Differential effects of anti-CD20 therapy on CD4 and CD8 T cells and implication of CD20-expressing CD8 T cells in MS disease activity

Significance CD20-expressing CD4+ and CD8+ T cells harbor proinflammatory and central nervous system (CNS)-homing attributes. The inverse relationship between levels of these cells (particularly the CD20dimCD8+ T cells) in the circulation of MS patients, with active and impending CNS inflammation, suggests that these cells participate early on in the cellular immune responses involved in relapse development. The differential effects of anti-CD20 treatment on CD4+ and CD8+ T cell subsets point to different contributions of direct removal of CD20-expressing T cells, as well as indirect effects likely reflecting the removal of B cells that alters in vivo T cell:B cell interactions.


Blood sample processing
In the Discovery cohort, PBMCs were processed using standard operating procedures for all steps of blood procurement, PBMC isolation (by density gradient centrifugation using Ficoll-Paque Plus; GE Healthcare, Little Chalfont, UK), cryopreservation and storage in liquid nitrogen vapor phase, as previously described (1). In the validation cohort, whole blood cell counts were measured using 6-color TBNK Reagent with Trucount tubes (BD, San Diego, CA) and PBMC were separated by density gradient centrifugation using LymphoPrep tubes (Axis Shield, Dunde, UK), frozen and stored in CTL-Cryo ABC Kit Freeze Medium (Cellular Technologies Limited, Shaker Heights, OH) in liquid nitrogen vapor phase by a certified central laboratory (Covance, Indianapolis, IN), with PBMC subsequently sent on dry ice for batched analyses at Penn using strict standardized operating procedures.

Flow cytometric and fluid marker analyses
Thawed PBMC were suspended in serum-free X-VIVO 10 media (Lonza, Basel, Switzerland) and subject to surface staining and intracellular cytokine staining which were standardized for cryopreserved PBMC to characterize the phenotype and functional profiles of immune subsets using antibodies and reagents as shown in Fig. S1 and S2, and Tables S4-6. For surface staining, cells were rested at 37ºC for 4 hours, labeled with Live/Dead Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Scientific, Waltham, MA), stained with antibodies at the optimal concentrations together with Brilliant Stain Buffer (BD Biosciences, Franklin Lakes, NJ) at room temperature for 30 minutes, and fixed using BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit (BD Biosciences). Chemokine receptors were stained at 37ºC for 15 minutes prior to surface staining of other markers. For intracellular cytokine staining, cells were cultured with 20 ng/ml of phorbol 12-myristate 13-acetate (Sigma-Aldrich, St. Louis, MO) and 500 ng/ml of ionomycin (Sigma-Aldrich) in the presence of GoldiStop (BD Biosciences) for 4 hours at 37°C. Intracellular staining was performed using BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit or eBioscience Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific). Stained cells were analyzed on a FACS Fortessa flow cytometer (BD Biosciences) on the same day. Rainbow Fluorescent Particles (Biolegend, San Diego, CA) were used to standardize the instrument settings and to minimize batch effects across experiments. Serial samples of the same individuals were always run in the same batched experiment. Data were analyzed using FlowJo software (BD Biosciences). As part of quality control, samples not meeting predefined criteria (at least 75% viability measured by Live/Dead staining) were excluded from analysis. Average PBMC sample viabilities were 93.6 % (SD 4.9, range 78.5-98.0) in the Discovery cohort and 92.4% (SD 4.5, range 77.6-98.7) in the Validation cohort. Pre-treatment (baseline) CSF samples were processed within 1 hour of lumbar puncture. All samples were centrifuged at 400xg for 10 minutes at room temperature with supernatant removed and stored at -80°C, until batched analysis for levels of neurofilament light chain (NfL) using Quanterix Simoa NF-light Advantage Kit (Billerica, MA, USA), and neurofilament heavy chain (NfH) as well as glial fibrillary acidic protein (GFAP), using ProteinSimple (San Jose, CA, USA). Fresh CSF cell pellets were resuspended in BD buffer (BD Biosciences, San Jose, CA, USA) and blocked with Fc block (BD Biosciences) for 10 minutes at room temperature. Zombie Aqua fixable viability dye (Biolegend, San Diego, CA, USA) and antibody staining cocktail containing CD3-PerCP-Cy5.5 and CD19-BV421 (Biolegend) were added to the cells and incubated at room temperature for 15 minutes in the dark, with cells then washed with BD buffer, resuspended in 0.5% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA, USA) and acquired via flow cytometry within 6 hours of collection.

MR imaging and identification of new disease activity
Validation cohort patients underwent standardized brain MRIs at baseline and at weeks 12, 24, and 52 following initiation of ocrelizumab. MRI acquisition protocol included pre-and post-gadolinium (Gd) injection (0.1 mmol/kg, 10-min post-injection delay) axial 3 mm T1-weighted slices (3D spoiled gradient-echo, repetition time = 28-30 ms, echo time = 5-11 ms, flip angle = 27-30), as well as axial 3 mm T2-weighted slices (2D fast spin-echo, repetition time = 4000-6190 ms, echo time = 74-91 ms and echo train length = 7-11), as previously described (2). Centralized reading (NeuroRx) was performed to ascertain the development of new disease activity defined as either presence of new Gd-enhancing T1 lesion/s or appearance of one or more new and/or enlarging T2 lesion/s compared to prior MRI.

Statistical analysis
Statistical analysis was performed using Prism 9 ver. 9.0.0 (GraphPad Software, San Diego, CA) or JMP Pro 15 ver. 15.2.0 (SAS Institute, Cary, NC). Normality was evaluated by Shapiro-Wilk test. Unpaired data were compared by parametric t-test and non-parametric Mann-Whitney test. Paired samples were compared using non-parametric Wilcoxon matched-pairs ranked-sum test to compare paired samples followed by multiple comparison correction using False Discovery Rate by Benjamini, Krieger and Yekutieli's two-stage step-up method. The repeated measures data containing missing values were analyzed using a repeated measures mixed-effect model, with the Geisser-Greenhouse correction for sphericity, followed by Holm-Sidak's multiple comparison test. Correlations were examined using Pearson or Spearman correlation coefficient test based on the normality examined by the Shapiro-Wilk test. A p-value < 0.05 was considered statistically significant.

Figure S4. Replication of immune-cell subset changes following anti-CD20 treatment -Validation cohort
Thirty-five patients with relapsing-remitting multiple sclerosis (RRMS) were enrolled in the validation cohort. Blood sampling was serially performed at baseline (W0), week 12 (W12) and week 24 (W24) after initiation of ocrelizumab. Whole blood cell counts were analyzed in all patients and cryopreserved peripheral blood mononuclear cells subsequently underwent comprehensive functional phenotyping using multi-parametric flow cytometry. (A) Major lymphocyte population counts in whole blood at W0, W12 and W24. (B) CD4/CD8 ratio in whole blood. (C) Absolute cell counts and (D) frequencies of naive and memory cells among CD4 + and CD8 + T cells, respectively. (E) Cell counts and (F) frequencies of cytokine-producing CD4 + and CD8 + T cells, respectively, after ex vivo stimulation with phorbol 12-myristate 13-acetate and ionomycin for 4 hours. (G) Counts and frequencies of CCR2 + CCR5 + CD4 + and (H) CCR2 + CCR5 + CD8 + T cells. (I) Counts and frequencies of adhesion molecule-expressing CD4 + and (J) CD8 + T cells . Statistical analysis was performed by fitting the mixed-effects model with the Geisser-Greenhouse correction (accounting for missing data and correcting the violation of sphericity), followed by Sidak's multiple comparison to compare the differences among all pairs. p value, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure S6. Replication of anti-CD20-induced depletion of CD20 dim T cells -Validation cohort (A) Cell counts and frequencies of CD20 dim T cells Pre-treatment (W0), and week 12 (W12) and week 24 (W24) after the first infusion of ocrelizumab. (B) Cell counts and frequencies of CD20 dim cells among CD4 + T cells and CD8 + T cells. (C) Comparison of surface and intracellular markers
between CD20 dim CD4 + and CD20 -CD4 + T cells prior to ocrelizumab treatment (n = 24). (D) Comparison of surface and intracellular markers between CD20 dim CD8 + and CD20 -CD8 + T cells prior to ocrelizumab treatment (n = 24). Abbreviations: Tcm, central memory T cells; Tem, effector memory T cells; Temra, terminally differentiated effector memory T cells; Tn, naive T cells. Statistical analysis was performed by fitting the repeated measures mixed-effects model with the Geisser-Greenhouse correction (accounting for missing data and correcting the violation of sphericity), followed by Sidak's multiple comparison to compare the differences among all pairs (A, B), or Wilcoxon matched-pairs signed-rank test followed by multiple comparison correction using False Discovery Rate by Benjamini, Krieger and Yekutieli's two-stage step-up method (C, D). p value, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Figure S7. Further characterization of CD20 dim CD4 + and CD20 dim CD8 + T cells (A) Comparison of intracellular and surface markers between CD20 dim CD4 + and CD20 -CD4 + T cells prior to ocrelizumab treatment (n = 14). (B) Comparison of intracellular and surface markers between CD20 dim CD8 + and CD20 -CD8 + T cells prior to ocrelizumab treatment (n = 14). Abbreviations: ALCAM, activated leukocyte-cell adhesion molecule; MCAM, melanoma cell adhesion molecule; NS, not significant. Statistical analysis was performed by Wilcoxon matchedpairs signed-rank test followed by multiple comparison correction using False Discovery Rate by Benjamini, Krieger and Yekutieli's two-stage step-up method. p value, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure S11. No association between on-treatment new disease activity and subsets of reemerging B cells and T cells except for CD20 dim T cells (A) Frequencies of B cell, CD4 + T cells and CD8 + T cells, (B) B cell-subsets, (C) naive/memory
CD4 + T-cell subsets, (D) naive/memory CD8 + T-cell subset, (E) IFN-γ or TNF-α-producing CD4 + T cells, (F) IFN-γ or TNF-α-producing CD8 + T cells at baseline, at week 12 and at week 24 after ocrelizumab treatment initiation, were compared between patients who did (n = 6, red symbols) or did not (n = 18, blue symbols) experience new disease activity beyond 12 weeks of ocrelizumab treatment. Statistical analysis was performed using multiple Mann-Whitney test. Abbreviations: CSM, class-switched memory B cells, nB, naive B cells; ns, not significant; nT, naive T cells; tB, transitional B cells; Tcm, central memory T cells; Tem. effector memory T cells; Temra, terminally differentiated effector memory T cells; USM, unswitched memory B cells.