Jenks2021 - B Cell Subset Composition in Cutaneous Lupus

Full citation: Jenks SA, Wei C, Bugrovsky R, Hill A, Wang X, Rossi FM, Cashman K, Woodruff MC, Aspey LD, Lim SS, Bao G, Drenkard C, Sanz I. B cell subset composition segments clinically and serologically distinct groups in chronic cutaneous lupus erythematosus. Annals of the Rheumatic Diseases. 2021;80(9):1190–1200. https://doi.org/10.1136/annrheumdis-2021-220349

Raw file: [[raw/Jenks2021.pdf]]

★ Why this paper matters to this wiki. Three things, none of which is the paper’s own headline. (1) It is a Sanz-lab primary that gates DN1/DN2/DN3 on CD21 × CD11c — making the Emory side of the wiki’s unresolved DN3 gating-axis conflict first-party rather than relayed through Sanz2025 - Human Atypical B Cells Overview. (2) The DN2:DN1 log2 ratio is again a reported outcome (Fig 1E) — the third setting in which the Sanz lab reports it, after SLE and acute COVID-19 — directly load-bearing for Thesis Objectives and Grant Pitch. (3) It shows that within SLE, presence of skin disease tracks lower DN expansion, so DN2 magnitude segregates with organ involvement rather than with the disease label — the strongest available argument that DN2 is a patient-stratum property, not a disease property.

⚠ Scope note. Roughly 60% of this paper (Figures 3–5) is serology: 9G4/VH4.34 idiotype, anti-dsDNA, anti-chromatin, anti-RNA, and anti-RNA-binding-protein antibodies by LIPS. Under the standing [2026-05-02] decision (9G4/VH4.34 tracking is out of scope absent a dengue context) no serology finding is propagated to any entity, concept or method page — all of it is recorded on this page as background only. (ELISA carries the method and the paired serology→immunophenotype design, flagged there as out of scope for its findings.) The specificity-level synthesis belongs to dengue-wiki/ and bridge-wiki/.

Summary

This is an Emory/Sanz-lab study of B cell composition across the clinical spectrum of lupus skin disease, drawing on the Georgia Organized Against Lupus (GOAL) population-based cohort. The question is whether B cells are pathogenically involved in primary chronic cutaneous lupus erythematosus (CCLE) — a condition where low serological activity and poor response to B cell depletion had cast doubt on a B cell role — and whether B cell profiling can identify the ~20% of CCLE patients who go on to develop systemic lupus. Three clinical groups were compared against healthy controls: primary CCLE without systemic disease (CCLE+/SLE−), SLE with cutaneous involvement (SLE+/CCLE+), and SLE without it (SLE+/CCLE−).

The cellular arm (Figures 1–2) is a conventional-flow B cell panel resolving naive+transitional, unswitched memory, switched memory, plasmablasts and the IgD⁻CD27⁻ double-negative compartment, with DN further split into DN1/DN2/DN3 on CD21 × CD11c and activated naive cells gated on the same two markers within the naive gate. All three lupus groups share the canonical SLE perturbation — loss of unswitched memory, expansion of plasmablasts, expansion of DN driven by DN2 and DN3 with reversal of the normal DN1 predominance, and expansion of activated naive cells. Primary CCLE shows the same abnormalities at lower magnitude. Unsupervised clustering of all subjects on subset frequencies yields five groups defined by an early/memory/effector axis; healthy donors fall exclusively in two of them.

The paper’s clinical payoff is that primary CCLE is immunologically bimodal: a substantial fraction of patients cluster with healthy donors and the rest cluster with SLE. The SLE-like fraction is more likely to have disseminated skin lesions, less likely to be male, and carries markedly more autoantibody — historical ANA/anti-dsDNA/anti-Ro/anti-La and contemporaneous anti-dsDNA, anti-chromatin, anti-RNP and anti-Ro52. The serological arm additionally shows that although autoreactive 9G4⁺ IgG is elevated in CCLE at SLE-like frequencies, it is uncoupled in CCLE from the anti-nucleic-acid antibodies with which it is tightly correlated in SLE — read by the authors as a limited, not-yet-disseminated breakdown of tolerance. The authors propose B cell phenotyping as a prognostic tool for systemic progression and as a way to select CCLE patients for B cell–targeted therapy.

Study Design

• Type: Cross-sectional human immunophenotyping + serology, nested in a population-based cohort. Primary data throughout. No longitudinal follow-up — the progression hypothesis is proposed, not tested.
• Sample size:
  • Flow cytometry: n=207 patients — CCLE+/SLE− n=69, SLE+/CCLE+ n=53, SLE+/CCLE− n=85 — plus n=46 healthy controls.
  • Serology: the above plus 39 additional SLE patients and n=69 healthy controls.
  • CLASI skin scores: n=32 of the CCLE+/SLE− group only.
• Setting: Georgia Organized Against Lupus (GOAL), a population-based cohort with validated lupus diagnoses; Emory University School of Medicine, Atlanta. Diagnoses validated by medical-record review, physician assessment and photograph review; SLE classified by 1997 Revised ACR criteria. No patient was on B cell depletion therapy.
• Population: Predominantly Black/African American (90.6–94.1% across groups) and female (84–93%). Within the two CCLE groups, discoid lupus (DLE) accounted for 65/69 (94%) and 52/53 (98%) of cases; the remainder were lupus panniculitis or tumidus. CCLE+/SLE− patients were significantly older (51.2 ± 13.7 vs 43.0 ± 12.7 and 47.6 ± 13.6 years, p=0.0047) and had shorter disease duration (9.9 ± 9.5 vs 10.1 ± 9.1 and 14.0 ± 9.7 years, p=0.013) than the SLE groups. Immunosuppressive use differed sharply: 10.6% / 27.0% / 40.5%, p=0.0002.
• Statistics: Mann-Whitney U for two groups; Kruskal-Wallis with Dunn’s multiple-comparison correction for more than two; Fisher’s exact and χ² for contingency; Pearson or Spearman for correlation.

Key Findings

Gating and panel

• ★ DN subsets are gated on CD21 × CD11c, with no CXCR5 anywhere in the paper. Figure 1B: DN1 = CD21⁺CD11c⁻, DN2 = CD21⁻CD11c⁺, DN3 = CD21⁻CD11c⁻. Activated naive cells are gated from the IgD⁺CD27⁻ naive+transitional gate on the same two markers (CD21⁻CD11c⁺); transitional T1+T2 on CD24^hi CD38^hi. The parent tree (Fig. 1A) is CD19⁺CD3⁻ → IgD⁺CD27⁻ (N+T) / IgD⁺CD27⁺ (USM) / IgD⁻CD27⁺ (SM+PB) / IgD⁻CD27⁻ (DN), with PB = IgD⁻CD27⁺⁺CD38⁺⁺ and switched memory derived by subtracting the PB frequency from the IgD⁻CD27⁺ compartment.
• The CD21⁺CD11c⁺ quadrant is left unnamed. The paper defines three DN subsets, not four. This is not the same as asserting that DN4 does not exist, and it does not arbitrate the DN4 marker conflict between Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues (DN4 = CXCR5⁺CD11c⁺) and Lamprinou2026 - ABCs and DN B Cells (CXCR5⁺CD11c⁻).
• ⚠ The antibody panel is not recoverable from this PDF. The fluorochrome-conjugate list is in online supplemental table 1, which is not part of the deposited manuscript. Markers named in the body text and figures are: CD19, CD3, IgD, CD27, CD38, CD24, CD21, CD11c, plus Fixable Viability Dye eFluor506. T-bet, CXCR5 and FCRL5 appear nowhere in the paper’s own data — T-bet and IL-21 occur only inside cited reference titles. Do not quote a fuller panel for this paper than these eight markers.
• Preanalytical conditions: cryopreserved PBMC (BD Vacutainer CPT, banked in liquid nitrogen), stained 30 min at 4 °C in PBS + 2% FBS, fixed in 0.5% formaldehyde, acquired on a BD LSRII, analysed in FlowJo.

B cell homeostasis across the lupus spectrum (Figure 1)

• All three lupus groups share the canonical SLE signature: contraction of unswitched memory, expansion of plasmablasts, expansion of the DN compartment, and expansion of activated naive cells. Primary CCLE shows every one of these at lower magnitude than SLE.
• DN expansion is accounted for in all groups by DN2 and DN3, with reversal of the normal DN1 predominance — the same pattern the authors report in SLE (Jenks2018 - DN2 B Cells and EF Pathway in SLE) and in severe COVID-19 (Woodruff2020 - EF B Cell Responses in COVID-19). DN1 is described as the population transcriptionally linked to resting switched memory.
• ★ Within SLE, the presence of cutaneous disease tracks lower DN expansion. DN frequencies were higher in SLE+/CCLE− than in CCLE+/SLE−, and the authors note explicitly that among SLE patients, CCLE involvement correlated with DN expansion of lower magnitude. They connect this to two independent facts: the association of DN2 with lupus nephritis (Jenks2018 - DN2 B Cells and EF Pathway in SLE), and the reduced incidence of nephritis in SLE patients who have discoid lesions.
• The DN2:DN1 ratio is reported as a log2-transformed ratio (Figure 1E) — a dedicated panel, not an incidental derivation. This is the third wiki source to report the metric, after Jenks2018 - DN2 B Cells and EF Pathway in SLE (SLE) and Woodruff2020 - EF B Cell Responses in COVID-19 (acute COVID-19, where the CoV-A ratio was indistinguishable from active SLE) — all three from the Sanz lab.
• Activated naive cells, described as “DN2 progenitors”, were expanded in both SLE groups relative to primary CCLE (Fig. 1F, as a proportion of both CD19⁺ and total naive).
• Transitional T1+T2 cells were expanded in CCLE relative to healthy donors and to SLE+/CCLE− (Fig. 1G) — a direction opposite to most other subsets, where CCLE sits between healthy and SLE.
• Every group is internally heterogeneous. For switched memory specifically, 17% of SLE+/CCLE−, 19% of SLE+/CCLE+ and 16% of CCLE+/SLE− had frequencies more than 2 SD above the healthy-donor mean.
• Significance is reported throughout as colour-coded bars (p<0.05 green, p<0.01 blue, p<0.001 red, p<0.0001 dark purple) rather than as numeric p-values, so exact p-values for the subset comparisons are not extractable from the text.

The five-cluster B cell fingerprint (Figure 2)

• Hierarchical clustering of all subjects on subset frequency yields five groups (I–V), defined by the relative frequencies of three families: early (resting naive, early T1+T2), memory (USM, switched memory, DN1) and effector (activated naive, DN2, DN3, PB).
• Healthy donors fall exclusively in clusters III and IV, distinguished by higher USM and resting-naive/T3 frequencies. No healthy donor clustered in I, II or V — so the authors could not run χ² comparisons of healthy donors against lupus patients for those clusters.
• SLE patients concentrate in clusters I, II and V, with only 15–16% expressing healthy-like profiles. Clusters I/II are the activated-effector profiles (activated naive, DN2, DN3, PB), most pronounced in cluster II. Cluster V is a distinct phenotype: coordinated expansion of T1/T2 and resting-naive/T3 with the largest USM decrease of any cluster.
• ★ Primary CCLE is the most heterogeneous group of all, with significant representation in every cluster: 58% SLE-like (I, II, V) and 42% healthy-like. Of the SLE-like fraction, 64% fell in clusters I/II; cluster II alone accounted for 12% of all CCLE patients.

Clinical and serological correlates of the CCLE B cell split (Figure 6)

• CCLE patients with an SLE-like B cell phenotype had more disseminated skin disease: 58% with generalised lesions vs 15% in the healthy-like group, and were less likely to be male (8% vs 31%).
• They carried more historical autoreactivity (ANA, anti-dsDNA, anti-Ro, anti-La), more contemporaneous anti-dsDNA and anti-chromatin, and higher titres of anti-RNP and anti-Ro52.
• Disease duration did not differ between the two CCLE phenotype groups — which is the paper’s own control against the split being an artefact of disease stage. CLASI skin activity did not differ either; there was a non-significant trend toward greater skin damage in the SLE-like group (n=32 subset only).

Serology — recorded as background, propagated nowhere (Figures 3–5)

• 9G4⁺ (VH4.34-idiotype) IgG and IgA were elevated in all three lupus groups at similar frequencies (48–57%); 9G4⁺ IgM was significantly elevated only in SLE+/CCLE−.
• ★ In CCLE the 9G4 response is qualitatively different, not merely smaller. A much larger fraction of 9G4⁺ antibodies in CCLE+/SLE− were reactive against apoptotic cell antigens (48% positive) with significantly lower anti-B-cell reactivity (17%). And 9G4⁺ IgG — tightly correlated with anti-dsDNA and anti-chromatin in both SLE groups — was sharply uncoupled from both in CCLE, even in patients with high levels of each.
• Anti-nucleic-acid antibodies were less frequent in CCLE: anti-dsDNA 13% (vs 33% SLE+/CCLE+, 44% SLE+/CCLE−); anti-RNA 46% (vs 56% and 72%).
• By luciferase immunoprecipitation (LIPS), RNA-binding-protein autoantibodies were present in CCLE but less frequent: anti-Sm 38% (vs 71% in SLE+/CCLE−, a rate the authors attribute to the cohort’s high African American representation), anti-RNP 17%, anti-Ro52 19%, anti-Ro60 49%. Only anti-Ro52 and anti-Sm titres exceeded healthy donors in CCLE.
• Breadth, not just magnitude, separates CCLE from SLE: 95% of healthy donors had no reactivity to any tested antigen; 33% of CCLE had none; but only 16% of CCLE had reactivity against multiple autoantigens, vs 44% of SLE+/CCLE− and 24% of SLE+/CCLE+. Anti-Sm and anti-RNP were highly correlated in SLE but dissociated in CCLE. The authors read the whole pattern as a limited breakdown of tolerance that has not yet spread to additional antigens.

Methods Used

• Conventional Flow Cytometry — 8 named markers on a BD LSRII, cryopreserved PBMC, FlowJo analysis; full conjugate panel not in the deposited PDF.
• ELISA — 9G4⁺ IgG/IgM/IgA, anti-dsDNA and anti-chromatin (QUANTA Lite, INOVA), anti-RNA.

Also used but not separately updated, as the findings they generated are out of scope under the serology note above: a luciferase immunoprecipitation system (LIPS) assay for anti-Sm/RNP/Ro52/Ro60, and an apoptotic-cell (Jurkat J45.1) serum-binding assay read by flow cytometry. Neither has a wiki method page and neither is being given one.

Entities Mentioned

Double-Negative B Cell, DN2 B Cell, DN3 B Cell, Activated Naive B Cell, Plasmablast, Switched Memory B Cell, CD19, CD21, CD24, CD27, CD38, CD11c, IgD

Concepts Addressed

Extrafollicular Response, Memory B Cell, Atypical B Cell Effector Output

Relevance & Notes

What this paper adds that the wiki did not already hold.

1. First-party Emory gating for the DN3 axis conflict. The wiki’s account of the Emory DN3 definition (CD21⁻CD11c⁻) has until now come through Sanz2025 - Human Atypical B Cells Overview and Woodruff2020 - EF B Cell Responses in COVID-19. This is a Sanz-lab primary stating it directly in a figure legend. ⚠ The supportable claim is “as of 2021 a Sanz-lab primary gates DN3 on CD21 × CD11c with no CXCR5” — not “the Emory axis is confirmed correct.” CXCR5 is absent from the paper’s own data, so this adds nothing toward reconciling the two axes; the reconciliation still requires running both in one tube, which B Cell Panel Variant 1 is designed to do.

2. A third Sanz-lab use of the DN2:DN1 ratio, and the first in a heterogeneous clinical cohort. Thesis Objectives and Grant Pitch proposes the DN2:DN1 ratio as the centrepiece outcome measure; this is precedent for that choice in a large cohort with heterogeneous patients — the same situation acute dengue presents.

3. The clearest evidence in the wiki that DN2 magnitude tracks organ involvement, not disease identity. Same disease label (SLE), different organ pattern, different DN expansion — in the direction predicted by the DN2–nephritis association. This is a cross-sectional association within a single cohort, not a mechanistic result, and the paper does not test it directly.

4. A stratum-level caution for the dengue design. In primary CCLE, 42–48% of patients had an entirely normal B cell profile. If a comparable bimodality exists in dengue, group-mean comparisons of DN2 frequency between severity categories will dilute the signal, and the analysis should anticipate patient-level clustering rather than assuming a unimodal shift.

How this connects to existing wiki content. The effector-axis result (activated naive → DN2 → DN3 → PB expanding together, DN1 falling) reproduces Jenks2018 - DN2 B Cells and EF Pathway in SLE in a much larger and clinically stratified cohort, and matches Woodruff2020 - EF B Cell Responses in COVID-19 in acute infection. The USM contraction is the same feature reported across Sjögren’s, rheumatoid arthritis, vasculitis and inflammatory bowel disease, and the authors raise the possibility that it reflects loss of a marginal-zone-equivalent with protective functions. The epigenetic explanation the discussion invokes for effector skewing is Scharer2019 - Epigenetic Programming in SLE B Cells, already ingested.

Attribution boundary. References 16, 19, 31, 33, 34 and 35 are Wei2007 - DN Memory B Cells in SLE, Jenks2018 - DN2 B Cells and EF Pathway in SLE, Sanz2019 - Consistent Classification of Human B Cell Populations, Tipton2015 - ASC Diversity and Origin in SLE, Scharer2019 - Epigenetic Programming in SLE B Cells and Woodruff2020 - EF B Cell Responses in COVID-19 — all already ingested. Where this paper relays them, wiki pages cite the primary. Content attributed to Jenks2021 is limited to: the GOAL CCLE cohort data, the CD21 × CD11c three-way DN gate as stated in Figure 1B, the five-cluster fingerprint, the within-SLE inverse CCLE↔DN association, the CCLE bimodality and its clinical/serological correlates, and the 9G4↔anti-nucleic-acid uncoupling.

⚠ Do not conflate with the other Jenks 2021. External Citation Audit entry #36 refers to a different paper — Jenks et al. 2021 in JCI, on B cell endotypes in SLE, cited via Sanz2025 - Human Atypical B Cells Overview and relied on at Extrafollicular Response. The titles are similar and both are Sanz-lab 2021. This ingest does not resolve audit #36.

⚠ Two internal inconsistencies — quarantined here under the [2026-08-23] rule

Neither is propagated to any other page.

1. The Figure 6 legend inverts its own cluster assignments. The legend reads “Primary CCLE from clusters III and IV (figure 2A) that resemble patients with SLE (red) are less likely than those from clusters I and II that resemble HCD (blue)…” — but the Results text and the Figure 2 legend both state that healthy donors clustered only in III and IV and that I/II were the SLE-enriched effector clusters. The Figure 6 legend is therefore the error; the direction of the clinical result is recoverable from the Results prose (SLE-like → 58% generalised lesions vs 15%), which agrees with Figure 2. This matters because Figure 6 carries the paper’s entire clinical-association result. Recorded, not silently corrected.
2. 42% or 48% healthy-like? The Results state that of primary CCLE, “a small majority (58%), expressed SLE-like profiles (I,II,V), 42% had HCD-like B cells.” The Discussion states “normal B cell signature in 48% of primary CCLE and 15% of SLE.” These do not reconcile. (The Discussion’s separate figure of 38% with a “highly activated SLE-like” profile is consistent — 58% × 64% in clusters I/II ≈ 37% — so 38% refers to clusters I/II specifically, not to all SLE-like patients.)

Limitations

Authors’ own: cross-sectional design cannot establish cause and effect; wide range of disease duration (though the two CCLE phenotype groups did not differ on it); skin scores available in only a subset; findings best generalised to the Southeastern USA where most lupus patients are Black. Curator’s additions: the progression hypothesis — that SLE-like B cells identify CCLE patients who will develop SLE — is entirely untested here, and is a proposal in the discussion rather than a result; subset comparisons are reported as colour-coded significance bands rather than numeric p-values; and the flow panel is not published with the manuscript.

Questions Raised

• Is the SLE-like B cell profile in primary CCLE prognostic for systemic progression, or is it a stable alternative phenotype? The authors call for longitudinal analysis; the cross-sectional design cannot distinguish the two.
• Why does unswitched memory contract? The paper documents it across four other autoimmune conditions and offers candidate explanations (loss of a marginal-zone-equivalent with apoptotic-clearance, IL-10 regulatory and autoreactivity-diluting functions) without evidence for any.
• Does the cluster V phenotype — transitional and resting-naive expansion with maximal USM loss — represent a distinct disease group with distinct implications, in either CCLE or SLE?
• Do CCLE-associated B cells run different differentiation programmes than SLE-associated ones, producing a less pathogenic effector output? The authors raise this to explain why CCLE within SLE tracks with less nephritis, and supply no data on it.
• Is the B cell profile in skin-infiltrating B cells the same as in blood? B cells and plasma cells infiltrate scarred CCLE lesions, but their phenotype relative to circulating and kidney-infiltrating cells is unknown — the same blood-vs-tissue gap Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues opened for DN2 and DN3.
• Does the same patient-level bimodality exist in acute infection? If a substantial fraction of dengue patients carry entirely normal B cell profiles, severity-stratified group means are the wrong first analysis.

Related Pages

Jenks2018 - DN2 B Cells and EF Pathway in SLE, Woodruff2020 - EF B Cell Responses in COVID-19, Scharer2019 - Epigenetic Programming in SLE B Cells, Sanz2019 - Consistent Classification of Human B Cell Populations, Wei2007 - DN Memory B Cells in SLE, Tipton2015 - ASC Diversity and Origin in SLE, Allard-Chamard2023 - DN3 B Cells Infiltrate Inflamed Tissues, DN2 Gating Strategy, B Cell Panel Variant 1, Thesis Objectives and Grant Pitch, Mechanistic Case for DN and DN2 Cells in Dengue, Why DN B Cells Matter - Disease Relevance and Infectious Disease Case, External Citation Audit