Extrathymic TCR Gene Rearrangement in Human Small Intestine: Identification of New Splice Forms of Recombination Activating Gene-1 mRNA with Selective Tissue Expression

Tissue

Specimens from apparently normal duodenum/jejunum were obtained from seven male and six female patients (median age: 64 37–76(37–76) years) undergoing bowel resection for cancer conditions, intestinal bleeding, ulcus, or strangi ileus. Duodenal/jejunal biopsies were collected from seven children as part of investigations of gastrointestinal symptoms, evaluation of asymptomatic growth-failure, or short stature. All had small intestinal mucosa with normal histology. Thymus tissues were from a pediatric patient who had to have part of the thymus removed during cardiac surgery for congenital heart disease and from Clontech Laboratories (Palo Alto, CA). Bone marrow aspirates were collected from three men (age: 62–72 years) with colorectal carcinoma but with no disseminated tumor cells in the bone marrow as determined by RT-PCR for carcinoembryonic Ag mRNA. One palatine tonsil was obtained from a child subjected to surgical treatment for idiopathic tonsillar hypertrophy. Informed consent was obtained from the adult patients and from the parents of pediatric patients.

Antibodies

mAbs used were: anti-CD1a mAb NA1/34, anti-CD2 mAb MT910, anti-CD3 mAb UCHT1, anti-CD7 mAb DK24, a mixture of anti-CD45 mAbs 2B11 and PD7/26, a mixture of anti-TdT mAbs HT-1/3/4, anti-epithelial cell Ag mAb BerEP4 (all from Dakopatts, Glostrup, Denmark), anti-CD117 mAb YB5.B8 (BD PharMingen, San Diego, CA), and anti-CD127 mAb R34.34 (Immunotech, Marseille, France). Polyclonal reagents were: IgG fraction of two rabbit anti-RAG1 antisera (Santa Cruz Biotechnology, Santa Cruz, CA and Scandinavian Diagnostic Services, Falkenberg, Sweden) and one rabbit anti-CD3 antiserum (Dakopatts), alkaline phosphatase-conjugated rabbit anti-digoxigenin (DIG; Boehringer Mannheim, Basel, Switzerland), HRP-conjugated F(ab′)₂ of sheep anti-mouse Ig (Amersham, Buckinghamshire, U.K.), FITC-conjugated goat anti-mouse Ig, FITC-conjugated swine anti-rabbit Ig, and biotinylated goat anti-rabbit Ig (all from Dakopatts).

Paramagnetic beads used for cell isolation were Dynabeads M-450 coated with goat-anti-mouse IgG (Dynal, Oslo, Norway) and charged with anti-CD3 mAb, anti-CD7 mAb, or BerEP4 and Dynabeads M-450 directly coated with anti-CD2 mAb (Dynal Biotech).

Isolation of leukocytes

Intestinal IEL and LPL were isolated from surgical samples as previously described (12, 13). Subpopulations of IEL and LPL were retrieved by sequential positive selection using paramagnetic beads charged with anti-CD3 mAb followed by beads charged with a mixture of anti-CD2 and anti-CD7 mAbs (14). IEL and LPL were isolated from intestinal biopsies of children as described (15) and cells of the T cell lineage were retrieved by positive selection using paramagnetic beads charged with a mixture of anti-CD2 and anti-CD7 mAbs. The isolation procedure was performed at 4°C and positively selected cells were frozen within 1 h after exposure to mAb. Microscopic inspection of all positively selected samples was performed to ascertain that only cells with surface-bound beads were present. Less than 4.6% of the unbound cells were CD3⁺ after treatment with anti-CD3 mAb-charged magnetic beads as determined by immunoflow cytometry. Calculations based on proportions of marker-positive cells and cellular yields showed that >98% of the marker-positive cells were bound to the relevant beads. Bone marrow cells (BMC) from bone marrow aspirates and PBMC were isolated by Ficoll-Paque (Amersham Pharmacia, Uppsala, Sweden) gradient centrifugation.

RNA preparation

Total RNA was extracted by the acid guanidinium thiocyanate-phenol-chloroform method as described (16).

Qualitative RT-PCR

RT-PCR analysis of different splice variants of RAG1 mRNA was performed using Moloney murine leukemia virus (MuLV) reverse transcriptase and random hexamers (Applied Biosystems, Foster City, CA) for reverse transcription. Specific primer pairs (see Table I⇓⇓) and recombinant thermostable Thermus thermophilus (rTth) DNA polymerase (Applied Biosystems) was used in the PCR. RT-PCR for RAG2 mRNA was performed using DNase-treated RNA (10) and the protocol and primers of Lynch et al. (11). Absence of signals from contaminating DNA was ascertained by PCR in which RNA was used as template. RT-PCR for β-actin mRNA served as control for intact quality of the RNA (for primer sequences, see Ref. 16).

Real-time quantitative RT-PCR (qRT-PCR)

Real-time qRT-PCR assays were constructed for different splice forms of RAG1 and pTα^a mRNA. Primers and probes are listed in Table I⇑⇑. Primers or probes hybridize over exon boundaries. Three different protocols were used: 1) one step: TaqManEZ technology using the 3′-primer for reverse transcription and the rTth DNA polymerase both for reverse transcription and in the PCR with specific primers and probe; 2) two steps: reverse transcription using MultiScribe reverse transcriptase and random hexamers followed by a PCR with specific primers and probe together with AmpliTaq Gold DNA polymerase; 3) nested: cDNA was synthesized using MuLV reverse transcriptase and random hexamers for 30 min at 42°C followed by 5 min at 99°C. Thereafter the outer primers were used with rTth DNA polymerase in a PCR with 20 amplification cycles of 1 min at 94°C, 1 min at 57°C, 2 min at 72°C, for RAG1, and 1 min at 94°C, 1 min at 64°C, 2 min at 72°C, for pTα^a. In both cases, the reaction was initiated by 2 min at 94°C and terminated with 5 min at 72°C. Finally, 2–5 μl of the 50-μl reaction mixture were subjected to real-time qPCR using rTth DNA polymerase. All assays showed linear correlation between log mRNA concentration and numbers of PCR cycles over a >5 log concentration range and the sensitivity varied between 10 and 250 RNA copies. None of the RT-PCR assays gave signals when nuclear DNA was used as a template. Release of the reporter dye from the probe during PCR was monitored by an ABI PRISM 7700 Sequence Detector (Applied Biosystems). For each system, a specific RNA copy standard was prepared (see below). Samples were analyzed in triplicates and expressed as copies of mRNA per microliter using the external copy standard. The concentration of 18S rRNA (Applied Biosystems) was determined in each sample and the results expressed as RAG1 or pTα^a mRNA copies per 18S rRNA unit. One 18S rRNA unit was defined as the signal obtained by 10 pg of a pool of total RNA extracted from PBMC stimulated with anti-CD3 mAb. This definition of a unit was chosen because it corresponds approximately to the 18S rRNA content of 100 freshly isolated intestinal IEL and LPL (mean ± 1 SD: 0.8 ± 0.5 U/cell; n = 14) and CD3⁺ cells thereof (1.1 ± 0.9 U/cell; n = 15), i.e., on average 1.0 ± 0.7 U/cell (n = 29). All analyses included in the study contained >125 18S rRNA units in each PCR.

Rapid amplification of the 5′cDNA end (5′-RACE)

Total RNA was converted into first-strand cDNA using the 5′-RACE cDNA synthesis primer (5′-CDS) (Clontech Laboratories), SMART II Oligo (Clontech Laboratories) and Superscript II MuLV reverse transcriptase (Life Technologies, Rockville, MD) according to the manual for samples containing <200 ng of RNA. The gene-specific primer was placed in the RAG1 exon 2 (5′-GCT TTG CCT CTT GCT TTC TCG-3′). The touchdown PCR was extended to 40 cycles. The 5′-RACE product was purified by electrophoresis in 1.2% agarose gel, cloned, and sequenced (see below).

Cloning and sequencing

After gel electrophoresis, RT-PCR products and the product from 5′-RACE were isolated using Qiaex II gel extraction kit (150; Qiagen, Hilden, Germany) and ligated into EcoRV-digested pBluescript SK II⁺ prepared with dT overhang. The ligated vectors were transformed into competent Escherichia coli XL-1 Blue. Recombinants were selected on Luria-Bertani agar plates containing 100 μg of ampicillin/ml, 12.5 μg of tetracycline/ml, 40 μg of isopropyl β-d-thiogalactopyranoside/ml, and 20 μg of 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside/ml. DNA was prepared from recombinant clones using the Quantum prep plasmid miniprep kit (Bio-Rad, Hercules, CA). Sequencing of dsDNA was performed using the AutoRead kit (Amersham Pharmacia) with Cy5-labeled T7 forward primer and reverse sequencing primer. The reactions were analyzed using the automated laser fluorescent sequencing system (Pharmacia).

Preparation of copy standard RNAs

Cloned and sequenced PCR products were expanded. DNA was purified from minicultures, the products were linearized, extracted from the agarose gel, and thereafter, in vitro-transcribed to RNA at 37°C for 2 h using the T7 in vitro transcription kit (Riboprobe System-T7; Promega). After DNase treatment, the RNA was purified by the protocol included in the kit and the number of RNA copies was calculated based on OD₂₆₀, molecular mass, and Avogadro’s number.

Immunoflow cytometry

Indirect single color staining of cell surface molecules was performed on live total IEL and LPL and unbound cells after treatment with magnetic beads charged with anti-CD3 or anti-CD2 plus anti-CD7 mAbs as described (10).

Metachromatic staining for mast cells

For identification of mast cells, sections were stained with toluidine blue according to the protocol of Luna (17) (18). Air-dried sections of frozen jejunal tissue were incubated with a mixture of 5 ml 1% (w/v) toluidine blue in 70% ethanol and 45 ml 1% sodium chloride for 1.5–2 min at room temperature. The sections were then rinsed in distilled water, quickly dehydrated in 95% and absolute ethanol and finally cleared in xylene and mounted. Mast cells stained red-purple and other cells stained blue.

Immunohistochemistry

Fresh tissue was snap-frozen in liquid nitrogen and stored at −80°C. Four micrometer-thick cryostat sections were subjected to staining by immunofluorescence or immunoperoxidase techniques. RAG1 and CD1a-expressing cells were stained by immunofluorescence performed as described (13), with two modifications. The slides were acetylated by incubation in 1.4% triethanolamide/0.18% HCl/0.25% acetic anhydride before addition of primary Abs. Primary Abs and conjugate were diluted in PBS supplemented with 0.05% saponin. Two immunoperoxidase techniques were used. c-kit/CD117-expressing cells were stained by indirect technique. Acetone fixed and air-dried sections were blocked for endogenous peroxidase activity by incubation in PBS/3% H₂O₂/2 mM NaN₃ at room temperature for 25 min and sticky sites were blocked by incubation in PBS with 0.2% BSA. Thereafter, the sections were incubated with mAb followed by peroxidase-conjugated F(ab′)₂ of sheep anti-mouse Ig. IL-7R/CD127 and TdT-expressing cells were stained using the enhanced ABC peroxidase technique according to the manufacturer’s protocol (Dakopatts) and blocking of endogenous peroxidase activity by incubation in PBS/0.03% H₂O₂/0.1 mM NaN₃ at 37°C for 1 h. Immunoperoxidase-stained sections were developed using the 3,3′-diaminobenzidine substrate kit (Vector Laboratories, Burlingame, CA) and counterstained with methyl green. Sections incubated with isotype and concentration matched irrelevant mAb or IgG fraction of normal rabbit serum (both Dakopatts) served as negative controls.

Morphometry

Morphometry analyses were performed on toluidine blue-, immunoperoxidase-, and immunofluorescence-stained tissue sections using a ×40 objective and the Leica Q500MC Computer Image Analysis System (Cambridge, U.K.). Frequencies of marker-expressing IEL were determined by calculating the ratio between the number of intraepithelial marker-positive cells and the number of epithelial cells in a defined length of the epithelium (19). Percentages of marker-expressing LPL were counted according to the Weibel method as described (13). Ten to 15 randomly chosen ocular fields were counted for both IEL and LPL. Final results are given as the percent marker-positive cells of CD45⁺ cells in sequential sections.

In situ hybridization

RNA probes for the 1B/2 RAG1 mRNA splice form and pTα^a mRNA were synthesized using the primers for real-time qRT-PCR. For information about primer positions and probe size/amplicon length see assay 5 for 1B/2 RAG1 and outer primers in the pTa^α assay for pTa^α (see Table I⇑⇑). An RNA probe for the 1C/2 RAG1 mRNA splice form was synthesized using the primers for qualitative RT-PCR (353 nucleotides, see Table I⇑⇑). The PCR products were cloned into the plasmid vector EcoRV-digested pBluescript SK II⁺ prepared with dT overhang, amplified, purified, and sequenced. Antisense and sense RAG1 and pTα^a cRNAs were prepared and labeled with DIG-UTP with the use of DIG RNA labeling kit (Boehringer Mannheim) according to the protocol of the manufacturer.

In situ hybridization was performed according to the protocol of Panoskaltsis-Mortari and Bucy (20) with the following modifications: 10-μm cryosections were air-dried, incubated in chloroform for 5 min, and fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). The hybridization was done at 56°C overnight. After RNase A treatment, the sections were incubated at 56°C for 30 min in 30 mM sodium citrate/0.3 M NaCl/50% formamide. The conjugate was diluted in 0.1 M Tris-HCl/0.15 M NaCl/4% normal horse serum. After incubation with substrate, the slides were rinsed in dimethylformamide, counterstained with methylgreen, and mounted in Canada balsam. The corresponding sense DIG-labeled RNA probes were used as negative controls.

Combined in situ hybridization and immunohistochemistry

In situ hybridization was conducted as described above. After the hybridization signal became visible, the slides underwent Ag retrieval by microwave cooking three times for 5 min in 1 mM EDTA (effect, 750 W). After blocking with PBS/20% normal horse serum, the sections were incubated with rabbit anti-CD3 Ab and thereafter treated with avidin followed by biotin. Endogenous peroxidase activity was blocked by incubation in 2 mM NaN₃/0.03% H₂O₂ for 1 h at 37°C. After blocking with PBS/20% normal horse serum, the slides were incubated with biotinylated goat anti-rabbit IgG and thereafter with ABC complex (Dakopatts). After incubation with 3-amino-9-ethyl-carbazole, the slides were mounted in glycerol jelly.

Statistics

The significance of differences in RAG1 and pTα mRNA expression levels was evaluated by one-way ANOVA with post-tests of either Newman-Keuls when comparing pTα mRNA or different RAG1 mRNA splice forms in cells from different tissue sources; Dunnett when comparing different mRNA species in a given cell type; or Bonferroni when comparing a particular mRNA species in pairs of cell types. The significance of differences in frequencies of samples with detectable levels of RAG1 mRNA splice forms was estimated by Fisher’s exact test. Analyses of correlation between mRNA levels for RAG1 and pTα were performed using the Spearman rank correlation test. Statistical analyses of differences in frequencies between cells positive for different markers in immunomorphometry were performed using the Student t test and the paired Student t test was used when comparing frequencies of marker-positive cells intraepithelially and in LP. The variance of groups was tested for equality by F test before ANOVA and Student’s t test analyses. Two-tailed analyses were used throughout. A value of p < 0.05 was regarded as statistically significant. Values obtained by immunomorphometry are given as mean ± 1 SEM.

New RAG1 mRNA 5′UTR exons are used in T cells from human jejunum

To quantify RAG1 mRNA expression, a real-time qRT-PCR assay with RNA copy standard was constructed on the basis of the published organization of the RAG1 gene (21). The 5′-primer was placed in the 5′UTR exon 1, the 3′-primer in exon 2, and the probe over the exon boundary (assay 2, Table I⇑⇑). Even though the assay was sensitive, detecting down to 250 RNA copies and giving strong signals in two samples of thymus RNA, no signal was obtained in 10 jejunal samples previously shown to contain RAG1 mRNA using DNase treatment and qRT-PCR assay 1. Furthermore, no signal was obtained even when a more sensitive, nested qRT-PCR assay, with the 5′-primer placed in exon 1 was used (assay 3, Table I⇑⇑).

A 5′-RACE was performed to investigate whether alternative 5′UTR RAG1 exon(s) were used by jejunal T cells. A sample of jejunal CD2⁺CD7⁺CD3⁻ cells was used. The product had a length of 607 nucleotides and contained the expected sequence of exon 2 (Fig. 1⇓a). However, the entire sequence 5′ of exon 2 was new. Four clones were sequenced with identical results. In the gene data bank part of the 5′UTR sequence (177 nucleotides) was present in one gene fragment and the other part (114 nucleotides) in the other fragment within contig no. 4 of the AC061999 working draft sequence of human chromosome 11. The previously published 5′UTR exon 1 and exon 2 were present within the same contig. The sequences of the new 5′UTR exons and the genomic organization of the RAG1 gene, as deduced from sequence comparison analysis, are given in Fig. 1⇓a. We named the new 5′UTR exons 1A and 1B according to their order relative to exon 2. The previously described exon 1 was located 3′ to the new exons. It was renamed to 1C. Intron/exon splice junctions, according to the GT/AG rule, flanked the intron between exons 1A and 1B, supporting the notion that the new 5′UTR sequence was indeed encoded in two different exons (Fig. 1⇓a).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

a, Organization and nucleotide sequence of exons in the 5′UTR of the human RAG1 gene. The order and nucleotide sequence of the two newly identified RAG1 5′UTR exons 1A and 1B are shown (capital letters) flanked by the two nearest nucleotides in the introns (small letters) as deduced from sequence comparison with the gene data bank. Sequence comparison also revealed that exons 1A and 1B both lie upstream of the previously published 5′UTR exon (1C). Introns are marked with —//— and their lengths are indicated. The arrow shows the position of the 3′-primer used in the 5′-RACE, and the open box indicates the start codon. b, Expression levels of different RAG1 mRNA splice forms in lymphocytes of different tissue origin. Expression levels of RAG1 mRNA splice forms in jejunal lymphocytes (jejunum), thymus, bone marrow, blood, and a T cell leukemia line (Jurkat) were determined by using real-time qRT-PCR assays with primers (arrows) and probe (plain lines) placed as indicated to the left. The amount of the respective mRNA species was determined for each mRNA splice variant, and normalized to the amount of 18S rRNA in the sample. Expression levels are shown as median RAG1 mRNA copies/18S rRNA unit based on determination of five jejunum samples (IEL + LPL), two thymus samples, three BMC samples, and four PBMC samples.

Thymocytes and jejunal T cells show distinct usage of RAG1 5′UTR exons

Jejunal CD3⁺ cells were analyzed for usage of different splice forms of RAG1 mRNA by qualitative RT-PCR in which the 5′-primer was placed either in exon 1A, exon 1B, or exon 1C, and the 3′-primer in exon 2 (Table I⇑⇑). Three major amplicons were obtained when using a 5′-primer placed in exon 1A. Their molecular sizes (≈235, 154, and 125 bp) agreed well with those calculated for the possible splice variants. The larger band had the sequence expected from amplification of the splice form obtained in the 5′-RACE, i.e., 1A/1B/2. One of the smaller bands had the expected sequence for amplification of exon 1A spliced directly to exon 2 (1A/2). The third band had no homology with the RAG1 gene. None of the bands contained the sequence of exon 1C. RT-PCR with the 5′-primer placed in exon 1B yielded only amplicons with 1B spliced directly to exon 2 (1B/2), and RT-PCR with the 5′-primer placed in exon 1C yielded no detectable amplicons. Thus, the two new exons can appear in at least two different splice forms of RAG1 mRNA in jejunal T cells, none of which includes the previously published exon 1 (here renamed 1C).

Two thymus RNA samples were subjected to similar analysis. No bands with sequence homology to the RAG1 gene were obtained with the 5′-primer placed in 1A. RT-PCR with the 5′-primer placed in exon 1B yielded only amplicons with 1B spliced directly to exon 2 (1B/2). RT-PCR with the 5′-primer placed in the 1C exon yielded only amplicons of 1C spliced directly to exon 2 (1C/2). Absence of a 1B/1C/2 splice form is in agreement with the observation that the intron sequence between exons 1B and 1C does not adhere to the GT/AG rule (Fig. 1⇑a).

qRT-PCR assays with RNA copy standards were constructed to analyze the expression levels of the new splice forms of RAG1 mRNA. Assays 4 and 5 were designed to discriminate between splice forms having exon 1A or 1B spliced to exon 2 (Table I⇑⇑, Fig. 1⇑b). Assay 6 was designed to estimate the frequency of mRNA containing the long splice form with 1A exon spliced to the 1B exon (Table I⇑⇑, Fig. 1⇑b). Results were normalized by calculating the ratios between RAG1 mRNA copies and units of 18S rRNA. One unit of 18S rRNA corresponds approximately to the 18S rRNA content in 100 intestinal lymphocyte. The results from analysis of jejunal IEL and LPL of adults are shown in Table II⇓ and summarized in Fig. 1⇑b. The majority of samples expressed splice forms containing either one of the 1A and 1B exons or both while no samples contained detectable amounts of the splice form containing exon 1C (p = 0.02 and 0.009 for IEL and LPL, respectively). All but one IEL sample expressed the 1B/2 splice form. The majority of IEL samples also expressed the 1A/2 splice form. No IEL samples expressed the 1A/2 form only. Two-thirds of the LPL samples expressed either the 1A/2 or the 1B/2 splice form or both. Several LPL samples expressed the 1A/2 form only. Eight of the nine samples that had exon 1B spliced to exon 2 expressed the long (1A/1B/2) form of RAG1 mRNA. Three of these expressed the long form only.

The new forms of RAG1 mRNA (1A/2 and 1B/2) were not detected in RNA extracted from purified small intestinal epithelial cells (n = 1).

Localization of RAG1 mRNA-expressing cells in jejunum

RAG1 mRNA-expressing cells were localized in three jejunal samples from adults by in situ hybridization. One RNA probe specific for the 1B/2 junction, recognizing cells expressing the 1B/2 and/or 1A/1B/2 splice forms, and one probe specific for the 1C/2 junction, recognizing cells expressing the previously described splice form, were used.

Cells expressing mRNA hybridizing with the 1B/2 probe were seen in all three samples. They were located both intraepithelially and in LP (Fig. 2⇓, a and b). RAG1 mRNA-positive cells were irregularly scattered throughout the epithelium, both in crypts (Fig. 2⇓b) and villi. RAG1 mRNA-expressing cells in LP were often found in small clusters (Fig. 2⇓a) and in the vicinity of vessels. RAG1 mRNA-positive cells were abundant in some villi, while LP of other villi were empty or sparse in RAG1 mRNA-positive cells (Fig. 2⇓a). Incubation with the sense probe gave no signal (Fig. 2⇓e), nor were any positive cells detected in sections incubated with the 1C/2 RAG1 mRNA probe, although this probe gave strong signals in tonsillar tissue (not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

RAG1- and pTα mRNA-expressing cells are present both intraepithelially and in LP of jejunal mucosa. In situ hybridization with DIG-labeled antisense probe for the 1B/2 and 1A/1B/2 RAG1 mRNA splice forms (a, b, and g) and pTα^a mRNA (c, d, and f) in normal jejunum of an adult. Both RAG1 mRNA and pTα^α mRNA-expressing cells were seen in the epithelium (b and d) and in LP (a and c). No positive cells were detected using a DIG-labeled RAG1 sense probe (e). Arrows indicate examples of mRNA-expressing LPL in a and c and IEL in b and d. f and g, Combined in situ hybridization for RAG1 (g) and pTα^α mRNA (f) (blue) and immunohistochemistry for CD3 (red). Arrows in g and f indicate double-positive cells and arrowheads indicate cells positive only for RAG1 or pTα mRNA. E, epithelium; L, lumen; C, crypt. Micrographs were taken by using a ×10 objective (a, c, and e) or a ×40 objective (b, d, f, and g). Bars indicate 100 μm in a, c, and e and 20 μm in b, d, f, and g.

RAG1-expressing cells constitute a minor population of IEL

The location and frequency of cells expressing the RAG1 protein was analyzed using two RAG1-specific antisera in immunohistochemistry. The two reagents gave consistent results. Positively stained cells were present both intraepithelially (Fig. 3⇓a) and in LP. The staining pattern suggested intracellular location. Between 1.4–6.1% of the IEL, defined as intraepithelial CD45⁺ cells, were positive for RAG1 (3.3 ± 0.6%; n = 7). RAG1⁺ cells with lymphoid morphology were also seen in the LP. However, no quantitative measurements were performed in LP, because cells with macrophage and dendritic cell morphology exhibited an unspecific staining that could not be totally abolished. No unspecific staining was seen intraepithelially, which is consistent with the absence of macrophages and dendritic cells in the epithelium (10).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

RAG1, c-kit, IL-7R, and CD1a-expressing lymphocytes are present in the jejunal mucosa. a, IEL in villous epithelium of jejunum positively stained for RAG1 (indirect immunofluorescence). b, Numerous IL-7R⁺ cells can be seen both intraepithelially (arrowheads indicate examples) and in LP (arrows indicate examples; immunoperoxidase enhanced by the ABC method). c and d, and e and f, Sequential sections of jejunal mucosa, stained for c-kit by indirect immunoperoxidase technique (c and e) and for mast cells by toluidine blue (d and f). Fat arrows in c and e indicate c-kit⁺ cells in LP not stained by toluidine blue. Arrowheads in c indicate c-kit⁺ IEL. Arrowheads in e and f indicate one c-kit positive mast cell. Thin arrow in f indicates a c-kit negative mast cell. g and h, Frequencies of c-kit⁺, IL-7R⁺, and CD1a⁺ IEL (g) and LPL (h) as determined by immunomorphometry. Numbers of c-kit⁺, IL-7R⁺, CD1a⁺, and CD45⁺ cells were determined in sequential sections and results are given as percent marker⁺ cells/CD45⁺ cells. Each dot represents the result from one individual and horizontal bars indicate the mean values.

Samples were also stained for TdT, which is commonly expressed in thymocytes during TCR gene rearrangement. No TdT⁺ cells were detected (n = 6).

pTα mRNA is expressed in jejunal T cells

pTα is expressed in two forms, a long form (pTα^a) with an extracellular domain that associates with the TCRβ chain during TCRα chain rearrangement forming a pre-TCR, and a short form which lacks the extracellular domain (pTα^b) (22, 23). A qRT-PCR assay with RNA copy standard was constructed for pTα^a mRNA (Table I⇑⇑) and expression levels were determined in IEL and LPL of adults (Table II⇑). pTα^a mRNA was detected in all IEL samples analyzed and all but one of the LPL samples. The levels of pTα^a mRNA correlated with the total levels of RAG1 mRNA, i.e., the sum of the levels of 1A/2 and 1B/2 splice forms in the sample (p = 0.002, r = 0.7). No signal for pTα^a mRNA was detected in RNA extracted from small intestinal epithelial cells (n = 1).

The localization of cells expressing pTα mRNA was investigated in three jejunal samples from adults by in situ hybridization, using a probe selective for the pTα^a form. pTα^a mRNA-expressing cells were detected in all three samples. The distribution of pTα^a mRNA-expressing cells showed great resemblance to cells hybridizing with the 1B/2 RAG1 probe (Fig. 2⇑c). Positive cells were seen both intraepithelially (Fig. 2⇑d) and in LP. Furthermore, while the LP of some villi harbored numerous pTα^a mRNA-expressing cells, other villi were devoid of such cells (Fig. 2⇑c). Incubation with the sense probe did not give any signal (not shown).

The new RAG1 mRNA splice forms are expressed in immature jejunal T cells

In the T cell lineage, CD2 and CD7 are expressed earlier in development than the TCR/CD3 complex and are retained on mature T cells (24). Previously, we showed that a significant proportion of human jejunal IEL are CD2⁺CD3⁻ (10). Immunoflow cytometry analyses showed that CD2⁺ and CD7⁺ cells outnumber CD3⁺ cells both in IEL and LPL (not shown) indicating that CD2⁺CD7⁺CD3⁻ cells are present in both compartments. The expression levels of RAG1 mRNA splice forms were assessed in lymphocytes of the T cell lineage retrieved from IEL and LPL of adults by sequential positive selection. First, CD3⁺ cells were retrieved. This population includes mature T cells and cells with immature forms of the TCR/CD3 complex on the surface. Cells expressing CD2 and/or CD7 were retrieved from the unbound cells by using magnetic beads charged with a mixture of anti-CD2 and anti-CD7 mAbs (CD2⁺CD7⁺CD3⁻ cells). This population includes immature cells of the T cell lineage with no surface expression of the TCR/CD3 complex and possibly mature T cells with Ag-induced surface down-regulation of the TCR/CD3 complex.

RAG1 mRNA was expressed both in CD3⁺ and CD2⁺CD7⁺CD3⁻ cells. All three new splice forms (1A/2, 1B/2, 1A/1B/2) were expressed but varied significantly in proportions between different samples (Fig. 4⇓, a–c). Five of six CD3⁺ IEL samples, four of six CD3⁺ LPL samples, all four CD2⁺CD7⁺CD3⁻ IEL samples, and three of four CD2⁺CD7⁺CD3⁻ LPL samples expressed the 1A/2 and/or 1B/2 RAG1 mRNA splice forms. The 1C/2 form was not detected.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

The new RAG1 mRNA splice forms and pTα^a mRNA are expressed in jejunal lymphocytes with mature T cell phenotype (CD3⁺) and immature, thymocyte-like phenotype (CD2⁺CD7⁺CD3⁻). Expression levels of the 1B/2 RAG1 (a), the 1A/1B/2 RAG1 (b), and the 1A/2 RAG1 mRNA splice form (c) and pTα^a mRNA (d) were determined in CD3⁺ cells (triangles) and CD2⁺CD7⁺CD3⁻ cells (circles) retrieved from IEL (filled) and LPL (open) by sequential positive selection. Lines connect CD3⁺ cells and CD2⁺CD7⁺CD3⁻ cells from the same individual. CD3⁺ IEL and CD3⁺ LPL were retrieved from samples J71, J72, J73, J74, J77, and J78; CD2⁺CD7⁺CD3⁻ IEL were retrieved from samples J71, J72, J73, and J77; and CD2⁺CD7⁺CD3⁻ LPL of samples J72, J73, J77, and J78 (see Table II⇑).

There was a tendency for increased expression levels of both the 1A/2 and 1B/2 RAG1 mRNA forms in CD2⁺CD7⁺CD3⁻ cells compared with CD3⁺ cells (Fig. 4⇑, a and c) and unfractionated IEL and LPL (not shown). Indeed, the expression level of the 1A/2 form was significantly higher in CD2⁺CD7⁺CD3⁻ IEL compared with the total IEL population (p = 0.05). In addition, five of the six CD3⁺ IEL samples expressed the long 1A/1B/2 form (Fig. 4⇑b). The long form was also expressed in CD3⁺ LPL samples from three of these individuals. There was a great variation between samples in the proportion of 1B/2 that was in the long 1A/1B/2 form, and as much as six samples had all the 1B/2 in the long form.

RAG1 1B/2 and/or 1A/1B/2 expression in CD3⁺ cells was also demonstrated by combination of in situ hybridization and immunohistochemistry for CD3⁺ cells. Double-stained cells were seen both intraepithelially and in LP (Fig. 2⇑g).

Independent evidence for TCR gene rearrangement was obtained by analysis of CD2⁺CD7⁺CD3⁻ cells for RAG2 mRNA expression. By RT-PCR, a single amplicon of the correct size and sequence was obtained.

pTα^a mRNA is preferentially expressed in immature T cells

Virtually all CD3⁺ and CD2⁺CD7⁺CD3⁻ IEL and LPL samples were positive when analyzed for pTα^a mRNA (Fig. 4⇑d). Expression levels were high in CD2⁺CD7⁺CD3⁻ cells both intraepithelially and in LP (Fig. 4⇑d). Indeed, the pTα mRNA levels were significantly higher in CD2⁺CD7⁺CD3⁻ cells than in CD3⁺ IEL, total IEL, CD3⁺ LPL, and total LPL, respectively (p = 0.001 in all four cases). Expression levels of pTα^a and RAG1 (1A/2 + 1B/2) mRNAs correlated in CD3⁺ cells (p = 0.04, r = 0.6) but not in CD2⁺CD7⁺CD3⁻ cells.

Combined in situ hybridization and immunohistochemistry demonstrated pTα^a mRNA CD3 double-positive cells (Fig. 2⇑f). However, the majority of pTα^a mRNA-positive cells were CD3-negative.

The new RAG1 mRNA splice forms and pTα^a mRNA are expressed in jejunal T cells of children

Table II⇑ shows the results from analysis of the total T cell lineage population retrieved from jejunal IEL and LPL of young children by using magnetic beads charged with a mixture of anti-CD2 and anti-CD7 mAbs. All three IEL samples expressed RAG1 mRNA, either the 1A/2 or the 1B/2 form, while no RAG1 mRNA was detected in LPL of the same individual. Only one of three additional LPL samples expressed RAG1 mRNA. The two IEL samples that had exon 1B spliced to exon 2 both expressed the long 1A/1B/2 splice form. The 1C/2 splice form was not detected.

pTα^a mRNA was detected in all three IEL samples and in the majority of LPL samples (Table II⇑). Notably two of the LPL samples expressed high levels of pTα^a mRNA although no RAG1 mRNA was detected. The expression levels of pTα^a mRNA did not correlate with the expression levels of RAG1 mRNA (1A/2 + 1B/2).

Expression of RAG1 mRNA splice forms and pTα^a mRNA in primary lymphoid organs and blood

Two thymus samples, mononuclear immune cells from BMC of three adults and PBMC of four healthy adults were analyzed for expression of RAG1 and pTα^a mRNA. The results are shown in Table II⇑ and summarized in Fig. 1⇑b.

The 1C/2 splice form of RAG1 mRNA was dominating in thymus and expressed at very high levels significantly above those of all analyzed cell types of other tissue origins (p < 0.001). The 1B/2 splice form was also expressed albeit at significantly lower levels (p < 0.001) and were comparable to the levels in IEL and LPL. mRNA for the 1A exon was not detected, either in the 1A/2 or the long 1A/1B/2 splice form. The expression level of pTα^a mRNA was high (median 254 copies/18S rRNA unit).

BMC expressed all four RAG1 mRNA splice forms. The 1C/2 splice form was dominating (p < 0.001 compared with both 1A/2 and 1B/2). Still, the expression level of this splice form was ≈700 times lower than in thymus (p < 0.001). The expression levels of the 1A/2, 1B/2, and 1A/1B/2 splice forms were not significantly different from the levels of these splice forms in jejunal lymphocytes. All splice forms of RAG1 mRNA could also be detected in PBMC. The average expression levels of the 1A/2 and 1B/2 splice forms of RAG1 mRNA were similar to BMC, while the 1C/2 splice form was significantly lower (p < 0.001). Interestingly, significant levels of pTα^a mRNA were detected in both BMC and PBMC (Table II⇑).

All four splice forms of RAG1 mRNA (Fig. 1⇑b), as well as pTα^a mRNA (15.5 copies/18S rRNA unit) were expressed in the human T cell leukemia cell line Jurkat, which has the phenotype of mature αβ T cells (TCR⁻αβ⁺CD3⁺CD4⁺CD8⁻CD2⁺CD7⁺).

Lymphocytes with immature phenotype are present in the jejunal mucosa

The frequency and location of possible progenitor cells was analyzed by morphometry analysis of anti-c-kit/CD117-stained jejunal mucosa. Because c-kit can also be expressed by mast cells (25), sequential sections were stained with toluidine blue for identification of this cell type. CD1a was used as a marker for lymphocytes with thymocyte phenotype. We also performed immunomorphometry analyses of cells expressing receptors for IL-7 (IL-7R/CD127⁺ cells), a cytokine that has been shown to be required for development of γδ T cells and a key factor for T cell development in general (26, 27).

c-kit⁺ cells were present both in LP and within the epithelium (Fig. 3⇑, c and e). Their frequency was significantly higher in LP than in the epithelium with a mean of 15% c-kit⁺ LPL, compared with 3.3% c-kit⁺ IEL (p < 0.05; Fig. 3⇑, g and h). Mast cells were detected in LP (Fig. 3⇑f) where they constituted 4.9 ± 1.9% (n = 6) of the CD45⁺ cells but were scarce within the epithelium (0.3 ± 0.3% of intraepithelial CD45⁺ cells; p < 0.01 comparing the frequencies of mast cells intraepithelially and in LP). The extremely low frequency of mast cells within the epithelium excludes the possibility that intraepithelial c-kit⁺ cells are mast cells (p < 0.001 comparing frequencies of intraepithelial c-kit⁺ cells and mast cells). Analysis of LPL in sequential sections revealed that c-kit⁺ cells significantly outnumbered mast cells (p < 0.05). Most c-kit⁺ cells were not mast cells as judged by staining with toluidine blue (Fig. 3⇑, c and d), and only a fraction of the mast cells were c-kit-positive (Fig. 3⇑, e and f). Thus, at least 10% of the LPL of the jejunal mucosa are likely to be c-kit⁺, immature progenitor cells.

Numerous cells (≈40% of the CD45⁺ cells) showing surface staining for IL-7R were present both intraepithelially and in LP (Fig. 3⇑b). However, full revelation of IL-7R⁺ cells required amplification by the ABC technique suggesting that the surface expression is low on a large proportion of these cells. IL-7R⁺ cells significantly outnumbered c-kit⁺ cells at both locations (p < 0.001; Fig. 3⇑, g and h). Lymphocytes expressing CD1a were also more frequent than c-kit⁺ cells and constituted approximately one-third of both IEL and LPL (Fig. 3⇑, g and h).