The Plasma Cell Signature in Autoimmune Disease

Authors

  • Katie Streicher,

    Corresponding author
    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Christopher A. Morehouse,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Christopher J. Groves,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Bhargavi Rajan,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
  • Fernanda Pilataxi,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Kim P. Lehmann,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Philip Z. Brohawn,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
  • Brandon W. Higgs,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Kathleen McKeever,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Steven A. Greenberg,

    1. Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
    • Drs. Greenberg and Fiorentino have received consulting fees, speaking fees, and/or honoraria from AstraZeneca/MedImmune (less than $10,000 each).

  • David Fiorentino,

    1. Stanford University School of Medicine, Stanford, California
    Search for more papers by this author
    • Drs. Greenberg and Fiorentino have received consulting fees, speaking fees, and/or honoraria from AstraZeneca/MedImmune (less than $10,000 each).

  • Laura K. Richman,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Bahija Jallal,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Ronald Herbst,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

    • Dr. Herbst is a named coinventor on a patent related to MEDI-551, an anti-CD19 monoclonal antibody.

  • Yihong Yao,

    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.

  • Koustubh Ranade

    Corresponding author
    1. MedImmune, Gaithersburg, Maryland
    Search for more papers by this author
    • Dr. Streicher, Mr. Morehouse, Mr. Groves, Ms Pilataxi, Ms Lehmann, and Drs. Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade own stock or stock options in AstraZeneca.


Abstract

Objective

Production of pathogenic autoantibodies by self-reactive plasma cells (PCs) is a hallmark of autoimmune diseases. We undertook this study to investigate the prevalence of PCs and their relationship to known pathogenic pathways to increase our understanding of the role of PCs in disease progression and treatment response.

Methods

We developed a sensitive gene expression–based method to overcome the challenges of measuring PCs using flow cytometry. Whole-genome microarray analysis of sorted cellular fractions identified a panel of genes, IGHA1, IGJ, IGKC, IGKV4-1, and TNFRSF17, expressed predominantly in PCs. The sensitivity of the PC signature score created from the combined expression levels of these genes was assessed through ex vivo experiments with sorted cells. This PC gene expression signature was used for monitoring changes in PC levels following anti-CD19 therapy, for evaluating the relationship between PCs and other autoimmune disease–related genes, and for estimating PC levels in affected blood and tissue from patients with multiple autoimmune diseases.

Results

The PC signature was highly sensitive and capable of detecting a change in as few as 360 PCs. The PC signature was reduced more than 90% in scleroderma patients following anti-CD19 treatment, and this reduction was highly correlated (r = 0.80) with inhibition of collagen gene expression. Evaluation of multiple autoimmune diseases revealed that 30–35% of lupus and rheumatoid arthritis patients had increased levels of PCs.

Conclusion

This newly developed PC signature provides a robust and accurate method of measuring PC levels in the clinic. Our results highlight subsets of patients across multiple autoimmune diseases who may benefit from PC-depleting therapy.

Considerable preclinical and clinical work focused on characterizing the role of B cells in human disease has revealed critical roles for B cells in numerous human inflammatory and autoimmune disorders as well as in hematopoietic malignancies ([1, 2]). Accordingly, completely depleting or modulating the numbers or activity of B cells by targeting cell surface antigens or other pathway components has emerged as an important therapeutic opportunity for multiple diseases ([3-8]).

Depletion of B cells with therapies targeting the B cell membrane protein CD20 has previously shown beneficial effects in rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and other autoimmune diseases ([7-9]). Additionally, in both open-label and controlled clinical studies in patients with relapsing–remitting multiple sclerosis (MS), depletion of B cells with anti-CD20 monoclonal antibodies (mAb) resulted in significantly decreased inflammatory lesions and relapse rates ([10-12]). Bolstered by the success of anti-CD20 therapy, there is growing interest in developing therapies targeted to other B cell membrane proteins, including CD52, CD22, and CD19. CD52 is expressed on T and B lymphocytes and functions as a costimulatory molecule important for T cell activation ([13, 14]). CD22 is a B cell transmembrane glycoprotein of the immunoglobulin superfamily playing a key role in the modulation of B cell activation and regulation of tolerance thresholds ([5, 14]). CD19 is a B cell–restricted transmembrane protein that functions to control the signaling threshold for B cell development and humoral immunity ([15-17]). Altering the density of CD19 on the cell surface in gene-targeted or transgenic mice significantly affects intrinsic B cell– and B cell antigen receptor–induced signaling thresholds and also increases B cell proliferation and autoantibody production ([18-22]). In contrast to CD20, CD22, and CD52, CD19 is expressed on a broader range of B cell subsets, including earlier-stage precursor cells and later-stage differentiated cells such as plasmablasts and some plasma cells (PCs), which are the major source of antibody production ([23, 24]).

The role of PCs in secreting pathogenic autoantibodies and maintaining a chronic autoreactive environment suggests that depletion of these cells could have a positive effect on clinical outcomes in multiple autoimmune diseases. Thus, regardless of the particular B cell–targeted therapy, having a method by which to sensitively and accurately assess PC levels in the blood and diseased tissue will be useful not only for monitoring the effectiveness of such therapies, but also for identifying subsets of patients who may benefit from these therapies. Consistent with this expectation, a recent study demonstrated that RA patients with high levels of plasmablast markers prior to therapy were less likely to respond to rituximab ([25]).

In order to evaluate the prevalence of PCs in autoimmune diseases and measure clinically relevant alterations, we first developed a robust gene expression signature to effectively evaluate PCs, because low numbers of PCs and high instability significantly hinder standard flow cytometry assays from accurately measuring PCs in a clinical setting. Additionally, we validated the sensitivity and specificity of this signature using ex vivo experiments and confirmed its utility in scleroderma patients enrolled in a phase I dose-escalation trial of MEDI-551, an anti-CD19 mAb with enhanced effector function ([26]). Furthermore, we identified multiple autoimmune diseases with increases in the PC signature, which may be indicative of a relationship between PCs and the pathogenesis of these diseases and which may aid in identifying patients who may benefit from PC-depleting therapy.

MATERIALS AND METHODS

Clinical samples

MI-CP200 is a phase I, randomized, double-blind, placebo-controlled study evaluating the safety and tolerability of escalating single intravenous (IV) doses of MEDI-551 in adult subjects with scleroderma. Five cohorts of subjects received 1 of 5 single IV doses of MEDI-551 (0.1, 0.3, 1.0, 3.0, or 10.0 mg/kg) or placebo. Whole blood and skin samples were collected from patients at the time points indicated in Supplementary Table 1 (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38194/abstract). The frozen tissue samples were split, and ∼1 mm of skin tissue was used for RNA isolation.

Whole blood, skin tissue, and synovial tissue from healthy donors or patients with autoimmune diseases were obtained from commercial vendors, external collaborations, and clinical trials. These samples have been described previously ([27]), and their sources are summarized in Supplementary Table 2 (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38194/abstract). All subjects provided informed consent, and all studies were approved by the institutional review boards from the different sites or commercial vendors.

RNA isolation

Total RNA was extracted from PAXgene blood tubes using a PAXgene Blood RNA kit (Qiagen). For skin tissue, samples were homogenized in RNA lysis buffer, and RNA was isolated using an RNeasy Fibrous Tissue kit (Qiagen). RNA purity and concentration were determined spectrophotometrically (the ratio of absorbance at 260 nm to that at 280 nm was >1.9). RNA quality was assessed on an Agilent 2100 Bioanalyzer using an RNA 6000 Nano LabChip.

Whole-genome microarray

For whole-genome microarray, biotin-labeled amplified complementary RNA (cRNA) from 2 μg of total RNA was generated with an Affymetrix GeneChip One-Cycle cDNA Synthesis kit and an Affymetrix GeneChip IVT Labeling kit. Twenty micrograms of cRNA was fragmented for hybridization on Affymetrix Human Genome U133 Plus 2.0 GeneChip arrays. Data capture and quality assessments were performed with the GeneChip Operating Software tool. The R statistical analysis tool was used to calculate probe-level summaries (frozen robust multiarray average analysis) from the array CEL files ([28]). All whole-genome microarray data have been deposited into GEO (accession numbers GSE45536 and GSE45537).

TaqMan quantitative polymerase chain reaction (qPCR).

For TaqMan qPCR, cDNA was generated using a SuperScript III First-Strand Synthesis SuperMix kit (Life Technologies) and random primers. Samples were prepared using a TaqMan Pre-Amp Master Mix kit and analyzed with a BioMark Real-Time PCR System. Cycle threshold (Ct) values above 28 were excluded from calculations. We calculated ΔΔCt values using the mean of 2 reference genes (β-actin and GAPDH) and each patient's baseline expression level as controls. Fold change values were determined by calculating 2–ΔΔCt.

Neutralization of genes/gene panels

Neutralization of gene signatures following treatment with placebo or MEDI-551 was calculated relative to each patient's baseline sample using microarray data for whole blood and TaqMan qPCR data for skin. B cell, PC, and collagen signature scores were calculated by determining the fold change for each gene in the panel, then calculating the median fold change for all panel genes. Median signature scores were calculated for all patients treated with either MEDI-551 or placebo at each time point. Error was calculated using the median absolute deviation (MAD), which is the median of the absolute deviations from the data's median. For a univariate data set X1, X2, …, Xn, the MAD is defined as mediani (|Xi − medianj (Xj|), where medianj represents the median of the original data set and mediani represents the new median absolute deviation calculated by subtracting the original sample median from each X value in that data set.

Flow cytometry

To examine gene expression in purified cellular fractions, normal human blood was collected from 4 donors in accordance with institutional policy. The granulocyte (CD15+), monocyte (CD14+), T cell (CD3+), B cell (non–PC gated, CD19+), and PC (CD27++CD38++) fractions from peripheral blood were separated. White blood cells were washed with fluorescence-activated cell sorting (FACS) buffer (phosphate buffered saline plus 0.5% bovine serum albumin plus 2 mM EDTA; Gibco) and incubated with 20% heat-inactivated fetal bovine serum for 10–15 minutes on ice. The following mAb and DAPI were added directly to the cells: anti-CD15 (HI98), anti-CD14 (M5E2), anti-CD3 (UCHT1), anti-CD27 (M-T271), and anti-CD38 (HB7) (all from Molecular Probes). Cells were sorted on a Becton Dickinson FACSAria II flow cytometer. All sorted fractions were collected in FACS buffer and centrifuged, and the resulting cell pellet was suspended in RNA lysis buffer (Ambion).

For PC spike-in, PCs were prepared and sorted as described above. The PC (CD27++CD38++) population was isolated from 4 independent donors. Sorted PCs were added to a second sample of peripheral blood mononuclear cells (PBMCs) from the cognate donor, and RNA was isolated. Samples with and without the PC spike were compared.

Statistical analysis

Gene expression fold change values were analyzed using Welch's t-test or the nonparametric Mann-Whitney U test. P values less than 0.05 were considered significant. All correlations were calculated with Spearman's rank correlation (indicated as r) using GraphPad Prism software, and significance was determined using a t distribution.

RESULTS

Development of the PC signature

To develop a gene expression signature with which to monitor PCs, whole blood was sorted into various cellular fractions using the specified cell surface markers, including granulocytes (CD15), T cells (CD3), monocytes (CD14), B cells (CD19), and PCs (CD27++CD38++). Whole-genome microarray analysis of each subset, as well as for unsorted whole blood, was performed to identify genes that are predominantly expressed in each subset. The percentage of contribution to the total gene expression signal from each cellular subset was determined and revealed a natural point of discontinuity at ∼85% enrichment of the gene expression signal in a particular cellular subset (see Supplementary Figure 1, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38194/abstract). Therefore, genes with ≥85% of the gene expression signal estimated to originate from either PCs or B cells were considered to be enriched for these specific subsets.

Additionally, we evaluated the ability of the PC-enriched genes to be consistently detected in whole blood to verify that they could be routinely monitored in a clinical setting. Those genes that had ≥85% specificity in either PC or B cell subsets and were consistently detectable in whole blood from healthy subjects and patients with scleroderma were selected for the PC or B cell signature gene panels (Table 1). These panels included IGHA1, IGJ, IGKC, IGKV4-1, and TNFRSF17 for PCs and CD22, FCRLA, MS4A1, VPREB3, TCL1A, EBF1, FCRL1, and BANK1 for B cells. We calculated the median expression value for all of the PC- or B cell–enriched genes to develop the PC or B cell signature scores. Multiple genes were used in each score to increase robustness of the assay by minimizing variability that may be introduced by relying on a single gene.

Table 1. Genes enriched in PCs or B cells identified by profiling specific cellular subsets in whole blood*
Gene nameGene symbolB cellsGranulocytesMonocytesPCsT cells
  1. The percentage of total gene expression arising from each cellular subset is shown for the indicated genes. Blood from 4 donors was used for this analysis. PCs = plasma cells.
Ig heavy constant alpha 1IGHA11.50.00.098.50.0
TNF receptor superfamily, member 17TNFRSF175.00.00.095.00.0
Ig kappa variable 4-1IGKV4-15.70.60.393.30.1
Ig kappa constant, Ig kappa chain V-I regionIGKC7.40.50.391.40.4
Ig J polypeptide, linker for Ig alpha and muIGJ15.00.00.085.00.0
Ig lambda variable 1-44IGLV1-4426.90.40.571.60.6
Ig heavy constant deltaIGHD28.70.10.570.20.5
Cyclosporin A transporter 1CYAT128.30.40.670.20.5
Ig heavy constant muIGHM41.40.30.158.00.2
POU class 2 associating factor 1POU2AF147.40.00.052.50.1
CD79a molecule, Ig-associated alphaCD79A35.15.33.751.34.6
CD22 moleculeCD221000.00.00.00.0
Fc receptor–like AFCRLA1000.00.00.00.0
CD20 moleculeMS4A198.00.00.02.00.0
Pre–B lymphocyte 3VPREB397.00.50.32.20.0
T cell leukemia/lymphoma 1ATCL1A96.00.00.04.00.0
Early B cell factor 1EBF195.50.00.04.50.0
Fc receptor–like 1FCRL193.30.00.06.70.0
B cell protein with ankyrin repeats 1BANK189.60.00.79.70.0
Spermatid perinuclear RNA binding proteinSTRBP78.71.00.519.10.7
CD19 moleculeCD1977.11.30.420.90.3
B lymphoid tyrosine kinaseBLK69.10.00.029.61.2
Purinergic receptor P2X, ion channel, 5P2RX568.80.51.324.74.7
Oxysterol binding protein–like 10OSBPL1061.90.00.038.10.0
B cell linkerBLNK61.70.00.338.00.0
Tetraspanin 13TSPAN1356.90.00.043.10.0

We used PC spike-in experiments to assess the sensitivity and specificity of detecting changes in PC numbers using the PC gene expression signature. PBMCs were obtained from 4 blood donors, and PCs (CD27++CD38++) were isolated using standard flow cytometry methods. For each donor, unsorted PBMCs were compared with PBMCs spiked with each donor's PCs (i.e., endogenous PCs plus spiked-in PCs), which varied in number from 360 to 4,600. RNA was isolated, and genes identified to be PC-enriched were examined by TaqMan qPCR along with the general B cell markers CD19 and CD20. Results revealed a specific increase in the PC gene signature (Figure 1A) that was linearly related to the number of PCs spiked into each sample (R2 = 0.99, P = 0.002) (Figure 1B). The ability to detect an increase in the PC signature with the addition of only 360 PCs indicates that this signature is highly sensitive to small changes in PC levels. In contrast, increasing PC spike-in had little effect on the expression of CD19 or CD20 genes, which is consistent with expectations as these genes are broadly expressed on B cells, and therefore their expression levels are not sensitive to small changes in PC levels. The specificity of the PC signature for detecting changes in PCs only was highlighted by this lack of change in the B cell markers CD19 and CD20 (Figure 1A), as well as by the poor relationship between the PC count and the B cell signature (R2 = 0.39, P = 0.37) (Figure 1B).

Figure 1.

Genes identified to be enriched in the plasma cell (PC) population are selectively increased with increasing spike-in of purified PCs. PCs were sorted from peripheral blood mononuclear cells (PBMCs) isolated from 4 healthy donors. Sorted PCs were added to a second PBMC sample from the cognate donor, and samples with and without a PC spike were compared. A, Fold changes of the PC signature score, CD19 transcript levels, and CD20 transcript levels, as measured by TaqMan quantitative polymerase chain reaction following the addition of varying numbers of PCs (indicated by the change [Δ] in PCs). B, Linear relationship between the fold change in the PC signature (linear regression: P = 0.002) or fold change in the B cell signature (linear regression: P = 0.37) following spike-in and the approximate number of PCs added to the sample.

Comparing the regression coefficients between the PC signature (R2 = 0.99) and the individual genes (R2 = ∼0.9) that comprise the signature with the PC counts suggests that using a score that encapsulates information from a panel of genes rather than expression levels of an individual gene from the panel may be a more robust method of assessing PC levels (Figure 1B) (see Supplementary Figure 2, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38194/abstract). Finally, the strong linear relationship between PC signature score and PC counts determined by flow cytometry (R2 = 0.99) indicates that this gene expression score is a good proxy for flow cytometry.

Depletion of the PC signature by an anti-CD19 mAb

We used this newly developed gene expression–based PC signature to monitor changes in this cell population in the clinic. We analyzed samples from patients enrolled in MI-CP200, a phase I dose-escalation trial of MEDI-551 in scleroderma (NCT00946699). At various time points after a single dose of MEDI-551, gene expression changes in the PC and B cell signatures were evaluated in blood and skin samples (for specific sample numbers and time points, see Supplementary Table 1, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38194/abstract). Changes in these signatures were calculated relative to each patient's baseline sample.

MEDI-551 caused a robust reduction of the PC gene signature in whole blood, with maximum depletion of ∼98% and sustained depletion out to day 85 posttreatment (Figure 2A). Differences between baseline and post–MEDI-551 treatment values of the PC signature in whole blood were statistically significant (P < 0.01) at all time points measured. In contrast, there was little or no change in the PC signature at all time points in placebo-treated patients. A statistically significant reduction of the PC gene signature in skin samples following MEDI-551 treatment was also observed (P < 0.01), reaching a maximum level of 90% and a median of ∼55% (Figure 2B). Furthermore, in patients with matched blood and diseased tissue specimens, there was concordant inhibition of the PC gene signature in whole blood and skin (Spearman's rank correlation r = 0.72, P = 0.002) (Figure 2C).

Figure 2.

Significant inhibition of plasma cell (PC) signature in whole blood and skin following MEDI-551 treatment. A, PC signature score in whole blood, as evaluated by whole-genome array. Median fold change values compared to baseline are shown for all patients at each time point. Error bars indicate the median absolute deviation. ∗ = P < 0.01 versus baseline, by Mann-Whitney U test. B, PC signature score in skin, as evaluated by TaqMan quantitative polymerase chain reaction. Bars show median fold change values compared to baseline for all patients on day 29. Dashed line represents the baseline value (set to 1). Symbols represent individual patients. ∗ = P < 0.01 versus baseline, by Mann-Whitney U test. C, Correlation scatterplot of PC signature inhibition in skin and whole blood from patients with scleroderma.

In comparison with the PC signature, the B cell signature was reduced by a maximum of ∼90% in whole blood (Figure 3A). Alterations in CD19 transcript levels in whole blood (Figure 3B) were very similar to those observed for the B cell signature. All differences between baseline and post–MEDI-551 treatment values of CD19 transcript levels and B cell signature were statistically significant (P < 0.05) at all time points measured. For the B cell signature and CD19 transcript levels, statistically significant (P < 0.05) reductions were also observed in the skin, similar to what was seen for the PC signature (Figures 3C and D). Additionally, in patients with matched specimens, concordant inhibition of CD19 and the B cell signature in whole blood and skin was observed, comparable to what was seen for the PC signature (for CD19, Spearman's rank correlation r = 0.80, P = 0.001; for B cell signature, Spearman's rank correlation r = 0.79, P = 0.0002). In both whole blood and skin, treatment with placebo did not affect the PC signature, the B cell signature, or CD19 transcript levels, indicating that the observed effects were specifically due to MEDI-551 administration. However, the small number of placebo-treated subjects is a limitation of this analysis.

Figure 3.

Significant inhibition of B cell signature and transcript levels of CD19 in whole blood and skin following MEDI-551 administration. A and B, B cell signature score (A) and CD19 transcript levels (B) were evaluated in whole blood from scleroderma patients by whole-genome array. Median fold change values compared to baseline are shown for all patients at each time point. Error bars indicate the median absolute deviation. C and D, B cell signature score (C) and CD19 transcript levels (D) in skin were evaluated by TaqMan quantitative polymerase chain reaction. Bars show median fold change values compared to baseline for all patients on day 29. Dashed line represents the baseline value (set to 1). Symbols represent individual patients. ∗ = P < 0.05 versus baseline, by Mann-Whitney U test.

Correlation of PC signature and collagen gene expression

Excess deposition of collagen eventually resulting in skin fibrosis is a characteristic feature of scleroderma, and positive correlations between levels of type V collagen and measures of scleroderma disease activity, including the Valentini Disease Activity Index and the modified Rodnan skin thickness score (MRSS) ([29]), have been reported ([30, 31]). The availability of skin biopsy samples before and after treatment with MEDI-551 in the MI-CP200 trial afforded us the opportunity to assess the impact of depleting PCs and B cells on collagen gene expression in the skin. Following anti-CD19 treatment, we found that median expression of 3 collagen genes, COL1A1, COL3A1, and COL5A1 (termed the collagen score), was inhibited in skin (Figure 4A). Interestingly, inhibition of the PC signature was positively correlated with inhibition of the collagen score in these patients (Spearman's rank correlation r = 0.80, P = 0.001) (Figure 4B). Inhibition of the B cell signature was also correlated with inhibition of the collagen score, although to a lesser extent (Spearman's rank correlation r = 0.59, P = 0.04). It remains to be determined whether PC depletion by MEDI-551 results directly in a reduction in collagen gene expression; nonetheless, this correlation demonstrates the potential for the PC signature to be a useful pharmacodynamic marker in trials of PC-depleting therapeutics.

Figure 4.

Concordant inhibition of plasma cell (PC) signature score and collagen score in skin from patients in the MI-CP200 study. Transcript levels of 3 collagen genes (COL1A1, COL3A1, and COL5A1) were evaluated in skin from patients in the MI-CP200 study by TaqMan quantitative polymerase chain reaction. Median inhibition of the 3 collagen genes tested compared to baseline values was calculated for each patient to generate a collagen gene score. A, Fold change in the collagen score on day 29 in placebo-treated and MEDI-551–treated patients. Bars indicate the median in each group. Dashed line represents the baseline value (set at 1). Symbols represent individual patients. B, Correlation scatterplot of inhibition of the PC signature in skin and inhibition of the collagen score in skin.

PC signature in blood and diseased tissue from patients with multiple autoimmune diseases

The large database of gene expression profiles from patients with SLE, RA, myositis, psoriasis, scleroderma, asthma, and osteoarthritis (OA) that we have previously described ([27, 32, 33]) permitted us to assess the distribution of the PC and B cell signatures across these diseases. Levels of the PC signature in whole blood from patients with SLE (n = 360), myositis (dermatomyositis [DM; n = 160], polymyositis [n = 27], and inclusion body myositis [n = 14]), scleroderma (n = 59), psoriasis (n = 29), RA (n = 167), and asthma (n = 133) were compared to those in whole blood from healthy donors (n = 53) (Figure 5A). Results indicated a statistically significant increase of the PC signature in blood from SLE patients compared to blood from healthy donors (P < 0.01), with a subset of patients expressing 5–8-fold higher levels of the PC signature than that observed in samples from healthy donors. No other statistically significant increases in the PC signature were observed in the other autoimmune diseases tested.

Figure 5.

Enrichment of plasma cell (PC) signature in whole blood and tissue from patients with various autoimmune diseases. A, Levels of the PC signature in whole blood from patients with systemic lupus erythematosus (SLE; n = 360), myositis (dermatomyositis [DM; n = 160], inclusion body myositis [IBM; n = 14], and polymyositis [PM; n = 27]), systemic sclerosis (SSc; n = 59), psoriasis (n = 29), rheumatoid arthritis (RA; n = 167), and asthma (n = 133) were compared to those in whole blood from healthy donors (n = 53). B, Levels of the PC signature in skin samples from patients with SLE (n = 29), DM (n = 52), SSc (n = 34), and psoriasis (n = 66) were compared to those in skin samples from healthy donors (n = 34). C, Levels of the PC signature in synovial tissue from patients with RA (n = 27) and osteoarthritis (OA; n = 31) were compared to those in synovium from healthy donors (n = 30). Bars show mean fold change values of the PC signature (log2 scale). Dashed line represents the baseline value (set to 1). Symbols represent individual patients. ∗ = P < 0.01 by Welch's t-test.

Levels of the PC signature in skin samples from patients with SLE (n = 29), DM (n = 52), scleroderma (n = 34), and psoriasis (n = 66) were compared to those in skin samples from healthy donors (n = 34) (Figure 5B). We found a statistically significant increase of the PC signature in skin from SLE and scleroderma patients (P < 0.01) compared to that in skin from healthy donors. As observed in whole blood, a subset of SLE patients displayed 6–20-fold higher levels of the PC signature than healthy donors (Figure 5B). Of the evaluated blood and tissue samples from SLE patients, 17 were matched. Overall, a moderate positive correlation (Spearman's rank correlation r = 0.67, P = 0.001) was observed across all matched samples; however, only 5 of the skin samples with the highest PC signature levels had a matched whole blood sample. Therefore, a complete comparison of this high PC signature expression subset between blood and tissue was not feasible. In addition to skin specimens, synovial tissue from patients with OA (n = 31) and RA (n = 27) was compared to synovium from healthy donors (n = 30) (Figure 5C). A statistically significant increase (P < 0.01) was observed in the PC signature in synovium from RA, but not OA patients, as compared to that in synovium from healthy donors (Figure 5C), which is consistent with the documented role of B cells in RA ([7]).

In contrast to the PC signature, no significant differences were observed for the B cell signature in whole blood or diseased tissue from patients with the autoimmune diseases evaluated, as compared to healthy donors (see Supplementary Figure 3, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38194/abstract). A trend toward increased B cell signature scores in skin samples from patients with SLE and scleroderma was evident, as observed for the PC signature, but the magnitude of increase was considerably lower, highlighting the value of specifically investigating PCs.

Lack of correlation of PC signature with other autoimmune disease–related pathways

To determine the correlation of the PC signature with other disease-related pathways, we examined genes from pathways known to be dysregulated in scleroderma and other autoimmune diseases, such as vascular function, chemokine/cytokine signaling, interferon (IFN) signaling, and fibrosis. These genes include the IFN signature ([27]) as well as chemokine (C-C motif) ligand 2 (CCL2), interleukin-8 (IL8), platelet-derived growth factor alpha (PDGFA), vascular endothelial growth factor A (VEGFA), chemokine (C-X-C motif) receptor 4 (CXCR4), transforming growth factor, beta 1 (TGFB1), angiopoietin-2 (ANGPT2), fibroblast growth factor 2 (FGF2), matrix metallopeptidase 9 (MMP9), matrix metallopeptidase 10 (MMP10), and chemokine (C-X3-C motif) ligand 1 (CX3CL1). No correlation was observed between PC levels and the IFN signature or genes involved in chemokine/cytokine signaling or vascular function.

DISCUSSION

Depletion of B cells with therapies targeting the B cell membrane protein CD20 has previously shown beneficial effects in RA, MS, and other autoimmune diseases. CD19 is expressed on a broader range of B cell subsets than is CD20, including plasmablasts and some PCs, which are the major source of autoantibody production ([23, 24]). The role of PCs in maintaining chronic autoimmunity suggests that depletion of these cells could have a positive effect on clinical outcomes in multiple autoimmune diseases. Consistent with this hypothesis are recent results indicating that bortezomib inhibited PCs and lupus-specific autoantibodies in 4 patients with refractory SLE, leading to clinical improvement ([34]).

To investigate PCs in autoimmune disease and to address the unmet need for a sensitive method for assessing PC levels in the clinic, we developed a robust PC gene expression signature and confirmed its sensitivity and specificity through ex vivo experiments with purified PCs. By using a panel of genes, the PC signature used here may be more robust than relying on single genes (see Supplementary Figure 1, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38194/abstract), such as IGJ, which was previously used as a marker of PCs in RA ([25]). Additionally, the utility of our PC gene signature was demonstrated by 1) showing that it could be easily and reproducibly measured in both blood and diseased tissue collected routinely as part of a clinical trial and 2) confirming that despite low levels of PCs in blood and skin, we could monitor specific changes in this population due to anti-CD19 treatment.

Excessive production of collagen and changes in the architecture of connective tissue components are considered the hallmark of scleroderma ([35]). Multiple studies have shown that increased deposition of types III and V collagen occurs early in scleroderma development, with increased deposition of type I collagen occurring in late stages of the disease ([31, 36, 37]). Furthermore, overexpression of type V collagen in the skin has been positively correlated with measures of scleroderma disease activity, including the MRSS and the Valentini Disease Activity Index ([31]). A correlation between fibrosis in scleroderma skin and the MRSS has also been identified, as well as an association between gene expression levels of type I collagen and the MRSS ([35]). To account for the dynamic process of collagen remodeling and its role in scleroderma progression, the collagen score measured in this study included types I, III, and V collagen. The positive correlation observed between inhibition of the PC signature and inhibition of this collagen score in scleroderma skin supports a role for PCs in the pathogenic mechanism of this disease. Additionally, the association of collagen deposition with skin fibrosis and disease activity suggests that inhibition of PCs by MEDI-551 may have an effect on scleroderma disease activity in some patients. These preliminary results need to be validated in a larger cohort of patients to determine if changes in the PC signature and the collagen score predict those who may respond to treatment with B cell– or PC-depleting therapy such as MEDI-551.

Identifying diseases with increased PC signatures may help to identify subsets of patients who may respond to anti-CD19 therapies such as MEDI-551. We identified increases in the PC signature in scleroderma skin as compared to healthy donors. Previous work characterized multiple molecular subsets evident in skin from scleroderma patients and identified various gene clusters, including a proliferation cluster, immunoglobulin cluster, and IFNγ gene expression cluster, among others ([38, 39]). We found that the PC signature was enriched in the immunoglobulin cluster, as expected, but was not consistently associated with any others. As these molecular subsets continue to be refined, understanding their relationship to the PC signature may reveal relationships important for scleroderma pathogenesis.

In addition to scleroderma, we identified an increased PC signature in RA synovium compared to normal synovium. In a subset (∼35%) of RA patients whom we evaluated, the PC signature was 2-fold greater than that found in unaffected subjects. The role of B cells in RA is well documented ([7, 8, 24]) and is consistent with the efficacy of the anti-CD20 therapy rituximab for the treatment of RA ([7, 8]). The presence of an increased level of PCs in the synovium may help explain why certain patients respond poorly to anti-CD20 therapies, which do not deplete PCs ([25]). The plasmablast marker gene IGJ used by Owczarczyk and colleagues is one of the genes in the PC gene signature described in this article, and it would be interesting to assess whether the predictive values derived by these authors could be further refined by the use of a composite PC signature that we developed. Regardless, these results highlight the importance of measuring PCs prior to therapy and, potentially, understanding their pattern of depletion and recovery following B cell depletion therapy.

Interestingly, elevation of the PC signature was also identified in ∼30% of blood and skin specimens from SLE patients analyzed in this study. The importance of B cells in SLE is supported by the recent approval of belimumab (BAFF blockade) for nonrenal lupus ([40]) and the off-label use of rituximab in these patients (summarized in ref.[9]). Additionally, the therapeutic relevance of PCs in SLE is supported by preclinical and clinical data with the proteasome inhibitor bortezomib, which depletes bone marrow PCs in a mouse model of SLE ([41]) and leads to clinical benefit with an associated reduction in autoantibodies and immunoglobulin levels in patients with refractory SLE ([34]). Therefore, the PC signature developed in this study has the potential to determine whether PC levels predict response to various targeted therapies, potentially providing clinical benefit to patients with difficult-to-treat autoimmune diseases.

Taken together, our results indicate that our newly developed PC signature is highly sensitive in detecting changes in the PC population. Additionally, this signature can be used to effectively monitor PC levels in a clinical trial setting where other methods of tracking PCs are not feasible, as well as to evaluate various inflammatory and autoimmune diseases for the existence of elevated PC levels. Finally, independent of the particular targeted therapy, the ability to sensitively and accurately assess PC levels in the blood and diseased tissue will be useful, not only for monitoring therapeutic efficacy, but also for evaluating the hypothesis that baseline PC levels or the extent of PC depletion can identify subsets of patients who are more likely to respond to these therapies.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Drs. Streicher and Ranade had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Streicher, Morehouse, Groves, Brohawn, McKeever, Richman, Jallal, Herbst, Ranade.

Acquisition of data. Streicher, Morehouse, Groves, Rajan, Pilataxi, Lehmann, Brohawn, McKeever, Fiorentino, Yao.

Analysis and interpretation of data. Streicher, Morehouse, Higgs, Greenberg, Ranade.

ADDITIONAL DISCLOSURE

Authors Streicher, Morehouse, Groves, Rajan, Pilataxi, Lehmann, Brohawn, Higgs, McKeever, Richman, Jallal, Herbst, Yao, and Ranade are employees of MedImmune.

Ancillary