Generation of Intestinal T Cells from Progenitors Residing in Gut Cryptopatches

See allHide authors and affiliations

Science  10 Apr 1998:
Vol. 280, Issue 5361, pp. 275-278
DOI: 10.1126/science.280.5361.275


Cryptopatches (CPs) are part of the murine intestinal immune compartment. Cells isolated from CPs of the small intestine that were c-kit positive (c-kit+) but lineage markers negative (Lin) gave rise to T cell receptor (TCR) αβ and TCR γδ intestinal intraepithelial T cells after in vivo transfer or tissue engraftment into severe combined immunodeficient mice. In contrast, cells from Peyer's patches and mesenteric lymph nodes, which belong in the same intestinal immune compartment but lack c-kit+Lin cells, failed to do so. These findings and results of electron microscopic analysis provide evidence of a local intestinal T cell precursor that develops in the CPs.

The gastrointestinal mucosa is one of the largest interfaces in the body with only a single layer of epithelium separating the exposed external lumen from the internal milieu, and in primitive vertebrates, only gut-associated lymphoid tissues are present (1). Therefore, from an evolutionary perspective, it is possible that the intestinal tract of higher vertebrates retains some self-supporting immune system to ensure internal integrity (2). The numerous intestinal intraepithelial T cells (IELs) have cellular and behavioral characteristics distinct from those of other peripheral T cells (2-6) and are enriched with TCR γδ T cells (7). In mice, most TCR γδ IELs and many TCR αβ IELs, unlike blood-borne T cells, use the CD8αα homodimer (2, 8) instead of CD8αβ and develop somewhere in the intestinal mucosa without passing through the thymus (2, 4, 8, 9).

Our immunohistochemical search (10) for anatomical sites of IEL generation revealed multiple tiny clusters [the cryptopatches (CPs)] filled with ∼1000 c-kit+IL-7R+Thy1+lymphohematopoietic cells in the lamina propria (LP) of the murine intestinal crypt. The presence of this phenotype suggested that the cells of the CPs could be T lineage–committed precursors (10). CP cells are widely distributed in small numbers throughout the length of the intestine (10) and are visible stereomicroscopically from the serosa, although in our study this was strain dependent—for example, observation was easiest in BALB/c mice (Fig. 1A) and most difficult in C57BL/6 mice. Subsequently, we isolated tiny fragments of the small intestine (∼0.13 mm3) containing one CP using an amputated and tapered 21 G needle (Fig. 1A). In one such fragment containing a CP and a fragment extracted from the region without CPs, we detected 1000 (range, 650 to 2000) and 60 (range, 25 to 150) lymphoid cells, respectively.

Figure 1

Isolation of CPs and flow cytometric analysis of CP, PP, and MLN cells obtained from 4-week-oldnu/nu mice, and in vivo exploration of precursor T cell activity in c-kit+Lin and c-kitsubpopulations of nu/nu CP cells. (A) Stereomicroscopic views of (a) jejunal PPs, (b) jejunal CPs (arrowheads), and (c) a CP at higher magnification. Circle indicates the region containing one CP to be extracted from the small intestine with the aid of a 21 G needle (inner diameter, 570 μm). (B) Flow cytometric profiles of CP, PP, and MLN cells. Lin (lineage markers; CD3, B220, Mac-1, Gr-1, and TER119). (C) Generation of TCR αβ and γδ IELs by c-kit+Lin CP cells from nu/numice. Five weeks after transplantation with 1 × 104c-kit+Lin (a) or c-kit (b) CP cells, IELs isolated from 2 Gy–irradiated SCID recipients and from age-matched wild-type C.B-17 mice (c) were incubated first with biotinylated anti-TCR αβ (H57-597; PharMingen) and then streptavidin-PE and FITC-conjugated anti-TCR γδ (GL-3; PharMingen). Dead cells were excluded by PI gating, and the percentage of positive cells in the corresponding quadrants is shown.

We analyzed CP, PP (Peyer's patch), and MLN (mesenteric lymph node) cells obtained from 4-week-old athymic (nu/nu) weanling BALB/c mice by flow cytometry (11). All locations had no CD3+ T cells (Fig. 1B) or CD8+ cells, and most CP cells were c-kit+Lin (Lin; CD3, B220, Mac-1, Gr-1, and TER119) (Fig. 1B), whereas cells from PPs and MLNs (Fig. 1B), and cells from the tissue fragments without CPs, did not have a c-kit+Lin population. Almost all c-kit+Lin CP cells were IL-7R+Pgp-1+ (12), 14% of c-kit+Lin CP cells were CD4+(Fig. 1B), and 68% of c-kit+Lin CP cells were Thy1+ (Fig. 1B). Four disinct lymphocyte subsets were present among c-kit+Lin CP cells as determined by the expression of CD4 and Thy1. In the three lymphoid tissues of euthymic nu/+ littermates, only CP cells contained a dominant population (60 to 70%) of c-kit+Lin cells (12). Thus, CPs differed from PPs and MLNs with respect to lymphoid residents, although they all constitute organized gut-associated lymphoid tissue (GALT).

To determine whether progenitors for intestinal T cells are present in CP cells from nu/nu mice, we transplanted intravenously 10,000 c-kit+Lin or c-kit CP cells (Fig. 1B), sorted by flow cytometry, into 2 gray (Gy)–irradiated C.B-17/severe combined immunodeficient (SCID) mice (13) (Fig. 1C). The c-kit+Lincells but not c-kit cells were able to generate αβ and γδ T cells in the IEL compartment. The c-kit+Lin cells also gave rise to substantial αβ T cells in MLNs and to a lesser extent in the spleen and thymus, but were unable to reconstitute B cells in any of the recipient's lymphoid tissues (see below). For a more comprehensive transplantation study on c-kit+Lin CP cells, we used the whole CP cell population, rather than sorted c-kit+Lin cells, because considerable time was required for isolation of CP cells, and the subsequent procedure of cell fractionation often resulted in unforeseeable impairment of cell viability.

For this purpose, 5 × 104 CP, PP, or MLN cells obtained from 4-week-old nu/nu mice (Fig. 1B) were transplanted intravenously into 2 Gy–irradiated SCID mice, because of the extremely small number of mature CD3+ T cells (Fig. 1B) that could expand and survive for extended periods in immunodeficient recipient mice (14). Three weeks after transplantation, CP cells generated T cells in the IEL and MLN compartments (Fig.2, A to C) but few in the spleen and thymus compartments (Fig. 2, A and C). Absolute numbers of T cells recovered from these anatomical sites reached 3 × 106to ∼4 × 106, indicating that repopulation of T cells by CP cells accompanied cell proliferation. CP cells gave rise to αβ T cells in both the IEL and MLN compartments, but to γδ T cells only in the IEL compartment (Fig. 2B). The relative proportions of CD8+ to CD4+ and of CD8α+ to CD8β+ cells in the reconstituted MLNs were 6 to 18 and 1.1 to 1.3, respectively. The collective reconstitution of T cells from CP cells in the posttransplantation time frame was quantitated (Fig.2C). However, CP cells were unable to generate B220+ or sIgM+ B cells (Fig. 2A), and cells from PPs (Fig. 2A, third column) and MLNs (Fig. 2A, fourth column) that lacked the c-kit+Lin subset (Fig. 1B) did not fill the empty T and B cell compartments as of 7 weeks after transplantation (12), indicating that the relevant T lineage–committed precursors in CPs are indeed c-kit+Lin cells.

Figure 2

CP but not PP and MLN cells isolated from 4-week-old nu/nu mice generate T cells in vivo after intravenous transplantation into 2 Gy– irradiated SCID mice. (A) Three weeks after transplantation with 5 × 104 CP, PP, or MLN cells, IELs and MLN, spleen, and thymus cells isolated from the corresponding 2 Gy–irradiated recipient mice, from 2 Gy–irradiated sham-transplanted SCID mice (negative control), and from age-matched C.B-17 mice (positive control) were incubated first with biotinylated anti-CD3 and then with streptavidin-PE and FITC-conjugated anti-B220. Dead cells were excluded by PI gating. FACScan profiles of inguinal LN cells isolated from the same SCID recipients of CP cells were comparable with those of MLN cells shown in this figure. A substantial number of IELs (first row) were B220 dull positive as reported elsewhere (21), but they were cell surface sIgM (12). (B) Three weeks after transplantation with 5 × 104 CP cells, IELs and MLN cells isolated from the SCID recipients and from age-matched C.B-17 mice (positive control) were incubated first with biotinylated anti-TCR αβ and then with streptavidin-PE and FITC-conjugated anti-TCR γδ. Dead cells were excluded by PI gating. (C) Reconstitution of T cells in IEL (○), MLN (•), spleen (□), and thymus (▪) compartments of 2 Gy–irradiated SCID recipients transplanted with 5 × 104 CP cells from 4-week-oldnu/nu mice. The data are shown as the mean ± SD of four independent experiments with three to five SCID mice per analysis.

It is conceivable that B cell (Fig. 2A) and probably myeloid cell progenitors are minimal, if present at all, in CP cells and that T cell progeny of c-kit+Lin CP cells actually develop in the extrathymic site (or sites) because intestinal epithelium and MLNs are already colonized with T cells 2 weeks after transplantation, although, at this time, T cells are absent in the recipient's thymus (Fig. 2C). Our preliminary in vitro study confirmed that sorted c-kit+Lin cells from CPs failed to generate not only B and myeloid cell colonies in cultures on a monolayer of a fetal thymus–derived stromal cell line TSt-4 (15) but also T cells in deoxyguanosine-treated fetal thymus lobes in a high-oxygen submersion (HOS) organ culture (15). In contrast, this was not true of those from fetal liver and adult bone marrow.

Next, we engrafted 10 tissue fragments extracted from the small intestine of weanling nu/nu mice with or without CPs (Fig.1A) into the kidney capsule of unirradiated SCID mice. Consistent with the observation in the cell transplantation study, engraftment of CP+ but not CP fragments resulted in the emergence of T cells in the recipient's intestine and MLNs. Five weeks after engraftment, CD3+ T cells were generated in the villus epithelium (IELs) and to a lesser extent in the villus LP (Fig.3A). There were, however, marked differences between implanted (Fig. 3, C and D) and in situ (10) CPs. (i) The cellular mass of implanted CPs was enlarged about two- to threefold in diameter (Fig. 3, C and D) compared with that of in situ CPs (10). (ii) A substantial number of CD3+ T cells were detected in the central region of implanted CPs (Fig. 3C), but few were detected in in situ CPs (Fig. 1B) (10). (iii) c-kit+ cells found at high density throughout the in situ CPs (10) were compartmentalized in the peripheral region of implanted and enlarged CPs (Fig. 3D). Therefore, extraintestinal administration of c-kit+Lin CP cells by tissue engraftment and intravenous transplantation per se appears to be responsible for the generation of T cells in the ectopic anatomical sites (Figs. 2 and 3) as well as in the entopic intestinal mucosa. Mosley and Klein (16) reported that ectopic engraftment of fetal intestine in the kidney capsule of athymic recipient mice promotes the generation of peripheral T cells, whereas peripheral lymphoid tissues lack T cells in sham-engrafted athymic mice although they have their own intestine.

Figure 3

Generation of T cells in the intestinal villi of SCID mice after engraftment of tissue fragments containing CPs. SCID mice were engrafted by bilateral implantation under the kidney capsule with 10 tiny tissue fragments (∼0.13 mm3 per fragment) with or without CP (Fig. 1A) extracted from the small intestine of 4-week-old nu/nu mice. Five weeks after engraftment, immunohistochemical analysis of fresh cryosections prepared from the recipient's small intestine and kidneys containing implanted CPs was carried out as described (10). (A) CD3+ T cells were generated extensively in the intestinal epithelia of SCID mice engrafted with the CP+ tissue fragments (magnification, ×200). A limited number of T cells were also located in the LP. (B) CD3+ T cells were not generated in the intestinal villi of SCID mice engrafted with CP tissue fragments (×200). Immunohistochemical visualization of (C) CD3+and (D) c-kit+ cells in two consecutive tissue sections of a representative CP graft (×200). (C) CD3+ T cells were generated in the central region and (D) c-kit+cells were compartmentalized in the peripheral region of an implanted CP graft, respectively.

Electron microscopy (17) showed that numerous lymphocytes cross the basement membrane (Bm) that comes in contact with the CPs (Fig. 4A) and that a large number of lymphocytes are present in the epithelium adjacent to the CPs as compared with the epithelium of the villi and crypts (12). Scanning electron microscopy of epithelium-detached specimens (18) demonstrated that Bm covering the CPs has numerous holes (one hole per 10 μm2) ranging in diameter from 3 to 6 μm (Fig. 4B) and that these holes are often filled with round lymphoid cells (Fig. 4C, arrows), whereas crypt Bm creates deep cavities, is continuous, and is virtually devoid of holes (Fig. 4B, asterisks). These results suggest that the Bm adjacent to the CPs is a busy anatomical front line of the mouse small intestine where migration of lymphocytes into the epithelium takes place.

Figure 4

Electron microscopic analysis of the epithelium adjacent to CPs and the basement membrane (Bm) covering CPs. (A) Electron micrograph showing a peripheral region of a cryptopatch (CP) facing the epithelium (E). Two lymphocytes (L) are passing through the Bm (arrows). Bar, 2 μm. (B) Scanning electron micrograph of epithelium-detached Bm extending over a CP. Numerous holes, which probably facilitate the migration of CP lymphocytes into the epithelium, are apparent. No such holes occur in the Bm of the cryptal region (asterisks). Bar, 10 μm. Although not shown, epithelial Bm encompassing villous LP also contains a substantial number of holes, which are usually no larger than 1 μm in diameter, and no cellular elements plug the smaller holes. (C) Scanning electron micrograph showing a peripheral region of a CP covered by Bm. Part of the epithelium (E) has been detached from the specimen exposing the highly perforated Bm. Lymphocytes in a CP have balloon-shaped cytoplasmic processes protruding through these holes into the epithelium (arrows). Bar, 10 μm.

On the basis of the cellular, functional, and structural properties described above, CPs fulfill the criteria for lymphoid tissues wherein precursor IELs develop. The ability of the intestine to induce T cell lymphopoiesis makes phylogenetic sense (1,2) because the enteric mucous membrane is the locale in the body exposed to the greatest danger, and it is here that external antigens continually enter the body. In this context, identification of CPs will not only shed light on the intestinal events underlying the development of IELs (2-6, 8,9, 19) but may also offer additional clues for understanding the distinctive features of intestinal immune responses to luminal antigens (20) such as the induction of oral tolerance and immunopathogenesis of inflammatory bowel disease.

  • * To whom correspondence should be addressed. E-mail: ishikawa{at}


View Abstract

Navigate This Article