluorescence was measured using the appropriate bandpass filters in a FACScan flow cytometer (Becton Dickinson). The analysis of three color populations was performed using the WinList program (Verity; Topsham, ME). In order to verify the cells harvested from the coculture were derived from the SC donor, CD34+CD38 SCs were prelabeled with 5-(and-6)-(((4-chloromethyl)benzoyl)amino)tetra-methyl-rhodamine (CMTMR; Molecular Probes, Inc; Eugene, OR) prior to CTEF coculture. CMTMR is retained in living cells through multiple cell divisions and is not transferred to adjacent cells. The fluorochrome can be detected using standard cytometry or fluorescence microscopy with excitation at 488 nm and emission at 580-610 nm bandpass however, the use of the cell tracker probe CMTMR precludes the use of double and triple labeling. Therefore a second series was examined in which the SCs were not prelabeled with CMTMR in order to assess the phenotype of the harvested cells using triple labeling.

Table Table 1.. Monoclonal antibodies and probes used throughout this study
  Thymic epithelia
TE-3MAS 251cortical thymic epithelium
TE-4MAS 252psubcapsular + cortical epithelium TNCs
CDR2 pan epithelium
KerCK-5Keratin intermediate filaments, epithelium
VimV-9Vimentin intermediate filaments, mesenchyme
  Mononuclear cells
CD3Leu4multichain receptor associated with TCR
CD4Leu3acytoadhesion glycoprotein binds major histocompatibility complex (MHC) II
CD8Leu2acytoadhesion glycoprotein binds MHC I
CD14Leu M3monocytes, DCs
CD16Leu IIcgranulocytes, NKs, FcRIII
CD20Leu 16B cells
CD34HPCA-2BM progenitor cells
CD38Leu 17committed SC progenitors
CD45HLe-1leukocyte protein tyrosine phosphatase family
CD56Leu 19NCAM isoform on NK cells
TCRαβWT31alpha/beta chains of TCR
TCRγδIIF2gamma/delta chains of TCR
  HIV probe
HIV ProteaseH2930fluorogenic peptide released upon HIV protease-specific cleavage of quenched fluorophore substrate

SC Migration Using Time-Lapse Videomicroscopy

TEC monolayers were established as previously described [32]. The TEC monolayer was placed in a Focht's Environmental Chamber where TEC dynamics were observed for several hours prior to the addition of SCs. Purified CD34+ SCs prelabeled with the cytoplasmic probe CMTMR were added to the chamber at a port distant from the TEC monolayer. The SCs had to traverse a distance up to 2 cm to encounter the TEC monolayer (a distance representing 2000× the cell diameter). CD34+ SC migration was videotaped using a Sanyo SVHS 4-head Time-Lapse Video Cassette Recorder over a 48-96-h period using a Zeiss Axiovert 135 microscope equipped with a Focht's environmental chamber and DIC optics. Analysis of average centroid movement (μm/h) using gradient reservoir chambers was not performed. An aliquot of the enriched CD34+ SCs was submitted for fluorescence-activated cell sorter (FACS) analysis both prior and subsequent to their addition to the TEC monolayer. Video stills were printed on a Sony Mavigraph UP-5200 MD color video printer (Figs. 1-5, Figure 2., Figure 3., Figure 4., Figure 5.).

Figure Figure 1..

Confocal immunofluorescence microscopy of the CD34+SC population within freshly acquired normal and HIV-affected thymus.The CD34+SCs are labeled with PE and are visualized in red. Structural detail of the thymic tissue was obtained by reflective imaging — visualized in green. Spectral overlap of the red channel and the green channel will result in color variation of the combined channel as orange to gold. A) Negative control — note red fluorescence (background) occurs primarily along vessel. B) Normal thymic tissue [100 × /N.A.1.4]. C) Close-up of CD34+/PE in subcapsular cortical area [63 × with 4 × Zoom]. D) SCs present in the subcapsular cortical area of HIV-affected thymic tissue — optical sectioning representing a 30 μ tissue section. E) and É) HIV-affected thymic tissue — two orthogonal Z sections showing cross sections in both XY and XZ axis. Note the presence of fused membranes between some of the CD34+cells (arrow). F) SCs from HIV-affected thymus demonstrating the presence of HIV protease. The reflective image is visualized in red while the positive signal for HIV protease is visualized in green. Negative controls with fluorophore alone, buffer alone and normal tissue contained no background fluorescence.

Figure Figure 2..

Normal TEC populations, visualized by confocal immunofluorescence microscopy.The TECs are labeled with PE and are seen as red. Structural detail of the thymic tissue obtained by reflective imaging is shown as green. A) Optical sectioning of TE-4+TECs along the subcapsular cortical region. B) Optical sectioning of CDR2+TECs located primarily in the cortex. C) Orthogonal Z section showing cross section in both XY and XZ axis. Note the stellate morphology of the TEC positive for CDR2.

Figure Figure 3..

Confocal immunofluorescence microscopy of partial TEC regeneration in HIV-affected thymic tissue in vitro.The TECs are labeled with PE and are visualized in red. Structural detail of the thymic tissue obtained by reflective imaging is shown in green. A) Normal thymic tissue TE-4+TEC in subcapsular cortical area [left] and HIV thymic tissue TE-4+TEC [right] (series collected with identical settings). B) Superimposed digital images from 100 μ HIV thymic tissue section labeled with CDR2-PE. C) A three-dimensional reconstruction with animated viewing from an oblique angle of (B) showing blunted stellate morphology of cells.

Figure Figure 4..

Confocal immunofluorescence microscopic view of a section through normal thymic tissue versus AIDS-infected thymus.Sections were dual labeled for vimentin intermediate filaments (Texas red), indicating cells of mesenchymal origin, and keratin intermediate filaments (fluorescein), delineating cells of epithelial origin.

Figure Figure 5..

Dynamic videomicroscopy of CD34+SC migration to TCD-TEC monolayers.A) Normal TEC monolayer. B) Migration of normal donor-derived CD34+SCs to normal host-derived TCD-TEC monolayer. C) Normal TEC monolayer. D) Migration of HIV+donor-derived CD34+SCs to normal host-derived TEC monolayer. E) TEC monolayer derived from HIV+host. F) Migration of normal donor-derived CD34+SCs to HIV+host-derived TEC monolayer. (Negative controls to nonthymic tissue not shown.)

Thymic Tissue Analysis Using LSCM

LSCM was selected for imaging the labeled tissue sections because of its ability to record images with enhanced XY resolution, with multiple fluorophores, in addition to recording reflected or transmitted images to provide high resolution, detailed morphology [35]. Orthogonal Z sections were obtained from several fields of each specimen to allow three-dimensional image reconstruction, quantitative image analysis and micrograph reproduction. Both normal thymic tissue and thymic tissue obtained at autopsy from children with end-stage HIV were analyzed using LSCM. Normal and HIV-affected CTEF were set up in parallel using dual chamber culture slides, which were maintained in coculture media with thymic-conditioned medium and T cell-conditioned media, prepared as previously described [31]. The conditioned media added as a 10% supplement served as a source of stem cell factor, IL-1β, IL-3, IL-7, transforming growth factor β and GM-CSF (determined by immunoblot analysis, data not shown). Small fragments of tissue were placed in chamber slides (Nunc) and stained with a panel of different mAbs at optimal concentrations after set intervals of culture. In addition to the above, mAbs included: TE-3, TE-4, CDR2 (a gift kindly provided by Dr. Richard Hong); HIV protease (Molecular Probes; Eugene, OR); and CK-5 (antihuman keratins 8, 18 antibody; Sigma) and V-9 (antihuman vimentin antibody; Dako; Carpenteria, CA). Appropriate isotype fluorochrome-conjugated secondary antibodies included goat α mouse F(ab)2 IgG or IgM conjugated to PE (Tago; Burlingame, CA), and rabbit α mouse F(ab)′2 IgG conjugated to FITC (Sigma).


Micrographs of tissue samples were obtained with a Zeiss LSM 410 scanning laser confocal microscope system built around a Zeiss 135 Axiovert inverted microscope. This particular system employs three photomultiplier tube (PMT) detectors, any two of which could be used for simultaneous recording of dual fluorescence labels at two different wavelengths, or for simultaneously collecting one fluorescence label image and one reflection image. The third PMT was used for simultaneously imaging the nonconfocal transmission image of the sample along with the two other PMT channels, and for separately recording a fluorescence image of a third fluorophore. An Omnicron argon/krypton dual gas laser set to emit the 488, 568 and/or 647 nm laser lines was engaged.

Fluorescence images of the FITC label were recorded using a 488 nm laser line, recording the image through a broad bandpass emission filter of 510-540 nm. PE fluorophore images were obtained with a 568 nm laser line, recording the image through a broad bandpass emission filter of 575-640 nm. PerCP fluorophore images were obtained with a 488 nm laser line, recording the image through a broad bandpass emission filter of 670-810 nm. Appropriate dual laser line beam splitters were used when recording single fluorophore or dual fluorophore images. Brightness (PMT gain) and contrast (PMT DC offset) were set on appropriate positive control samples to obtain a full 8 bit grey scale rendering of each image. In the case of sample versus control comparisons, the brightness and gain settings for the sample were held fixed for the subsequent recording of the image of the control sample. Microscope images were obtained using either a Zeiss 63× oil/NA = 1.25 Neofluor objective or a Zeiss 40× Achroplan 0.60 Korr Ph2 objective. The pinhole of the confocal system was adjusted for maximal spatial resolution which for the 63× objective yielded an ultimate XY resolution of ∼0.20 μm in the XY plane. Simultaneous recording of either single or dual fluorescence-labeled images plus epireflection images were primarily utilized in this study. Digital image data (with some files exceeding 14 Mb in size) were stored on a 256 Mb MO drive. Three-dimensional image reconstruction from Z sections was performed with the software provided by Zeiss for the LSM 410 system.


Statistical analysis was performed using Student's two-tailed t-test comparing normal allogeneic and HIV thymus cultures.


Morphological Assessment of End-Stage HIV Thymus

HIV thymus was available for evaluation from three children with AIDS who were infected either in utero or during the newborn period. Each thymus was small and notably dysplastic. Structurally, there were signs of significant stromal dysgenesis throughout the tissue. The capsular area was thickened with discontinuous patches of subcapsular tissue and effacement of the cortical epithelium. There was a prominent hyaline change with fatty infiltration throughout the tissue. The lobular architecture was collapsed and depleted of lymphocytes. Corticomedullary differentiation could not be appreciated. Hassals corpuscles were absent. In an effort to further define the thymic stromal damage induced by HIV, a detailed histologic study was performed using laser scanning confocal microscopy. Generation of reconstructed three-dimensional images with viewing of stereo images, orthogonal Z sections and animated rotational scanning allowed through tissue viewing of otherwise structurally compromised thymic tissue.

Immunohistochemical Localization of CD34+ SCs by Laser Confocal Imaging

Normal thymic tissue was initially evaluated using LSCM. Tissue sections of 50 μ (maximum of 100 μ) were able to be visualized without structural distortion. Transitional areas were sought in an effort to orient the optical dissection. In an effort to visualize detailed cell morphology underlying the fluorescence images, epireflection images were simultaneously recorded with the fluorescence images. Figure 1A represents the negative controls. With regard to the SC compartment of the thymus, the subcapsular areas in normal thymic tissue contained CD34+ SCs (Figs. 1B, 1C). The HIV-affected thymic tissue contained a predominantly thickened and reticular subcapsular cortex. The discontinuous patches of subcapsular cortical tissue contained scant patches of TE-4+ TEC (image not shown) and predominant clusters of stem cells (Fig. 1D). Amid these dense tufts of SCs, there were clusters of cells with fused cell membranes between adjacent cells (Figs. 1E and 1É). No multinucleated giant cells were appreciated. Several specimens were evaluated in vitro for their ability to cleave an HIV protease-specific peptide. The clusters of CD34+ SCs in the subcapsular cortex were presumably infected with HIV, based on their ability to cleave an HIV protease-specific probe (Fig. 1F). Further quantitation of the HIV-infected cells was not performed. The cortical areas of the HIV tissues were effaced and devoid of both lymphocytes and epithelial cells. In addition, a distinct loss of Hassall's corpuscles was observed. Both natural killer (NK) and B cells were present in the HIV tissue in slightly higher proportions than in normal thymic tissue; however, this may be reflective of the significant T cell depletion present (Table 2).

Table Table 2.. Thymic tissue histology and immunocytochemistry
ArchitectureCell types identified
SourceTypeCMJInitial TE markersStem cell/LymphocytesIn vitro TE markers
 G/L/H+ or –TE-4TE-3CDR2CD34T cellB cellNK cellTE-4TE-3CDR2
  1. a

    HIV thymus obtained from children with end-stage AIDS, normal thymus obtained from children undergoing corrective cardiac surgery. Architecture evaluated at low power as G = guirlandes, L = lobular/involuted, H = hyalinized, CMJ = presence or absence of corticomedullary junction. In vitro culture of thymic tissue maintained in conditioned media as described for a minimum of two months prior to staining. Other mAbs TE-7, TE-15, MHCI, MHC-II tested were negative in the HIV tissue.

HIVG + Hscant+scant++++
HIVL + H+++++

In regard to the thymic stroma, the subcapsular area of the normal thymus contained modest numbers of TE-4+ TEC (Fig. 2A). The cortical stromal tissues predominantly stained with TE-3 and CDR2 (Figs. 2B and 2C). Lymphocytes were present throughout the cortical and medullary regions. Corticomedullary differentiation was easily appreciated as was the presence of Hassall's corpuscles.

When thymic epithelial fragments from HIV+ children were cultured in vitro in the presence of supplemental thymus hormones (conditioned media), a partial regeneration of certain subpopulations of the thymic epithelium occurred. The subcortical epithelium composed primarily of TE-4 and CDR2 were present in increasing numbers during the in vitro culture period (Fig. 3). However, the cortical epithelium defined by the cell population TE-3 was not detected initially or following in vitro culture. This may represent irreversible damage to certain subpopulations of the thymic epithelium induced by HIV, or may reflect constraints imposed by the culture conditions.

Vimentin and Keratin Immunostaining

The epithelial and mesenchymal cellular components of thymic tissue were further characterized by vimentin immunostaining of cells in the stromal compartment, and keratin immunostaining of epithelial cells throughout the tissue (Fig. 4). Disruption of morphological architecture was the major finding in the HIV tissue compared to matched normal tissue.

SC Migration Visualized by Time-Lapse Videomicroscopy

Although significant T cell depletion was present in all HIV thymi examined, CD34+ SCs were present in the subcapsular regions of the tissue. The presence of BM-derived SCs suggests the capacity of the BM-ME to export pro-T cells, and their ability to home in significant numbers, and to marginate into the thymus remains functional, despite significant thymic end organ damage. Chemotaxis and chemokinesis of CD34+ SCs are mediated by soluble factors. Thymic hormones such as thymotaxin, thymulin and thymopoietins have been implicated in mediating pro-T migration from the BM and immigration into the thymus.

In the present study, chemokinesis of CD34+ SCs was evaluated and recorded using time-lapse videomicroscopy. Baseline studies were performed using normal CD34+ SCs migrating toward a TCD normal donor-derived TEC monolayer (Fig. 5A). Analysis of the cells obtained after culture on the thymic epithelial monolayer (Fig. 5B) by flow cytometry verified the acquisition of early T cell differentiation markers (data not shown). Studies were performed on SCs derived from an HIV+ donor migrating to a normal host TCD-TEC monolayer (Figs. 5C and 5D). In addition, TCD-TEC derived from an HIV host were capable of attracting normal CD34+ SCs and maintained a small percentage of the CD34+ phenotype over the recorded period (Figs. 5E and 5F). Analysis of cells obtained after culture on the HIV donor-derived thymic epithelial cell monolayer failed to show acquisition of CD3, CD4 or CD8 T cell differentiation markers.

Arrest of T Cell Differentiation in HIV Thymus

Thymocytes freshly isolated from both normal thymic tissue and HIV-affected thymic tissue were analyzed by flow cytometry. The cells derived from the HIV-affected thymic tissue were significantly T cell-depleted. Both CD34+ SCs and CD3CD4CD8 triple negative thymocytes were present, as were small populations of CD3+CD8+, NK cells and B cells. The proportions of double and triple positive thymocyte subpopulations were significantly decreased (Table 3A).

Table Table 3A.. Phenotypic analysis of freshly isolated thymocytes show an arrest of T cell differentiation in HIV thymus
 Mean Percent Distribution
PhenotypeNormal thymus (6)*HIV Thymus (3)*
  1. a

    *Number of experiments in parenthesis


In Vitro Thymopoiesis of CD34+CD38 SCs in HIV CTEF

Enriched CD34+CD38 SCs isolated from normal BM were cocultured with thymic epithelial fragments from either normal children or children with AIDS (Table 3B). The ability of the thymic epithelia derived from an HIV+ host to support thymopoiesis was markedly diminished compared to normal controls. There was decreased cellular differentiation; 13,000 cells/CTEF in HIV compared to 58,000 cells/CTEF in normal controls (p < 0.01). Examination of thymocytes and T cell populations by flow cytometry revealed markedly abnormal differentiation and maturation in the HIV thymic organ cultures. This was most evident in CD4+-bearing cells, as shown in Table 3B. Abnormal thymocyte populations were manifested by decreased expression of CD3+ cells in HIV versus controls, 36% versus 75% (p < 0.03), respectively; and in CD4+ cells, 20% versus 52% (p < 0.05), respectively (Table 3B). Furthermore, there were increased triple negative CD3CD4CD8 thymocytes in the HIV thymus compared to normal, 28% versus 1.5%. Subpopulations of CD4+-bearing thymocytes were also decreased in the HIV thymic organ culture. Expression in normal versus HIV thymus, respectively, of double positive CD4+CD8+ thymocytes (24% versus 9%, p = 0.053); triple positive CD3+CD4+CD8+ thymocytes (21% versus 12%, p < 0.01); and single positive mature CD4+ T cells (28% versus 4%, p = 0.011) were markedly decreased in the HIV TECs. The acquisition of CD8 and TCRαβ was normal in the HIV+ thymus, suggesting that HIV+ thymic epithelia can support CD8+ lineage development (Fig. 6).

Table Table 3B.. Abnormal thymopoiesis of CD34+CD38 SCs in HIV + CTEFs
PhenotypeNormal thymusaHIV thymusap valueb
  • a

    a% of lymphoid cells stained with listed mAbs ± 1 SD.

  • b

    bp value using Student's t-test.

CD3+75 ± 1136 ± 100.03
CD4+52 ± 1920 ± 60.05
CD8+67 ± 964 ± 20 
CD3CD4CD8(TN)1.5 ± 0.428 ± 20 
CD3CD4+CD8+(DP)24 ± 119 ± 40.05
CD3+CD4+CD8+(TP)21 ± 1312 ± 6 
CD3+CD4+(SP)28 ± 94 ± 40.02
CD3+CD8+(SP)25 ± 919 ± 14 
CD3+ TCRαβ84 ± 581 
CD3+TCRγδ1.0 ± 0.50.5 
Figure Figure 6..

FACS analysis of CD3, CD4 and CD8 expression of T cells obtained from the coculture of lineage negative CD34+SCs with either normal allogeneic CTEF or HIV CTEF.There is a notable decrease in CD3+and CD4+expression, but normal expression of CD8+thymocytes in the HIV thymic organ culture (A-C), compared to the normal control (D-F).


How is the thymus gland damaged during HIV infection? Is there irreversible damage produced by protective host responses such as TCTL, Ab-mediated responses with bystander damage to the thymic epithelium, autoimmune damage due to cross-reactive antibody with thymic stromal elements, direct cytopathicity, or HIV thymotropic variant strains which affect the epithelial or dendritic networks of the thymus? One fundamental issue which needs to be resolved regarding thymopoiesis in HIV is whether the virus induces a primary acquired thymic stromal defect which compromises lymphopoiesis, or whether there is a secondary and contingent effect on the thymic stroma which is associated with the SCs entering and populating the thymus which have become HIV-infected. If the latter hypothesis holds, then early antiviral gene therapy of SCs should provide sufficient thymocyte protection from infection with HIV. If the former is true, then T cell reconstitution in HIV disease will require protection of the lymphoid SC compartment, as well as the thymic stroma and supporting elements from HIV infection. At this time, the data presented is preliminary and further studies which can include different AIDS patient populations and age matched “stressed thymus” controls will be necessary.

The thymic tissues studied in this series had significant end organ damage, but still allowed several important observations. Migration studies show that HIV-affected thymic tissue is capable of attracting CD34+ SCs in vitro. During HIV infection, the migration of SCs to the thymus in vivo is supported by the demonstration of CD34+ SCs in the subcapsular cortical areas of HIV+ thymus. The SC present in the subcapsular cortical areas of the HIV thymus can be infected by HIV, as evidenced by their ability to cleave an HIV-specific protease probe and develop distinct changes in cell morphology (syncytia). Therefore, the SC compartment of the thymus which is responsible for maintaining both the integrity of the lymphoid compartment and for providing signals for the regeneration of the epithelial compartment is damaged during infection with HIV. There are significant alterations in the thymic stromal and supporting cell populations present in the HIV thymus. The thymic stromal SCs present in epithelial guirlandes are able to differentiate in vitro, but the functional status of these cells remains to be determined. Our data suggest that the HIV thymus does support thymopoiesis; however, it is compromised both in rate and the diversity of thymocyte subpopulations which can be generated. Unfortunately, HIV antiretroviral drugs were not available for use in this study. Thus, it is unclear from these experiments whether the decreased expression of CD4+ thymocytes was due to destruction or a developmental arrest at the triple negative stage of thymocyte development. Further analysis of CD44 and CD25 expression and viral load in the presence of antiretrovirals is planned.

The normal thymus is capable of maintaining both a pro-T (multilineage) and a pre-T (committed) CD34+ SC population [36]. The integrity of the thymic epithelial ME is dependent on a continual supply of SCs. The pro-T cell population provides not only lymphocyte precursors but also DCs which support the stromal cell populations and serve to select developing lymphocytes. Thymic stromal dysgenesis has been noted when there are alterations induced in the ingressing SC populations. An example of the inductive ME provided by SCs has recently been described by Hollander et al. in transgenic mice defective in the CD3ε chain [37]. In this model, the thymic architecture lacked distinct cortical and medullary compartments. Failure to form even a rudimentary corticomedullary ME in these CD3ε –/– tg is presumably due to a change in soluble cytokines or direct signals transmitted to TECs by transiting SCs. Normal thymopoiesis could only be reconstituted by the transplantation of wild-type SCs in fetal CD3ε –/– tg thymic stroma, further indicating a defined developmental window for inducing corticomedullary differentiation. A broader developmental window may exist in humans, as reflected in severe combined immunodeficiency syndrome (SCIDS). Long-term T cell reconstitution has been achieved in SCIDS patients transplanted with normal BM SCs [26].

The results presented in this report are preliminary, but establish essential background information regarding the morphological components and utility of the thymic culture model. Indeed, further studies can now address what parameters define thymic failure. If partial thymic stromal regeneration can be appreciated in vitro, is thymic stromal regeneration possible in vivo? With respect to future clinical applications in cellular reconstitution — will the regeneration (partial or complete) of an autologous thymus affect the long-term engraftment of a transplanted allogeneic thymus? Considering the highly stressed turnover of CD4+ T cells in the peripheral lymphocyte pools during HIV infection in addition to the reduced capacity for thymopoiesis, therapeutic intervention should address not only arresting HIV replication but reestablishing a lymphopoietic steady state. Therapeutic strategies directed at providing protection of the BM and thymic ME could provide multilineage hematopoietic protection from HIV and may improve T cell reconstitution, thereby serving to inhibit the progression of HIV disease.


The authors extend their condolences to the families who have lost their children to AIDS and wish to express their heartfelt appreciation for the donation of thymic tissues. Acknowledgment is extended to the BMT Division for bone marrow specimens and to Dr. Andrew Fiore who provided normal thymus tissue; Pat Whelan for typing assistance; and to Dr. C. George Ray for editorial advice and review. A special acknowledgment is extended to Dr. Richard Hong for technical advice.

These studies were supported in part by the Minority Investigator Supplement Award to NIH Grant CA59702-03S2; a Fleur de Lis Fellowship Research Support Fund; and IMMUNO-US, Inc.