Background
Ulcerative colitis (UC) belongs to the chronic inflammatory diseases of unknown etiology. Manifestations of the disease differ highly with regard to onset, severity, course and response to therapeutics and it is likely, that various forms of the disease are covered by the umbrella diagnosis of UC. It is presently thought that a combination of genetic, environmental and microbiotic factors is responsible for an uncontrolled immune response characterized and driven by the typical Th2 cytokine Interleukine (IL)-13 (for review see [
1]). We have recently taken a slightly different view which is based on the assumption that the inflammatory response partially resembles an uncontrolled wound healing process caused by epithelial damage [
2]. In this model an active role is assigned to epithelial cells as they release damage associated molecular pattern molecules (DAMP) such as TSLP (thymic stromal lymphopoietin) to direct immune cells towards a Th2 characterized immune response ultimately resulting in increased mucus production, epithelial cell hyperplasia and fibrosis. DAMPs comprise heterogeneous molecular entities ranging from toxic agents such as uric acid to proteases and allergens. The release of TSLP by epithelial cells can be induced by damage, sensing microbial products or dsRNA via toll like receptors (TLR) [
3‐
9]. This results in macrophages, dendritic cells (DC) and monocytes having the center stage because as the preferred targets of TSLP and as a highly volatile cellular population they have the capacity to further fuel or dampen inflammation depending on the inflammatory milieu. The crucial role of macrophages in UC is supported by the efficacy of infliximab an anti-tumor necrosis factor (TNFα) monoclonal human antibody with IgG1 effector function which is thought to be deleterious to macrophages expressing TNFα on their surfaces [
10].
Wound healing processes are also governed by macrophages. Here, neutrophils and inflammatory macrophages (M1) infiltrate damaged tissue as a first step to protect the organism from pathogens. Once the wound is sealed macrophages engulf apoptotic neutrophils in a process called efferocytosis and in consequence alter their phenotype from inflammatory to healing (M2) [
11,
12]. Macrophages are main sources of hepatic growth factor (HGF), transforming growth factor (TGF) ß1 and thymus and activation regulated chemokine (TARC), all of which play a crucial role in wound healing processes and have been described as elevated in biopsies and/or sera of UC patients [
13,
14]. Hepatic growth factor induces re-epithelization and is expressed in fibroblast, monocytes, dendritic- and endothelial cells. HGF leads to proliferation, motility, IL-1ß, IL-4, GM CSF secretion and suppresses CD4+ and CD8+ cells [
15]. TGFß1 induces fibrosis and scarring and additionally acts as an immune modulator. It is expressed in regulatory T-cells, M2 macrophages, fibroblasts and platelets [
16,
17]. TARC is expressed in epithelial cells, monocytes and macrophages and attracts CCR4 bearing Th2 cells, regulatory T-cells and macrophages into the damaged tissue [
18,
19]. Finally, periostin is a protein associated with remodeling of the colon and lung architecture and expressed under the control of IL-13 which has been identified as a biomarker for patients responding to IL-13 therapy [
20,
21]. Expressed by epithelial cells and fibroblasts and acting in an autocrine manner it induces collagen disposition, TGFß1 secretion, cell migration and epithelial to mesenchymal transition.
CD1a has been known for decades as a phenotypic marker of human epidermal Langerhans cells (LC). Like other members of the CD1 family CD1a presents lipids to evoke T cell activation resulting in the release of IL-22, IL-13 and interferon (IFN)γ [
22,
23]. In the healthy skin auto reactivity is thought to be prevented by physical segregation of LC located in the epidermis and the corresponding lipid ligands concentrated in the stratum corneum. Upon injury or inflammation segregation breaks down and LC activate autoreactive T-cells, notably Th22 cells which are the major source of IL-22 production, which is thought to be instrumental in wound healing processes and antimicrobial defense [
24‐
26]. The supply of fatty acid ligands can also be provided by phospholipase A2 (PLA2) an important component of inflammation and bee- and wasp venom [
27]. PLA2 releases fatty acids from phospholipids in the extracellular space where loading of CD1a takes place. To date, CD1a bearing macrophages have not been detected in the colon.
To test our hypothesis of UC resembling an uncontrolled wound healing process we analyzed the immunological profile of UC patients and compared these to those of Non UC donors by determining the frequency of immune cells in the blood and serum levels of TARC, HGF, TGFß1 and periostin. Cluster- and correlation analysis was performed to portray individual profiles and to understand the correlation of factors and immune cells in the inflammatory milieu of UC patients. In a second step, we analyzed immune cells isolated from colons of UC patients undergoing colectomy to verify that the cells identified as relevant in the blood reflected the inflammatory condition in the colon.
Methods
Ethical considerations
All donors gave informed written consent and the study was approved by the Institutional Review Board (IRB) of the Medical Faculty at the University of Munich (2015–22).
Colon samples and annotated data were obtained and experimental procedures were performed within the framework of the non profit foundation HTCR, including the informed patient’s consent [
28].
Isolation of PBMC
Peripheral blood was collected from the arm vein of patients suffering from UC and Non UC donors. The Non UC donors had no apparent infection, inflammation and did not suffer from other chronic inflammatory diseases. As we could not exclude hidden infections they were considered as Non UC. 10 ml of blood in trisodium citrate solution were diluted with Hank’s balanced salt solution (Thermofisher, Waltham, USA) in a 1:2 ratio and 30 ml of the solution loaded onto Leukosep tubes (Greiner Bio One, Frickenhausen, Germany). Cells were separated by centrifugation with 800g for 30 min. The interphase containing PBMC was in Hanks balanced salt solution and centrifuged with 1400g for 5 min. The cell pellet was resuspended in sterile phosphate-buffered saline (PBS) at a concentration of 4 × 106 cells in 100 µl.
Isolation of lamina propria leucocytes
For isolation of lamina propria mononuclear cells (LPMC) from human colons which were extracted from UC and colon cancer patients, a protocol modified of Hansson et al. [
29] was used. A piece in the size of approximately 2 × 2 cm of colonic wall were kept in 1× RPMI (Thermofisher, Waltham, USA) containing 10% FCS on ice until preparation. The mucosa was dissected of underlying muscular layers and fat with scissors and cut into small pieces <5 mm.
The tissue was predigested for 4 × 15 min in 15 ml predigestion solution containing 1× HBSS (Thermofisher, Waltham, USA), 5 mM EDTA, 5% FCS, 100 U/ml Pencillin–Streptomycin (Sigma-Aldrich Co.,St. Louise USA) in an orbital shaker with slow rotation (40 g) at 37 °C. To remove epithelial cells, cell suspensions were filtered through a nylon filter. Following removal of excess EDTA with RPMI the pieces were cut into finer pieces of <1 mm and digested for 60 min in digestion solution containing 1× RPMI, 10% FCS, 1 mg/ml collagenase A (Sigma-Aldrich Co., St. Louise USA), 10 KU/ml Dnase I (Sigma-Aldrich Co., St. Louise USA), 100 U/ml Pencillin–Streptomycin (Sigma-Aldrich Co., St. Louise USA) in an orbital shaker with slow rotation (40 g) at 37 °C.
Isolated LPMC were collected by centrifugation with 177 g for 10 min and resuspended for FACS analysis. Cell suspensions were filtrated one more time using a 35 µm cell strainer for further purification before labelling the cells for flow cytometry analysis.
Flow cytometric analysis
Cellular markers to phenotype UC patients and healthy controls are depicted in Table
1.
Table 1
Cellular markers used in phenotyping of UC patients
CD19+ CD27+ IgD+ | Unswitched memory B-cell | |
CD19+ CD27+ IgD− | Switched memory B-cell | |
CD19+ CD38+ | Plasma cell | |
CD4+ CD25+ | Activated CD4+ T-, regulatory T cells, Th2 cells | |
CD14+ | MC | |
CD14+ CD80+ CD86+ | MC, mature | |
CD14+ CCR2+ | MC, tissue penetrating, inflammatory | |
CD14+ TSLPR+ | MC, expressing TSLPR | |
CD14+ CD64+ | M1MC | |
CD14+ CD163+ | M2 MC, scavenging cells | |
CD14+ CD1a | MC CD1a expressing | |
CD11b+ | cDC1 | |
CD11b+ CD80/86+ | cDC1, mature | |
CD11b+ CD1a+ | cDC1 CD1a expressing | |
CD11b+ TSLPR+ | cDC1 TSLPR expressing | |
CD3+ CD56+ | NK T-cell | |
CD3− CD56+ | NK cell | |
CD3− CD56−, CRTH2+ CD127+ | ILC2 | |
Table 3
Immunological profiling leads to identification of specific cellular markers for UC and to markers signifying therapeutic responses
CD19+ | −8.80
|
0.05
| −0.1 to 17.7
| −10.78 | | |
24.07
|
0.01
|
5.1 to 43.0
| −3.99 | | |
25.85
|
0.05
| −1.0 to 52.78
|
CD19+ CD27+ IgD+ | −6.07 | | | 5.50 | | | −13.67
|
0.009
| −23.74 to −3.58
| 6.47 | | | −16.27
|
0.006
| −27.45 to −5.08
|
CD19+ CD27+ IgD− | 10.57 | | | 3.47 | | |
18.65
|
0.002
|
7.0 to 30.27
| −1.56 | | |
20.36
|
0.002
|
7.82 to 32.88
|
CD19+ CD38+ | −37.10
|
5E−03
|
23.4 to 52.5
| −29.82
|
8E−04
| −45.8 to −13.2
| 14.75 | | | −4.10 | | |
27.99
|
0.05
|
0.46 to 55.5
|
CD4+ | −8.56
|
0.006
|
2.7 to 14.9
| 2.00 | | | −2.71 | | | −6.64 | | |
8.63
|
0.03
|
0.8 to 16.4
|
CD4+ CD25+ | −7.16
|
3E−05
|
4.1 to 10.4
| −2.20 | | | −0.68 | | | −2.64 | | | 5.53 | | |
CD8+ | −0.19 | | | −4.75 | | | 1.30 | | | 0.97 | | |
11.60
|
0.008
|
4.0 to 19.78
|
CD11b+ | −6.31
|
0.005
|
2.1 to 11.2
| −4.12 | | | 2.85 | | | −2.93 | | | −0.86 | | |
CD11b+ TSLPR+ | 3.73 | | | −15.36
|
0.04
| −30.26 to −0.44
|
23.92
|
0.03
|
2.4 to 45.3
| 0.08 | | | 23.55 | | |
CD11b+ CD1a+ |
17.73
|
5E−04
| −27.9 to −8.3
| 4.10 | | | −12.65 | | | 8.76 | | | −6.59 | | |
CD11b+ CD80/86+ | −0.97 | | | −12.86
|
0.04
| −25.3 to −0.35
| 0.38 | | | 9.17 | | | 3.95 | | |
CD14+ | −0.14 | | | −0.86 | | | 0.37 | | | −0.83 | | | −0.67 | | |
CD14+ CCR2+ | 15.58 | | | −5.78 | | | 11.75 | | | −1.05 | | | 1.59 | | |
CD14+ CD80/86+ | −10.69 | | | 8.03 | | | −26.71
|
0.03
| −49.95 to −3.4
| −2.07 | | | −20.97 | | |
CD14+ TSLPR+ |
3.53
|
0.03
| −8.4 to −0.5
|
6.74
|
0.03
|
0.8 to 12.4
| 1.67 | | | 1.79 | | | 3.00 | | |
CD14+ CD1a+ |
20.35
|
0.02
| −40.8 to -4.5
| 6.91 | | | 4.38 | | | 6.40 | | | 3.76 | | |
CD14+ CD64+ | −9.87
|
0.06
| | −0.14 | | | −0.51 | | | −0.66 | | | 0.50 | | |
CD14+ CD163+ | −3.54
|
0.05
|
6.0 to 2.5
| 0.15 | | | −2.03 | | | 0.00 | | | −0.99 | | |
CD3− CD56+ | −0.65 | | | −0.58 | | | 1.80 | | | −2.59 | | | 4.39 | | |
CD3+ CD56+ |
6.35
|
0.001
| −10.3 to −2.6
| −1.41 | | | −4.94 | | | 3.22 | | | −4.36 | | |
Factors [µg/ml] | | | | | | | | | | | | | | | |
TARC | 1.79 | | | −6.69
|
1E−04
|
32.57 to 99.45
| 4.25 | | | −1.34 | | | 5.91 |
3E−03
|
23.92 to 94.34
|
HGF |
1.41
|
6E−04
| −21.98 to −0.62
|
2.02
|
0.001
| −32.20 to −0.82
|
2.05
|
0.03
|
0.18 to 39.11
| −0.11 | | | 1.63 | | |
TGFß1 |
16.76
|
3E−03
|
23.53 to 38.14
| −22.13
|
0.01
|
52.94 to 30.81
| −24.15
|
0.03
| −45.31 to −29.97
| −4.36 | | | −26.47 |
0.05
| −53.10 to −0.16
|
Periostin | −13.43 | | | −21.17 | | | −29.87
|
0.02
| −53.84 to −59.07
| −1.47 | | | −17.06 | | |
Table 4
Leucocytes characterizing the two inflammatory conditions
CD19+ CD27+ IgD+ | − | + |
CD19+ CD27+ IgD− | + | − |
CD19+ CD38+ | + | − |
CD4+ CD25 | + | |
CD11b+ TSLPR+ | + | − |
CD11b+ CD1a+ | | + |
CD14+ TSLPR+ | − | + |
CD14+ CCR2+ | + | − |
CD14+ CD1a+ | ? | ? |
CD14+ CD64+ | + | |
CD3+ CD56+ | | + |
CD3− CD56− CD294+ CD127+ | | + |
Factors [ng/ml] | | |
TARC | + | − |
HGF | + | − |
TGFß1 | − | + |
Periostin | − | + |
Labelling of human leukocytes was performed as described in Table
1. All anti human antibodies were purchased from Biolegend (San Diego, USA) and used according to manufacturer’s instructions. Samples were measured using a BD FACS Canto II™ and analysed with FlowJo 10.1-Software (FlowJo LLC, Oregon, USA).
ELISA analysis
Human serum TARC, HGF, TGFß1, periostin, levels were measured via Enzyme linked Immunosorbent Assay (ELISA) (Biotechne, Minneapolis, USA) according to the manufacturer’s instructions. Sample was measured in duplicates.
Statistical analysis
Statistical analysis was performed with R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria (3.2.2). URL
https://www.R-project.org/. Variables were represented with mean, standard deviation, median, and IQR values. A two-sided t.test and a significance level = 0.05 was used to compare two groups and for more than two groups ANOVA followed by TukeyHSD was conducted. Correlation analysis was performed using spearman rank correlation analysis. ROC curves were performed using R. Heatmap was performed using R (default).
Discussion
Most studies to characterize inflammatory processes in UC are based on defining roles of single cell types or interleukins, chemokines or growth factors in the human disease and to validate their roles in an animal model. This approach has tremendously improved the understanding of inflammatory mechanism underlying the disease and to develop therapeutics targeting previously identified as crucial components. However, this approach does not take into consideration that inflammatory processes are rarely one dimensional and static but are better described as ‘mobilées’ in which there is crosstalk between static cells such as epithelial-, endothelial-muscle cells, fibrocytes and mobile inflammatory cells which can shift balances affecting the entire system in a time dependent manner. Therefore, we took a different approach.
First we analyzed frequencies of inflammatory cells in the blood and serum levels of certain factors of UC patients and compared them to Non UC patients. The selection of the subtypes of cells and the factors had a bias, as it was based on the hypothesis that part of the inflammation in UC resembled an uncontrolled wound healing process and therefore might be strongly influenced by macrophages and monocytes. Based on this immune-cell- and ELISA panel several significant differences in frequencies or concentrations between Non-UC donors and UC patients could be identified. These cells included cell of the adaptive immunity as plasma cells, cells of the innate immunity acting as mediators for the adaptive immunity as dendritic cells, CD11b+ macrophages and CD14+ monocytes and subtypes thereof as CD11b+ CD1a+ , CD14+ TSLPR+ , CD14+ CD1a+ , CD14+ CD163+, and cells of the innate immunity as NK T-, NK-cells and ILC2. In addition, HGF and TGFß1 which also have been previously described as elevated displayed a significant increase in serum levels [
13,
43].
With the exception of NK T-cells age did not affect these cell types. Age and duration of disease resulted in decreased frequencies of naïve CD8+ and increase of frequencies of central memory CD8+ T-cells. This observation is in agreement with previous results, which also described a loss of naïve T-cells and an accumulation of experienced T- cells with age [
40]. Thus, age might diminish the effect of the disease on these cell types. The observed increase of NK T-cells, however, might not be ascribed solely to the disease phenotype but also to age.
Gender did not affect expression levels of the analyzed immune cells. Since few of the patients were therapeutically naïve, one could not rule out that results were masked by a therapeutic effect on these cell types and factors. When UC patients were subdivided according to the respective therapy, the analysis revealed that TNFα-blockers (infliximab or adalizumab) had a profound effect on plasma cells, dendritic cells, TSLPR expressing CD11b+ macrophages and TSLPR expressing CD14+ monocytes. In this patient group, TARC and HGF serum levels were reduced, whereas TGFß1 levels were elevated. It is noteworthy that these patients—although most of them were in remission—did not regain the profile of a non UC subject but seem to represent a specific inflammatory condition. Opposing effects were observed in the glucocorticoid treated group. Cells of the adaptive immunity were not affected and in contrast to the group treated with TNFα-blockers frequencies of TSLPR expressing CD11b+ macrophages were elevated. Likewise, TARC and HGF serum levels were elevated as opposed to the group treated with TNFα-blockers. Mesalazine and immuno-suppressive drugs induced minor effects in this analysis. The group treated with immuno-suppressors was similar to the glucocorticoid treated group. As some of these patients were unresponsive to treatment, the observed effect might be due to the inflammatory condition and not the response to treatment. The cellular subtypes unaffected by all treatments and significantly different in UC patients and Non UC donors were CD1a expressing CD11b macrophages and monocytes.
CD1a expressing CD11b+ macrophages and CD14+ monocytes emerged as cell types significantly associated with UC. Both cell types differentiated between UC and Non UC donors with high discriminating potential as shown by an AUC value of 0.86 and 0.75, respectively. To our knowledge, CD1a expressing monocytes and macrophages have neither been associated with the gut nor with UC. The opposing effects of the treatment with glucocorticoids and TNFα-blockers suggested that these treatments might sustain different inflammatory conditions. This idea was supported by cluster- and correlation analysis. Unlike the previous analysis, cluster and correlation analysis can give information about individual inflammatory signatures and the inter dependencies of cells and factors. Both analyses suggested that two distinct inflammatory conditions prevail in UC patients: Treatment with TNFα-blockers induced a condition signified by CD14+ TSLPR+ monocytes which correlated positively with unswitched CD19+ B-cells and TGFß1 and periostin, indicating that this condition represents an inflammatory condition characterized by remodeling of the colon architecture (Table
4). The fact that most patients in this group responded to treatment and were considered as in remission further supported the idea. Of note, this condition does not reflect the homeostatic condition found in Non UC subjects, but more of an uncontrolled remodeling condition. These data might be consistent with studies in patients with Crohn’s disease which describe increased stenosis in response to treatment with infliximab. NK T-cells are considered the main source of IL-13 in UC [
44] and our data suggest, that IL-13 might be the cytokine playing a key role in the remodeling condition.
The second inflammatory condition could be more characteristic of an acute inflammation signified by TSLPR+ CD11b+ macrophages that correlated positively with the SCCAI-Score, cells of the adaptive immunity, HGF and TARC. It is noteworthy that TSLPR+ CD11b+ and TSLPR+ CD14+ monocytes correlated negatively with each other, suggesting that treatment with TNFα-blockers clearly suppresses the acute condition while promoting the remodeling condition (Table
4). Our model might explain why anrukinzumab, an anti-IL-13 monoclonal antibody was found to have no effect on clinical activity score mucosal healing, rectal bleeding or clinical remission rates [
45]. It might be that IL-13 may exert its activity in the remodeling condition by inducing tissue fibrosis [
46] and not during the acute phase of the disease.
Furthermore, in Crohn’s disease it has been described that patients progress from inflammation to stenosis over a long period of time [
47], suggesting that these two inflammatory conditions might also be found in Crohn’s patients. Finally, as an increased obstruction was observed in patients treated with infliximab, this drug is contraindicated in patients with stenosis. Our model might give also an explanation to these findings.
In this snapshot study, the dynamics of the disease had not been taken into consideration. Thus, we cannot conclude from our data that the inflammatory conditions relate to different stages such as relapse or remission. Longitudinal studies and studies with therapeutic naïve patients in remission might provide more insight into the dynamics of the disease.
In the intestine the inflammatory milieu in homeostasis and inflammation mainly depends on hematopoietic stem cell derived macrophages [
48]. In the gut, macrophages are replenished from the blood and infiltrate the mucosa in case of inflammation. According to a new concept macrophages are considered accessory cell types which support their client cells—notably mucosal epithelial cells in the gut- with various functions. One major function is the clearance of apoptotic cells; however, in addition to this housekeeping function macrophages act as important sensors and can acquire functions on demand adapted to the inflammatory milieu ranging from pro-inflammatory to healing or regulatory [
49]. This postmodern behavior challenges our view of the one-cell one-function approach in which the function of a cell can be identified by certain surface markers. Instead, we are confronted with a highly volatile cell population which shapes and is shaped by the inflammatory milieu. Thus, in the inflammatory condition of the gut epithelial cells might relay signals to macrophages and monocytes and depending of the presence or frequency of the TSLPR expressing macrophages or monocytes the outcome is tipped towards ‘acute’ or ‘remodeling’. Analysis of the second subtype of macrophages and monocytes, namely CD1a expressing CD11b+ macrophages, and CD14+ monocytes did not reveal a clear-cut functional segregation although CD1a+ macrophages correlate mainly with cell types and factors of the remodeling condition. Further studies have to be performed to analyze whether both cell types evoke different or similar responses.
The identification of potentially significant markers in the blood of UC patients is based on the assumption that the influx of inflammatory cells into the colonic mucosa is not a unidirectional route. Due to disrupted endothelial layers or an active transport mechanism, cell trafficking occurs to and from both compartments. If this assumption is correct one has to find the cells identified as relevant in the blood also in the colon of UC patient. Analysis of leucocytes from colons of UC identified monocytes, and CD1a+ CD11b+ macrophages and NK T-cells as significantly different in comparison to normal tissue derived from cancer patients undergoing colectomy. CD14+ CD1a+ monocytes just failed to reach significance. All three had been identified as significantly elevated markers in the blood of UC patients. Whether this observation reflects increased emigration to the colon or increased adherence to the colon has to be elucidated.
Authors’ contributions
MF flow cytometric analysis of PBMC; PP flow cytometric analysis of human colon; UM statistical analysis; FB recruitment of patients, patient history; MS conception, analysis; RG conception, analysis of data, writing of the manuscript. All authors read and approved the final manuscript.