Hyperactivated B cells in human inflammatory bowel disease

Patients

BU IRB (Boston, MA, USA) approved this study. Upon informed, written consent, blood was drawn into heparinized vacutainers from patients with CD (n= 100), UC (n=30), T2DM (n=15), or volunteers with no clinical symptoms of chronic inflammatory disease (healthy; n=50). Nonspecific inflammatory markers, CRP and ESR, were assayed as part of routine IBD disease activity assessment by the outpatient clinical laboratory at Boston Medical Center (Boston, MA, USA). Patient data are summarized in Table 1.

We used two clinical indices to evaluate disease severity in CD, the IBDQ and CDAI [18, 19]. For UC patients, we used the IBDQ, which consists of 32 questions that are primarily subjective in nature and quantifies the frequency of bowel function, the patient’s feelings about bowel function and frequency, the disease’s effect on their social function, and the psychological effect of the disease upon their lives. IBDQ values correlate strongly with disease severity (r=–0.5; P<0.001) with a test-retest reliability of 0.7 [20] and is a common tool used in clinical trials. A higher IBDQ value indicates a better quality of life and an implicit reduced disease activity.

The CDAI uses subjective and objective data, such as bowel frequency, general well-being, and degree of abdominal pain, and clinical data, such as hematocrit, body weight, and presence of extraintestinal manifestations of disease. Although we evaluate CDAI as a linear continuum, in general, CDAI values below 150 represent patients in remission or less-active disease, values of 150–220 indicate active mild-moderate disease, and values over 221 reflect severe disease.

The IBDQ can at times be discordant with the physician’s determination of disease severity, as the psychological effect that the disease has on the patient is variable depending on their coping mechanism and support systems. The same can be true of the CDAI, although given the more objective portions of this clinical indicator, the discordance tends to be less variable. Nevertheless, there is a strong correlation between the two disease measures (see Fig. 1A).

Whole blood analysis

Heparinized whole blood (100 μl/tube) was incubated with fluorescently labeled antibodies purchased from BD PharMingen (San Diego, CA, USA) or eBioscience (San Diego, CA, USA) at 4°C for 30 min. RBCs were lysed with 1× FACS lysing buffer (BD PharMingen). Assessment of surface expression on B cells was performed with gates generated with anti-CD19 + the appropriate isotype controls for each sample. Intracellular IL-8 was labeled with Fast-Immune® (BD PharMingen). “Phospho-flow” was performed with cell lysis buffer and antibodies purchased from BD PharMingen [21]. All antibodies were pre-titrated to determine optimal concentrations. Flow cytometric data were acquired with equipment maintained by the BU Medical Campus Flow Cytometry Core Facility.

Human tissues

Fresh ileal, colonic, and appendix tissues from patients with CD or UC were obtained following surgical resection in accordance with the BU IRB or were purchased from the NDRI (Philadelphia, PA, USA). The initial pathologic designation of resected tissues as inflamed or noninflamed (healthy) was based on the gross appearance, as determined by the attending gastrointestinal pathologist at the time of surgery. The diagnosis was subsequently confirmed following examination of histological sections obtained from these tissues. Gross colonic inflammation in patients with UC was characterized by mucosal erosions and ulcerations, erythema, hyperemia, and granularity. The mucosal surface of the inflamed small intestine from the patients with CD showed an irregular nodular appearance with hyperemia and focal superficial ulcerations. Fresh, surgically discarded tonsils from tonsillitis patients and appendices from appendicitis patients were purchased from the NDRI.

Cell isolation and culture

PBMCs were isolated from whole blood by Ficoll-density gradient from patients who were able to give 50 ml blood. B cells were purified from PBMCs by a magnetic bead negative isolation kit (Invitrogen, Carlsbad, CA, USA, or Miltenyi Biotec, Auburn, CA, USA) [7]. For some experiments, monocytes were purified from Source Leukocytes (NY Biologics, New York, NY, USA) by magnetic bead separation (Invitrogen). MCs were purified from tissues as described [7]. Briefly, whole tissues were snipped into small pieces with scissors and then subjected to gentle homogenization using a closed tissue-grinder system (Fisher Scientific, Pittsburg, PA, USA). Single-cell suspensions were obtained by running the homogenate over a 70-μm BD Falcon cell strainer. MCs were purified from the single-cell suspensions by Ficoll-density gradient as above and labeled immediately with fluorescently labeled antibodies as above.

Blood B cells were plated at 1 × 10⁶ cells/ml RPMI supplemented with 10% FCS, 100 U/ml penicillin, 100 U/ml streptomycin, and 2 mM L-glutamine (Fisher Scientific). B cells were stimulated for 18 h–6 days with Pam3CSK4 (1 μg/ml). Cell-free supernatants were collected and stored at −20°C until assayed. Monocytes were stimulated with 100 ng/ml Escherichia coli LPS prior to biochemical analyses.

Measurement of cytokine protein levels

Cell-free supernatants from B cell cultures or serum samples were quantitatively assayed for cytokines by cytometric bead assay human inflammatory mediators or standard ELISA for IL-8 (BD PharMingen).

Biochemical analysis

Quantification of protein/DNA interactions by ChIP was performed as described using 500,000 purified B cell or monocytes/precipitation [22]. Transcription factor-associated DNA was quantified using TLR2-specific primers: 5′-CTTTCCCTGTGGTTGCCAAT-3′ and 5′-GAATCGAGACGCTAGAGGCG-3′. ChIP-competent antibodies from Santa Cruz Biotechnology (Santa Cruz, CA, USA) were: anti-histidine tag antibody (sc-803) as a nonspecific-matched control, anti-cJun (sc-45), and anti-PU.1 (sc-352). Changes in gene expression in mRNA extracted from freshly isolated B cells were evaluated with the JAK/STAT PCR array per the manufacturer’s instructions and reagents (SABiosciences, Frederick, MD, USA).

Statistical analysis

Pearson’s correlation coefficient or Spearman’s rank and linear regression was used to calculate linear relationships. Unpaired two-tailed t-test with Welch’s correction was used to determine differences in group means. Odds ratio was calculated using contingency tables and P values using Fishers exact test. Sample sizes vary for different tests as a result of the unavailability of certain patient samples. All statistical analyses were performed with GraphPad Prism software (GraphPad Software, San Diego, CA, USA).

B cells from CD patients constitutively secrete IL-8

Previous studies have shown that B cells produce cytokines in murine models of colitis [23]. We therefore hypothesized that B cells would also represent a significant source of cytokines in human IBD [24]. To test this possibility, we purified peripheral blood CD19+ B cells (>99% pure, Fig. 1B) from IBD patients and cultured cells to assay spontaneous cytokine production. Surprisingly, B cells from CD patients with active disease (IBDQ <168) constitutively secreted IL-8 (Fig. 1C) but not IL-1β, IL-2, IL-6, IL-10, IL-12, or TNF-α (not shown). The amount of IL-8 produced by B cells rivaled that secreted by monocytes, up to 10,000 pg/ml/million cells [25].

B cells from healthy donors and CD patients with less active disease (IBDQ>168) constitutively secreted relatively low levels of IL-8 (Fig. 1C). Furthermore, spontaneous IL-8 production by B cells was positively correlated with a clinical measure of CD, CDAI (Fig. 1D), which is highest in active disease, and inversely with IBDQ, which is lowest in active IBD (Fig. 1E). Remarkably, these results demonstrate that B cell IL-8 production is correlated directly with the quality of life of a CD patient. The clinical significance of these findings raised the possibility that spontaneous IL-8-producing B cells in the blood might elevate systemic IL-8 levels, thereby intensifying patient symptoms [25, 26]. Increased levels of serum IL-8 in CD patients compared with healthy serum support this likelihood (Fig. 1F).

Interestingly, B cells from UC patients failed to secrete significant IL-8 spontaneously, regardless of disease activity (Fig. 1C), and this was reflected in the lower levels of serum IL-8 (Fig. 1F). However, although not a quantitative measure, the identification of intracellular IL-8-positive B cells in the blood of UC patients suggested that B cells from these patients would readily secrete IL-8 upon stimulation (Fig. 1G). For comparison, we assessed B cells from T2DM patients, another chronic disease with inflammatory parallels to IBD [27], and found that unstimulated B cells did not actively produce IL-8 (Fig. 1C).

As GC B cells have been shown to secrete abundant IL-8 [28], we asked whether peripheral blood B cells in CD patients might have phenotypic markers of GC B cells, perhaps as a result of aberrant cellular trafficking. However, circulating B cells expressed high levels of CD23, moderate CD38, and low CD77, consistent with a non-GC phenotype (not shown) [29].

Circulating B cells from IBD patients demonstrate increased levels of basal activation

Spontaneous IL-8 production by circulating B cells from CD patients suggested that B cells had encountered ligand or other stimuli in vivo to induce IL-8 production. We tested whether peripheral blood B cells from patients had signatures of ligand- or cytokine-induced activation. Freshly isolated B cells from CD patients or healthy donors were first analyzed for expression of genes generally associated with cellular activation using PCR arrays [30]. We found that B cells from CD patients expressed higher levels of transcript for NF-κB1 (p50) and FAS compared with B cells from healthy volunteers, illustrating an amplified level of general activation in patient B cells [29, 31]. Additionally, elevated expression of the IL-10R and reduced expression of the IL-4R compared with B cells from healthy volunteers suggest a differential responsiveness of CD patient B cells to IL-10 and IL-4 (Fig. 2A).

Changes in gene expression are triggered by activation of kinase signaling pathways. Thus, untouched B cells in fresh whole blood were assessed for the presence of phospho-kinase+ molecules involved in ligand/receptor-induced signal transduction, including Btk (T551), Syk (Y348), p38 (T180), and ERK1/2 (T202/Y204) [21, 32, 33]. B cells from healthy blood exhibited low levels of phospho-Btk, -p38, and -Syk (Fig. 2, C and D). However, the majority of circulating B cells in healthy and IBD blood samples was positive for phosphorylated ERK1/2 (Fig. 2C). B cells from CD and UC patients contained phospho-p38 and -Syk (Fig. 2, B and C), but only B cells from CD patients had elevated levels of phosphorylated Btk (Fig. 2, B and C). Although disease activity was not associated with the level of phospho-Syk and -p38 in circulating B cells, patients presenting with clinical disease had fewer phospho-Btk-positive B cells (Fig. 2D). Overall, these data are consistent with the interpretation that B cells from IBD patients have increased basal levels of activation, as measured by multiple parameters: constitutive IL-8 production, gene expression, and activation of kinases, and that the degree of CD activity affects B cell phenotype and function.

B cells from CD and UC patients express elevated TLR2

A potent inducer of IL-8 in multiple cell types is the TLR family, particularly TLR2, which has been implicated in IBD pathogenesis and involves the signaling molecules demonstrated to be activated above, such as Btk [34,35,36,37]. Additionally, we have reported previously that IL-4 reduces TLR2 expression and responsiveness to TLR2 ligands by B cells, and IL-10 induces an increase in these measures [7]. These data imply that the changes noted in cytokine receptor expression in Figure 2A might augment TLR2 expression in B cells from IBD patients.

Indeed, peripheral blood from CD and UC patients had a significantly increased percentage of TLR2+ B cells compared with healthy subjects (Fig. 3A, middle and right panels, and B). All circulating B cell subsets expressed TLR2, including CD27+ memory B cells (not shown). For comparison, T2DM patients had primarily TLR2 low B cells, similar to healthy donors (Fig. 3B).

To demonstrate TLR2 gene expression independently, we quantified molecular signatures of an active TLR2 promoter using B cells from CD patients. We focused our analysis on two factors with binding sites in the monocyte TLR2 promoter: PU.1 and AP-1 [39]. ChIP showed that PU.1 constitutively associated with the TLR2 promoter in B cells from CD but not healthy donors (Fig. 3C). This finding recapitulates the pattern of PU.1/TLR2 promoter association in resting and stimulated monocytes (Fig. 3D), further supporting our hypothesis that circulating B cells in CD patients have encountered stimuli. By contrast, the cJun component of AP-1 associated with the TLR2 promoter in monocytes as well as in B cells from CD patients and healthy donors, suggests that this interaction does not play a critical role in promoter activity (Fig. 3, C and D) [40]. Thus, B cells from CD patients increase TLR2 expression as measured by two independent criteria: promoter activation and surface expression.

TLR2 expression by B cells is related to disease activity

As the level of spontaneous IL-8 production by B cells correlated with a poor quality of life in CD, we assessed the relationship between expression of TLR2 and clinical measures of disease activity. Figure 3E shows that the percentage of TLR2+ B cells had a positive linear relationship with the CDAI. Similarly, percentages of TLR2+ B cells correlated inversely with IBDQ values for CD (Fig. 3F), indicating that the percentage of TLR2+ circulating B cells is a predictor of disease severity in CD. By contrast, the relationship between UC disease activity and TLR2 expression demonstrated that peripheral blood TLR2+ B cells are expanded when the disease is quiescent and reduced during periods of active disease reflecting a lower quality of life (Fig. 3G). These results suggest that IBD, in general, is associated with an increase in circulating TLR2+ B cells, but based on relationships with disease activity measures, TLR2+ B cells likely have an important, although undoubtedly different, role in UC and CD pathogenesis. Defining the clinical significance of these differences may help to better understand the etiology of CD and UC.

TLR2 stimulation induces B cell IL-8 secretion from IBD patients

Constitutive IL-8 secretion by circulating B cells in flaring CD patients is likely induced by microbial-derived TLR2 ligands in vivo. We tested whether purified B cells were responsive to the TLR2 ligand Pam3CSK4 by measuring secreted IL-8 levels. B cells from UC and CD patients inducibly secreted IL-8 in response to TLR2 ligand (Fig. 4A). In contrast, B cells from healthy donors secreted relatively little IL-8 in response to Pam3CSK4 (Fig. 4A) [7]. More detailed analysis of IL-8 secretion by B cells from CD patients demonstrated that B cells from patients with active disease produced higher amounts of IL-8 in response to Pam3CSK4 (Fig. 4A). B cells from patients with milder disease responded to Pam3CSK4 by secreting higher levels of IL-8 compared with healthy donors (Fig. 4A). The disease state in UC was not associated with the potential to secrete inducible IL-8, although the samples size was too small in this study to make a definitive conclusion. Interestingly, despite elevated TLR2 expression [7], tonsil B cells produced relatively little IL-8 in response to TLR2 stimulation (Fig. 4A), suggesting that B cells from IBD patients have a distinct propensity to secrete copious IL-8.

Kinetic studies revealed that IL-8 production by B cells from CD patients is relatively long-lived, as evidenced by the profound accumulation of IL-8 in cells cultured for up to 6 days (up to 80 ng/ml IL-8/million B cells; Fig. 4B). Thus, populations of B cells in CD patients demonstrate the ability to chronically secrete IL-8, a generally tightly regulated and inducible inflammatory chemokine [41]. These observations might explain how B cells, which are circulating in the blood, secrete IL-8 spontaneously in CD patients.

TLR2+/IL-8+ B cells are increased in inflamed gastrointestinal tissues

Systemic inflammation characterizes both major forms of IBD, but the major site of inflammation is in the gastrointestinal mucosa. Although B cells are a small subset of cells within blood, they are in substantial numbers in the lymphoid tissues [9]. Thus, B cells represent a previously unappreciated source of inflammatory mediators in both the blood and tissues in CD. To determine whether the elevated percentages of TLR2+ B cells in circulation reflected mucosal B cell phenotypes, we assayed levels of TLR2 on CD19+ B cells in fresh diseased ileum and/or colon and adjacent healthy tissue from CD and UC patients, respectively. Figure 5, A, left panel, and B, demonstrates that adjacent healthy tissue contains few TLR2+ B cells. In contrast, inflamed tissue from UC and CD contains an abundance of B cells, which expressed TLR2 (Fig. 5, A, middle panel, CD patient, and B). The ileum tissue may also contain CD19low TLR2+ B cells (Fig. 5A, middle panel, arrow). These cells were phenotypically GC B cells, as evidenced by expression of CD77 (Fig. 5A, follow arrow to right panel) and likely arise in the Peyer’s patches abundant in the ileum [29, 42]. It should be noted that although we described circulating CD19low TLR2+ B cells previously in healthy volunteers [7], we did not note an increase in these cells in IBD patient blood.

We assessed TLR2 expression in acutely inflamed appendices as a basis for comparison. Appendicitis tissues (without IBD) also contained increased levels of TLR2+ B cells, suggesting that TLR2 expression may arise during an acute inflammatory response and perhaps become chronic in IBD inflammation (Fig. 5B).

We next explored whether B cells from inflamed gastrointestinal tissue would have the potential to secrete IL-8. As the yield of purified B cells from small tissue snips is low, we evaluated freshly isolated CD19+ B cells for the presence of intracellular IL-8 by flow cytometry. Figure 5C demonstrates that although healthy adjacent tissue contains few intracellular IL-8+ B cells (left panel), inflamed colon had increased levels (right panel; tissue from UC patient). In addition, B cells from appendicitis tissue, but not B cells from tonsillitis (not shown), were intracellular IL-8+ (Fig. 5D), suggesting that gastrointestinal inflammation promotes B cell IL-8 production.

The clinical significance of TLR2 expression by circulating B cells

To better understand the clinical significance of B cell phenotypes in IBD, we evaluated the relationship between percentages of TLR2+ B cells and clinical measures of general inflammation. We found that the percentage of TLR2+ B cells was not correlated with the level of CRP or ESR (Fig. 6, A and B) [16]. Notably, TLR2+ B cell percentages were not related to medication use in UC or CD or disease location in CD (not shown).

The increased basal level of B cell activation, such as spontaneous IL-8 production by circulating B cells from CD patients, suggests that B cells are encountering antigen actively. In contrast to UC, CD is generally characterized by transmural inflammation, where the entire wall of the gut is affected by inflammation [43], thereby predisposing the patient to bacterial translocation into the blood [44]. Transmural inflammation may also predispose a patient to fistulizing disease, which is characterized by the development of abnormal passageways or connections between the bowel and other structures, such as the skin, vagina, bladder, or other parts of the bowel. We therefore sought to determine whether TLR2 expression or spontaneous IL-8 secretion by B cells was associated with fistulizing disease. Approximately 33% of CD patients had developed fistulas (Table 1), but there was no association between fistulizing disease and TLR2 or IL-8 expression by B cells (P=0.77). These findings support the interpretation that TLR2 expression on B cells in peripheral blood is specifically reflective of gastrointestinal activity as a result of the reliance of IBDQ and CDAI on bowel frequency rather than general systemic inflammation, medication, disease location, or type.

Nevertheless, the data presented in Figure 3, E–G, demonstrating a relationship between TLR2+ B cells and disease activity, are cross-sectional. CD and UC are chronic, progressive conditions, whose natural histories are characterized by periods of disease-free remission followed by relapses of disease activity [2]. This prompted us to collect longitudinal data on some patients to characterize within-person variations of B cell activity during periods of disease remission and disease relapse. Figure 6C demonstrates that TLR2+ B cells indeed expand during periods of illness and retract during periods of remission in five CD patients. Additionally, Figure 6D depicts three CD patients who were evaluated at three or more time-points between 6 and 12 months. Regardless of TLR2+ B cell percentages at baseline, a decrease in TLR2+ B cells was reflected by an increase in the quality of life of the patient. These data reinforce our finding that levels of TLR2+ B cells are positively correlated with disease severity in CD patients.