Plasmablasts During Acute Dengue Infection Represent a Small Subset of a Broader Virus-specific Memory B Cell Pool

Abstract

Dengue is endemic in tropical countries worldwide and the four dengue virus serotypes often co-circulate. Infection with one serotype results in high titers of cross-reactive antibodies produced by plasmablasts, protecting temporarily against all serotypes, but impairing protective immunity in subsequent infections. To understand the development of these plasmablasts, we analyzed virus-specific B cell properties in patients during acute disease and at convalescence. Plasmablasts were unrelated to classical memory cells expanding in the blood during early recovery. We propose that only a small subset of memory B cells is activated as plasmablasts during repeat infection and that plasmablast responses are not representative of the memory B cell repertoire after dengue infection.

Keywords: Antibodies; Dengue; Memory B cells; Plasmablasts.

Fig. 1

Study setup and time points of sample collection. Dengue patients with a primary or secondary infection were enrolled into the study between 14 and 71 h after onset of fever. The same patients were recalled 4–7 days and 15–166 days after onset of fever. Each patient is color coded. Samples from patients with a black box were used for single B cell sorting, sequencing and Ab expression. Samples from patients with a grey-shaded box were used for pooled B cell sorting and 454 sequencing.

Fig. 2

B cell repertoires change between acute phase and early convalescence. A and B) Flow cytometry plots representative of a dengue patient (Patient 3) to illustrate the sorting strategy. A) Sorting of CD19⁺, CD20⁻, CD27^hi, CD38^hi plasmablasts. B) Sorting of CD19⁺, CD20⁺, CD27⁺ DENV-binding memory B cells (MBC) (indicated in blue squares) and non-DENV-binding MBCs (indicated in green square). Sequences from DENV-2 and DENV-3-binding MBCs were pooled as “DENV-binding MBCs”. C) VH gene family usage of plasmablasts (PB), DENV-binding MBCs, and non DENV-binding MBCs. The total numbers of unique CDR3 sequences per sample is indicated in the center of pie charts. Analysis included IgG sequences, IgM sequences or total sequences, as indicated next to the charts. Cells from patients 3 and 8 were analyzed with 454 sequencing; for all other patients, single B cells were analyzed by Sanger sequencing. A Wilcoxon matched-pairs signed rank test showed a statistically significant difference only for VH3 and VH4 usage between PB and MBC-neg cells (p = 0.03).

Fig. 3

Few shared sequences between plasmablasts and memory B cells. A) Identical unique CDR3 Sequences shared between plasmablasts (PB), DENV-2-binding MBCs (MBC-DENV) and non-binding MBCs (MBC-neg) for two patients with secondary DENV-2 infection (E1291, E1392) and two patients with primary DENV-2 infection E1414, E1465. B) Numbers of mapped reads for sequences shared between PBs MBC-DENV and MBC-neg. C) number of mapped reads and isotype for PB and MBC-DENV shared sequences that are at least 10 × expanded in the memory compartment. D) Shared clones based on at least 85% identity in the CDR3 amino acid sequence. E) CDR3 mutation analysis for the three clones that were expanded and mutated in the MBC-DENV repertoire. The MBC-DENV CDR3 sequence in bold in the x-axis represents the expanded clone according to the number of mapped reads. The CDR3 amino acid sequence and mutation is indicated above the graphs, with bold characters indicating the dominant sequence. F) Isotype distribution of PB, MBC-DENV and MBC-neg sequences. Total mapped reads was used for the analysis.

Fig. 4

Specificity of antibodies derived from plasmablasts and memory B cells after secondary infection is markedly different. A) Overview of the assays used to assess the specificity of single B cell-derived monoclonal Abs. Abs were characterized into four groups: EDIII-specific, E-specific, prM-specific, and complex epitope specific antibodies. B) DENV-binding Abs as fractions of all Abs tested per patient and per cell type. The average of DENV-binding Abs across all patients is illustrated for PB-derived Abs and (MBC)-derived antibodies, with the total number of Abs tested indicated in the center of the pie charts. C) Percentages of Abs binding to complex epitopes, recombinant E protein (Rec. E), or prM were defined for plasmablasts (PB) from patients 1 to 7. All E-specific Abs of patient 6 bound to EDIII, whereas no EDIII-specific antibodies were observed for all other patients and time points analyzed. Binding specificity was defined for memory B cells (MBCs) from patients 1 to 4. One MBC-derived Ab could not be characterized (black bar in Fig. 2C, due to low expression levels (≤ 0.1 μg/ml).

Fig. 5

E-specific antibodies have a higher neutralizing capacity compared to complex-epitope-specific antibodies. A) Neutralizing capacity of 51 mAbs representing the different antibody groups for plasmablasts and MBCs. The patient origin and mAb ID is indicated on the left. The NT50 for PB-derived mAbs was determined for DENV-1,-2 and -3. DENV-4 was not tested since this serotype is very rare in Singapore. NT50 for the current and the previous serotype of infection (based on the neutralizing profile of the patient plasma) were tested for MBC-derived mAbs. Each NT50 value represents the mean from three independent experiments. B) Virus particle (UV-DENV) ELISA for complex-epitope-specific mAbs. C) Cross-reactivity of antibodies per patient and cell type, assessed by UV-inactivated PEG precipitated virus particle ELISA, rec.E ELISA, immunohistochemistry and western blot assays for DENV-1 to -4. Abs were grouped based on their ability to bind across all four, three, two or one serotype(s) of DENV in at least one or multiple assays.

Fig. 6

DENV-binding antibodies derived from PB and memory B cells use distinct VH genes but show similar mutation rates. A–B) VH gene family usage for epitope-specific antibodies derived from plasmablasts (PB) (A) and memory B cells (MBCs) (B). Ab specificity is indicated with different colors, and the numbers of Abs per gene family is indicated on the y axes. Significant differences between PB and MBCs were calculated with a Fisher's Exact test: prM: p = 1.00, Complex epitope: p = 0.298, Rec.E: p = 0.022. C) Comparison of the CDR3 length of all plasmablasts (primary and secondary infection; n = 603) with the CDR3 length of DENV-binding MBCs (n = 245) and DENV-non-binding MBCs (n = 163). P < 0.0001 (one-way ANOVA with Tukey's multiple comparison test). D) CDR3 lengths of plasmablast (PB) and MBC-derived Abs across Ab specificities. E) Content of neutral non-polar amino acids (AA) in CDR3 sequences from prM-specific Abs compared to CDR3 sequences from rec. E-specific Abs. F) VH nucleotide mutation frequencies of plasmablast-derived Abs and MBC-derived Abs of different specificity. A one-way ANOVA analysis comparing all five groups of Abs (both plasmablast and MBC-derived) showed no significant differences between plasmablasts and MBCs. G) Number of CDR3 nucleotide (N) additions for all Ab groups. For scatter plots each dot represents one Ab and horizontal lines indicate the means. Whiskers in the box and whisker graphs indicate min to max values.