Published online before print July 5, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Antigen Presentation Research Group and
St. Mark’s Institute for Intestinal and Colorectal Disorders, Imperial College London, Northwick Park and St. Mark’s Campus, Harrow, United Kingdom; and
Obstetrics & Gynaecology, St. Thomas’ Hospital, London, United Kingdom
1 Correspondence: Antigen Presentation Research Group, Imperial College London, Northwick Park and St. Mark’s Campus, 7W, Watford Road, Harrow HA1 3UJ, UK. E-mail: s.knight{at}imperial.ac.uk
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: mucosa • inflammatory bowel disease • fatty acids
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(TNF-) and interleukin-6 [3
, 4
]. Pre-adipocytes of the omentum are also reported to mature into macrophages, which contribute to local immune and inflammatory responses [5
, 6
]. Cells within adipose tissue are also now known to have stem cell activity [7
8
9
]. However, accumulation of macrophages within adipose tissue in obesity and their involvement in inflammation have been highlighted recently [10
, 11
]. Dendritic antigen-presenting cells (DC; APC) initiate and regulate innate and adaptive immune responses. They can be derived from stem cells shared with macrophages [12
], and monocytes may give rise to DC [13
]. We therefore hypothesized that DC, like macrophages, might be obtained from adipose tissue, and to test this hypothesis, we characterized myeloid cells migrating from human omental tissue. We further proposed that DC derived from omentum could contribute to or reflect local inflammatory changes and so, looked for changes in omental DC in the presence of inflammation in Crohn’s disease. Hypertrophy of adipose tissue and fat-wrapping characterize Crohn’s disease. This adipose tissue shows overexpression of peroxisome proliferator-activated receptor 
and increased production of TNF- [14
]. Increase in adipose tissue with its inflammatory changes forms the basis for theories that these changes represent a fundamental part of the pathogenic processes producing Crohn’s disease. We therefore used omental adipose tissue from these patients to identify whether changes in DC in adipose tissue might contribute to or reflect changes in inflammatory disease. |
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Cells were obtained by a walkout technique developed for enriching DC populations from gut biopsies [15 ]. Tissue samples were cut into small pieces (1–2 mm3) and put in tissue-culture medium (RPMI 1640, Dutch modification, Sigma, Poole, UK) with 10% fetal calf serum, L-glutamine (2 mM), penicillin (100 IU ml–1), and streptomycin (100 µg ml–1) overnight at 37°C in 5% CO2. Tissue was removed on a 100-µm sieve, and any red cells were removed by lysis. We originally looked at collagenase-treated cells obtained using a technique for separating "pre-adipocytes". The small number of cells obtained had a phenotype similar to those obtained with the walkout technique. We chose to study only walkout cells more extensively to avoid release of excessive amounts of fat, which cause problems in quantitating and handling cells.
Flow cytometry
Cells were labeled with monoclonal antibodies to cell surface markers directly fluorochrome-conjugated and fixed in 1% paraformaldehyde. Antibodies were fluorescein isothiocyanate (FITC)-conjugated CD45 and CD80, human leukocyte antigen (HLA)-DR, and phycoerythrin-conjugated CD45, CD3, CD19, CD14, CD16, CD34, and CD56 (Becton Dickinson, Oxford, UK); FITC-conjugated CD1a, CD40, and CD86 (Serotec, Oxford, UK); FITC-conjugated CD11c (Dako Cytomation, Ely, UK); and FITC-conjugated CD83 (Alexis Biochemicals, Nottingham, UK). Samples were acquired immediately or stored at 4°C and acquired within 36 h. Histogram analysis on the live cell gate used Winlist analysis software (Verity Software House, Topsham, ME). Percentage-positive cells and the intensity ratio, which indicates the ratio of positive-to-background staining, were determined using the super-enhanced D max subtraction routine in Winlist.
Morphological studies
For electron microscopy between 2 and 10 x 105, freshly isolated, omental cells were processed for electron microscopy. Some cells were surface-labeled with biotinylated anti-major histocompatibility complex (MHC) Class II antibody followed by goat anti-biotin 10 nm gold prior to processing for electron microscopy. At least 100 cells were counted per sample, viewed using a JEOL JEM-1200 EX electron microscope (Osaka, Japan), and assessed for labeling for MHC Class II by counting gold particles per cell. In controls, gold labeling was generally negative and always <5 gold grains per cell; >10 grains were considered positive. Macrophages and DC were identified from their characteristic morphology [16
].
Frozen sections were also prepared and stained with CD11c and MHC Class II using the avidin-biotin complex (ABC) method (Vector Laboratories, Peterborough, UK). Slides were air-dried, fixed in cold acetone (–20°C) for 5 min, washed in phosphate-buffered saline (PBS), quenched for 30 min in 3% hydrogen peroxide in methanol, washed in PBS, transferred to a humidified chamber, and incubated for 1 h at room temperature with 5% goat serum in PBS. Slides were incubated with mouse anti-CD11c or HLA-DR primary antibodies (Dako Cytomation) overnight at 5°C, washed and stained with secondary biotinylated antibody for 1 h, stained with ABC for 30 min at room temperature, and developed with peroxidase substrate solution (Vector Laboratories) and were counterstained in hematoxylin, washed, cleared in xylene, and mounted under cover-slips. Control slides were prepared by replacing the primary antibodies with normal serum. Some slides were double-labeled by immunofluorescence microscopy; frozen tissue sections (4 µm) were fixed in cold methanol:acetone (1:1) for 10 min, incubated with normal mouse serum for 1 h, washed, and incubated with primary antibody to CD83 (Beckman Coulter, High Wycombe, UK) overnight. Sections washed in PBS were incubated with secondary antibody conjugated with tetramethyl rhodamine isothiocyanate (TRITC) for 2 h and washed again in PBS. To block nonspecific binding sites, sections were incubated with normal mouse serum for 1 h and then incubated with FITC-conjugated primary antibody to CD86, CD80, or CD40 (Serotec) for 4 h, washed in PBS, and mounted under cover-slips using 4',6-diamidino-2-phenylindole-conjugated mounting medium (Vector Laboratories). Negative controls were incubated with appropriate isotype controls.
Mixed leukocyte reactions (MLR)
Large mononuclear cells of low density were enriched from omental cells over a metrizamide gradient (14.5% w/v, Sigma) [17
]. These cells were >85% DC from electron microscopy and phenotyping and were used in varying numbers (500–2000) with 6.25–100 x 103 allogeneic peripheral blood mononuclear cells in triplicate, 20 µl hanging drops in Terasaki plates. The stimulatory capacity of these cells was compared with that of DC enriched from mononuclear cells of peripheral blood, and human peripheral blood DC were enriched from the Ficoll-Paque-separated mononuclear cell population by first removing the adherent macrophage population and after 24 h, isolating nonadherent, low-density cells on a metrizamide gradient (14.5%, Sigma) [18
]. The cell population contained <5% B and T cells from labeling with CD20 and CD3. The large mononuclear cells were approximately 30% classical CD14– DC and 70% CD14 low (intensity of CD14 staining, <5 times isotype background vs. approximately >40 times background for freshly isolated monocytes). CD14– and CD14 low populations have veiled morphology, express costimulatory markers CD80, -86, and -40, and are potent stimulators of primary T cell proliferation; these cells, added at a ratio of less than 1:200 allogeneic T cells, stimulate a MLR. Plates containing cell cultures of DC plus allogeneic T cells were inverted and cultured for 3 days over sterile saline at 37°C and pulsed for 2 h with 3H thymidine to give a final concentration of 1 µg ml–1 (specific activity, 2 Ci/mM, Amersham Biosciences, Amersham, UK). These conditions allow accurate reflection of DNA synthesis in the cells, as they ensure free availability of the alternative pathway precursor throughout the short pulse-time with low levels of radiation-induced damage in cells taking up the thymidine [19
]. Cells were then blotted onto filters, washed with saline, trichloroacetic acid (5%), and methanol, and dried. The filter was exposed to a tritium screen for 4 h and imaged in a phosphor imager (Amersham Biosciences). The accumulated counts on the storage phosphor screen were quantitated using ImageQuant software (Molecular Dynamics, Amersham, UK) [17
]. This 20-µl culture technique produced earlier proliferative responses than those obtained in conventional 200 µl cultures. Counts were lower using imaging than those produced from scintillation counting, but the results give >0.97 correlations between scintillation counts and the counts obtained by imaging [20
]. Counts stimulated by concanavalin A, used routinely as a positive control, were 1000–5000 counts. Significance was tested using two-way ANOVA of log transformed, normally distributed data for each of five omental samples studied in each patient group.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
 View larger version (10K): [in a new window] |
Figure 1. Cells obtained from omental samples. The total numbers of cells migrating out of fragments of omental adipose tissue after overnight incubation are shown ( ) and the percentages of these cells expressing CD45 as measured by flow cytometry ( ) in samples from controls and from Crohn’s patients. There was no significant difference in the proportion of CD45+ve cells obtained from control or Crohn’s omentum. There were significantly more cells obtained from the Crohn’s than from control omentum (P=<0.001).
|
 View larger version (19K): [in a new window] |
Figure 2. Phenotypic profile of cells from control omentum by flow cytometric analysis. (a) A typical profile of forward- and side-scatter of cells, which were CD45+. (b–f) The open histograms show the positive staining for the cell surface markers indicated. The isotype control (not shown) has been subtracted using the Winlist algorithm; cells that are positive after subtraction of the isotype control are shown in the filled portions of the histograms, and those that are negative remain unshaded. The profile for each antibody is typical of that obtained in at least five experiments.
|
 View larger version (20K): [in a new window] |
Figure 3. Profiles of staining for different markers of the CD45+ cells from omentum. Cells migrating from omental fragments were double-labeled for CD45 and for the markers indicated. The profile of the CD45+ cells positively stained for different markers is shown. The portion positive after subtraction of the isotype control is shown in the filled portions of the histograms, and those negative remain unfilled. The percentage positive and the positive/control intensity ratio denoting the level of staining are shown on each histogram. The labeling of omental cells from a control and a Crohn’s is shown. These plots are representative of the labeling obtained from 11 controls and nine Crohn’s patients.
|
 View larger version (12K): [in a new window] |
Figure 4. Numbers of CD45+ omental cells positive for different markers. Cells migrating from omental fragments were double-labeled for CD45 and for the markers indicated and the proportions positive shown after subtraction of the isotype control using the Winlist algorithm. The percentage-positive values for cells from controls and Crohn’s patients were compared, and significant differences are indicated.
|
Morphological identification of DC
In cytospins of omental walkout cells, we confirmed that there were only occasional lymphocytes and that large mononuclear cells formed the major cell population, some containing obvious lipid droplets; the latter cells appear to be developing adipocytes and may account for the presence of CD45-negative cells within the samples. In control samples, granulocytes were not identified, but in samples from Crohn’s patients, granulocytes formed 20 ± 8% of the cells (range 5–45%). As DC are difficult to identify securely using light microscopy, we sought to confirm the presence of DC using electron microscopy. DC enriched using metrizamide gradients were studied in two control samples and two from Crohn’s patients. The majority (>90%) of cells from the controls appeared myeloid cells, of which >70% were identified as DC (Fig. 5a
and 5c
); the remainder was not clearly identifiable or was macrophages (with many round vacuoles of different sizes, heterochromatic nuclei with a rounded or horseshoe shape, and multiple organelles of different types present in a cytoplasm, which often appeared darker than that of the DC). Cells of three morphological types of DC previously identified in DC separated from human peripheral blood were identified in each sample [16
]. They included cells with electron dense nuclei, some vacuoles, and numerous short pseudopods as shown in Figure 5
. DC with a smoother cell boundary and fewer processes, resembling Langerhans cells and cells with longer processes, which had a less electron-dense nuclear morphology, similar to that of veiled cells in the afferent lymph, were also present. Over 70% of myeloid cells labeled positively for MHC Class II by immunogold labeling (Fig. 5b
and 5c)
, and many had high levels of gold staining (Fig. 5d)
. Amongst mononuclear cells in the samples from Crohn’s patients, there was again a high preponderance of veiled cells clearly identified as DC by electron microscopy (65% and 80% in the two studied), and there was no evidence of staining of these cells for MHC Class II, confirming the loss of MHC Class II observed by flow cytometry. The presence in omentum of mononuclear cells of dendritic morphology was also confirmed by immunostaining in light microscopy. Clusters of cells of veiled/dendritic morphology were seen in the omentum in interstices between the adipocytes, and specific staining of these cells for CD11c is shown in Figure 5e
and 5f
. In the Crohn’s disease samples, there was staining for CD11c but not for MHC Class II. In the Crohn’s omentum, there were also scattered granulocytes visible (not shown). In frozen sections, double-staining of cells was confirmed using fluorescently labeled antibodies to CD83, which gave the highest level of labeling, and each of the costimulatory molecules (CD40, CD80, and CD86), as exemplified for CD86 in Figure 5g
and 5h
.
 View larger version (39K): [in a new window] |
Figure 5. Morphology of omental DC. (a) A typical DC from a control omentum is shown. (b) An example of immunogold staining of the veils of a DC. The original bars indicate 1 µm. In controls, gold labeling was always <5 gold grains per cell; >10 grains was considered positive. (c) The types of myeloid cells identified in the omental cells from two control patients; the proportion positive for HLA-DR staining is shown in the solid portion of the histograms, and those negative are hatched. M , Macrophage. (d) The gold grains per cell for the surface DR-labeling in the two experiments performed are shown. No evidence of DR-labeling was seen in either of the two Crohn’s patients studied. (e) A light microscopy picture of a cluster of putative DC immunostained with CD11c on a frozen section of omentum from a Crohn’s patient showing the dark, positive-peroxidase staining, which was absent from control slides. Cell nuclei, which were stained with hematoxylin, were distinguishable, showing that the veils were associated with mononuclear cells in the section. (f) Control for e using serum instead of specific antibody and with only hematoxylin staining visible. (g) Frozen section of control omentum, showing two cells specifically labeled for CD83 (TRITC-labeled). (h) The same section as g with label specific for CD86 (FITC-labeled).
|
 View larger version (11K): [in a new window] |
Figure 6. MLR stimulated by omental DC. Peripheral blood DC were enriched from 24 h nonadherent Ficoll-separated mononuclear cells on metrizamide gradients. Omental DC were also enriched over metrizamide. Small numbers, 1000 ( ) or 2000 ( ), of these cells were added to different numbers of responder cells in 20 µl hanging drop cultures, and the stimulation was measured by uptake of tritiated thymidine into the cells in a 2-h pulse of thymidine on Day 4 of culture after blotting them onto a filter and imaging the filter in a phosphor imager. (a) The uptake in a normal MLR in this microculture system using peripheral blood DC. (b) Stimulation with control omental DC and (c) effect of adding DC from Crohn’s omentum. The background proliferation of unstimulated responder cells () is shown in each graph. The control and omental DC each caused a significant stimulation (P=<0.001), but the Crohn’s omental DC caused little (P=0.05, n=2) or no (n=6) significant stimulation.
|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
DC can be obtained as they mature and migrate out of skin or gut tissue fragments [15 ]. We adapted this method to obtain enriched DC from human omental samples. There were few adipocytes present, as these cells floated and were removed, so avoiding problems of having large quantities of fat present. There was some blood contamination within omental samples, and some cells could possibly derive from this contamination. However, the preponderance of myeloid cells and the presence of higher numbers of DC/mg tissue than there are DC/ml blood ruled out blood as the source of the majority of cells obtained. Occasional omental samples showed pale areas, which may have been "milky spots", but these areas proved impossible to dissect out, as they were fluid rather than structured. The lack of lymphocytes also suggested that we were not obtaining immune cells from structured lymphoid tissue within the fat. All control, omental samples were from females, whereas the Crohn’s patients were evenly split between male and female. However, analysis of the numbers of cells from male and female Crohn’s patients showed no significant differences in numbers of omental cells. We also saw no evidence that the treatment of some patients was significantly altering the cells that we obtained or their functions.
Expression of MHC Class II is generally considered a defining characteristic of DC and was well represented on a high proportion of the cells from control omentum. The lower MHC Class II found on cells in Crohn’s could indicate a loss of DC from omentum in Crohn’s. However, the similarity of the phenotype of cells from controls and Crohn’s patients argues that these are related to the DC and have lost or lack MHC Class II. The information that control omentum contains stem cells, including hematopoietic stem cells, as indicated from the CD34 labeling, however, could mean that these cells represent cells developing down an alternative maturation pathway. The presence of granulocytes in omentum in Crohn’s could indicate that some omental stem cells are maturing down a granulocytic pathway, although recruitment of additional cells from the circulation is an alternative possibility. There is evidence of a close relationship between granulocytes and DC. Thus, granulocytes can be driven to acquire DC characteristics [25 ], and an imbalance between granulocytes and DC, favoring production of the former, may contribute to production of autoimmune disease [26 ].
The lack of Class II and loss of T cell stimulatory function of cells with the phenotype and morphology of DC in omentum were unexpected. Innate immune mechanisms to bacteria mediated by PRR such as TLR on DC might then underlie the immune activity in the gut [27
]. A reduction in the capacity of cells to express MHC and to initiate adaptive immune responses could perhaps represent a mechanism indicating failure of DC to mature and elicit appropriate adaptive immune activity to deal with infecting organisms entering via a gut with an increased permeability. Systemic changes in numbers and functions of DC in Crohn’s disease have recently been reported [28
], as well as DC changes in the gut [29
], and it will be of interest in future experiments to examine in more detail DC from these other compartments. The changes in putative DC in Crohn’s include an increase in CD40 expression, which has also been observed in DC isolated from the lamina propria of patients with Crohn’s, providing a link between the activity of cells in the colon in Crohn’s and those in the omentum. The CD40 levels on DC in the colon normalize after treatment with anti-TNF antibody [29
]. The elevated production of TNF- in the adipose tissue in Crohn’s thus provides a further link between the aberrant activity in the colon and expression of costimulatory markers and the function of DC. The involvement of adipose tissue in the development of Crohn’s disease rather than the concept that the changed fat distribution is merely a consequence of the disease process has been postulated [14
]. The presence of primary antigen-presenting DC migrating from the omentum suggests that adipose tissue may be a source of local APC. Significant changes in cellular components of adipose tissue in Crohn’s disease, taken together with the increased TNF- production in adipose tissue in Crohn’s, lend weight to the idea of greater involvement of adipose tissue in disease pathogenesis than has previously been considered, although questions of cause and effect have yet to be elucidated.
Finally, different fatty acids can modulate immune functions differentially, and some evidence suggests that fatty acid distribution is altered in Crohn’s disease and that manipulation of dietary fats can be beneficial in its treatment [30
]. One possible mechanism of action of fatty acids is via modulation of macrophages, perhaps via effects on TLR [31
]. Fatty acids may also block DC function, an effect that is independent of nuclear factor-
B activation [32
]. Fats are taken up into developing DC, and this uptake is modulated by the effects of cytokine, antigen, or location [24
]. The presence of 
-3 fatty acids particularly can down-regulate immune activity, possibly via the effects of downstream molecules such as prostaglandins [33
]. -3 fatty acids fed to rats down-regulate the stimulatory functions of DC [34
], and prostaglandins modulate DC function [35
], suggesting that DC are affected by environmental fatty acids and their downstream products. Perinodal adipose tissue is high in unsaturated fatty acids, which can down-regulate the activity of stimulated lymphocytes [36
]. The DC derived from adipose tissue itself may therefore provide a conduit for effects of fatty acids on the immune system. There is loss of -3 fatty acids in the plasma in Crohn’s disease [37
, 38
] and consequent changes in downstream molecules such as prostaglandins with their effects on DC function [35
], providing possible mechanisms via which fatty acids within the adipose tissue may influence APC and produce changes in immune activity in Crohn’s disease. We showed that the usual preponderance of unsaturated fatty acids in perinodal adipose tissue was absent in tissue from patients with Crohn’s disease with a major loss in -3 fatty acids [39
]. However, changes in environmental fatty acids may not be mirrored in lymphoid cells, and the interactions between environmental fatty acids and the cells of the immune system may be critical in determining the metabolic and immune activity [40
]. We observed that the fatty acids in lymph node cells from patients with Crohn’s disease, unlike those in plasma and adipose tissue, showed a disproportionate loss in -6 fatty acids, leading to a significantly lower-than-normal -6/-3 ratio [39
]. Such an increase in the relative proportion of anti-inflammatory -3 fatty acids in immune cells from Crohn’s patients may provide a mechanism for the loss in immune activation observed.
In conclusion, these novel studies of DC showing their presence in omental adipose tissue and changes in their number, phenotype, and function in Crohn’s disease identify a number of mechanisms by which these omental DC may be involved in local immune responses and altered in inflammatory disease. These data, establishing that DC are present in adipose tissue, also pave the way for evaluation of their source, development pathways, lineage relationships, and a more detailed examination of the way in which adipose tissue DC may be involved in local immune responses in health and in disease.
Received September 7, 2005; revised April 11, 2006; accepted May 19, 2006.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
release from human adipose tissue in vitro J. Endocrinol. 163,33-38[Abstract], interleukin 4 and interleukin 6 in the control of lipolysis from adipocytes around lymph nodes Cytokine 11,334-346[CrossRef][Medline] J. Exp. Med. 179,1109-1118B activation J. Biol. Chem. 280,14293-14301
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
