Background
There is currently a significant increase in allergic diseases around the world, especially in the developed countries [
1]. These diseases include atopic dermatitis, asthma, allergic rhinitis and food allergy [
1]. Arising more often in the childhood, they tend to progress over time manifesting in more severe forms in the adulthood, the phenomenon known as the atopic march [
2]. The mechanism of allergy is most often associated with the production of IgE to harmless antigens [
1]. The only method for its etiotropic, rather than symptomatic, therapy is allergen-specific immunotherapy (ASIT). This method is associated with the stimulation of the production of blocking allergen-specific IgG4 antibodies [
3].
In order to develop new methods of ASIT dedicated to stimulate allergen-induced B-cell switching to IgG4 but not IgE isotype, it is clearly necessary to understand specific mechanisms triggering selective B-cell immunoglobulin class switching to these isotypes. For the development of the general strategy to eliminate IgE-producing B-cells and their precursors, it is important to know the exact site where B-cell IgE isotype switching occurs. We have previously shown that in young (1–8 years old) patients with allergy to house dust mite specific IgE production is not associated with specific IgG, IgG4 or IgA1 production [
4] and may occur outside the germinal centers of secondary lymphoid organs (SLOs). Possibly it can occur in the tertiary lymphoid structures (TLSs) or so-called tissue-associated lymphoid clusters. Because of the close proximity to epithelial or endothelial barrier tissues, immune processes in TLSs may be somewhat different from those in SLOs. For example, type 2 innate lymphoid cells (ILC2) activated by tissue damage-induced cytokines interleukins (IL) 25, 33 or thymic stromal lymphopoetin (TSLP) [
5], instead of T helper-2 cells, may act as the main producers of pro-allergic cytokines like IL-5, 13 [
5] and, to a lesser extent, IL-4 [
6], in TLSs during the early phases of immune response.
Lifestyle can significantly affect the severity of atopic diseases. One of the characteristics of the developed country population is the high prevalence of obesity [
7]. Epidemiological data indicate a direct relationship between the degree of obesity and the severity of asthma. The molecular and cellular mechanisms of this association can be diverse. For example, in obese individuals, the production of pro-inflammatory IL-6 in tissues may be increased. Adipokine effects are also suggested. Very often, although not always, the level of leptin in patients with obesity is increased, and the level of adiponectin is reduced [
7,
8]. It was shown that the administration of leptin to BALB/c mice increased the intensity of allergic inflammation and production of IgE while the administration of adiponectin inhibited it [
8‐
10]. It is especially important to note that adipose tissue contains TLSs, although their number varies depending on the anatomical location [
11]. The role of TLSs in the humoral immune response to allergens has not been studied.
Despite a large number of successively developed mouse allergy models, most of them require high allergen doses [
12‐
17] sometimes with adjuvants like Alum [
12‐
15] or CFA [
16], resulting not only in high IgE levels but also high IgG1 [
12‐
15] and sometimes IgG2a [
17] production. So, in these models, humoral immune response differs from that in allergic patients. Among allergic individuals some of them have significant IgG4 production [
4,
18,
19] which is not, however, associated with specific IgE levels [
4,
18,
19]. Elevated IgG4 levels detected in atopic patients can represent secondary reaction in response to multiple antigens entering the mucosal surfaces. The presence of already existing specific IgE makes it possible to induce IgG4 production due to the immune recognition of IgE-antigen complexes [
20‐
22].. Indeed, recent data reviewed in [
23,
24] show that high antigen doses induce robust germinal center response in SLOs which could, in certain circumstances, block the development of IgE producing B-cells. It is not surprising that B-cell IgE class switch is usually detected in the SLO-like lymph nodes and specifically within germinal centers but this process is transient, since it occurs mainly at early stages of its development and not in mature structures [
25,
26]. Moreover, numerous studies [
27‐
29] provide the evidence that in allergic patients B-cell IgE class switching usually occurs in TLSs such as nasal polyps or inducible bronchial-associated lymphoid clusters [
28,
30].
The aim of this work was to clarify the hypothesis that B-cell IgE class switching in response to low, but not high, administrated antigen doses occurs mostly in the specific milieu of TLSs but not in SLOs. In clinical practice allergens usually pass through the respiratory barrier tissues where inducible bronchial-associated lymphoid tissue or nasal polyps represent TLSs [
28,
30]. In humans these clusters are induced during early infancy, especially among children due to their higher susceptibility to viral and bacterial infection, or to the exposure to certain substance such as diesel-derived particles [
31]. These structures can be generated in young mice by the administration of specific substances such as lipopolysaccharide [
32]. To avoid the difficulties of small antigen doses delivery to mouse lungs as well as to induce respiratory tract linked TLSs formation in adult animals, we chose to use a principally novel model in which antigen was administrated s.c. in the withers (s.c.w.). It is well known that the withers of mice contains significant volume of white adipose tissue and many so called fat-associated lymphoid clusters (FALCs) [
33] which resemble those associated with the respiratory epithelium. Importantly, tissue damage-induced cytokines, such as IL-33 and TSLP, could be produced not only in barrier epithelium cells [
5] but also in adipose tissue, particularly, in adipocytes and adipose-linked endothelium [
34,
35]. Of note, at first ILC2 were discovered in the association with FALCs [
36], though later ILC2 were identified in SLOs as well [
5]. In adipose tissues, ILC2 not only promote allergic type 2 inflammation but also function as metabolic and homeostatic regulators [
37]. In addition, the introduction of antigen into the subcutaneous adipose tissue allows one to check the role of the fat associated TLSs in the synthesis of pro-allergic antibodies. The results can be relevant to understanding the mechanisms linking the obesity and atopic march.
We demonstrate here that direct IgE class switch in response to subcutaneously administrated allergen occurs in local TLSs, and adipose tissue provides unique environment for B cell activation.
Discussion
Previously we have shown that in young human individuals allergic to house mite dust allergen specific IgE production is not associated with the production of other Ig classes [
4]. Consequently they can be triggered in the sites other than SLOs germinal centers, for example in TLSs. In the present study, we demonstrated that model antigen OVA at low dose induced specific IgE production only when administrated in TLSs-enriched region, the withers in particular. Fat-associated lymphoid clusters (FALCs) are relatively constant structures [
48] in comparison with the inducible bronchial-associated lymphoid tissues (iBALTs) [
32,
49] and, so, immunization in FALCs may represent an adequate model of chronic allergen instillation into respiratory tract of patients containing iBALTs. In such patients, iBALTs can be formed due to chronic bacterial or virus infections or even due to inhalation of different particulate matter, though it was shown only in mouse models [
32,
49]. As a matter of fact, FALCs are not the same structures as iBALTs and supporting tissues are completely different. While not denying it, we assume that the basic functional principles for these two types of TLSs are generally the same. Due to the presence of ILC2s in FALCs as well as T helper 2 and M2 macrophages [
48] the basic mechanisms of IgE production triggering could be similar both in FALCs and iBALTs. It should be emphasized that the main innate immunity linked pro-allergic cytokines such as IL-33 and TSLP are expressed not only in epithelial barriers but in adipose tissue as well [
34,
35].
As it was shown low dose of chronically administrated antigen induced more pronounced IgE response not linked with specific IgG1 or IgG2a response. Our allergic model is considered as clinically relevant representative reconstruction of allergy progression in humans. It is also interesting that specific IgE and IgG1 levels in this model were relatively weakly interconnected if any relation even occurred. Despite this fact, low antigen doses trigger IgE and IgG1 class switching in the same anatomical location – white adipose tissue. Withers adipose tissue rather than regional lymph nodes was the only site of B-cell activation upon low dose antigen administration. It was probably due to the lack of antigen delivering to SLOs in case of low doses. In the absence of pattern recognition receptor (PRR) stimuli, dendritic cells are either not activated or activated in a tolerogenic but not immunogenic way, and their maturation and following migration from tissues to SLOs are rare [
50]. Therefore, low antigen doses accumulate only in tissues but not in SLOs.
Low antigen doses are very weak immunological stimulus which does not engage PRRs and very weak triggers of FALCs immune cell activation. So, cells activated by low antigen doses in these structures would be stimulated occasionally in the presence of very low amounts of proliferation supporting stimulus (like IL-2, IL-4, TNF-α etc.) and cell attracting chemokines. The latter may intensify activation-induced B cell migration from tissues to SLOs. This hypothesis was confirmed by reduced expression of
cd19 gene despite induction of genes corresponding to B-cell activation and Ig class switching. Because
cd19 gene expression should correspond to the B-cell numbers [
51,
52] and the decrease in this number could not be due to toxicity of administrated antigen or due to the cell differentiation into plasma cells, the only remaining explanation is migration of B-cells from withers tissue. Besides Ig class switching could not occur without germline transcripts induction [
47] and so lymph node cells that express postswitch transcripts should arise in the place where germline transcripts expression was detected. The withers adipose tissue seems to be the source of such cells in our allergy model.
Similarly, the data on postswitch ε and γ1 transcripts expression indicates that despite the local tissue-restricted B-cell activation IgE production occurs mostly in SLOs, i.e. regional lymph nodes. It is low rather than high antigen doses induced high levels of specific IgE production, according to ELISA, which enhanced the expression of postswitch ε transcripts in the lymph nodes. Also the expression of postswitch ε transcripts in the lymph nodes normalized to Gapdh was relatively weak which probably corresponded to larger numbers of mature antibody producing cells. This reflects the presence and availability of a higher number of B-cell niches in the lymph nodes than in the withers. In mice, upon high dose antigen administration, the migration of IgE-switched B-cells from TLSs to SLOs does not occur or occurs relatively weakly, therefore, the IgE switched B-cells, that not completely differentiate to plasma cells, accumulate in TLSs. These cells are probably responsible for low level IgE production observed in high dose group.
Supposing that (i) B-cell proliferation in TLSs, namely FALCs, especially in response to weak immunological stimulus, is hampered due to low numbers of B-cell supporting niches and (ii) it is FALCs B-cells that are switched to IgE production in a response to low antigen doses, it is logical to speculate that hampered proliferation in FALCs is responsible for shifting the balance towards IgE production. Indeed, in this work, we showed that anti-proliferative drugs, especially hydrophobic (and apparently locally acting) etoposide, enhanced specific IgE production. But the effects of the same drugs on IgG1 production were opposite. These data are in agreement with the recent observations. For example, in reporter transgenic mice expressing IgE fused with fluorescent protein, the isotype switching occurs mostly in early stages of anti-helminthic immune response when B-cells proliferation in germinal centers is not so rapid as at the later stages [
25,
26]. IgE
+ B-cells have diminished expression of anti-apoptotic proliferation supporting transcription factor Bcl6 [
25]. In the other hand, recently switched IgE
+ B-cells have low levels of costimulatory molecules expression and were more apoptosis prone than IgG1
+ B-cells [
26]. It should be noted that these drugs might have other effects on living cells, for example, on the induction of double stranded breaks that are powerful inducers of class switch recombination [
53]. However in this case, it would be logical to expect the induction of both IgE and IgG1 class switching. In addition, these drugs may induce genotoxic stress signaling pathways activation and pro-inflammatory cytokine production [
54]. The same genotoxic stress double strand breaks activated pathways also responsible for the inhibition of cell proliferation. Although this issue has remained unexplored in this work it can be assumed that activation of signaling pathways associated with genotoxic stress per se may also be responsible for the shifting the balance towards IgE production. It is interesting that in some groups throughout all the experiments, and especially in the experiments with anti-proliferative drugs, we have observed significant data outliers in humoral immune response. We suggest that it may be due to uneven distribution of FALCs in withers adipose tissue.
In conclusion, in our work we introduced a new low-dose reproducible mouse allergy model which can be an alternative to standard allergy models with high antigen dose usage. In this new model all mice produceed significant specific IgE and IgE isotype switching was triggered in the damaged tissue but not in the lymph nodes. So this new model may be more clinically relevant that commonly used high antigen dose and adjuvant based mouse allergy models. We hope that this model may be interesting and useful for further investigations of IgE production triggering. This model may be useful in studying the mechanisms responsible for the more severe course of atopic diseases in overweight individuals in which the excess of adipose tissue may correspond to the excess of numbers of FALCs in which B-cell proliferation relative to SLOs is hampered and IgE switching occurs more effective. Although the authors do not deny that the other mechanisms may also be responsible for IgE+ cells generation in such structures, for example chronic expression of pro-inflammatory cytokines and adipokines, that may be the subject for future investigations.
Methods
Mice
Female wild type BALB/c mice (6–8 weeks old) were obtained from Andreevka Center (Stolbovaya, Russian Federation). Mice were housed in SPF conditions about 2 weeks before experiments. During all experimental procedures mouse were kept in 12-h light dark cycle at room temperature in a special room in vivarium. Mice were housed in plastic cages (10–12 mice for 1 cage) with filings as a bedding material. Mice were fed ad libitum by granular feed. Mice were treated and their health status was monitored according to an approved by IBCh RAS IACUC protocol.
Immunization and blood sampling
Ovalbumin (OVA, Sigma Aldrich, Darmstadt, Germany) was used as a model antigen. Mice received OVA by intraperitoneal (i.p.) and subcutaneous injections in foot pad (f.p.) or in the withers areas (W). OVA was administrated repeatedly 3 times a week for 4–5 weeks (total 14 immunizations) in low (100 ng/injection) or high (10 μg/injection) doses in total volume of 100 μl saline. Before study the randomization procedure was performed. Two control groups were commonly used in the study, namely intact mice and saline immunized mice (n = 5) We used 2 experimental groups for each immunization protocol when studied the dependence of antibody production from the site of immunization and antigen dose (n = 5). The number of animals in groups was determined so that for U-test quantification p < 0,01 values could be obtained. Blood sampling from retro-orbital sinus was performed after 7th (17th day) or 14th (33th day) of immunization. Blood was collected, coagulated for 15 min at 37 °C and spinned (600 g, 20 min). Serum was collected and stored at − 20 °C prior to use.
In some experiments, mice (
n = 5) were immunized with low OVA dose with anti-proliferative drugs doxorubicin and etoposide (Sigma Aldrich) in doses of 2 and 20 μg/ml which corresponded to doses 0.1 and 1 mg/kg. These doses are too low in comparison with usually used for treatment of malignancies [
55,
56] and very unlikely cause systematic harmful effects. Doxorubicin was administrated in saline and etoposide in the vehicle containing 0,25% DMSO+ 10% PEG-400 because of water insolubility. Two control groups were used, i.e. saline immunized mice and mice immunized with antigen alone. Two experimental groups for each drug (with different drug dose) were used (
n = 6–7).
Elisa
Microtiter plates (MaxiSorb®, Costar, USA) were coated with 50 μl of 5 μg/ml OVA solution in PBS (pH = 7.2) and incubated overnight at + 4 °C. Following extensive washing with PBS containing 0.05% Tween-20 (PBS-T) and subsequent blocking with 5% BSA in PBS, plates were incubated with serially diluted serum samples at + 4 °C overnight. Next day plates were subsequently exposed to primary biotin-labeled anti-mouse antibodies (BioLegend, San Diego, CA, USA) to either IgEa (clone UH297) or IgG1 (clone RMG1 1) or IgG2a (clone RMG2a 62), and streptavidin-conjugated HRP. Plates were further processed with substrate solution (3,3′,5,5′-tetramethylbenzidine) Optical densities (OD) were measured with automatic 96-well plate reader (Thermo Fisher Scientific, Waltham, MA, USA) at 450 nm with subtraction of optical density at 620 nm that does not correspond to the reaction colored product. Antibody quantities were estimated as OVA-specific antibody titers defined as the serum dilution, in accordance according to 4-PL regression analysis.
For estimation of total IgE levels, plates were coated with unlabeled anti-mouse IgE (clone RME1, BioLegend). Serum samples were added in 1:100 dilution in duplicates with different dilutions of mouse IgE κ (MEA-36, BioLegend). In following steps biotin labeled anti-mouse IgE (UH297) and streptavidin-HRP were added subsequently as described above.
Systematic anaphylactic response measurement
For measurement of systematic anaphylactic response in mice after full immunization protocol, OVA in challenging dose (0.2 ml of 2.5 mg/ml solution) was administrated i.p. to animals. Body temperature was measured with infrared thermometer CEM DT-8806S, (SEM Test Instruments, Moscow, Russia) as it was performed in [
57] before OVA administration and 45 min after. The difference in the temperature between two time points was defined. The experiment was performed in laboratory.
Local anaphylactic response measurement
Standard Evans Blue based test was used to measure local anaphylaxis response. Briefly, abdominal region of each mouse was shaved and after that different doses stating from 10 μl of 1 mg/ml OVA were administrated subcutaneously in this area. After 3 min 200 μl of 0.5% Evans Blue solution were administrated in the tail vein. After 15 min reaction intensity was evaluated.
Gene expression measurement by qPCR
After 14th immunization mice were sacrificed in laboratory by cervical dislocation. Lymph nodes were harvested, homogenized and cells were spinned in PBS (600 g, 10 min), followed by the RNA extraction using phenol chloroform based extraction protocol. (ExtractRNA, Evrogen, Moscow, Russia). In the case of white adipose tissue small pieces were harvested and washed from blood in the PBS and then subjected to RNA extraction with the same reagent. cDNA was prepared from the samples with RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). Quantitative PCR was performed with SybrGreen I based kits from BioLabMix (Novosibirsk, Russia). Expression of target genes was estimated by normalizing to expression of house-keeping gene GAPDH and was calculated as 2^-d(dCt) in comparison with expression in the tissues of intact control mouse. In the case of postswitch ε transcripts for analysis of absolute IgE-producing B cell numbers in samples, normalization to GAPDH was performed. For measurement of IgE-producing B cell content in B cell fraction, normalization to CD19 was performed. Primers were designed and synthesized by EvroGen. The following primers were used:
for GAPDH F: GGTGCTGAGTATGTCGTGGA; R: TGGAAGAGTGGGAGTTGCTG;
IL-25 F: CCCAGCAAAGAGCAAGAACC; R: ATCCTCTAGCAGCACAAGCG;
IL-33 F: GTCTCCTGCCTCCCTGAGTA; R: GTGGTGCCTGCTCTTCTGAA;
TSLP F: CTGCCTGAATCAAACCTCACAA; R: TGACTGCCCGAACTGTCATT;
CD19 F: TTCACTACTGAGCCTAAGCCTTG; R: CAACAGCCAGAGCCACACT;
AICDA F: ACAGCACTGAAGCAGCCTT; R: GCCCAGCGGACATTTTTGAA;
Bcl6 F: CAGTCCCCACAGCATACAGA; R: CCTCAGAGAAACGGCAGTCA;
EBI2 F: CATAAAAGGACGCCTGCTCG; R: TTGCCAGTGGGGTAGTGAAA;
germline ε F: GCACAGGGGGCAGAAGAT; R: CCAGGTCACAGTCACAGGAT;
germline γ1 F: CACGGGAGATTGGGAAGGAG; R: TTTGGGCAGCAGATCCAGG;
Gata3 F: TGGAGGAGGAACGCTAATGG; R: GATGTGGCTCAGGGATGACA;
NMUR1 F: ATGACTCCTCCCTGCCTCA; R: GAGCACCAGCATATCGGACA;
postswitch ε F: CAGGCTGCTGCTGGGTAG; R: GCTGGTGGTGACCTTGAGTT;
postswitch γ1 F: ACATGCTCTGTGTGAACTCCC; R: AGGTCAGACTCCAGGACAGC. Ct values were determined for each sample. Reaction was performed in DT Prime Amplificator (DNA Technology, Moscow, Russia) according to the following protocol: initial denaturation at + 95 °C for 3 min and then 50 cycles with: 15 s denaturation at + 95 °C, 45 s annealing and elongation at + 60 °C.
Statistics
For comparisons between experimental groups and samples, nonparametric Mann-Whitney test was used. Levels of p < 0.05 were considered statistically significant. For correlation coefficients determination, Spearman test was used. The data from each individual animal was considered as a single point. Mean and standard deviations for each compared group were calculated.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.