¹¹¹Indium-labelled human gut-derived T cells from healthy subjects with strong in vitro adhesion to MAdCAM-1 show no detectable homing to the gut in vivo

Abstract

Integrin α4β7 is the principal gut-homing receptor, and it is assumed that expression of this specific integrin directs lymphocytes to the gut in vivo. Adoptive cellular immunotherapy against inflammatory bowel disease (IBD) may depend on the expression of integrin α4β7 to accomplish local delivery of intravenously injected regulatory T cells in inflamed gut mucosa. The present study aimed to investigate whether in vitro expanded human T cells from the colonic mucosa maintain integrin expression, show in vitro adhesion and retain in vivo gut-homing properties during cultivation. Whole colonic biopsies from healthy subjects were cultured in the presence of interleukin-2 (IL-2) and IL-4. The integrin expression of the cultured T cells was determined by flow cytometry and in vitro adhesion was assessed in a mucosal addressin cell adhesion molecule 1 (MAdCAM-1) adhesion assay. We studied the homing pattern after autologous infusion of 3 × 10^{8 111}Indium (¹¹¹In)-labelled T cells in five healthy subjects using scintigraphic imaging. The cultured CD4⁺CD45RO⁺ gut-derived T cells express higher levels of integrin α4β7 than peripheral blood lymphocytes (PBLs) and show strong adhesion to MAdCAM-1 in vitro, even after ¹¹¹In-labelling. Scintigraphic imaging, however, showed no gut-homing in vivo. After prolonged transit through the lungs, the T cells migrated preferentially to the spleen, liver and bone marrow. In conclusion, it is feasible to infuse autologous T cells cultured from the gut mucosa, which may be of interest in adoptive immunotherapy. Despite high expression of the gut-homing integrin α4β7 and adhesion to MAdCAM-1 in vitro, evaluation by ¹¹¹In-scintigraphy demonstrated no gut-homing in healthy individuals.

INTRODUCTION

Lymphocyte homing is the preferential return of activated lymphocytes to the organ where they once were activated and a fundamental immunological principle in host defense of antigen-challenged mucosal surfaces. The interaction between integrin α4β7 and mucosal addressin cell adhesion molecule 1 (MAdCAM-1) is believed to play a pivotal role in the multistep process of lymphocyte homing to intestinal mucosa [1]. Yet, tissue specific chemokines such as thymus-expressed chemokine (TECK) interacting with chemokine receptor 9 (CCR9) are also involved in lymphocytes trafficking to the small intestine [2]. The homing concept adds important aspects to understanding the etiopathogenesis of inflammatory bowel disease (IBD) and has interesting therapeutic implications [3]. Local delivery in the gut of T cells with interleukin(IL)-10-dependent regulatory properties has been introduced in murine models of colitis [4], and recent antiadhesion studies with α₄-antibody, natalizumab, have shown promising results in Crohn's disease [5].

The homing process is governed by overlapping functions of the adhesion molecules, and MAdCAM-1 may mediate rolling in the absence of l-selectin as well as firm adhesion and transmigration [6]. MAdCAM-1 expression in adults is restricted to high endothelial venules (HEV) in Peyer's patches and mesenterial lymph nodes and to flat-walled venules of the lamina propria and submucosa in the gut [7]. For unexplained reasons, MAdCAM-1 is also expressed in sinus-lining cells of the spleen [8]. MAdCAM-1 expression is up-regulated in lamina propria venules in experimental colitis [9] and in Crohn's disease [10] where it may contribute to the transmural accumulation of inflammatory cells.

Upon antigen challenge in intestinal mesenterial lymph nodes T cells acquire high levels of integrin α4β7 [11] and populations of α4⁺β7⁺ memory lymphocytes may represent a well-defined intestinal lymphocyte pool with specialized cellular functions [12,13]. According to the ‘area-code’-paradigm, in vivo trafficking may be predicted from the specific integrin expression. This view is challenged by results showing that up-regulation of integrins may occur after tissue homing [14] and that the expression of specific integrins on circulating lymphocytes does not result in selective organ entry [15].

In sheep, gut derived immunoblasts were shown to return preferentially to the small intestine [16] and by intravital fluorescence microscopy it was shown in mice that T cells from the lamina propria adhered selectively to the microvessels of the villous mucosa [17]. Most evidence in humans has been obtained from in vitro adhesion studies. The few in vivo homing studies that have been conducted in humans present conflicting results concerning the ability of ex vivo expanded tumour-infiltrating lymphocytes to traffic specifically to tumour masses [18,19]

We have previously shown that gut-derived T cells can be propagated in high numbers from colonic biopsies by the use of high concentrations of IL-2 and IL-4 [20]. The present study aims to investigate whether the supposed intrinsic gut-homing potential of such in situ activated gut-derived T cells is maintained during cultivation. We have studied the integrin expression and in vitro adhesion properties of cultured mucosal T cells and their actual in vivo migration behaviour.

MATERIALS AND METHODS

Biopsy specimens

T cell cultures were established from colonic biopsies from 5 healthy individuals (3 males and 2 females, age 20–47). Histological evaluation revealed normal mucosa. Further, T cell cultures were established from bronchial biopsies from 1 healthy male, age 58. Approval of the study was obtained from the local ethical committee in Aarhus County, Denmark (J.nr. 20020150 and 20020252).

Cell culture

Biopsies of whole colonic mucosa were cultured as previously described [20]. Briefly, the biopsies were washed twice in RPMI 1640 and placed in 5 ml culture medium consisting of 90% RPMI 1640, 10% human AB-serum, Penicillin G 100 U/ml, streptomycin 100 µg/ml, supplemented with 2000 U/ml IL-2 and 500 U/ml IL-4 to facilitate the growth of CD4⁺ T cells. Neither antigen nor feeder cells were added to the T cell cultures in order to expand them; activation was only dependent on in vivo-derived antigen contained in the biopsies. In the culture medium, T cells migrated from the mucosal biopsies and the cytokine milieu promoted the preferential proliferation of CD4⁺CD45RO⁺ memory T cells [21]. Within 2–4 weeks the T cell cultures proliferated to 50 × 10⁶ cells. The bronchial biopsies were cultured in the same way. Application of this culture system makes it possible to preserve T cell function during continuous long-term cultivation [20]. The cultured gut-derived T cells should be categorized as blasts. They are slightly larger than resting lymphocytes and are generally activated due to the culture conditions.

Only T cell cultures with a low cytokine production (<500 pg/ml IL-10, IFN-γ and TNF-α in the supernatant) were used for infusion. The T cells were propagated to a final volume of 3 × 10⁸ T cells from aliquots of cultures that had been stored for approximately 18 months at minus 135°C.

Prior to infusion, samples were taken for detection of contamination by aerobic and anaerobic bacteria, fungi as well as Mycoplasma pneumoniae by polymerase chain reaction (PCR).

Cultured T cells were tissue typed to assure that only autologuos T cells were infused and the cell culture was filtered in a sterile filter (30 µm) from Miltenyi Biotec (Bergisch Gladbach, Germany).

The whole sequence is shown as a flow chart in Fig. 1.

Fig. 1.

Flow chart illustrating the sequence of events from colonic biopsy to injection of cultured T cells.

Antibodies

The following antibodies were obtained from Becton Dickinson: CD25 (mouse IgG₁, 2A3), CD69 (mouse IgG₁, L78), CD4 (mouse IgG_1, SK3), CD3 (mouse IgG₁, SK7), CD62L (mouse IgG_2A, SK11), Integrin β₁ (mouse IgG₁, MAR4), Integrin β₇ (rat IgG_2A, FIB504); from Immunotech: Integrin α₄ (mouse IgG₁, HP2/1), CD45RO (mouse IgG_2A, UCHL1), CD11a (mouse IgG₁, SK7); and ACT-1 (mouse IgG) against heterodimer α4β7 [12], donated by LeukoSite (Cambridge, MA, USA), was labelled with secondary antibody Alexa(488)Fluor®-goat-anti-mouse (Molecular Probes, Leiden, the Netherlands). The amount of each antibody was optimized against the recommended IgG subtype specific isotype controls from the same suppliers with staining of 5 × 10⁵ cells in 100 µl culture medium.For some experiments peripheral blood mononuclear cells were isolated using standard density centrifugation with Ficoll (Pharmacia, Uppsala, Sweden) and subjected to CD4 isolation by immunomagnetic separation.

Flow cytometric analysis

Data (20 000 events) was obtained using a FACScalibur® flow cytometer. Necrotic cells (1–2% of the cultures) were excluded by propidium iodide staining. Results are expressed as mean fluorescence intensity (MFI) and the percentage of positive cells was calculated by subtracting the positivity at 2% in the isotype controls.

MAdCAM-1 adhesion assay

Dr S. Fong, Genentech, Inc. (San Francisco, CA, USA) kindly provided human 293 embryonic kidney cells stably secreting recombinant soluble human MAdCAM-IgG fusion protein. Subject to minor modifications, the cell adhesion assay was performed as previously described [9].

High binding Delfia® microtitration plates (Wallac, Turku, Finland) were precoated with 200 ng/well goat-anti-mouse IgG₁ (Caltag, Burlingame, CA, USA) overnight at 4°C. The wells were blocked with RPMI 1640 with 5 mg/ml BSA for 20 min and washed with Tween 0·2% before incubating the supernatants from the MAdCAM-IgG producing cells for 2 h at room temperature.

The T cells were labelled with carboxyfluorescein (CFSE) (Molecular Probes) in PBS for 10 min at 37°C and washed and resuspended in block solution. The cells were plated at a density of 5 × 10⁴ cells/well and incubated at 37°C for 20 min. Unbound cells were removed by gentle washing with PBS followed by centrifugation of the inverted microtest plate sealed with adsorbent paper at 200 g for 5 min. The fluorescence from bound cells was analysed on a Victor2 plate reader (Wallac). The experiments were performed in quadruplicates and data is presented as the proportion of plated cells bound. The integrin α4β7 negative K562 cell line served as negative control and was arbitrarily assigned a binding index of 1 [22]. Antibodies against MAdCAM-1 (Oncogene, San Diego, CA, USA) and integrin α₄ (BD Biosciences) were used in receptor-blocking experiments. For some experiments PBLs and cultured T cells from the same individual were separated by CD4 isolation Kit (Miltenyi Biotec) according to manufacturer's protocol. Most purities obtained exceeded 95%. Magnetic beads have been shown not to interfere with static adhesion assays [7].

¹¹¹In-labelling of cultured lymphocytes

Cultured lymphocytes (2·6 ± 0·8 × 10⁸) were labelled with 37·8 ± 2·5 Mbq ¹¹¹In-tropolone (local hospital pharmacy). The labelling efficiency was 90·9 ± 7·9% and the labelled cells were washed and resuspended in 5 ml of autologous plasma. In the fifth test person (JK) the lymphocytes (2·3 × 10⁸) were labelled with 2·3 MBq ¹¹¹In-tropolone in order to lower the risk of cellular radiation damage [23].

After labelling a cell aliquot was tested in vitro for cellular integrity by staining with trypan blue and for expression of surface integrin α₄β₇ and MAdCAM-1 adhesion.

Test persons

Four healthy test persons, 20–47 years old (2 females; 2 males), had an intravenous injection with autologous, cultured ¹¹¹In-labelled lymphocytes (15·8 ± 0·8 MBq). One subject (male (JK)) received 1·2 MBq ¹¹¹In-labelled lymphocytes.

Gamma camera imaging

The distribution of the labelled lymphocytes was visualized by a Philips-Marconi Axis® (Marconi, Cleveland, USA) dual-head gamma camera 1 h, 3 h and 24 h after intravenous injection. The thorax and the abdomen were visualized in anterior and posterior projections and regions of interest (ROIs) were placed over the lungs, the liver, the spleen and over the lumbar vertebrae L3 and L4.

The total activity in the organs was registered (as counts) in a 256 × 256 matrix and corrected for attenuation by calculating the geometric mean of the counts measured in the anteroposterior (AP) and posteroanterior (PA) projections. The total counts in the bone marrow were extrapolated from the activity measured in the L3 and L4 vertebrae, based upon the assumption that the activity in these vertebrae corresponds to 4·6% of the activity in the entire bone marrow [24,25]. A positive signal was defined as 3 standard deviations above the background level. The scintigraphy was estimated to be sufficiently sensitive to allow detection of about 0·04% of the injected number of cells (about 80 000 cells), assuming an even distribution of these cells in the gut mucosa underlying an abdominal surface area of about 4 cm².

Assessment of blood activity

Blood samples from 3 of the subjects were collected in heparinized tubes at the same times as the gamma camera examinations. Three ml of blood was centrifuged at 300 g for 12 min and the cell-free plasma was carefully separated from the pellet.

The cell pellets and the cell-free plasma samples were then counted in a gamma-counter and the amount of cell-bound activity was calculated as the ratio between activity in the pellet and total activity in the blood samples.

RESULTS

Gut-homing potential evaluated by in vitro techniques

Prototypical lamina propria-derived CD4⁺ T cells were propagated from colonic biopsies and their characteristics mainly matched those of CD45RO⁺ CD62L^low/negative memory T cells. The gut-derived T cells have a high expression of gut-homing integrin α4β7 and LFA-1, and show variable expression of activation markers CD25 and CD69 (Table 1 and Fig. 2).

Table 1.

Phenotype of infused cultured gut-derived T cells (unseparated cultures)

Culture	CD4	α4β7	Act-1	LFA	CD62L	CD25	CD69
1	62	62	53	86	0·2	1·5	17
2	72	40	39	92	0·2	0·6	44
3	82	55	53	87	0·5	8·5	8
4	87	63	57	90	1·5	0·7	51
5	74	41	n.d.	92	1·1	1·2	7

Percentage of positive cells (isotype subtracted) in the 5 infused cultures (n.d. not determined).

Fig. 2.

Cultured gut-derived T cells have a high expression of gut-homing integrin α4β7 and LFA-1 (CD11a/CD18), but no detectable expression of l-selectin (CD62L) in memory T cells. Specific antibodies represented by filled histograms in one representative of the five gut-derived T cell cultures.

Presumably due to their organ of origin, the cultured gut-derived T cells had a significantly higher expression of integrin α4β7 than peripheral blood lymphocytes (PBLs) from matched individuals, both with respect to the percentage of positive cells (54% ± 12 vs. 37% ± 4) and integrin α4β7 density on the cell surface determined by the MFI.

In the five cultured gut-derived T cell cultures we observed a MFI (isotype fluorescence subtracted) integrin α₄ = 67 ± 7 (s.e.m.) and MFI integrin β₇ = 62 ± 19. This should be compared with the values in fresh PBLs from the same individuals, where MFI integrin α₄ = 25 ± 11 and MFI integrin β₇ = 25 ± 3.

We addressed the question of differential organ-specific integrin expression applying the same culture principle on biopsies from bronchial mucosa. The bronchial T cells predominantly expressed integrin α₄β₁(Fig. 3) suggesting that the integrin expression is not merely induced by the culture conditions but influenced by the organ of origin.

Fig. 3.

The expression of integrin α4β7 on gut-derived CD4⁺ T cells (left dot plot) outperforms the expression on freshly isolated CD4⁺ PBLs from the same individual, both with respect to number of positive cells and MFI. Results shown are from a single representative of three subjects in the in vivo study. CD4⁺ T cells cultured from the bronchial mucosa (right 2 dot plots) from another healthy individual show a lower expression of integrin α4β7 than gut-derived T cells, mainly expressing integrin α₄β₁, visualized by gating on the integrin α₄⁺β₇^– population.

We investigated whether the integrin expression on cultured cells could be correlated with the cellular function in vitro by performing a MAdCAM-1 adhesion assay (Fig. 4). The cultured gut-derived T cells uniformly displayed a strong adhesion to recombinant human MAdCAM-1 in vitro. Out of 5 × 10⁴ cells added to each well 4·270 ± 695 of the gut-derived T cells adhered to the well after washing and centrifugation compared to 217 ± 37 cells of the integrin α4β7-negative K562 control cell line. Thus, the adhesion of gut-derived T cells was generally 15–20 fold higher than the control cell line K562. Adhesion of the cells was inhibited by MAdCAM-1 antibody and even more so with an α₄ antibody. As demonstrated by others [26], the adhesion could be enhanced by a factor 2–10 by addition of 1 m m manganese chloride (MgCl₂), inducing a conformational change in the integrin α4β7 (data not shown).

Fig. 4.

Cultured gut-derived T cells show strong adhesion to MAdCAM-1. The adhesion was inhibited by integrin α₄ antibody (1 µg/5 × 10⁵ cells) and to a lesser extent by the commercially available MAdCAM-1 antibody (0·1 µg/well) not raised against the specific recombinant MAdCAM-1 protein applied in this assay. The integrin α4β7-negative cell line K562 was arbitrarily assigned a binding index value of 1. Results are expressed as mean of quadriplicate experiments in the same five gut-derived T cell cultures used in the in vivo studies.

Furthermore, we investigated whether the observed higher expression of integrin α4β7 in gut-derived T cells than in PBLs was reflected by a higher adhesive function in vitro. The adhesion of freshly isolated CD4⁺ PBLs was remarkably homogenous between individuals, possibly reflecting the observed narrow interval of α4β7 expression on PBLs. Of special interest was the 5–10-fold higher adhesion of cultured gut-derived CD4⁺ T cells than of freshly isolated PBLs (Fig. 5), which indicates that the culture principle selects gut-derived lymphocytes with intrinsic features that are clearly discernible from PBLs.

Fig. 5.

Cultured gut-derived CD4⁺ T cells show 5–10 fold stronger MAdCAM-1 adhesion than freshly isolated peripheral blood CD4⁺ T cells from the same individual. Results are expressed as mean of quadriplicate experiments in 3 of the participants in the in vivo study. Error bars indicate SEM.

Radiolabelling with ¹¹¹In-chloride

We observed 97–100% viability of the radiolabelled and filtered T cells evaluated by trypan blue staining. Gut-derived T cell lines from three of the test persons were labelled with ¹¹¹In to evaluate the possible influence of radiation toxicity on surface integrin α4β7 expression 4 and 24 h after labelling. The expression level of integrin α4β7 was stable within the first 4 h after ¹¹¹In-labelling. In accordance with the flow cytometric changes in integrin α4β7 expression, the MAdCAM-1 adhesion also remained unaffected by ¹¹¹In-labelling within the first 4 h.

However, integrin α4β7 down-regulation to a MFI level of 50% of the starting value was observed in one of three cultures 24 h after ¹¹¹In-labelling, while the two remaining cultures were unaffected. Nevertheless, in all three cultures the in vitro MAdCAM-1 binding capacity was reduced to a mean level of 65% at 24 h (Fig. 6). Thus, when evaluating the noxious effect of ¹¹¹In-labelling, the level of integrin expression on the cell surface did not mirror the functional in vitro adhesion capacity.

Fig. 6.

¹¹¹In-labelling does not affect the MAdCAM-1 adhesion in vitro within the first 4 h after labelling. Cultured T cells from three individuals in the study were labelled with ¹¹¹In and the MAdCAM-1 binding capacity was assessed. Four hours after labelling the in vitro adhesion was not affected, but declined to a mean level of 65% at 24 h. As control the binding of unlabelled cells from the same individuals was assigned a binding index value of 1. To rule out nonspecific adhesion of activated cells, integrin α₄ antibody was added as control in all experiments.

Gamma camera imaging

In the four healthy subjects, the cultured lymphocytes were labelled with 15·8 ± 5·2 Mbq/10⁸ cells and the relative distribution of activity at 1, 3 and 24 h after injection is shown in Fig. 8. At 24 h the activity in the lungs had almost disappeared and most was now found in the liver, spleen and bone marrow. In other areas of the body only background activity was seen. Neither focal nor diffuse activity could be found in the intestines in any of the subjects. We did not observe any activity in lymph nodes (Fig. 7).

Fig. 8.

Relative organ distribution of injected gut-derived lymphocytes 1, 3 and 24 h post injection (h.p.i.). At each time point the activities in the lungs, liver, spleen and bone marrow were summarized and assigned index 100 after correction for ¹¹¹In decay. Only background activity was seen in other parts of the body.

Fig. 7.

Gamma camera images of thorax and abdomen in a healthy subject following infusion of ¹¹¹In-labelled cultured autologous T lymphocytes from colonic mucosa. At 1 h.p.i. we noted retention in lungs, uptake in liver and spleen and excretion of ¹¹¹In in bladder. At 3 h.p.i. we observed gradual clearing of the lungs together with an increasing uptake in the liver, spleen and bone marrow. By 24 h.p.i. the lung activity was almost cleared and bone marrow uptake was clearly visible, but overall no detectable gut-homing. Hours post injection (h.p.i.).

In the fifth subject examined with cells labelled with 2·2 MBq/10⁸ cells, the analysis was less reliable due to a very low count rate. Lung activity was retained longer than in the other subjects and activity at 24 h was distributed as follows: lungs 69%, liver 16%, spleen 5%, and bone marrow 11%.

¹¹¹In-activity in blood

The results for the cell bound activity are presented in Table 2. The bound fraction seems to be rather constant in the individual subjects although the values vary between them.

Table 2.

Percentage of activity bound to the cellular fraction of blood

Culture	1 h.p.i.	3 h.p.i.	24 h.p.i.
2	41·6%	39·2%	37·2%
3	31·0%	25·7%	31·0%
5	14·9%	19·0%	14·7%

DISCUSSION

In the present paper we have investigated whether cultured gut-derived T cells with presumed organ-specific homing properties display in vitro adhesion and homing to the gut upon intravenous infusion. We found that these cells had a higher expression of gut-homing integrin α4β7 and LFA-1 than PBLs from the same individual and the cultured T cells likewise had a 5–10-fold higher binding capacity than the same individual's PBLs in a MAdCAM-1 adhesion assay. For comparison, T cells cultured from bronchial mucosa preserved a high expression of integrin α₄β₁ although cultured in the same manner as gut-derived T cells, indicating that the culture system does not invariably select for integrin α4β7 expression irrespective of organ origin.

Labeling with ¹¹¹In did not affect the in vitro adhesion of the gut-derived T cells during the first 4 h, and a mean 65% of the MAdCAM-1 binding capacity was preserved after 24 h. However, we were surprised to learn that ¹¹¹In-scintigraphy was unable to detect gut-specific homing of the infused gut-derived T cells. After infusion we observed accumulation in the liver, spleen and bone marrow. The strong MAdCAM-1 adhesive properties of the cells suggest that they may belong to an intestinal pool of lymphocytes. Alternative in vitro adhesion assays such as Stamper-Woodruff frozen section assay or binding assay on cultured endothelial cells [27] may mimic in vivo conditions more accurately, but our in vivo findings do not support the notion of organ-specific gut entry in healthy individuals.

The present study was motivated by reports suggesting that evidence from animal and in vitro studies may not be extendable to human in vivo conditions [28,29]. Ex vivo generated T cells with regulatory properties represent an option for adoptive immunotherapy in IBD. Studies are in progress to isolate gut-derived (organ-specific) CD4⁺CD25⁺ regulatory T cells, which in our culture system have a high expression of integrin α4β7, contrary to their counterpart in peripheral blood [30].

The idea of using T cells cultured from colonic mucosa was to take advantage of their presumed intrinsic gut-homing potential and we evaluated the feasibility of such a strategy. Importantly, none of the participants experienced any kind of side-effects.

Long-term cultivation may induce chromosomal aberrations [31] and the main concern would be induction of lymphoma. With the applied T cell culture system we have previously observed chromosomal aberrations when the T cells were propagated beyond 120 population doublings with a population doubling time of approximately 36 h (e.g. 200 days) [20]. In the present experiments the T cells were only cultured for 8 weeks to minimize the risk of chromosomal aberrations. However, we cannot rule out the possibility that an inflammatory compartment could provide a cytokine milieu sufficient for continued proliferation of infused T cells. Thus, a future adoptive immunotherapy should only be considered for IBD patients, which do not respond to conventional regimens.

Lymphocyte migration studies in humans are, of course, limited to noninvasive techniques such as ¹¹¹In-scintigraphy. This technique, however, carries a risk of radiation-induced toxicity, which may obscure migration studies. Lymphocytes are more vulnerable to radiation-induced toxicity than granulocytes, but a combination of ¹¹¹In and the chelator tropolone is believed to ensure intact cells, so that random migration and chemotaxis of lymphocytes will not be affected [32].

The distribution of ¹¹¹In-labelled, cultured lymphocytes in our study was qualitatively comparable to that of conventional PBLs in healthy individuals [23,33] and the same pattern of preferential migration to liver, spleen, bone marrow and lungs was also observed in pigs after intravenous injection of PBLs [34]. The transit time in the lungs for cultured gut-derived T lymphocytes was longer than for granulocytes and native lymphocytes [23,35,36], with more than 50 ± 15% of the activity being retained in the lungs at 3 h.p.i. However, new evidence suggests that lung accumulation is due to an active process retaining living cells [37].

The cell-bound ¹¹¹In-activity in the blood was reasonably high and averaged around 28% of the circulating activity at all time points. Release of cell-bound ¹¹¹In is an active process and dead cells do not increase soluble activity significantly until 6–24 h after cell death [38], indicating that the ¹¹¹In-labelled T cells in the experiments were viable at all time points.

Reducing the amount of used ¹¹¹In did not change cell behaviour except increasing lung transit time and necessitating increased examination time.

Both ¹¹¹In and tropolone cause dose and time dependent toxicity in vitro, reducing the proliferative response to mitogenic stimuli [39], which does not allow conclusions about interference with cell surface expressed integrins [36]. We therefore evaluated both the integrin expression and the in vitro adhesion properties after ¹¹¹In exposure. Neither integrin down-regulation nor impaired MAdCAM-1 binding was observed within the first 5 h, which should leave a sufficient window for in vivo migration studies performed by ¹¹¹In-scintigraphy.

In the present study we focused on the interaction between integrin α4β7 and MAdCAM-1, since clinical studies in Crohn's disease have shown that treatment with α₄-antibody, natalizumab, is effective in active disease [5]. Involvement of chemokine receptors is also important for cellular interaction with endothelium during the homing process and ultimate transmigration [40]. In retrospect we have analysed the studied gut-derived T cell cultures from individuals which uniformly have a high expression of CXCR3, CCR4 and CCR5, whereas CCR9 was only detected on a small proportion (2–8%) of the colon T cells. Due to the homogenous expression observed in several T cell cultures, we have reason to believe that the injected T cells preferentially expressed CXCR3, CCR4 and CCR5. The low expression of CCR9 on colon T cells is in accordance with an earlier study [41].

In contrast, the expression of CCR4 on our cultured gut-derived T cells is difficult to reconcile with observations by others that freshly isolated intestinal T cells consistently were negative for CCR4, and that high levels of CCR4 were found on skin-infiltrating CD4⁺ T cells [42]. Further, It was demonstrated that virtually all freshly isolated lamina propria lymphocytes, like our cultured T cells, express CCR5 and CXCR3 [43], although the ubiquitous expression of these receptors on lymphocytes from different tissues argue against a mucosa-specific role in lymphocyte homing [42].

The adequate signal for transmigration may be missing in the healthy noninflamed gut. We hypothesize that in the absence of an appropriate MAdCAM-1 expression in the gut of healthy individuals, the accumulation in the spleen may partly be due to specific homing to MAdCAM-1 expression on splenic sinus-lining cells, although no such functional role of MAdCAM-1 could be demonstrated in mice [8].

However, this does not exclude that a low level of continuous recirculation to the gut may occur below the detection level of the scintigraphy and animal studies have shown that only 5% of injected PBLs may be recovered in the gut [34]. We calculated that the scintigraphy is sufficiently sensitive to allow detection of about 0·04% of the injected number of cells (about 80 000 cells). If homing is, indeed, taking place, then the cellular concentration in the mucosa of the colon must be less than that. The time would therefore seem to be ripe for considering alternative labelling methods for human migration studies, for instance gene marking [19], although such a strategy would require a second biopsy from the target organ after infusion.

Experiments were performed in healthy individuals in order to circumvent the complex scenario in the inflamed gut where selectivity of gut-homing may be lost [3]. Indeed, murine models of colitis suggest that expression of β7 integrin is dispensable for localization to the intestine [44].

It is possible that gut-homing of cultured gut-derived T cells would have occurred in the inflamed gut, which is the real target for adoptive immunotherapy. Consequently, future studies should include in vivo homing in patients with active IBD in order to decide whether an inflammatory stimulus will direct cultured gut-derived T cells to the gut mucosa.