Endoscopic photoconversion reveals unexpectedly broad leukocyte trafficking to and from the gut

Establishment of Minimally Perturbant Methodology to Monitor Leukocyte Trafficking to and from the Gut.

The Kaede tg mouse line provides the possibility of noninvasively tracking immune-cell trafficking over a timescale of 1 to 3 wk (12). As anticipated, given that expression of the Kaede reporter is driven by β-actin promoter/enhancer elements, all CD45⁺ cells fluoresced green and not red at all sites examined in unmanipulated Kaede tg mice (Fig. 1A).

To study leukocyte immigration to the gut, we monitored movement of selected phenotypes from a distal lymphoid organ. Preliminary studies revealed very similar immigration of cells from the inguinal LNs (ILNs; lateral iliac LNs per ref. 11) and cervical LNs (CLNs, or mandibular LNs according to the nomenclature of ref. 11) to the gut and gut-draining LNs (Fig. S1). We chose to photoconvert cells in the CLNs in subsequent experiments because of several practical considerations: (i) the increased reproducibility of photoconversion at that site given that CLNs, but not ILNs, can be visualized by eye once the fur has been removed; (ii) even lower nonspecific photoconversion of cells in the peritoneum and gut tissues by targeting the more distant CLNs; and (iii) the larger cell population in the CLNs (see below). Similar to what has been reported for other LNs (12, 14), exposure of the CLNs, through depilated overlying skin, to violet light for 3.5-min photoconverted Kaede in 30–50% of CD45⁺ cells at that site, although it remained unconverted at all distal locations examined (Fig. 1B and see below).

To examine immune-cell emigration from the gut, we developed an endoscopic procedure to stereotypically label cells in the descending colon, which is the enteric region richest in microbial colonization (16) and the site of ulcerative colitis (17). This gut segment is known to host functionally relevant populations of both T helper (Th)17 (18) and regulatory T (Treg) (19) cells. In set-up experiments, transcutaneous exposure of the enteric region to violet light led to weak and diffuse photoconversion, and the endoscope was not able to access the small intestine (SI) and other regions of the colon. Endoscopic exposure of violet light to the descending colon for a total of 5 min resulted in photoconversion of Kaede in 25–35% of CD45⁺ cells in the distal large-intestinal lamina propria (LI-LP), but not detectably at other sites, including other regions of the LI-LP, the SI-LP, the Peyers Patches (PP), and nearby LNs (Fig. 1C and see below).

Subsequent to violet light exposure, most of the Kaede-red⁺ CD45⁺ cells progressively exited the CLNs or LI-LP and were replaced by incoming Kaede-green⁺ cells (Fig. 1 D and E and Table S1; see also Table S2), a process previously demonstrated to reflect steady-state leukocyte circulation (12, 14) (and see below). Efflux of Kaede-red⁺ immune cells was distinctly more rapid from the LI-LP than from the CLNs, the former near-complete after only 1 d. In general, the major lymphoid subsets circulated more extensively and rapidly than did the myeloid subsets, although this appeared not to be the case for all T-cell classes. For example, Kaede-red⁺ conventional T (Tconv) cells exited the CLNs much more effectively than did Kaede-red⁺ Treg cells, and photo-tagged Th1 and Th17 cells were both restrained in their exodus (Fig. 1 D and E). The various observations are unlikely to reflect substantial violet light-induced cell death in either tissue given near-identical scatter profiles before and after photoconversion.

Tracking Immigration of Leukocytes to the Gut from a Distant LN.

We began with a time-course experiment designed to evaluate the frequency and rapidity of cell trafficking from the CLNs to various regions of the gut, in comparison with their circulation through extragut lymphoid tissue. CLNs of 13-wk-old Kaede tg mice were exposed to violet light, and tissues were excised for flow-cytometric analysis of cells containing photoconverted Kaede, immediately or 1–3 d later. That the resulting data reflect cell proliferation to any substantial degree is unlikely given the invariable mean fluorescence intensity of Kaede-red staining in cells examined from 1 to 7 d after exposure to violet light (Fig. S2). Kaede-red⁺ CD45⁺ cells were detected at all of the locations examined, although relatively few, fractionally, were found in the SI-LP and PPs (Fig. 2A). The kinetics of Kaede-red⁺ leukocyte appearance in the LI-LP was very similar to that in the MLNs (jejeunal and colic LNs according to ref. 11) or in the ILNs (Fig. 2B).

In general, and as expected, appearance of cells of the major myeloid and lymphoid subsets in the ILNs reflected exodus rates from the CLNs (Fig. 1D vs. Fig. 2 C and D). To assess selective migration of the various subsets to the gut, we quantified, over time, the content of photo-tagged cells in the different compartments of gut-associated tissues compared with in the same compartment of the ILNs, the latter being a measure of systemic circulation pools (Fig. 2 E and F and Table S3). All of the major myeloid subsets immigrated to the colon to at least some degree, although accumulation of CD11b⁺c⁻ and CD11b⁻c⁺ cells was relatively disfavored and of CD11b⁺c⁺ favored (Fig. 2E). These differences did not simply parallel variations in relative numbers of cells photoconverted in the CLNs (Tables S1 and S2), exodus rates (Fig.1D), or differential compartment sizes in the LI-LP vs. CLNs (Table S4). Each of the major lymphoid subsets also trafficked to the LI-LP, with only subtle variations in effectiveness (Fig. 2F).

Examination of CD4⁺ T-cell subsets revealed certain more striking differential migratory behaviors. Reflecting their restrained egress from the CLNs (Fig. 1D), the fractional accumulation of Kaede-red⁺ Treg cells was lower than that of their Tconv cell counterparts in all tissues examined, levels of Tregs in the LI-LP being particularly low (Fig. 2G). (Note: The Th1 and Th17 compartments in the LI-LP were too small for accurate assessment of immigration from the CLNs.)

Monitoring Leukocyte Emigration from the Colon.

Next, we tracked the movement of immune cells from the LI to nearby and distant lymphoid organs. The descending colon of 13-wk-old Kaede tg mice was endoscopically exposed to violet light, and various tissues were removed and cytofluorimetrically examined for cells containing photoconverted Kaede, at various time points up to 72 h. By 48 h, cells photo-tagged in the colon were detected at all locations examined, including the SI-LP and other (ascending and transverse) regions of the colon (Fig. 3A, Lower). A comparison with the corresponding dot-plots for the 0-h time points indicates that spurious photoactivation within the nearby regions cannot account for the appearance of cells with photoconverted reporter (Fig. 3A, Upper). Kaede-red⁺ CD45⁺ cells from the descending colon emerged first and foremost in the caudal LN (CaLN; caudal mesenteric LN per ref. 11), well before their appearance in the MLNs (Fig. 3B), consistent with the aforementioned dye-tracking experiments (10). The rapid accumulation of photo-tagged cells in the CaLN was specific as such cells were not detectable in the adjacent medial iliac LNs. The lower fractional representation of labeled cells in the MLNs and ILNs after flashing the LI-LP versus the CLNs (Fig. 2B) parallels the fewer CD45⁺ cells photo-tagged in the former vs. latter tissue (Table S1 vs. Table S2).

Particular immune-cell subsets exhibited distinct patterns of emigration from the colon to neighboring and outlying lymphoid tissues (Fig. 3 C–E and Table S5). Unfortunately, the small size of the CaLN precluded studies on subset-specific migration to this node, but it was possible to track the major myeloid and lymphoid subsets to all of the other sites examined. For the myeloid lineage, Kaede-red⁺ CD11b⁺c⁺ cells consistently accumulated to the highest fractional levels, which, interestingly, did not reflect either a particularly large starting population of labeled LI-LP cells (Table S2) or particularly effective colonic egress (Fig. 1E). In contrast, photo-tagged CD11b⁻c⁺ cells always showed the lowest fractional representation in the lymphoid tissues (Fig. 3C), again not reflective of either the size of the starting labeled LI-LP population or the egress rates (Fig. 1E and Table S2). The fraction of photo-tagged lymphoid cells in nearby and distal lymphoid organs was, in general, distinctly lower than that of myeloid cells (compare Fig. 3 C and D). This result was unexpected given that more lymphocytes were labeled within (Table S2) and that lymphocytes egressed faster from (Fig. 1D) the LI. However, the differences might be explained—at least partially—by the fact that, in general, the relative sizes of lymphoid cell compartments in lymphoid organs versus in the LI-LP are greater than is the case for the corresponding myeloid cell compartments (Table S4). In contrast to what had been seen for migration to the LI-LP from the CLNs, circulation of Tregs from the distal LI to other locations, although quite low, was slightly more effective than trafficking of Tconv cells (Fig. 3E). This difference is also likely to reflect relative compartment sizes: the relative size of the Tconv cell compartments in lymphoid organs versus in the LI-LP is greater than is the case for the corresponding Treg compartments (Table S4). Again, the Th1 and Th17 compartments in the LI-LP were too small to permit accurate tracking after dispersal through the body.

Linking Colonic T Cells to a Gut-Distal Autoinflammatory Disease.

Finally, we were interested in potentially tying leukocyte trafficking from the gut to pathogenesis. K/BxN TCR tg mice are a model of autoinflammatory arthritis (reviewed in ref. 20). The K/BxN disease depends on excessive production of autoantibodies (autoAbs) to glucose-6-phosphate isomerase (GPI), primarily in the spleen (21). Circulating GPI and anti-GPI autoAbs form immune complexes that deposit on joints, where they induce activation of the innate immune system and ultimately inflammation (22). The requisite high level of circulating anti-GPI autoAbs depends critically on an interaction between B and Th17 lymphocytes in the spleen which, in turn, relies on induction of a robust Th17 compartment by the intestinal microbiota (4). We sought evidence that gut-derived Th17 cells can contribute to the gut-distal autoinflammatory disease of K/BxN mice.

Violet light was shone on the descending colon of 3.5-wk-old K/BxN mice, the age at which Th17 cells start to accumulate in the spleen and anti-GPI autoAb production begins (4). Seventy-two hours later, few CD4⁺ T cells with photoconverted Kaede remained in the LI-LP, although slightly more IL-17A producers than nonproducers (Fig. 4 A and B). Although we could pick up both types of CD4⁺ cell in the spleen at that time point, there was a higher fractional representation of Th17 cells (Fig. 4C). Moreover, the fraction of Kaede-red⁺ Th17 cells in the spleen of individual mice correlated well with the titer of serum anti-GPI autoAbs (Fig. 4D). Thus, we have evidenced gut-derived Th17 cells in the spleen of K/BxN mice. Note that, in contrast, there was minimal exodus of Th17 cells from the gut in the above-described experiments on nonarthritic mice.

Mice.

The Kaede tg mouse line (Kaede.NOD) was obtained from O. Kanagawa (RIKEN, Wako, Japan). K/BxN animals carrying the Kaede reporter were generated by crossing KRN T-cell receptor tg mice on the C57BL/6 background (KRN/B6) (26) with Kaede.NOD mice. All animals were bred and maintained in our specific pathogen-free facility at Harvard Medical School. Experiments were conducted under protocols approved by Harvard Medical School’s Institutional Animal Care and Use Committee. Typically, six to eight male mice were used per condition, spread over more than two independent experiments. In our colony, male mice of the NOD strain rarely develop diabetes and only at an advanced age.

Photoconversion Procedures.

Mice were anesthetized with ketamine:xylazine in combination (10 mg/kg:2 mg/kg i.p.) or 3% (vol/vol) isoflurane, delivered in 2 L/min of 100% O₂. For photoconversion of Kaede in CLN cells, overlying fur was removed using depilatory cream. Mice were placed on their backs with an aluminum foil blanket covering all but the depilated area, and violet light (Electra Pro Series Violet Handheld Laser; Laserglow Technologies) was shone (405 nm; peak power <5 mW; sustained power: 0.5–4.9 mW) onto the exposed area for a period of 3.5 min (direct exposure) or 5 min (via fiberoptics). To manipulate the size of the light field so that both CLNs could be exposed, we attached a lens to the laser to defocus the beam, and the source of the defocused light beam was positioned 28 cm above the mouse; beam diameter was 1.5 cm. For cell photoconversion in the descending colon, a custom-built fiberoptic endoscope (ZIBRA Corporation) was coupled to the handheld 405-nm laser, via an in-house, custom-made connection device (fixed mounts were purchased from ThorLabs). After cleansing the colon of fecal pellets with PBS, we inserted the fiberoptic endoscope through the anus into the descending colon to a depth of 2.5 cm. The laser was switched on, thereby exposing the inner colon to violet light (beam diameter was 3.5 mm). Subsequently, the endoscope was gently retracted, pausing at 2-mm increments for 30-s light pulses at each interval (for a total of 5 min).

LI-LP, SI-LP, and PP Cell Isolation.

LI and SI were harvested and fat and PPs removed. LI and SI were incubated in extraction medium (0.01% DTT, 1 mM EDTA, 2% FCS, RPMI-1640 medium for 15 min with continual stirring, at 37 °C). Tissues were then washed and residual mucus removed. Next, all tissues, including PPs, were cut into 1-mm² pieces and incubated in digestion medium (2% FBS, 0.05% Dispase, 0.15% Type II Collagenase, RPMI-1640) for 15–30 min at at 37 °C. Following digestion, the preparation was filtered and washed twice.

Flow Cytometry.

For surface phenotype analyses by flow cytometry (LSRII), fluorophore-labeled mAbs specific for CD45, CD4 (BD Biosciences), CD8, CD19, CD11b, or CD11c were obtained from BioLegend (unless otherwise noted). Intracellular Foxp3 was detected with a staining-buffer set (eBioscience), according to the manufacturer’s instructions. For intracellular cytokine analysis, cells were stimulated with 50 ng/mL phorbol 12 myristate 13-acetate and 1 μM ionomycin (both from Sigma) in the presence of BD GolgiPlug (1:1,000 dilution) for 4 h at 37 °C, 5% CO₂. Intracellular cytokines were then stained using a BD Cytofix/Cytoperm kit, as per the manufacturer’s directions.

ELISAs.

Anti-GPI titers in blood collected from 4-wk-old K/BxN.Kaede mice were quantified as previously reported (27). For correlation analysis, serum dilution at a change in absorbance of 1 was plotted against proportion of photoconverted Th17 cells in spleen.

Statistical Analyses.

Statistical significance, indicated by asterisks, was determined by Student t test (two-tailed, unpaired). P values < 0.05 were considered significant: *P < 0.05, **P < 0.01, ***P < 0.001.