Clonal evolution and stereotyped sequences of human IgE lineages in aeroallergen-specific immunotherapy

Abstract

Background:

Allergic disease reflects specific inflammatory processes initiated by interaction between allergen and allergen-specific IgE. Specific immunotherapy (SIT) is an effective long-term treatment option, but the mechanisms by which SIT provides desensitization are not well understood.

Objective:

To characterize IgE sequences expressed by allergen-specific B cells over a 3-year longitudinal study of patients with aeroallergies undergoing SIT.

Methods:

Allergen-specific IgE-expressing clones were identified using combinatorial scFv libraries and tracked in PBMC and nasal biopsies over a 3-year period with antibody gene repertoire sequencing. Characteristics of private IgE-expressing clones were compared with those of stereotyped or “public” IgE responses to the grass pollen allergen Phl p 2.

Result:

Members of the same allergen-specific IgE lineages were observed in nasal biopsies and blood, and lineages detected at baseline persisted in blood and nasal biopsies after 3 years of SIT, including B cells that express IgE. Evidence of progressive class-switch recombination (CSR) to IgG subclasses was observed after 3 years of SIT. A common stereotyped Phl p 2-specific antibody heavy chain sequence was detected in multiple donors. The amino acid residues enriched in IgE stereotyped sequences from seropositive donors were analyzed with machine learning and k-mer motif discovery. Stereotyped IgE sequences had lower overall rates of somatic hypermutation and antigen selection than scFv-derived allergen-specific sequences or IgE sequences of unknown specificity.

Conclusion:

Longitudinal tracking of rare circulating and tissue-resident allergen-specific IgE+ clones demonstrates persistence of allergen-specific IgE+ clones, progressive CSR to IgG subtypes, and distinct maturation of a stereotyped Phl p 2-clonotype.

Capsule summary

Aeroallergen-specific IgE+ clones of allergic individuals subjected to SIT are persistent in blood and tissue, undergo progressive class-switch recombination, and include a stereotyped Phl p 2-specific clonotype that has a distinct maturation pathway.

INTRODUCTION

Allergic symptoms are produced by specific inflammatory processes initiated by interaction between allergen and allergen-specific IgE^1,2. However, human B cells or plasma cells expressing IgE remain poorly understood in terms of their localization, clonal persistence, diversity, and relationship to memory pools, in large part due to a rarity of IgE-producing cells in the blood. Flow cytometric sorting and antibody sequencing have provided some data directly from IgE-expressing B cells^3–5, or from B cells expressing other antibody isotypes that belong to allergen-specific clones containing IgE-expressing members^{6, 7}. Cloning of recombinant IgE using phage display technology has been a tractable approach for identifying human allergen-specific IgE, although such technology does not ensure native heavy/light chain variable domain pairing or recapitulation of the avidity of the antibodies in the in vivo context^{8, 9}. Next generation DNA sequencing (NGS) has provided new insights into IgE responses^{6–8, 10–15} by enabling analysis of the antibody genes expressed by thousands to millions of B cells in human blood or tissue specimens. Through analysis of antibody repertoires of allergic patients, several examples have been found of “convergent”, “public”, or “stereotyped” antibody responses against viruses, autoantigens, and food allergens¹⁶.

Specific immunotherapy (SIT), where increasing doses of an allergen are delivered to an allergic patient until a dose is reached that induces immunologic desensitization, is often, but not always, an effective treatment for aeroallergies. Patients typically need to undergo many years of treatment before they are desensitized, with tolerogenic effects often sustained past cessation of immunotherapy¹⁷. The mechanisms for how SIT induces desensitization are somewhat unclear, but likely involve changes of immune cell populations including B cells (reviewed in¹⁸). One proposed mechanism is that SIT stimulates the generation of allergen-specific IgG antibodies that prevent the binding of IgE antibodies to allergen. It is unclear what the effect of SIT is on the B cells and plasma cells that produce the pathogenic allergen-specific IgE antibodies.

Longitudinal analysis of antibody repertoires in blood and other tissues of allergic patients is needed to understand the evolution of allergen-specific IgE+ B cell clones and their responses to SIT over time. As antibody isotype class-switching to IgE can occur locally in immunologically important sites such as the nasal mucosa^{19, 20}, bronchial tissues^{21, 22}, and gastrointestinal mucosa¹⁵, it is also important to characterize IgE responses in these difficult-to-assay tissues. Prior studies have identified antigen-specific IgE+ clones persisting for at least one year in subjects undergoing allergen-specific immunotherapy^{8, 14}, and have indicated class-switched members of some of these allergen-specific IgE+ clones. Here, we analyze the immunoglobulin heavy chain (IGH) repertoires of peripheral blood and nasal biopsies after three years of SIT, and characterize the sequence characteristics of allergen-specific clones derived from patients’ individual IgE repertoires as well as clones associated with stereotyped Phl p 2-specificity^23–25.

METHODS

Patient samples

Blood and nasal biopsy samples were collected from 8 allergic subjects undergoing SIT (donors 1–8) and 8 allergic controls (donors 9–16) at 4 and 3 timepoints, respectively, during a period of 3 years (Fig 1A, Table I). The study was approved by the regional ethics board at Lund University. Details of the sample collection are provided in the Methods of this article’s Online Repository. Serum levels of allergen-specific IgE, IgG, and IgG4 have been determined in a previous study⁸.

FIG 1.

Study design and determination of allergen-specific clones from NGS sequencing data. (A) Study design: blood samples at Week 8, Year 1, and Year 3, and nasal biopsies at Year 1 and Year 3, were obtained approximately one week after the preceding SIT injection. Plasma and blood were not obtained from the nonvaccinated group at Week 8. (B) Mean frequency (of all samples from a donor) of heavy chain sequences that originate from IGHV4-30-4 or IGHV4-31, and have a CDRH3 that are 9 or 10 residues long with glycine as the fifth residue, in the NGS data of allergic individuals. Such sequences have previously been identified as Phl p 2-stereotyped rearrangements. Stereotyped heavy chain sequence frequencies in IgG (left panel) and IgE (right panel) populations are shown for Phl p 2-seropositive (grey bars) and Phl p 2-seronegative (black bars) individuals, as determined by ISAC analysis (Table EIII). Frequencies in other populations are presented in Fig E1. (C) Counts of allergen-specific clones that also contained at least one IgE-expressing member in donors undergoing SIT and donors who did not undergo SIT. One scFv-derived clone, IT8-P220, also met the criteria for being a stereotyped Phl p 2-specific clone, and is here counted in both categories. Additional details on stereotyped clones found in donors who did not undergo SIT may be found in Table EIV.

TABLE I.

Donor characteristics.

Donor ID	Age (y)	Sex	Skin prick test	SIT	Year 3 Tissues	Year 1 scFv libraries selected on:
1	38	F	BP, GP	BP	-	-
2	30	M	BP, GP, CD, DD	BP, GP*, CD	PBMC, Nasal biopsy	-
3	30	M	BP, GP	BP, GP*	PBMC, Nasal biopsy	-
4	36	F	BP, GP, HDM, CD, DD	BP, GP*, HDM	PBMC	-
5	23	M	HDM, AF, CD, DD	HDM	PBMC, Nasal biopsy	-
6	41	F	BP, MP, HDM, MM	BP, MP, HDM	PBMC, Nasal biopsy	Der p 1
7	27	M	BP, GP, MP	BP, GP*, MP	PBMC, Nasal biopsy	-
8	24	F	BP, GP, CD, DD, HD	BP, GP*, CD	PBMC, Nasal biopsy	-
9	25	M	BP, GP, MP, DD, CD	No	-	-
10	30	F	BP, GP, MP, DD	No	-	-
11	23	M	BP, GP	No	PBMC, Nasal biopsy	-
12	24	F	HDM, BP	No	PBMC, Nasal biopsy	-
13	38	M	BP, GP, DD	No	PBMC, Nasal biopsy	-
14	22	M	BP, DD	No	-	-
15	24	F	BP, GP, HDM, CD, DD	No	-	-
16	47	M	BP, GP, CD, DD	No	PBMC, Nasal biopsy	-

AF, Aspergillus fumigatus; BP, birch pollen; CD, cat dander; DD, dog dander; F, female; GP, grass pollen; HD, horse dander; HDM, house dust mites; M, male; MM, mold mix; MP, mugwort pollen.

^*Five-grass mixture containing Dactylis glomerata, Festuca pratensis, Lolium perenne, Phleum pratense, and Poa pratensis was used for SIT.

Analysis of antibody heavy chain variable domain-encoding transcriptome and of allergen-specific clones

IGH repertoire NGS data from collected PBMCs and nasal biopsies were obtained using the 454 platform (Roche, Mannheim, Germany), as previously reported⁸, and/or Illumina MiSeq (San Diego, CA, USA). The number of sequence reads are shown in Supplemental Table EI. NGS data can be accessed through the SRA archive with Bioproject ID PRJNA391821. Allergen-specific scFv sequences had previously been identified from samples collected 8 weeks after SIT initiation using combinatorial scFv library technology⁸. Here, additional combinatorial scFv libraries were generated from a 1-year sample and panned to identify additional allergen-specific sequences. Clones expressing scFv-derived sequences (Table II) or Phl p 2-stereotyped rearrangements were identified in patient antibody repertoires. Details of antibody repertoire and scFv data generation, and of downstream analyses, are provided in the Supplementary Methods of this article’s Online Repository.

TABLE II.

Allergen-specific scFvs isolated after initiation of SIT.

Donor	Clone ID	Allergen	Library	Timepoint used to generate library	IGHV	IGHV mutations (%)	CDRH3	Identified by using HTS	Identified as an IgE-containing clone by using HTS	Reference
1	IT1-B215	Bet v 2	PBMC, IgE	8 weeks	1–18	0.0	ARVRSSGYGNWFDP			Levin et al., 2016
1	IT1-B228	Bet v 2	PBMC, IgE	8 weeks	1–69	0.0	ARDNGGEGY			Levin et al., 2016
1	IT1-B237	Bet v 2	PBMC, IgE	8 weeks	3–21	4.2	ARGGTRYFAS			Levin et al., 2016
1	IT1-B248	Bet v 2	PBMC, IgE	8 weeks	3–11	0.0	ARDRSPIAAARLAFDI			Levin et al., 2016
2	IT2-B116	Bet v 1	PBMC, IgE	8 weeks	3–11	2.6	ARDAYFDWSLDY	Yes	Yes	Levin et al., 2016
2	IT2-B12	Bet v 1	PBMC, IgE	8 weeks	5–51	9.8	ARLGGGSRGWYYYYGMDV	Yes	Yes	Levin et al., 2016
2	IT2-B15	Bet v 1	PBMC, IgE	8 weeks	3–48	5.3	ARGYYDGRKLRN	Yes	Yes	Levin et al., 2016
2	IT2-B227	Bet v 2	PBMC, IgE	8 weeks	3–74	3.8	ARRDVAVVPGATGDNYYYGLDV	Yes	Yes	Levin et al., 2016
2	IT2-B23	Bet v 2	PBMC, IgE	8 weeks	3–66	5.0	ARGYYDGRKLRN	Yes	Yes	Levin et al., 2016
2	IT2-B239	Bet v 2	PBMC, IgE	8 weeks	1–18	2.7	ARSLFLLSPKARGYYYYGMDV	Yes	Yes	Levin et al., 2016
2	IT2-P11	Phl p 1	PBMC, IgE	8 weeks	3–23	4.5	ANLGFDY	Yes	Yes	Levin et al., 2016
2	IT2-P514	Phl p 5	PBMC, IgE	8 weeks	1–69	8.4	ARALIPLESQAAD	Yes		Levin et al., 2016
2	IT2-P536	Phl p 5	PBMC, IgE	8 weeks	1–69	5.7	ASMGRGYCSGDNCYNFDH	Yes		Levin et al., 2016
2	IT2-P614	Phl p 6	PBMC, IgE	8 weeks	3–48	3.4	ARDRGTPWHYDGMDV	Yes	Yes	Levin et al., 2016
2	IT2-P64	Phl p 6	PBMC, IgE	8 weeks	5–51	3.8	ARRFGGDWFGSLGWYYFDH	Yes	Yes	Levin et al., 2016
3	IT3-B11	Bet v 1	PBMC, IgE	8 weeks	1–46	10.2	ASDIAEGLGQHLFDH			Levin et al., 2016
3	IT3-B116	Bet v 1	PBMC, IgE	8 weeks	3–30	6.8	ASSLTSTGMGRY			Levin et al., 2016
3	IT3-B229	Bet v 2	PBMC, IgE	8 weeks	1–69	12.1	ARDRKFYYDRSGVPYFDH	Yes		Levin et al., 2016
3	IT3-P11	Phl p 1	PBMC, IgE	8 weeks	4–4	3.4	ARDGKNGSSDY	Yes	Yes	Levin et al., 2016
4	IT4-B119	Bet v 1	PBMC, IgE	8 weeks	3–48	4.2	ARGPGYSSSWYGYYFDS	Yes	Yes	Levin et al., 2016
4	IT4-B148	Bet v 1	PBMC, IgE	8 weeks	3–48	11.3	ASPPTSYDFWSDYSDYDYYYMDV	Yes	Yes	Levin et al., 2016
4	IT4-B225	Bet v 2	PBMC, IgE	8 weeks	3–30	3.4	ARDLTGNFAN	Yes	Yes	Levin et al., 2016
4	IT4-B229	Bet v 2	PBMC, IgE	8 weeks	3–20	9.8	ARDLIGNFAN	Yes	Yes	Levin et al., 2016
4	IT4-P517	Phl p 5	PBMC, IgE	8 weeks	3–7	4.5	ARDSRQWWFHIEGDAFDI	Yes	Yes	Levin et al., 2016
4	IT4-P62	Phl p 6	PBMC, IgE	8 weeks	3–33	7.5	ARDFFPLAVLSPPLGY	Yes	Yes	Levin et al., 2016
5	IT5-nD134	Der p 1	PBMC, IgE	8 weeks	3–53	0.8	AREGGVAARPNPDAFDI	Yes	Yes	Levin et al., 2016
5	IT5-rD120	Der p 1	PBMC, IgE	8 weeks	3–23	5.7	AKAHGFGSPGWGSGWHRKTPSRYPYYFDY	Yes	Yes	Levin et al., 2016
5	IT5-rD13	Der p 1	PBMC, IgE	8 weeks	3–11	5.3	ARDGALVWFGNKSYGMDV	Yes	Yes	Levin et al., 2016
6	IT6-B15	Bet v 1	PBMC, IgE	8 weeks	3–30-3	11.3	ARGRPRAYYYDDSGSKQYWDEHYFDY	Yes	Yes	Levin et al., 2016
6	IT6-B22	Bet v 2	PBMC, IgE	8 weeks	3–30	10.6	ARDIFASQGAADY	Yes	Yes	Levin et al., 2016
6	IT6-nD12	Der p 1	PBMC, IgE	8 weeks	3–53	8.4	ARLYYDFWSGHAYFFYYMDV	Yes	Yes	Levin et al., 2016
6	IT6-nD13	Der p 1	PBMC, IgE	8 weeks	3–23	8.7	AKSERPIVATITGYYYYYMDV	Yes	Yes	Levin et al., 2016
6	IT6-nD16	Der p 1	PBMC, IgE	8 weeks	3–30	9.1	AKAAYSYGMKNSLDF	Yes	Yes	Levin et al., 2016
6	IT6-nD17	Der p 1	PBMC, IgE	8 weeks	3–30	17.0	TRDRSPVSGVLQH	Yes	Yes	Levin et al., 2016
6	IT6-rD111	Der p 1	PBMC, IgE	8 weeks	3–64	6.0	VKNMVRGVITDAFEI	Yes	Yes	Levin et al., 2016
6	IT6-rD116	Der p 1	PBMC, IgE	8 weeks	3–49	14.4	SRGLWFGKLWGPPREH	Yes	Yes	Levin et al., 2016
6	IT6-rD128	Der p 1	PBMC, IgE	8 weeks	3–30	10.9	ARDEGGDSSGNH	Yes	Yes	Levin et al., 2016
6	IT6-phage-clone01	Der p 1	biopsy, IgA	1 year	3–7	7.8	ARSKAFY	Yes		This study
6	IT6-phage-clone05	Der p 1	biopsy, IgE	1 year	1–3	5.4	ARGFWDDGFDI	Yes	Yes	This study
6	IT6-phage-clone06	Der p 1	biopsy, IgA	1 year	3–49	5.1	NSQGGADVFDL	Yes		This study
6	IT6-phage-clone06	Der p 1	biopsy, IgA	1 year	3–49	9.5	NSQGGPDVYDL	Yes		This study
6	IT6-phage-clone06	Der p 1	biopsy, IgA	1 year	3–49	9.5	NSQGGPDVYDL	Yes		This study
6	IT6-phage-clone07	Der p 1	PBMC, IgE	1 year	3–33	6.5	ARPTVGVVIRLDH	Yes	Yes	This study
6	IT6-phage-clone09	Der p 1	PBMC, IgE	1 year	3–48	12.2	ARDSGTYREDAFDI	Yes	Yes	This study
6	IT6-phage-clone09	Der p 1	PBMC, IgE	1 year	3–48	11.8	ARDSGTYREDAFDI	Yes	Yes	This study
6	IT6-phage-clone09	Der p 1	PBMC, IgE	1 year	3–48	12.2	ARDSGTYREDAFDI	Yes	Yes	This study
6	IT6-phage-clone15	Der p 1	PBMC, IgE	1 year	5-a	10.9	VRHRSEHWLGRPFDY	Yes	Yes	This study
6	IT6-phage-clone16	Der p 1	biopsy, IgE	1 year	3–23	10.9	AKAQLWRREMYYGMDV	Yes	Yes	This study
6	IT6-phage-clone16	Der p 1	biopsy, IgE	1 year	3–23	10.8	AKAQLWRREMYYGMDV	Yes	Yes	This study
6	IT6-phage-clone16	Der p 1	biopsy, IgE	1 year	3–23	10.2	AKAQLWRREMYYGMDV	Yes	Yes	This study
6	IT6-phage-clone16	Der p 1	biopsy, IgE	1 year	3–23	10.5	AKAQLWRREMYYGMDV	Yes	Yes	This study
6	IT6-phage-clone16	Der p 1	biopsy, IgE	1 year	3–23	10.8	AKAQLWRREMYYGMDV	Yes	Yes	This study
6	IT6-phage-clone17	Der p 1	biopsy, IgE	1 year	1–18	8.2	ARSPFTHYGSGRGGPI	Yes	Yes	This study
6	IT6-phage-clone20	Der p 1	biopsy, IgG	1 year	3–30	11.5	AKDFVRGDYVWESYRYTRDYIPDY			This study
6	IT6-phage-clone21	Der p 1	PBMC/biopsy, IgE	1 year	3–30-3	10.9	ARGRPRAYYYDNSGSKKYWDEFYFDN	Yes	Yes	This study
6	IT6-phage-clone21	Der p 1	PBMC/biopsy, IgE	1 year	3–30-3	9.2	ARGRPRAYYYDSSGSKHYWDEFYFDY	Yes	Yes	This study
7	IT7-B14	Bet v 1	PBMC, IgE	8 weeks	1–2	13.6	ARGLRSQLWYLDV	Yes	Yes	Levin et al., 2016
7	IT7-B148	Bet v 1	PBMC, IgE	8 weeks	1–2	7.5	ARGLRSQLWYMDV	Yes	Yes	Levin et al., 2016
7	IT7-B210	Bet v 2	PBMC, IgE	8 weeks	3–23	9.1	VKEKGWQQLPKGGHNWFDP	Yes	Yes	Levin et al., 2016
7	IT7-B214	Bet v 2	PBMC, IgE	8 weeks	3–7	6.0	TRQSGWLPFKD	Yes	Yes	Levin et al., 2016
7	IT7-B215	Bet v 2	PBMC, IgE	8 weeks	3–21	6.0	ARDLSYSGGD	Yes	Yes	Levin et al., 2016
7	IT7-B22	Bet v 2	PBMC, IgE	8 weeks	3–23	0.4	APSRYCSGGSCYSGY			Levin et al., 2016
7	IT7-B222	Bet v 2	PBMC, IgE	8 weeks	3–30	0.0	AKAPRSSSFRYFQH			Levin et al., 2016
7	IT7-P51	Phl p 5	PBMC, IgE	8 weeks	3–21	2.6	ARERSPWSEEAFDV	Yes	Yes	Levin et al., 2016
7	IT7-P517	Phl p 5	PBMC, IgE	8 weeks	3–23	9.8	AKDGISEYCSGGSCHSRGWVYFDY	Yes	Yes	Levin et al., 2016
7	IT7-P61	Phl p 6	PBMC, IgE	8 weeks	3–21	1.5	TRSPYYSGSGSYLDN	Yes	Yes	Levin et al., 2016
8	IT8-B13	Bet v 1	PBMC, IgE	8 weeks	3–49	4.8	TRRATLYYDTSGYSYYFDY	Yes	Yes	Levin et al., 2016
8	IT8-B216	Bet v 2	PBMC, IgE	8 weeks	3–11	0.0	ARVGRVRGVIRFDP			Levin et al., 2016
8	IT8-B22	Bet v 2	PBMC, IgE	8 weeks	1–18	0.0	ARGVGSRRQVYYYGMDV	Yes	Yes	Levin et al., 2016
8	IT8-P243	Phl p 2	PBMC, IgE	8 weeks	3–23	1.5	AKGGSGTISTP	Yes		Levin et al., 2016
8	IT8-P220	Phl p 2	PBMC, IgE	8 weeks	4–31	7.1	ARGAGDFDS	Yes	Yes	Levin et al., 2016
8	IT8-P621	Phl p 6	PBMC, IgE	8 weeks	3–30	9.4	ARGVGDYVWGPKGDY			Levin et al., 2016

RESULTS

Identification of allergen-specific antibody fragments from Year 1 blood and biopsy repertoires by phage display technology

We previously isolated 37 allergen-specific clones using phage display libraries generated from PBMCs of 8 individuals during SIT vaccine up-dosing and tracked these clones in IGH repertoire data obtained by NGS of blood or nasal biopsy samples⁸. To evaluate the effects of SIT on B cell populations at longer time intervals, we collected blood samples and nasal biopsies after Year 3 of SIT for NGS analysis of IGH repertoires. In the previous study, only a minor portion of the isolated allergen-specific clones were detected in NGS data from nasal biopsies, and those were all Der p 1-specific clones identified in donor 6⁸. Therefore, we used a nasal biopsy as well as blood from donor 6 at the Year 1 timepoint to generate additional IgE, IgA, and IgG-derived phage display libraries, and panned these against house dust mite allergen Der p 1. Thereby, we identified 19 new Der p 1-specific sequences from 10 clonal lineages defined as sharing the same IGHV gene, CDRH3 sequence length, and 90% identity of CDRH3 amino acids (Table II). Year 1 Der p 1-specific scFv sequences showed diverse IGHV, IGHD, and IGHJ germline gene usage. There was no bias towards utilization of IGHV genes from the VH5 family in allergen-specific clones derived from Year 1 (Table II) or in the Week 8-derived clones⁸, as was previously observed in the IgE-positive B cells found in the nasal mucosa of allergic rhinitis patients²⁶. NGS IGH repertoires contained 9 of 10 additional clones from Year 1 transcripts and members of 42 of 53 previously generated scFv binders⁸, recognizing allergens Bet v 1, Bet v 2, Phl p 1, Phl p 2, Phl p 5, Phl p 6, and Der p 1 (Table II).

Identification of stereotyped IgE sequences specific for grass pollen allergen Phl p 2

We and others have previously identified clones^{23, 27} representing a stereotyped IgE antibody response against the grass pollen allergen Phl p 2^{24, 28}. The heavy chains of these antibodies are derived from IGHV4-31 or the highly related gene IGHV4-30-4, and have CDRH3s of 9 or 10 amino acids with a glycine as the invariable fifth residue^{24, 28}. Such transcripts have been identified in three donors seropositive for Phl p 2-specific IgE, but not in three seronegative donors²⁸, and notably, addition of recombinant stereotyped antibodies was sufficient to block basophil activation by Phl p 2 allergen²⁵. To further define the characteristics of these antibodies, we searched for similar Phl p 2-stereotyped sequences in the heavy chain gene repertoires of the 3-year cohort of SIT-recipients and control allergic donors. Stereotyped clones detected in IgE repertoires of 7 of 8 donors seropositive for Phl p 2-specific IgE (all but Donor 10, who lacked the required IGHV germline genes [Table EII]), but only in 3 of 8 seronegative donors (Fig 1C). One scFv-derived Phl p 2-binder, IT8-P220 from Donor 8, which utilizes IGHV4-31 and has a CDRH3 of 9 residues with the sequence ‘ARGAGDFES’, also met the criteria for this stereotyped class of IgE sequences. In non-IgE repertoires, stereotyped sequences were found in most individuals, regardless of seroconversion status (Fig 1B, Fig E1). The frequencies of stereotyped sequences in IgM, IgD, IgG, and IgA repertoires were generally lower compared to the IgE repertoire of seropositive donors, where stereotyped Phl p 2 sequences accounted for over 1% of reads in 5 individuals. These data indicate that the stereotyped Phl p 2 heavy chain signature is common in individuals with allergies to grass pollen.

Persistence and tissue expression of allergen-specific IgE+ B cell clones in patients undergoing 3 years of SIT

We searched the total IgE repertoire data from all SIT recipients for members of allergen-specific clones and identified 77 IgE-containing clones closely related to the 53 scFv-sequences. We identified 28 IgE+ stereotyped clones from donors who received grass pollen (GP) SIT (donors 2, 3, 4, 7 and 8). 20 additional IgE-containing stereotyped clones were identified in donors not undergoing GP SIT, with 16 clones from seropositive donors (donors 11 and 16), and 4 from donors who did not seroconvert (donors 1, 6 and 15) (Fig 1C, Table EIV). Members of allergen-specific IgE-containing clones were identified mainly in blood samples, but several clones had members in nasal biopsies or in both tissues. IgE-expressing members of allergen-specific clones were identified in blood as well as nasal biopsies even after 3 years of SIT (Fig 2). Stereotyped clones for donors who did not undergo SIT are summarized in Fig E2.

FIG2.

Dotplots illustrating persistence of isolated allergen-specific clones with an IgE-expressing member after 3 years of SIT, as tracked by NGS of antibody heavy chain rearrangements from blood and nasal biopsies. The isotypes expressed by allergen-specific clones at each timepoint and in each tissue are indicated on the x-axis and by color and shape, respectively. Allergen-specific scFv-derived and stereotyped lineages, and their specificities, are shown on the y-axis.

The frequencies of IgE-expressing, allergen-specific clones with clone members of any isotype found in blood or nasal biopsies after 3 years of SIT were compared with the frequencies of IgE+ clones of unknown specificity (Fig 3A). At Year 1, the frequencies of scFv-derived allergen-specific IgE+ clones detected in blood, nasal biopsies, or both were increased compared to baseline and were significantly greater than frequencies of IgE+ clones of unknown specificity. At Year 3, IgE+ allergen-specific scFv-derived clones were still detected in the blood, but at decreased frequencies compared to IgE+ clones of unknown specificity, while the frequencies of scFv clones detected in nasal biopsies or both tissues remained significantly higher (Fig 3A). Stereotyped clones were enriched in biopsies or both tissues before SIT initiation, but at Year 1 and 3 they were more often detected only in the blood. In non-SIT participants, stereotyped clones were enriched in nasal biopsies through all three years sampled (Fig 3B).

FIG3.

Analysis of isotype expression, tissue localization, and clonal persistence of B cells belonging to clonal lineages containing IgE-expressing members in donors undergoing SIT and untreated donors. Tissue localization of members of IgE-containing B cell clones whose specificity is known based on binding of phage-displayed scFv to allergen, or stereotyped Phl p 2-specific sequences, compared to IgE-expressing clones of unknown specificity, in treated (A) and untreated (B) donors. Clones found in both blood and nasal biopsies were also counted in the blood category and the biopsy category. Persistent detection of clones of known and unknown specificities that contain IgE-expressing members in treated (C) and untreated donors (D). Isotypes expressed by allergen-specific clones containing IgE-expressing members in treated (E) and untreated (F) donors. Significance was determined by using the Fisher exact test. * p-value < 0.05; ** p-value <0.01; *** p-value < 0.001. White bars, phage display-derived allergen-specific scFv clones; grey bars, Phl p 2-specific stereotyped clones; black bars, clones of unknown specificity. Numbers on top of the bars indicate clone counts.

We also examined the persistence of allergen-specific IgE+ clone members of any isotype in blood or nasal biopsies and detected scFv-derived and stereotyped clones more often at 2 and/or 3 timepoints compared to IgE+ clones of unknown specificity in SIT donors (Fig 3C). In non-SIT donors, stereotyped clones were also more often detected at more timepoints than IgE+ clones of unknown specificity (Fig 3D).

It is possible that IgE-expressing clones generated during SIT have different properties than clones that exist before the start of SIT. We thus examined IgE-expressing clone members observed at Day 0, Week 8, Year 1, and Year 3, and assessed what percentage of these clones had IgE-expressing members detected at other timepoints. Allergen-specific clones (scFv-derived and stereotyped) showed greater persistence compared to clones of unknown specificity. Of the IgE-expressing clones detected at Day 0 before SIT initiation, IgE-switched members from the scFv-derived lineages were more likely to persist after 3 years of SIT than IgE lineage of unknown specificity (Fig E3A). Stereotyped IgE-expressing clones of non-SIT donors also showed greater persistence than unknown specificity IgE clones (Fig E3B).

Inferred class-switch recombination (CSR) of allergen-specific IgE+ clones after 3 years of SIT

Generation of allergen-specific IgG antibodies during SIT is associated with successful desensitization to allergen after SIT²⁹ as well as with naturally acquired tolerance³⁰, potentially due to blocking of IgE binding to allergen. It is less clear whether the IgG-producing cells are members of the same clones as the IgE-producing cells, or if these are newly generated allergen-specific lineages that do not contain IgE. We assessed the frequency of non-IgE-expressing members of IgE+ clones in patients undergoing SIT. The frequency of IgG-switched members of Phl p 2-stereotyped IgE+ clones was higher than at baseline at all SIT timepoints, and elevated compared to IgG clone member frequencies for unknown specificity IgE+ clones (Fig 3E). Similar results were seen for IgG members of stereotyped IgE+ clones compared to unknown specificity IgE+ clones in donors who did not undergo SIT (Fig 3F), but IgG clone members were at lower frequencies than those seen in SIT participants. IgG-switched isotype frequencies were significantly greater in scFv-derived IgE+ clones compared to IgE+ of unknown specificity after 8 weeks of SIT, decreased slightly at Year 1, but were significantly enriched again at Year 3 (Fig 3E). IgG4 members of stereotyped clones or scFv clones were seen at Year 1 in both SIT and non-SIT participants, but were at Year 3 only detected in the SIT participants (Fig 3E, Fig 3F). Class-switching to IgA1 and IgA2 was infrequent in allergen-specific clones in both SIT and non-SIT donors, and was not observed in SIT donors until the third year of treatment.

SHM of allergen-specific IgE+ clones after 3 years of SIT

We and others have shown that IgE sequences from allergic individuals accumulate variable region SHM^{8, 31}, but it is less clear to what extent affinity maturation occurs in IgE-switched cells or in the B cells expressing other isotypes that class switch to IgE. IgE sequences from Phl p 2-stereotyped and scFv-derived allergen-specific clones were substantially hypermutated before the start of SIT, consistent with repeated exposure to seasonal and environmental allergens, and showed expected higher mutation frequencies in complementarity-determining regions 1 and 2 (CDRH1 and CDRH2) compared to framework regions (Fig 4A–B). SHM levels of IgE sequences of scFv-derived allergen specific clones were strikingly decreased at Week 8, then increased through Year 1 and 3. In contrast, stereotyped sequences were decreased during treatment and were significantly lower compared to the scFv-derived clones or clones of unknown specificity at Year 3.

FIG4.

Somatic hypermutation and antigen selection in IgE transcripts from allergen-specific clones in donors undergoing SIT. (A) Total IGHV SHM levels for IgE-switched members of allergen-specific scFv-derived (white boxes), stereotyped Phlp2-specific (light grey boxes) clones, and clones of unknown specificity (black boxes). Each point represents the mean of the clonal median SHM for a donor. (B) Frequency of mutations occurring per base in framework (FR1, FR2, FR3) and complementarity-determining (CDR1, CDR2) regions in IgE-expressing members of allergen-specific clones and clones of unknown binding specificity. Each point represents the mean of the IgE transcripts in a donor. (C) Evidence of antigen selection in allergen-specific IgE transcripts. The frequency of amino acid replacement mutations occurring in CDRH1 and CDRH2 as a proportion of total IGHV gene mutations (R_CDR/Mv) is overlaid on the 95% confidence interval (grey area) for an IGHV sequence with Mv mutations. The size of the filled circles indicates the number of observed reads for each plotted sequence. Red, phage-selected allergen-specific scFv sequences; blue, stereotyped Phl p 2-specific sequences; green, sequences of antibodies of unknown antigen specificity. (D) Percent of IgE transcripts with evidence of antigen selection by donor (points) and antigen-specificity (box color). Significance was determined by Wilcoxon rank sum test. * p-value < 0.05; ** p-value <0.01; *** p-value < 0.001.

We looked for evidence of antigen-driven selection in the amino acids encoded by the IGHV gene segment by examining the proportion of IGHV replacement mutations that occurred in CDRH1 and CDRH2³² of IgE transcripts (Fig 4C). Evidence of antigen selection was low in stereotyped clones compared to scFv-derived clones or clones of unknown specificity (Fig 4D). For scFv-derived clones, evidence of antigen selection dropped at Week 8, but thereafter increased to levels higher than for clones of unknown specificity, potentially reflecting a recruitment of new low mutation clones to the responding allergen-specific pool, or a generation of new low-mutation IgE members of clones. For 3 of 5 scFv-derived lineages (IT4–796269, IT6–369403, and IT7–801811), and 2 of 5 stereotyped Phl p 2-specific lineages (IT2–363211, IT2–443075), median IgE SHM levels decreased over time (Fig E4).

Increasing the specificity of a stereotyped CDRH3 Phl p 2-binding motif

Although in levels lower than for Phl p 2-seropositive donors, some seronegative individuals expressed stereotyped IgE sequences with features previously associated with Phl p 2-specificity (using IGHV4-31 or IGHV4-30-4, with 9 or 10 amino acid CDRH3 and glycine as the fifth CDRH3 residue) (Fig 1B). We hypothesized that the non-seroconverted participants’ stereotyped sequences were not specific for Phl p 2, and attempted to identify sequence features distinguishing these sequences from those found in the Phl p 2-seropositive participants. We compared CDRH3s of stereotyped IgE from seropositive individuals to stereotyped non-IgE sequences of seronegative individuals and stereotyped IgM sequences of another dataset³³. We trained a random forest and a decision tree classifier using the properties of CDRH3 residues and found that these could distinguish between putative Phl p 2-specific and non-specific stereotyped sequences (Fig 5A). Analysis of feature contribution (Fig 5B) suggested that residues 108 and 113, both adjacent to the conserved glycine, were the main positions used to distinguish between the two groups, and the decision tree (Fig 5C) classified any sequence expressing N/D108 or W/Y113 as presumably Phl p 2-specific. We extended our analysis of the stereotyped CDRH3 motifs using the KpLogo³⁴ tool, which allows for the detection and visualization of position-specific ultra-short sequence motifs. Seven of the 14 significantly enriched CDRH3 motifs employed either D108 or Y113 as part of the motif (Fig E5A). The frequencies of amino acids residues at position 108 and 113 in stereotyped IgE transcripts of seropositive subjects in this study, and in the bone marrow of Phl p 2 IgE seropositive subjects in another study²⁸, confirmed that N108 and/or D108 and Y113 were significantly higher than in similarly stereotyped sequences of naïve B cells (Fig 5D). IgG and IgA-encoding transcriptomes of seronegative subjects of the present study were not enriched for D108, N108, or Y113 (Fig 5D). Analysis of the occurrence of Phl p 2-stereotyped sequences with the N/DGY motif showed that such sequences were present in high levels in IgE repertoires of seropositive donors (Fig 5E), but very rare in Phl p 2-seronegative individuals and in IgM, IgD, and IgA repertoires of all donors (Fig E5B).

FIG5.

Redefinition of a stereotyped Phl p 2-binding motif. Performance metrics (A) and feature contribution (B) of a random forest model and a decision tree model (C), which had been trained to distinguish between stereotyped IgE antibodies of seropositive individuals and antibodies with, presumably, a large diversity of specificities, based on properties of their CDRH3 amino acids. Non-IgE sequences of this study, and IgM sequences (from naïve B cells) of another study³³ were used as the Phl p 2-negative sequence pool. (D) Positional analysis of CDRH3 amino acids of stereotyped sequences from the IgM transcripts of naïve B cells of another study³³ (far-left bars), IgG/IgA transcripts from Phl p 2-seronegative donors of this study (left bars), and IgE transcripts from Phl p 2-seropositive donors of this study (right bars) and another study²⁸ (far-right bars). Highlighted residues at positions 108 and 113 were significantly enriched in IgE sequences from seropositive subjects, compared to both IgM and IgG/IgA transcripts (**** P-value < 0.0001). (E) Mean frequency, based on all samples of an individual, of heavy chain IgG (left panel) and IgE (right panel) sequences expressing both characters previously associated with Phl p 2-stereotyped rearrangements, and the herein described N/D108 and W113, in Phl p 2-seropositive (grey) and Phl p 2-seronegative individuals (black). Frequencies in other populations are shown in Fig E4B.

In Phl p 2-seropositive subjects 3, 11, and 16, IGHD5–24 was the most commonly used IGHD gene in stereotyped IgE (Fig 6A), indicating a potential genetic basis for the enrichment of the N/DGY motif, as this gene can encode a complete DGY-motif in reading frame 3 (Fig 6B). Analysis of templated and junctional nucleotides as inferred by IgBLAST indicated that N/D108, Y113, and/or the invariant G109 in several cases were IGHD-gene encoded (Fig 6B). Mapping of the stereotyped CDRH3 residues onto a solved structure of a human IgE Fab-Phl p 2 complex (PDB: 2VXQ)²⁵ showed that residues D108, G109, and Y113 were in close proximity to Phl p 2 (Fig 6C–E), supporting their importance for generation of the antigen binding site.

FIG6.

Structural validation of predicted Phl p 2 contact residues on stereotyped IgE sequences. (A) IGHD gene usage by subject in stereotyped IgE transcripts with 108N/D and 113Y in Phl p 2-seropositive donors. Increased color opacity indicates higher frequency by unique read counts. (B) Inferred contribution of gene-templated (black) and non-templated (orange) bases to stereotyped amino acid CDRH3 motifs. Nucleotides encoded by the IGHD gene are shown in purple. (C) Amino acid sequences of variable gene rearrangements of stereotyped IgE and IgG sequences, and an IgM sequence that does not contain the stereotyped CDRH3 motif (bottom sequence in alignment). Sequences are aligned against germline IGHV4-31*03. Letters indicate disagreements with the reference germline gene. Stereotyped CDRH3 residues Asp108, Gly109, and Try113 are indicated by filled circles. The flanking cysteine at position 104 is marked for reference. (D) Interaction of stereotyped CDRH3 residues with antigen (grey) in structure 2VXQ. CDRH3: green. The Asp108, Gly109, and Tyr113 motif is highlighted as stick representations. A lysine residue (above) and an arginine residue (below) that stack against Tyr113 are coloured (carbon: cyan; nitrogen: blue; oxygen: red). (E) 2VXQ structure. Antigen (yellow) on top. Complementarity-determining regions of the heavy chain are coloured: CDRH1: orange; CDRH2: blue; CDRH3: green.

DISCUSSION

Here, by combining combinatorial antibody phage technology and NGS, we report on changes in rare aeroallergen-specific IgE clonal lineages of allergic patients after 3 years of SIT. We find that IgE members of clonal lineages present in allergic individuals at the start of treatment may persist even after years of immunotherapy. This result is consistent with studies that demonstrate persistence of serum antibody epitope specificity profiles over time in subjects undergoing SIT^{35, 36}. We also find that aeroallergen-specific IgE-containing clones in individuals undergoing SIT are enriched for containing IgG members, similar to the results of a recent study of 40 adults with seasonal allergic rhinitis who underwent sublingual immunotherapy for 12 months¹⁴. After 3 years of SIT, the frequency of IgG members was significantly higher in both scFv-derived and stereotyped allergen-specific IgE clones than in IgE+ clones of unknown specificities, supporting a mechanistic role for IgG-IgE competition for allergen binding in SIT desensitization³⁷. We note that these clonal relationships between IgE and IgG subtype expressing B cells could reflect switching either from IgG-expressing clones or from other members of the same clone expressing an isotype located upstream of IgG subtypes and IgE, such as IgM. For IgE clonal lineages carrying a previously described Phl p 2-stereotyped rearrangement, we find that the maturation profiles differ from other aeroallergen-specific IgE clonal in several aspects, including clonal localization, and CSR and SHM profiles. The reported findings are based on heavy chain transcripts rather than paired heavy/light chain transcript data and allergen-specific clones were identified from phage libraries with non-native heavy/light chain pairing. Considering the high impact of heavy chains, in particular their CDR3s, on antibody specificity³⁸ and the relatively constrained diversity of light chains³⁹, we however envisage that heavy chain sequence data capture critical features required for allergen-specific binding. We also expect that our scFv phage display analysis will not have captured all allergen-specific IgE-expressing clonotypes present in the patient samples, for example, those that require correct pairing of heavy and light chains with rarely used V genes or unusual CDR3 characteristics to form the allergen-specific paratope.

Maintenance of secreted allergen-specific IgE may occur by the ongoing production of IgE+ antibody secreting cells via stimulation of allergen-specific memory B cells expressing IgE or other isotypes^{40, 41}, or by continuous production of IgE antibodies by long-lived plasma cells residing in tissues^{15, 22, 42, 43}. The nasal mucosa is rich in IgE-expressing B cells^{44, 45}, DNA circles excised from the genome during IgE class switching can be detected in the nasal mucosa of allergic patients¹⁹, and exposure of the nasal mucosa to allergen leads to an increase of allergen-specific IgE levels in the blood⁴⁶. We found scFv-derived allergen-specific IgE+ clones at higher frequency in blood and nasal biopsies than IgE+ clones of unknown specificity after 1 year of SIT, but by 3 years, only the scFv-derived IgE+ clones found in both tissues or only in biopsy remained at elevated frequency. This suggests greater persistence of the IgE+ lineages in tissue sites such as the nasal mucosa, either due to the presence of long-lived clone members, or from generation of new clone members either locally or at distal sites followed by trafficking to the nasal mucosa. Patients not undergoing SIT also had persistent IgE+ B cell lineages in nasal mucosa and blood, suggesting that not only SIT, but seasonal or chronic exposure to environmental allergens may contribute to IgE+ lineage persistence, or alternatively, that IgE+ lineage persistence can be independent of allergen exposure. The current analysis did not assess clinical outcomes in participants of this study, however, persistence of IgE-expressing allergen-specific clones in patients who have undergone 3 years of SIT suggests that further treatment might be necessary in these patients. SIT treatment courses are often continued for up to 5 years. Study of patients over a longer treatment duration could address whether IgE-expressing allergen-specific clone frequencies decrease further over longer time periods, and could assess correlations with treatment response and resolution of symptoms of allergy. This monitoring approach could also be used to compare correlates of treatment efficacy for different immunotherapy modalities, such as SCIT vs. SLIT.

Exposure to grass pollen has been shown to transiently increase the SHM rates of IgE sequences detected in blood and nasal biopsies of allergic patients³¹ and lead to increased allergen-specific IgE protein concentrations in blood⁴⁶. In contrast, we observed a temporary decrease in allergen-specific IgE+ heavy chain SHM during SIT up-dosing at Week 8, followed by increased SHM frequencies at Year 1 and 3 in scFv-derived clones, but not in the stereotyped Phl p 2 binding clones. Given that daughter cells inherit SHM changes accumulated in clone members that they divided from, these data indicate that new or lower-SHM IgE+ clone members are recruited during the initial weeks of SIT, but that subsequent maintenance SIT dosing is associated with increasing SHM in allergen-binding scFv-derived clones. The stereotyped Phl p 2 IgE+ clones did not show higher SHM at Year 1 or 3, potentially indicating that these clones may turn over more rapidly with new low-SHM clones replacing former high-SHM clones, or persistence of low-SHM members of the clones seen at Week 8.

Stereotypic antibody rearrangements are increasingly being recognized as important contributors to the immune response against pathogens as well as in autoimmunity and allergy^{3, 15, 47–50}. We now extend the definition of a stereotyped antibody heavy chain specific for grass pollen allergen Phl p 2, and propose that a N/D108-G109-Y113 motif in the 10 amino-acid long CDRH3, and possibly also in 9-residue long CDRH3, together with multiple polar and apolar interactions with Phl p 2 by germline-encoded tyrosine residues in IGHV4-31, form the basis for Phl p 2-specific binding by these antibodies. D108 and Y109 were highly enriched in these IgE-encoding transcriptomes, but other residues surrounding the conserved G109 were also seen here and previously reported²⁴. The N/D108-G109-Y113 residues, or substantial parts thereof, may be encoded by IGHD gene segments, increasing the motif’s likelihood of being observed. Compared to other IgE+ clones, there was relatively little evidence for antigen-selection in the stereotyped IgE sequences, suggesting that even the low-mutated V gene residues, together with the conserved N/DGY motif, may bind with sufficient affinity to the allergen. Consistent with this, the stereotyped response has been strictly associated with the highly similar IGHV4-30-4 and IGHV4-31 genes. However, other solutions to Phl p 2-specificities exist, as evidenced by the Phl p 2-binding IgE detected in subject 10 of this study, who lacks IGHV4-30-4 and IGHV4-31 genes. Such germline IGHV gene deletions are not uncommon^{33, 51}, and we hypothesize that they may affect the nature of the immune response to Phl p 2. One of the most abundant IgE clones of donor 10 is derived from a rearrangement involving IGHV4-34 and expresses a D108-G109-W113 motif. This may represent an alternative approach to form stereotyped Phl p 2-specific antibodies when neither IGHV4-30-4 or IGHV4-31 are available, a hypothesis that needs further evaluation. We envisage that as the list of identified, stereotypic allergen-specific immunoglobulin sequences increases^{3, 7, 15}, they may form a new class of biomarkers for allergen-specific immunity that could aid in, for example, patient and treatment selection, or that could facilitate the development of new allergy-vaccine candidates that further support the development of protective antibodies that may antagonize clonotypes that make up a substantial fraction of the allergen-specific IgE population⁵².

Key messages.

• Recombinant phage-displayed antibody scFv libraries generated from IgE, IgG and IgA sequences from patients undergoing specific immunotherapy for aeroallergies were screened to identify allergen-specific antibody sequences

• Tracking of allergen-specific IgE+ clones using immunoglobulin repertoire sequencing demonstrates the persistence of clones in blood and nasal mucosa, progressive class-switch recombination, and a distinct maturation of a stereotyped Phl p 2-clonotype.

Abbreviations:

CDR

complementarity-determining region

CSR

class-switch recombination

scFv

single-chain variable fragment

SHM

somatic hypermutation

SIT

specific immunotherapy