key: cord-0277861-g8cxx4dd
authors: Szelinski, F.; Stefanski, A.-L.; Wiedemann, A.; Schrezenmeier, E.; Rincon-Arevalo, H.; Reiter, K.; Lettau, M.; Dang, V. D.; Fuchs, S.; Frei, A. P.; Alexander, T.; Lino, A. C.; Dörner, T.
title: Antigen-experienced CXCR5- CD19low B cells are plasmablast precursors expanded in SLE
date: 2021-05-27
journal: nan
DOI: 10.1101/2021.05.25.21257784
sha: 65b4f731e8a60f33176d42dcc8d1e6090533a095
doc_id: 277861
cord_uid: g8cxx4dd

B cells play a critical role in the pathogenesis of systemic lupus erythematosus (SLE). We analysed two independent cohorts of healthy donors and SLE patients using a combined approach of flow and mass cytometry. We have found that IgD-CD27+ switched and atypical IgD-CD27- memory B cells, which are increased in SLE, represent heterogeneous populations composed each of three different subsets, such as CXCR5+CD19int, CXCR5-CD19high and CXCR5-CD19low. Here, we characterize a hitherto unknown antigen-experienced CXCR5-CD19low B cell subsets enhanced in SLE and carrying a plasmablast (PB) phenotype enriched for switched immunoglobulins, and expressing CD38, CD95, CD71, PRDM1, XBP-1, and IRF4. CXCR5-CD19low resemble activated B cells with a characteristically diminished B cell receptor responsiveness. CXCR5-CD19low B cells increased with PB frequencies in SLE and upon BNT162b2 vaccination suggesting their interrelationship. Our data suggest that CXCR5-CD19low B cells are precursors of plasmablasts, thus co-targeting this subset may have therapeutic value in SLE.

More recently, it became evident that DN B cells represent a heterogeneous subset, 37 including age-/autoimmune-associated B cells (ABCs) 11, 12, 13 , Syk ++ B cells 14 and 38 double negative 2 (DN2: IgD -CD27 -CXCR5 -CD11c + ) populations 15 . All these subsets 39 share partially overlapping characteristics and are linked to disease activity and 40 autoantibody formation. Besides these alterations of B cell subsets residing among 41 DN B cells, impaired chemokine receptor expression 16 and reduced responsiveness 42 upon BCR stimulation 17, 18, 19 have been reported for these B cells in patients with 43 SLE suggesting their distinct involvement in chronic autoimmunity. 44

This study further delineates the heterogeneity of switched memory (mem) and DN 45 memory B cells in healthy donors and SLE patients. Using a combined flow and 46 mass cytometry approach, we identified an enhanced CXCR5 -CD19 low B cell subset 47 in the switched memory and DN compartments (mem low and DN low ) in SLE carrying 48 (excluding CD24, CXCR5 but including CD22) and equilibrated for 1h at 37°C in 113 RPMI before stimulation with IgG/IgA/IgM (H+L) F(ab') 2 (15 µg/ml) for 5, 8 or 15 min. 114 For baseline, control cells were incubated for 5 min with RPMI. Adding pre-warmed 115 Lyse/Fix buffer (BD) at the respective time-points stopped the stimulation. After 116 washing Phosflow Perm II Buffer (BD) was added to permeabilize cells overnight at -117 20°C. Cells were then stained for CD24, CXCR5 and the intracellular targets Syk and 118 pSyk (Y352). To investigate BCR response and phosphorylation kinetics, median FIs 119 of Syk, pSyk (Y352) and CD22 were determined. 120

Gene expression analysis of B cell subsets. Isolated PBMCs were enriched for B 121 cells by depleting CD3, CD14 and CD235a via MACS microbeads (Miltenyi Biotec) 122 according to the manufacturer's instruction. Cells were stained with CD19, CD20, 123 CD27, CD38, CD3, CD14, CXCR5 and IgD for sorting. Naïve, pre-switched, total 124 memory and PBs as well as DN int , DN low and DN high subsets were sorted and purity 125 check was performed using Sony Sorter MA900. Cells were counted using 126 MACSquant (Miltenyi Biotec). Cell suspension were spin down, resuspended in HTG 127 lysis buffer at a concentration of 200 cells/µl and stored at -80 °C until further 128 processing by HTG (Tucson, AZ). Samples with less than 28% of counts allocated to 129 positive controls probes, read depth above 750,000 and expression variability above 130 0,094 were analyzed. 131 We found that DN low B cells are significantly increased among CD19 + B cells in SLE 156 compared to HD (Fig. 1A) . While only mem high B cells were increased in the switched 157 memory compartment, all of the three subsets within the DN fraction were 158 significantly increased in SLE patients (Fig. 1B) . DN B cells were enriched for 159 CXCR5 negative CD19 low and CD19 high B cell subsets compared to the CD27 + (mem) 160 compartment and in general remarkably expanded in SLE ( Fig. 1 C) . Next, we 161 applied the dimension reduction algorithm UMAP 21 to cluster IgD -B cells. Using this 162 approach, we identified mem/DN CD19 int , mem/DN high and mem/DN low as distinct 163 populations (Fig. 1D ). In this UMAP, both CD19 low populations clustered together with 164 CD27 ++ CD38 ++ PB (Fig. 1D) . Comparison of the clusters obtained from HD with SLE 165 patients revealed an increased density of the corresponding subsets in SLE (Fig.  166 1E), consistent with their significant expansion in this condition. 167

Next, we characterized DN int , DN high and DN ow B cell subsets for the expression of 169 several surface molecules including lineage, differentiation and activation markers. 170

The resulting expression patterns of CD27, CD19, CXCR5, CD24, CD71, CD95, 171 CD38 and CD11c is visualized by colour code in the dimension reduced UMAP (Fig. 172 2A) . With this analysis, we found that the subsets defined by their distinct CD19 and 173 CXCR5 expression allowed further differentiation by different expression profiles of 174 CD24, CD71, CD95, CD38 and CD11c among the IgD -CD27subsets ( Fig. 2A) . 175

We found that CD24, a marker with a dynamic expression pattern throughout B cell 176 maturation and absent in antibody producing cells 23 , was not present in both subsets 177 with low CD19 and lacking CXCR5 expression (mem/DN low ) in contrast to the DN int 178 population. 179

Besides CD19 expression the main discriminators between the DN low and DN high 180 populations were CD38 and CD11c ( Fig. 2A Since we saw differences in the frequencies we also checked for qualitative 204 differences between HD and SLE. When comparing expression profiles of the three 205 DN B cell subsets between HD and SLE patients, we found that expression levels of 206 proliferation marker CD71 and frequencies of activation markers CD95 + and CD38 + 207 cells were increased within the DN int ( Fig. 2G ) but not the mem int (Supplementary 208 Fig.1D ) population of SLE patients. The expression of CD71 was enhanced in both 209 DN low and mem low in SLE (Fig. 2G, Supplementary Fig. 1D ). Regarding Ig isotype 210 expression, we found that mem/DN int and mem/DN high mainly express IgG while 211 mem/DN low express IgG and IgA to a similar extend. (Fig. 2H) . 212

profile 214

Next, we validated our findings in an independent cohort of 27 SLE patients and 18 215 healthy donors using CyTOF including expression of activation markers and 216 checkpoint molecules (Fig. 3, Supplementary Fig. 2 ). Initially, we confirmed the 217 increased frequency of all three subsets within the DN compartment in SLE patients 218 as shown by the UMAP plot by increased DN int , DN high and DN low clusters (Fig. 3A) . 219

Besides the significantly increased DN subsets in patients, mem low B cells were also 220 substantially increased in SLE (Fig. 3B Fig. 2 C) . When comparing the 236 expression in SLE, we found that CD45RA expression was reduced on DN high cells 237 from patients in comparison with HD ( Supplementary Fig. 2 D) . 238

Co-stimulatory and co-inhibitory immune checkpoints (CPMs) regulate and modulate 239 immune cells and play an important role in fine tuning the immune response 20, 28, 29, 30, 240 31 . Therefore, we analyzed the expression profiles of various CPMs among the 241 subsets of interest. Of particular note, mem/DN high upregulated immune checkpoint 242 molecules, such as BTLA, VISTA and CTLA-4 ( Since mem low and DN low are increased in SLE and correlated with PB, we asked 253 whether an acute immune response resulting in PB formation would be accompanied 254 by alterations and expansion of those subsets in HD. Therefore, we monitored B cell 255 subsets in HD on day 0, 7, 14 and 21 after single-dose administration of BNT162b2 256 vaccine and 7 days after boost (Fig. 4A ). Of particular note, this vaccine is able to 257 elicit a striking T cell dependent immune activation 32 . As a result, the frequencies of B 258 cell subsets did not change significantly except reduction of mem int and DN int on day 259 21 after vaccination compared to day0 and day 7, respectively. Although not 260 significant, a trend of increased plasmablast formation was detected at day 7 after 261 boost (Fig. 4A ). Of particular note, frequencies of mem low B cells correlated strikingly 262 with PB on day 21 and 7 days after boost. DN low B cells also showed a correlation 263 with PB 7 days after boost (Fig. 4B ). These findings supported further that mem low 264 and DN low expansion follows kinetics of PB induction and depends on T cell 265 instruction as evidenced by BNT162b2 vaccination. 266

To evaluate B cell receptor responsiveness of newly identified subsets as a read-out 268 for their functional competence, we studied phosphorylation kinetics of Syk (Y352) 269 upon anti-BCR stimulation (Fig. 5A was an overall lower phosphorylation in B cells from SLE patients, both mem/DN low 276 subsets from SLE patients showed a BCR response kinetics as found for 277 CD27 ++ CD38 ++ plasmablasts (Fig. 5A) . 278

Subsequently, we evaluated whether differences in Syk protein levels in steady state 279 may account for this difference between subsets and SLE vs HD (Fig. 5B ). We found 280 that basal Syk levels were highest in mem/DN high . SLE patients showed significantly 281 decreased Syk levels in mem low B cells. Interestingly, Syk expression was 282 comparable in mem/DN low subsets compared with mem/DN int cells (Fig. 5A ,B). 283

To further understand the distinct nature of the analyzed B cell subsets, 286 transcriptome analysis was performed on naive, pre-switched, total memory (IgD -287 CD27 + ), DN int , DN high , DN low and PBs of patients with SLE and HD (Fig. 6) . 288

Transcripts of IRF4 a transcription factor crucial for differentiation 33 and survival 34 of 289 plasmablast and plasma cell was not only upregulated in plasmablast but also 290 increased in DN low cells. IRF4 is known to regulate Blimp-1 (encoded by PRDM1) 291 expression a regulator of plasma cell differentiation 35 Herein, we identified two novel CXCR5 -CD19 low populations, mem low and DN low , 301 residing within conventional switched memory and IgD -CD27atypical memory B 302 cells, respectively. Additionally, we found a not yet described CXCR5 -CD19 high 303 population in switched memory B cells (mem high ) which shared characteristics, such 304 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. corresponds to the DN high population which is further characterized here and has 306 been described as precursor of antibody secreting plasmablasts generated by 307 extrafollicular activation, and found to be increased increased in autoimmune 308 conditions such as SLE 15 but also acute viral infections such as SARS-CoV-2 37 . The 309 expression profile of CD19 high , CD38 -, CD95 + and Ki-67together with the high BCR 310 responsiveness and increased Syk expression also suggest that mem/DN high 311 represent the previously described Syk high population 14 . 312

Although various groups reported an overall reduction of CD19 expression on B cells 313 in SLE, a specific CD19 low population has not been characterized so far. In the study 314 by Culton et al. 38 , the SLE cohort was subdivided into CD19 lo and CD19 hi patients 315 based on global CD19 expression and the presence or lack of a CD19 hi B cell 316 population. Autoantibodies were detected in both patient groups. The majority of the 317 CD19 low B cells were described as IgD + , CD38 + , and CD27 − . Additionally, a decrease 318 in CD19 expression on B cells was reported in anti-neutrophil cytoplasmic 319 autoantibody associated small vessel vasculitis (ANCA-SVV) patients, suggesting 320 that downregulation of CD19 might be a common feature of antibody driven 321 disease 38 . Other studies reported lower CD19 expression in CD27but also in CD27 + 322 B cells 39, 40 . These studies did not discriminate populations based on their IgD 323 expression. Overall decrease of CD19 expression was seen in both active 39 and 324 quiescent 40 patients with SLE and also in patients with ANCA-SVV, suggesting 325 variation in CD19 expression as an intrinsic abnormality linked to autoimmunity rather 326 than driven by antigen specificity or disease severity. 327

We found that CD19 low B cell subsets, mem low and DN low expressed co-stimulatory 328 molecule and activation marker CD86, proliferation marker CD71 41 and the majority 329 were CD38 + and CD95 + while early B cell stage marker CD24 42 and CD10 43 were 330 absent. In combination with surface expression of class switched immunoglobulins 331

IgG and IgA allows the conclusion that mem/DN low B cells were antigen experienced. 332

Although, Syk itself was not reduced, phosphorylation kinetics of Syk upon anti-BCR 333 stimulation were lower in DN low , similarly to PB. This could be caused either by an 334 anergic post-activation phenotype similar to the one seen in general naïve and 335 memory B cells of patients with autoimmune conditions like SLE, RA and pSS 18 . An 336 alternative explanation would be the downregulation of the BCR including BCR-337 associated surface molecules like negative regulator BTLA which we found 338 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The features of the three subsets were comparable between corresponding subset in 359 IgD -CD27 + switched and IgD -CD27atypical memory B cells supporting the idea that 360 the increase of DN B cells in chronic immune conditions can also be largely related to 361 loss of CD27 expression 45 which is supported by their comparable increase in SLE 362 and in particular by lack of CXCR5 expression. The latter can result from their a) 363 post-GC b) extrafollicular or c) activation status also known to be related with CD27 364 shedding including increased solubleCD27 in autoimmune diseases 45, 46 . 365 It is widely known that B cell homeostasis is altered in patients with SLE and B cell 366 targeted interventions are promising in this disease. However, the heterogeneity of 367 the switched memory and DN populations is only partially understood, and it remains 368 unclear how it is induced and/or maintained or how it contributes to the course of the 369 disease. Using CD19 and CXCR5 clearly allows the differentiation of switched 370 memory and DN B cells not only into three distinct subsets each but also suggest 371 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Collectively, the data presented here, including surface marker expression, 387 correlation analysis, BCR kinetics and transcription analysis strongly indicate that 388 mem/DN low cells are precursors of PB and directly contribute to plasmacytosis upon 389 immune activation. These mem/DN low cells reflect a subset of pre-plasma cells that 390 may not require to undergo full or incomplete memory B cell differentiation. 391

Understanding of the involved selection mechanisms will be important not only in 392 terms of their immunobiological features but also for potential treatment strategies. In 393 this regard, the current data suggest that there could be selective treatment 394 approaches not only for certain B cell subsets but also distinct PB/PC compartments, 395

including the possibility to leave protective PB/PC untouched. 396 397 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Rheumatology 66, 3424-3435 (2014) . 479 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. International Immunology 25, 129-137 (2012) . 521 522 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (IgD + CD27 -), pre-switched (IgD + CD27 + ) and plasmablast (PB). Red indicating 643 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2021. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

Disturbed peripheral B lymphocyte homeostasis in 429 systemic lupus erythematosus

Plasmablasts With a Mucosal Phenotype Contribute to 432 Plasmacytosis in Systemic Lupus Erythematosus

Distinct Effector B Cells Induced by Unregulated Toll-like 481

Receptor 7 Contribute to Pathogenic Responses in Systemic Lupus 482 Erythematosus

Type I and II interferons commit to abnormal expression 485 of chemokine receptor on B cells in patients with systemic lupus 486 erythematosus

Signaling and Promotes Survival of Human Lupus B Cells

Identification and Characterization of Post-activated B 494 Cells in Systemic Autoimmune Diseases

Programmed Death Ligand 1 Up-Regulation Capacity of Post-Activated Lupus 498 B Cells

Deep Phenotyping of CD11c+ B Cells in Systemic 501 Autoimmunity and Controls

Uniform Manifold Approximation and 504 Projection for Dimension Reduction

Activated memory B cell subsets correlate with disease 507 activity in systemic lupus erythematosus: delineation by expression of CD27