Correlated dynamics of serum IGE and IGE+ clonotype count with allergen air level in seasonal allergic rhinitis

About authors

1 Skoltech, Moscow, Russia

2 Shemyakin and Ovchinnikov Institute of Bioorganic Chemistry, Moscow, Russia

3 Pirogov Russian National Research Medical University, Moscow, Russia

Correspondence should be addressed: Ivan V. Zvyagin
Miklukho-Maklaya, 16/10, Moscow, 117997; moc.liamg@nigayvzi

About paper

Funding: the study was supported by the Grants Council under the President of the Russian Federation (grant MK6000.2018.4).

Acknowledgments: we are very grateful to all the donors who participated in the study.

Author contribution: Mikelov AI — antibody levels determination, IGH cDNA libraries preparation, sequencing data and results analysis, research design, drafting; Staroverov DB — cell subpopulations isolation (flow cytofluorometry); Komech EA — blood samples collection, cell subpopulations isolation (flow cytofluorometry); Lebedev YB — results analysis and discussion, advisory support; Chudakov DM — results analysis and discussion, advisory support (cDNA libraries preparation); Zvyagin IV — IGH cDNA libraries preparation,sequencing data and results analysis, research design, drafting, research organization.

Received: 2019-10-09 Accepted: 2019-10-23 Published online: 2019-10-31

Seasonal allergic rhinoconjunctivitis (SAR), or seasonal allergic rhinitis, is one of the most common allergies. According to various studies, 17 to 28% of the population of Europe suffer from this disease [1]. A significant part of SAR patients (8–16% of the total population) react to tree pollen, and birch pollen is the most common tree-produced allergen [2]. Same as other types of allergy, SAR is a chronic disease for most patients, and the probability of development of concomitant pathologies, such as asthma and food allergies, is quite high. While the most widely used therapeutic agents only aim to alleviate the symptoms (antihistamine and glucocorticosteroid drugs, mast cell membrane stabilizers), allergen-specific immunotherapy is the only currently practiced approach to SAR treatment that is designed to bring a long-term change in the chronic course of the disease. This is a long-term therapy that requires strict patient compliance and yields complete disappearance of the symptoms only in some patients [3]. Therefore, development of the new therapeutic approaches designed to prevent SAR from turning chronic is an extremely urgent task.
Rational design of such approaches requires a deep understanding of the mechanisms of maintenance of immunological memory that determine chronic course of allergy. A number of recent unsuccessful clinical trials of the allergic rhinitis treatment modalities demonstrate that the currently available information about these mechanisms is insufficient [4, 5]. Basic research efforts implying deeper investigation of the mechanisms of immunological memory preservation, which turns the allergy chronic, is a necessary basis both for the development of the new approaches to SAR treatment and improvement of the existing therapies.
It is known that IgE, a class E immunoglobulin, is one of the key agents in the development of an allergic response. However, there is little information about cell subpopulations producing IgE in allergic rhinitis patients: their localization, lifespan and required survival factors are not clear. Recently, using a mouse model of food allergy it was demonstrated, that shortly after sensitization IgE-secreting plasma cells can be found in the bone marrow and have a limited lifespan (up to 1 year) [6]. IgE class antibodies can remain bound at mast cell membranes for up to 100 days. It was also shown that at least some of the long-term immunological memory that turns allergy chronic is maintained by the subpopulation of IgG1+ allergen-specific memory B-lymphocytes, which serves as a "reservoir" from which the population of IgE producers is replenished in case of contact with the allergen [6, 7].
This study aimed to investigate the dynamics of serum IgE levels in greater detail compared to the previous research efforts [810]. We enrolled patients allergic to birch pollen and studied the dependence of the aforementioned dynamics on the intensity of their contact with the allergen that is the derivative of the pollen concentration in air. As part of the study, we sought to identify the time point that, relative to the moment of contact with the allergen, marks replenishment of the IgE secreting cell fraction. We also aimed to characterize the subpopulation of IgE-secreting plasmablast and plasma cells aand describe their relationship to other isotypes and memory B-cell population during seasonal exacerbation of the disease.


Patient cohort and collection of serum samples

The studied cohort included three patients (22, 22, and 28 years old). The inclusion criteria were: 1) any gender; 2) previously diagnosed seasonal birch pollen allergic rhinitis; 3) no bronchial asthma; 4) no other chronic inflammatory, autoimmune, oncological and infectious diseases. The exclusion criteria were: 1) ongoing or recently finished allergen-specific immunotherapy course; 2) vaccination during the study period. Venous blood samples were taken in medical laboratories and by medical personnel of the Invitro CDL chain of labs; the personnel used the Vacuette Serum Gel Z and Vacuette K3EDTA tubes (BD Biosciences; USA). Further experimental work was carried out in the research laboratory of the Institute of Bioorganic Chemistry, Russian Academy of Science.
To measure the total level of IgE and the level of birch and alder pollen-specific IgE we collected the patients' blood samples every fortnight from March to May 2019, thus taking 6 samples from each donor. From one of the donors (28 years old) we also collected four additional serum samples collected in March and May 2017, and in March and May 2018


We followed the protocol provided by the tube manufacturer (BD Biosciences; USA) to isolate blood serum. To establish the total IgE level and the level of birch and alder pollen-specific IgE we used the commercially available ELISA kits (Alkor Bio; Russia) according to the manufacturer's protocol. Hidex Sense reader (Hidex Oy; Finland) was used to measure the optical density. Each serum sample was measured in two independent replicates.

Isolation of subpopulations of memory B-lymphocytes, plasmablasts and plasma cells

For one of the donors (28 years old), we obtained samples of three cell B-cell subpopulations: memory B-lymphocytes, plasmablasts and peripheral blood plasma cells. These samples were collected at three time points, March and May 2017 and March 2018.
Peripheral blood mononuclear cells from whole blood were isolated using standard Ficoll density gradient protocol: centrifugation at 310g for 20 min, Eppendorf 5804 centrifuge (Eppendorf; Germany). Then, the cells were stained with the set of fluorescently-labeled antibodies: anti-CD19-APC, anti- CD20-VioBlue, anti-CD27-VioBright FITC, anti-CD138-PE-Vio770 (Myltenyi Biotec; USA). Next, using the fluorescence-activated sorting equipment (BD FacsAria III, BD Biosciences; USA) we isolated the target B-cell subpopulations: memory B-lymphocytes (CD19+ CD20+ CD27+), plasmablasts (CD20– CD19+ CD27++ CD138–) and plasma cells (CD20– CD19low/– CD27+ CD138+). At each time point, we collected two samples of memory B-lymphocytes (50,000 cells), plasmablasts and plasma cells (1000 and 500 cells, respectively).

Sequencing and analysis of B-cell receptor repertoires

Preparing the immunoglobulin (IgH) heavy chain cDNA libraries, we followed the previously published protocol [11] with some modifications. At the reverse transcription stage we introduced the adapter with a unique molecular identifier (UMI) and sample-specific barcode (tab. 1). Further amplification was done in two stages using the primers described in tab. 1. For sequencing we used Illumina Hiseq 2000/2500 (Illumina; USA) in pair-end mode with 310 + 310 read length.
MIGEC software [12] was used for raw data set demultiplexing and consensus assembling on the basis of unique molecular identifier and sample-specific barcode sequences [12]. MIXCR software [13] was used to annotate the V-, D-, J-, and C-segments, to identify the unique clonal IGH sequences (clonotypes) and to assess the level of hypermutations. To assemble the IGH clonal sequences, we used the IGH cDNA sequences supported by at least two reads based on the UMIs analysis. Clonotypes were defined by the nucleotide sequence of the immunoglobulin heavy chain starting from V segment framework region 1 (FR1) to the end of the J segment, with respect to the antibody isotype determined by the 5' fragment of the C segment 15–16 nucleotides long.
R programming language [14] was used for IGH clonal repertoire analysis, statistical processing of the results and regression analysis; the Figures were prepared using the ggplot2 package [15]. Change-O software [16] was used to identify the clonal groups on the basis of the following criteria: one clonal group contains IGH sequences which have the same CDR3 length and IGH V-segment, and at least 85% of CDR3 nucleotide sequence identity 15% R. Phylogenetic analysis was performed using MEGA 7 software [17] (maximum likelihood phylogenetic trees, evolutionary GTR model).

Regression analysis

Regression model was used to test whether there is a relation between the birch pollen sIgE serum level dynamics and the allergen air level. Table 2 shows the data used for the model. Regression equation:

sIgE_level = b + k1 × pollen_level + k2 × donorMRK + k3 × donorLK,

where sIgE_level is a dependent variable, representing serum level of the birch pollen-specific IgE; pollen_level — predictor variable, representing average level of birch pollen in the air (class) registered through the two weeks before sampling blood to measure the serum IgE level; donor MRK — dummy variable, can be 1 (MRK donor) and 0 (not MRK donor); donorLK — dummy variable, can be 1 (donor LK) and 0 (donor not LK); b is the intercept of the regression model that reflects the basic level of the dependent variable, relating to MS donor; k1, k2, k3 are the regression coefficients (k1 reflects the change in the dependent variable when the pollen level in the air changes by one pollination class; k2 reflects the difference in the basic level of birch pollen-specific IgE between MRK and MS donors; k3 reflects the difference in the basic level of birch pollen-specific IgE between LK and MS donors.
Calculated coefficients: b = 207.813 (p < 0.01); k1 = 6.474 (p < 0.05); k2 = –194.4 (p < 0.01); k3 = –208.15 (p < 0.01). Adjusted R2 = 0.98; p < 0.01; F test.
Similarly, we used a regression model to analyze the relation between the serum IgE level dynamics and the air level of birch pollen. Total IgE serum level was selected as the dependent variable; predictor variables were the same as in the described model for sIgE. For the model where the total IgE serum level is the dependent variable, the coefficients were as follows:
b = 1299.05 (p < 0.01); k2 = –1056.4 (p < 0.01); k1 = 51.45 (p < 0.05); k3 = –1313.65 (p < 0.01). Adjusted R2 = 0.95; p < 0.01; F test.

Pollen monitoring data

Pollen monitoring data for Moscow were obtained from open sources (Allergophone). tab. 3 is the pollination classification table that shows the number of pollen grains per cubic meter of air for each class.


The total and birch pollen-specific IgE serum levels increased as the birch pollen level grew in the air

We measured total and birch pollen-specific serum IgE levels at 6 time points before and after the start of birch pollination season. The absolute values reflecting serum levels of both specific (fig. 1A) and total IgE (fig. 1B) significantly differ among patients, and the individual differences significantly exceed differences between IgE levels of the same donor at different time points. In all donors, the absolute values of both birch pollen sIgE and total IgE increased between time point 1 and time point 6. The level of birch pollen-specific IgE correlates with the total IgE level in each donor; it exceeds the reference level at all time points (fig. 1).
The change in the levels of birch-pollen sIgE and total IgE registered through the period of the study is significantly smaller than the differences in the total and specific IgE serum levels between donors (coefficient of variation between all time points, one donor, birch pollen-sIgE: MS — 0.130, MRK — 0.228, LK — 0.282; total IgE: MS — 0.189, MRK — 0.154, LK — 0.485; the smallest variation coefficient between donors, sIgE — 1.39,
total IgE — 1.17). Taking this into account, in order to characterize the seasonal dynamics of total IgE and sIgE in serum we considered these levels with regards to the initial level of total IgE and sIgE, measured before the birch pollination season had started (time point 1). For each donor, each total and birch pollen-specific IgE level measurement was normalized to the corresponding values of total and birch pollen-specific IgE level at time point 1.
Our findings are consistent with the previously published research on the seasonal dynamics of serum levels of IgE specific to other pollen allergens [810]: birch pollen-sIgE level grows significantly between the off-season time point (1) and the season peak time point (6). The dynamics of birch pollen sIgE correlates with the level of birch pollen in the air (fig. 2). The maximum increase of sIgE was observed at the peak of the birch pollen level, between time points 3 and 5: the increase reached 90.7% for donor LK, for MRK — 63.7%, for MS — 36.9%.
A linear regression model was developed to statistically test the hypothesis that the level of birch pollen in the air and the serum level of birch pollen-specific IgE are related. According to the model, a high proportion of variance of the dependent variable (sIgE level) depends on the selected predictors (adjusted R2 = 0.98), i.e. pollen air level and donor identifier (the latter accounts for significant differences in the absolute sIgE level between donors). All coefficients are significantly not equal to 0, (p < 0.05 for all coefficients), and the coefficient at the predictor-variable reflecting birch pollen air level is greater than 0 (k1 = 6.47), which serves as a statistical confirmation of the consistency of serum sIgE level dynamics with the birch pollen air level. Similarly, the regression model for total IgE level also demonstrates that its level grows together with the growth of the allergen's concentration in the air (adjusted R2 = 0.95; p < 0.05 for each of the coefficients).
It should be noted that a relatively small (compared to the maximums registered through the observation period) birch pollen sIgE level increase occurs before the start of pollination season or when the allergen concentration is low (time points 1–3). The possible reasons could be a random fluctuation in birch pollen sIgE level (two out of three donors exhibited the absolute level increase below 1 IU/ml between time points 1–2 and 2–3) or a contact of the patient with birch pollen that occurred early, before pollen level growth was registered by pollen monitoring stations. Such an increase in the birch pollen sIgE level can also be the result of its cross-reactivity to alder pollen antigens as alder pollination season precedes birch pollination season and there is a high level of structural similarity between birch and alder pollen antigens. To assess the contribution of cross-reactive IgE we measured the level of alder pollen sIgE in all collected serum samples (fig. 3). In all donors, the level of alder pollen sIgE followed birch pollen air level more than alder pollen air level. Alder pollen sIgE peaked between time points 4 and 5, when the concentration of birch pollen in the air was at its maximum, but not at time points 1 and 2, when alder pollen level showed most of its growth. Therefore, we can assume a high level of birch pollen sIgE cross-reactivity to alder pollen antigens in all participating donors.
In 2 donors (MRK, LK), birch pollen sIgE grew slightly (<1 IU/ml) while alder pollen sIgE went down. Thus, the insignificant increase in birch pollen sIgE level shown for MRK and LK donors cannot be explained by its cross-reactivity to alder pollen sIgE. In MS donor, we registered growing alder pollen sIgE between time points 1 and 2, which may explain the increase in birch pollen sIgE level before the beginning of the birch pollination season.

The abundance of IgE-expressing cells in peripheral blood B-cell subpopulations is relatively low and increases during seasonal exacerbation of allergies

Based on the data obtained from the mouse model of food allergy, we can assume that birch pollen sIgE serum level and total IgE level grow with the increase of the number of IgE-producing cells. In order to assess the seasonal dynamics of IgE+ cell abundance in different B-cell subpopulations, we analyzed the immunoglobulin heavy chain (IGH) clonal repertoires of memory B-cells, plasmablasts and plasma cells from peripheral blood at three time points (1_2017, 3_2018 — time points outside the birch pollination season, 2_2017 — season peak).
For all three time points, sequencing yielded a total of 50,550,291 reads representing 1,616,747 unique IGH cDNA molecules. To eliminate the majority of errors in determining the clonal sequence, we included only those IGH cDNA sequences which were supported by at least two independent reads. A total of 116,437 IGH clonotypes were identified (a clonotype is defined as unique nucleotide sequence of IGH starting from the framework region (FR1) of the V segment to the end of the J segment, with antibody isotype determined on the basis of 5’ fragment of C segment). IgE clonotypes were extremely rare (~0.01% out of all detected clonotypes). For comparison, the frequency of IgM clonotypes was 48.4%, IgD — 14%, IgG — 17.4%, IgA — 19.8%.
All IGE+ clonotypes were detected in IGH repertoires of memory B-lymphocytes, plasmablasts and long-lived plasma cells exclusively in samples obtained during the birch pollination season (fig. 4), while the sample size (in the number of cells) for each subpopulation does not differ between time points. This result demonstrates the increase in abundance of IGE+ cells in the patient’s peripheral blood during birch pollination season.
The technology applied to prepare immunoglobulin heavy chain cDNA libraries allows estimating the number of hypermutations along the entire length of the IGH sequence, from the FR1 region to 5’ fragment of the constant region. For each clonotype, we identified the number of hypermutations and compared it between clonotypes with different isotypes. As expected, the lowest level of hypermutations was detected among IgD and IgM clonotypes (median — 2.3 and 3.7 bp per 100 bp of the V segment sequence, respectively), while a higher level — among IgG and IgA clonotypes (median — 6.8 and
7.1 bp per 100 bp of the V segment sequence, respectively) (fig. 5), which are typically expressed in B-lymphocytes that matured in germinal centers. The average number of hypermutations of IgE clonotypes was not lower (median — 7.4 bp per 100 bp of the V segment sequence) compared to that of IgG and IgA clonotypes. Similar level of hypermutations between IgE and IgG clonotypes may reflect the origination of the IgE clonotypes of antibody-secreting cell subpopulations from memory IgG+ B lymphocytes, but also does not exclude the independent isotype switching and accumulation of hypermutations in IgE+ and IgG+ B lymphocytes.
To analyze the relationship between IgE clonotypes in immunoglobulin heavy chain repertoires and clonotypes of other isotypes in greater detail we applied the previously described approach for identification of sets of clonotypes most likely derived from a single precursor and having a similar but not necessarily identical nucleotide sequence encoding a variable domain of the B-cell receptor heavy chain (clonal groups) [16]. A total of 13 IgE-containing clonal groups were identified; they included 154 clonotypes. In 4 of them we have identified clonotypes of other isotypes: IgG — 82, IgA — 19, IgM — 4 and IgD — 1. Phylogenetic analysis that we did for clonotypes of each of these clonal groups reveals a close relationship between IgE and IgG clonotypes (fig. 6): for most IgE clonotypes, the closest neighbour in the phylogenetic tree was an IgG clonotype. It should be noted that in each of the considered IgE-containing clonal groups there were IgG-clonotypes representing the IgG+ fraction of memory B-lymphocytes.


Previously published studies describe an increase in the serum level of total and allergen-specific IgE during the seasonal allergic rhinitis exacerbation [:lit_8–10;]. However, these studies present the serum levels seasonal dynamics data that were collected once every several months. Moreover, no connection was made to the air level of pollen that causes the allergy. In this work, we measured serum levels of total IgE and birch pollen/alder pollen-specific IgE with a higher time resolution (immunoglobulin levels were measured once a fortnight) during a period including the birch pollination commencement. Collation of the immunoglobulin levels and air level of the pollen at each time point allowed to demonstrate a correlation between birch pollen concentration in the air and level of birch pollen-specific and total IgE. Interestingly, all donors had the alder pollen sIgE growing together with the concentration of birch pollen in the air, but not with the level of alder pollen. Most likely, this is due to the birch pollen sIgE's cross-reactivity to alder pollen antigens; this fact may indicate that the donors that participated in our study lack specific antibodies to unique alder epitopes. However, since we did not register the alder pollen sIgE level before alder pollination season (it was a little over a week away from the beginning of sIgE level registration, and within this week the alder pollen air concentration was as low as 1–10 grains per 1 m3), we cannot exclude the possibility of alder pollen sIgE peaking during this season (time points 1–3).
In all donors, birch pollen sIgE and total IgE levels began growing at the same time when the concentration of birch pollen in the air started to increase. This simultaneity may be the result of the growing count of IgE-secreting plasmablasts and plasma cells differentiating from memory B-lymphocytes in response to stimulation by the allergen. Analysis of the IGH repertoire dynamics and properties of sequences of identified IGH clonotypes generally confirms this assumption. The data obtained show that the abundance of peripheral blood IgE-expressing plasmablasts and plasma cells grows when birch starts pollinating, which reflects the active differentiation process of B-lymphocytes. The high level of hypermutations observed in IgE clonotypes is the sign of a long history of receptor selection in germination centers, which suggests that these IgE-secreting cells may originate from memory B-lymphocytes.
Overall, the obtained results allow assuming that the long-term immunological memory maintenance model proposed for food allergies [6] may also be valid for seasonal allergic rhinitis. In this model, the long-lived IgE-secreting plasma cells have a limited life span (~100 days in the mouse model), while memory IgG1 B-lymphocytes serve as the long-term "reservoir" from which the pool of IgE-secreting plasma cells is replenished. The replenishment occurs when the patient contacts the allergen, the process itself implies rapid proliferation, isotype switching, and differentiation of memory IgG1 B-lymphocytes under the action of Th2-associated cytokines (IL4, IL13, etc.). The previously published study of the B-cell receptor repertoires in respiratory allergies [18] did not allocate clonotypes to specific subpopulations, however, there were indications of clonal connections between IgG and IgE clonotypes in the peripheral blood immunoglobulin heavy chain repertoires, which is an indirect confirmation of correctness of the model suggested by R. Jiménez-Saiz and co-authors. The phylogenetic analysis of IgE-containing clonal groups, which was part of our study, demonstrates that the IgE-secreting plasma cells may originate from memory IgG+ B lymphocytes in allergic rhinoconjunctivitis patients. However, the analysis does not exclude other IgE-producers origination patterns.
Our results also do not exclude the possibility of serum IgE level increase being caused by the boost in production and secretion of antibodies by existing IgE-secreting plasma cells. The average lifespan of IgE+ plasma cells in human red bone marrow remains unexplored. Allergic rhinitis patients exhibit high IgE concentration in serum even outside the pollination season, which suggests the lifespan exceeding the time between two pollination seasons. However, taking into account the previously published data on correlation between serum IgE and peripheral blood IgE+ plasma cells count [19], the growing number of IgE-secreting cells seems to be a more likely explanation for growth of IgE and sIgE levels, which is consistent with the results of this work.


Compared to the previously published study, we monitored the level of antibodies in serum at a higher temporal resolution, which allowed to demonstrate that level of sIgE starts to increase at the same time with the increase of birch pollen in air for patients with seasonal birch pollen allergy. The dynamics of total and allergen-specific IgE correlates with the level of birch pollen in the air.
Combined with the data obtained by other researchers in a mouse model of food allergy, our results of high-throughput sequencing and analysis of repertoires of cells of B-cell lineage suggest that this growth is most likely caused by the growing abundance of IgE-secreting plasma cells. The analysis of IGH nucleotide sequences of IgE-containing clonal groups of different B cell subpopulations showed a high similarity of IgE and IgG clonotypes and presence of IgG clonotypes representing the IgG+ memory B lymphocyte fraction in these groups, which suggests that IgE-secreting cells could originate from a pool of IgG+ B lymphocytes.
The small sample size does not allow to generalize the revealed characteristics to all patients since there is an alleged variety in endotypes of allergic rhinoconjunctivitis. The results we obtained in our study highlight similarities in the long-term immune memory maintenance mechanisms for SAR and food allergies. To validate this assumption, it is necessary to conduct further research with larger cohorts and more elaborate clinical characterization of patients. Such research efforts are needed to develop new therapies aimed at breaking the allergen-specific immunological memory. For example, such new therapy may imply blocking isotype switching to IgE during differentiation of IgG+ allergen-specific memory B cells, which is achievable through suppressing the effect of Th2 T cell cytokines by using a monoclonal antibody to IL4R. The obtained information on the dynamics of serum antibodies level with changing the allergen air concentration facilitates further investigation of the seasonal dynamics of concentration of IgE-secreting allergen-specific cells.