B lymphocyte stimulator (BLyS; a trademark of Human Genome Sciences, Inc., Rockville, MD, USA) is a 285-amino-acid member of the tumor necrosis factor ligand superfamily 123. A causal relation between constitutive overproduction of BLyS and development of systemic lupus erythematosus (SLE)-like illness has incontrovertibly been established in mice. BLyS-transgenic mice often develop SLE-like features as they age 345, and SLE-prone (NZB × NZW)F₁ (BWF₁) and MRL-lpr/lpr mice respond clinically to treatment with BLyS antagonists (decreased disease progression and improved survival) 36.

Considerable inferential evidence points to a role for BLyS overproduction in human SLE as well. Cross-sectional studies have demonstrated elevated circulating levels of BLyS in 20–30% of human SLE patients tested at a single point in time 78. Moreover, a 12-month longitudinal study documented persistently elevated serum BLyS levels in about 25% of SLE patients and intermittently elevated serum BLyS levels in an additional 25% of patients 9. Remarkably, circulating BLyS levels did not correlate with disease activity (measured using the SLE Disease Activity Index [SLEDAI]) in these cross-sectional or longitudinal studies 789. Although a statistically significant correlation between circulating BLyS levels and SLEDAI has been appreciated in a more recent 24-month longitudinal study of 245 SLE patients (with >1,700 plasma samples analyzed) 10, the correlation remains weak.

The limited correlation between circulating BLyS protein levels and disease activity in these studies may have exposed an inadequacy of the former to reflect faithfully endogenous BLyS overproduction. In addition to the rate of BLyS protein production, several other factors (for example, utilization and excretion) can affect circulating BLyS protein levels. Although there are no practicable means of directly measuring in vivo BLyS production per se in humans, the level of BLyS mRNA may serve as a better surrogate marker of in vivo BLyS production than does the level of BLyS protein. Candidate BLyS mRNA isoforms include the full-length BLyS mRNA isoform, which encodes the full-length protein, and the alternatively spliced ΔBLyS mRNA isoform, which encodes a protein with a small peptide deletion 11. (ΔBLyS does not bind to cells expressing BLyS receptors, and therefore it has no agonistic activity. Moreover, ΔBLyS can form heterotrimers with full-length BLyS, thereby actually functioning as a dominant-negative antagonist of BLyS activity.)

In this report we demonstrate that peripheral blood leukocytes from SLE patients express elevated mRNA levels of both full-length BLyS and ΔBLyS relative to those levels expressed by patients with rheumatoid arthritis (RA) or by normal control individuals. In the SLE patients, both full-length BLyS and ΔBLyS mRNA levels are more closely associated with disease activity (SLEDAI) than are BLyS protein levels. Accordingly, BLyS mRNA levels may be a helpful biomarker in the clinical monitoring of SLE patients.

Elevated blood levels of BLyS protein and mRNA are well described features of human SLE 789. We confirmed these observations in our study and extended them by documenting increases not just in levels of full-length BLyS mRNA but also in levels of ΔBLyS mRNA (Figure 1). Of note, BLyS mRNA levels were elevated in SLE but not in RA, raising the possibility that BLyS overproduction in SLE is systemic whereas BLyS overproduction in RA may be more focused to the affected arthritic joints 24. The modestly elevated plasma BLyS protein levels in RA patients may reflect, at least in part, release of locally overproduced BLyS into the circulation.

The relationship between circulating BLyS protein levels and disease activity was addressed in several previous studies, but significant correlations between the two measures did not emerge 789. In the largest study to date, a 2-year longitudinal study of 245 patients in which more than 1,700 plasma samples were analyzed, a significant but weak correlation between the two was appreciated 10. In the present study, a significant correlation between plasma BLyS protein levels and disease activity was again not realized (Figure 3a,d).

The weak, at best, correlation between circulating BLyS levels and disease activity is seemingly rather surprising. There is a clear-cut association in BlyS transgenic mice between BLyS overexpression and development of SLE-like features 345, and treatment of SLE-prone mice with BLyS antagonists retards the progression of disease and improves survival 36. Moreover, development of precocious glomerular pathology in autoimmune-prone mice correlates strongly with circulating BLyS levels 25.

The likely explanation for the weak correlation between circulating BLyS levels and disease activity in human SLE is not that disease activity in SLE patients is insensitive to the degree of BLyS overproduction. Rather, a more tenable explanation is that circulating BLyS levels in human SLE do not always accurately reflect excessive endogenous BLyS production. We can identify at least three nonmutually exclusive mechanisms to explain a dissociation between the two.

First, SLE patients frequently develop autoantibodies to a myriad of self-targets (for example, erythrocytes, lymphocytes). Indeed, we detected circulating IgA/IgM/IgG anti-BLyS autoantibodies in 10% (6/60) of the tested SLE patients, and we detected circulating IgG anti-BLyS autoantibodies in 5% (3/60). These percentages may be underestimates of the true prevalence of anti-BLyS autoantibodies, because some of these autoantibodies may be saturated in vivo with circulating BLyS, rendering them incapable of binding to BLyS in the in vitro detection assay. We do not yet know whether the anti-BLyS autoantibodies are functionally neutralizing but, regardless, such autoantibodies could enhance the clearance of BLyS and/or interfere with in vitro detection of BLyS, thereby masking endogenous BLyS overproduction.

Second, increased urinary excretion of BLyS has been reported in SLE patients, especially among those with clinically overt renal involvement 26. At least four of the patients we studied manifested nephrotic-range proteinuria (≥3 g/24 hours), and so urinary loss of BLyS was probably substantial in at least these patients. A validated assay for urinary BLyS detection has not yet been developed so we were unable to quantify urinary BLyS levels. Once an assay for urinary BLyS levels is validated, we should be able to assess the effect of urinary BLyS excretion on circulating BLyS levels.

Third, BLyS promotes in vivo expansion of B cells 1. Freshly isolated SLE B cells, despite intact surface expression of BLyS receptors, bind less biotinylated BLyS ex vivo than do freshly isolated normal B cells 27. Although other interpretations are possible, the most likely explanation is that BLyS receptors on B cells in SLE patients are occupied in vivo by soluble BLyS. Accordingly, it is likely that BLyS receptors expressed by the expanded B cell population do bind BLyS and remove it from the circulation, resulting in a homeostatic pathway that modulates the effects of BLyS overproduction on circulating BLyS levels. Indeed, circulating levels of BLyS rise with peripheral blood B cell depletion and fall with re-emergence of peripheral blood B cells in rituximab-treated RA or SLE patients 2829, highlighting this inverse relationship between circulating BLyS levels and B cell load. Moreover, one of the hallmarks of active disease in human SLE is the increased percentages of activated B cells and plasma cells in peripheral blood 3031323334, probably reflecting increased systemic numbers of activated B cells and plasma cells. Although not yet formally tested, differential BLyS receptor expression by these cells compared with expression by nonactivated B cells may result in increased peripheral BLyS utilization, further dampening the effects of BLyS overproduction on circulating protein levels.

To circumvent these confounding processes, we used BLyS mRNA levels as a surrogate marker of endogenous BLyS production. Overall, the correlations between disease activity and either full-length BLyS or ΔBLyS mRNA levels were much stronger than that between disease activity and BLyS protein levels (Figures 3 and 4). This was the case regardless of whether we used the standard SLEDAI or the modified SLEDAI as a measure of disease activity. These correlations were not spurious ones consequent to shifts in percentages of leukocyte subpopulations in peripheral blood, because BLyS mRNA levels did not correlate with percentages of blood neutrophils, monocytes, or lymphocytes (Figure 5).

A similar pattern was observed between plasma immunoglobulin levels and the BLyS parameters, with plasma levels of total immunoglobulin, IgG, and IgA correlating significantly with full-length BLyS and ΔBLyS mRNA levels but not with plasma BLyS levels (Figure 2). These significant correlations between full-length BLyS or ΔBLyS mRNA levels and plasma immunoglobulin levels again highlight the greater ability of BLyS mRNA levels, compared with plasma BLyS protein levels, to reflect ongoing BLyS overproduction.

At present, it is not known whether soluble ΔBLyS protein is present in the circulation of SLE patients or of normal individuals. Although full-length BLyS protein is readily cleaved and released from cells transfected with a vector containing murine full-length BLyS, ΔBLyS protein is not cleaved or released from murine ΔBLyS transfectants 11. Given the strong similarities between murine and human full-length BLyS and ΔBLyS, it is likely that human soluble ΔBLyS protein is also not cleaved from the cell surface and released into the circulation. Moreover, soluble ΔBLyS protein is not released from cells transfected with a vector containing just the extracellular domain of human ΔBLyS (which encodes the soluble protein; A.L. Gavin, unpublished observations). Whether this reflects rapid intracellular degradation of soluble ΔBLyS or some other impediment to its release remains unknown. Regardless, if the inability to release soluble ΔBLyS in vitro faithfully recapitulates in vivo biology, then the stronger associations between SLE disease activity and full-length BLyS or ΔBLyS mRNA levels compared with that between SLE disease activity and BLyS protein levels could not be attributable to interference by biologically inactive (inhibitory) ΔBLyS protein in the BLyS protein detection ELISA. Importantly, even if soluble ΔBLyS protein is present in the circulation and is detected by the BLyS protein detection ELISA, then the stronger correlations between SLE disease activity and full-length BLyS or ΔBLyS mRNA levels than that between disease activity and total BLyS (including ΔBLyS) protein levels suggest that full-length BLyS and/or ΔBLyS mRNA levels may operationally serve as useful biomarkers of disease activity in SLE. Although the complexity and labor intensiveness associated with quantitative real-time PCR may render measurement of BLyS mRNA levels impracticable for routine clinical practice, such measurement could readily be incorporated into clinical trials and yield valuable information.

Longitudinal observations in large numbers of SLE patients will be necessary to establish or refute the utility of full-length BLyS and/or ΔBLyS mRNA to subserve this clinically vital function.

Although expression of the two major BLyS isoforms was highly coordinate among SLE patients, there were several patients in whom ΔBLyS mRNA levels were markedly greater than or less than the expected values based on full-length BLyS mRNA levels (data not shown). This raises the possibility that dysregulation of ΔBLyS may contribute to overall BLyS dysregulation in at least some SLE patients. It is known that interferon-γ, interleukin-10, interferon-α, and CD154 can upregulate full-length BLyS mRNA levels 142235, but it is not known what effects these or other cytokines/cell-surface structures have on ΔBLyS expression. Further investigation of the regulation of ΔBLyS and the differential expression of BLyS isoforms is certainly warranted.

Although the associations between full-length BLyS and/or ΔBLyS mRNA levels and disease activity in SLE were usually strong when the SLE cohort was analyzed in aggregate, there were several SLE patients in whom BLyS mRNA levels were quite high despite little objective ongoing disease activity, and there were several SLE patients in whom BLyS mRNA levels were low despite considerable ongoing disease activity. One must recognize that the bulk of the pathogenic autoimmune responses probably takes place in the spleen and lymph nodes, rather than in the peripheral blood, where myeloid lineage cells (for example, dendritic cells) produce BLyS and support B cell survival and expansion 36. Local BLyS production in the secondary lymphoid tissues will be more important to the development and maintenance of an autoimmune response than will remote BLyS levels in the circulation. Because at least some autoreactive B cells may be more sensitive to BLyS-mediated survival signals than non-autoreactive B cells 3738, local increases in BLyS production could preferentially promote expansion of autoreactive B cells. These cells, in turn, could activate autoreactive T cells by presenting autoantigen to them, and some of the autoreactive B cells would respond to T cell derived signals and mature into (pathogenic) autoantibody secreting plasma cells. In contrast to murine studies, in which investigators can readily harvest and analyze lymphoid and myeloid lineage cells from any site (for example, spleen, bone marrow), such is not the case for human studies. Peripheral blood is the only site readily accessible for human studies, and it is possible that, at least in some patients, BLyS mRNA levels in circulating leukocytes do not reflect local BLyS production in the secondary lymphoid tissues.

One must also recognize that disease activity in SLE is not solely driven by B cells. Systemic inflammation and SLE flares can be triggered via B cell independent means. Not all SLE patients treated with a B cell depleting course of rituximab experience clinical remission 39, strongly pointing to the importance of non-B cells in disease pathogenesis/maintenance. Conversely, not all pathogenic B cells necessarily require high levels of BLyS to effect their pathogenicity. Murine studies have unequivocally documented B cell subpopulations that do not depend upon BLyS for their survival 404142. Although mice completely devoid of BLyS have reduced numbers of mature B cells and harbor reduced levels of immunoglobulin, these reductions are incomplete. Thus, it is possible that some SLE patients harbor pathogenic B cells that are relatively insensitive to BLyS and could drive considerable disease activity even in the context of low endogenous BLyS production. Conversely, patients with high BLyS mRNA levels may be those patients whose disease is strongly driven by BLyS and may be especially helped by BLyS antagonist therapy. Future clinical trials should be able to establish whether the BLyS mRNA levels are good predictors of response to such agents.

Plasma total immunoglobulin, IgG, and IgA levels and disease activity (as measured by SLEDAI) in SLE patients correlate more closely with peripheral blood leukocyte levels of BLyS mRNA than with plasma levels of BLyS protein. These findings suggest that BLyS mRNA levels better reflect in vivo BLyS production than do circulating BLyS protein levels, and may be a useful biomarker in the clinical monitoring of SLE patients. These findings also support the premise that BLyS overexpression not only promotes development of disease but also actively contributes to the ongoing maintenance of disease in SLE patients. This reinforces the rationale underlying clinical trials with BLyS antagonists in SLE.

anti-dsDNA = anti-double-stranded DNA; BLyS = B lymphocyte stimulator; bp = base pairs; Ct = threshold cycle; ELISA = enzyme-linked immunosorbent assay; PCR = polymerase chain reaction; RA = rheumatoid arthritis; SLE = systemic lupus erythematosus; SLEDAI = SLE Disease Activity Index.

TSM and DMH were employees of Human Genome Sciences (HGS) at the time the investigation was conducted. (DMH has since left the company.) WS has received research support from HGS and has served as a consultant to HGS (<$10,000). CEC, ALG, and DN declare that they have no competing interests.

CEC identified and recruited all participants; collected all the blood samples and reviewed all the medical charts; and wrote the initial working draft of this manuscript. ALG developed and performed all the real-time PCR assays and assisted in the interpretation of the results and in writing the final version of the manuscript. TSM performed the plasma BLyS protein and anti-BLyS assays and assisted in the interpretation of the results and in writing the final version of the manuscript. DMH assisted in the design in the study, in the interpretation of the results, and in writing the final version of the manuscript. DN assisted in the design in the study, in the interpretation of the results, and in writing the final version of the manuscript. WS conceived the study, supervised the recruitment of participants, performed the statistical analyses, assisted in the interpretation of the results, and supervised the editing of the manuscript to its final form. All authors read and approved the final manuscript version.