To determine the role of APRIL in the development of systemic lupus erythematosus (SLE) in mice.
To determine the role of APRIL in the development of systemic lupus erythematosus (SLE) in mice.
Wild-type (WT) NZM 2328, NZM. April–/–, NZM.Baff–/–, and NZM.Baff–/–.April–/– mice were evaluated for lymphocyte phenotype by flow cytometry, for serum total IgG and IgG autoantibody levels by enzyme-linked immunosorbent assay, for glomerular deposition of IgG and C3 by immunofluorescence, for renal changes by histopathology, and for clinical disease by laboratory assessment (severe proteinuria).
In comparison to WT mice, NZM.April–/– mice harbored increased spleen B cells, T cells, and plasma cells (PCs), increased serum levels of IgG antichromatin antibodies, and decreased numbers of bone marrow (BM) PCs. Glomerular deposition of IgG and C3 was similar in NZM.April–/– mice and WT mice, renal changes on histopathology tended to be more severe in NZM.April–/– mice than in WT mice, and development of clinical disease was identical in NZM.April–/– mice and WT mice. BM (but not spleen) PCs and serum IgG antichromatin and anti–double-stranded DNA antibody levels were lower in NZM.Baff–/–.April–/– mice than in NZM.Baff–/– mice, whereas renal immunopathology in each cohort was equally mild.
APRIL is dispensable for the development of full-blown SLE in NZM mice. Moreover, the elimination of both APRIL and BAFF had no discernible effect on the development of renal immunopathology or clinical disease beyond that of elimination of BAFF alone. The reduction in BM PCs in hosts doubly deficient in APRIL and BAFF beyond that in hosts deficient only in BAFF raises concern that combined antagonism of APRIL and BAFF may lead to greater immunosuppression without a concomitant increase in therapeutic efficacy.
The clinical diagnosis of systemic lupus erythematosus (SLE) is based on a constellation of signs, symptoms, and clinical laboratory–based abnormal findings. This inherent heterogeneity notwithstanding, certain common threads tend to run across all “varieties” of SLE. One of these common threads is B cell hyperactivity. A priori, any factor that enhances B cell survival and/or function may play a pathogenetic role in SLE.
One such factor is BAFF (commonly also known as B lymphocyte stimulator [BLyS]), a 285–amino acid type II transmembrane protein member of the tumor necrosis factor (TNF) ligand superfamily (1, 2). Cleavage of surface BAFF by a furin protease results in the release of a soluble, biologically active 17-kd molecule (2, 3) that binds to 3 receptors (BCMA, TACI, and B lymphocyte stimulator receptor 3 [BR3; also known as BAFF receptor]) on the surface of B cells (4–7).
The connection between BAFF and the development of SLE is very strong. In mice that otherwise are not autoimmune-prone, constitutive overexpression of BAFF consequent to the introduction of a BAFF transgene leads to the expression of SLE-like features, including elevated circulating titers of multiple autoantibodies and immune complex–mediated glomerulonephritis (8–10). Moreover, introduction of a BAFF transgene into mice that otherwise have an incomplete diathesis to SLE (i.e., mice that develop SLE-associated autoantibodies but rarely develop renal disease) results in precocious development of glomerular disease (11). Indeed, SLE-associated autoantibodies develop in BAFF-transgenic (Tg) mice even in the absence of class II major histocompatibility complex expression (12), and SLE-like features develop in otherwise non–autoimmune-prone BAFF-Tg mice even in the complete absence of T cells as long as myeloid differentiation factor 88–mediated signaling is intact (13).
Observations in humans are consonant with those in mice. Circulating BAFF levels are elevated in as many as 50% of SLE patients (14–16), and a large 2-year longitudinal study and 3 independent smaller studies documented a significant correlation between BAFF expression and clinical disease activity (17–20).
Importantly, prevention/amelioration of SLE is associated with elimination/neutralization of BAFF as well. Treatment of SLE-prone (NZB × NZW)F1 (NZB/NZW) mice or MRL/lpr mice with a BAFF antagonist retards disease progression and improves survival (10, 21–23). Furthermore, very limited clinical disease develops in SLE-prone NZM 2328 (NZM) mice (an inbred recombinant strain derived from NZB/NZW mice that closely mirrors the parental SLE phenotype and shares many features with human SLE ) in which the Baff gene has been disrupted (25). Indeed, NZM.Baff–/– mice are completely protected from the virulent SLE disease that rapidly develops in young (preautoimmune) wild-type (WT) NZM mice upon overexpression of interferon-α (26). Finally, the anti-BAFF monoclonal antibody (mAb) belimumab has been shown to be efficacious in two separate phase III trials in humans with SLE (27, 28).
In contrast to the large body of evidence that links BAFF with SLE, the evidence linking the closely related APRIL to SLE is limited at most. Although APRIL does not bind to BR3 (6), its 3-dimensional structure is sufficiently similar to that of BAFF to permit APRIL to bind to the other two BAFF receptors (BCMA and TACI) (29–32). Of potential importance, BAFF and APRIL naturally form heterotrimers that are biologically active in in vitro assays and that circulate at elevated concentrations in patients with SLE (33, 34). What regulates the formation of heterotrimers rather than BAFF or APRIL homotrimers and whether the potency of the heterotrimers under in vivo conditions is greater than, equal to, or less than those of the BAFF or APRIL homotrimers remain unknown. Nevertheless, given that APRIL can costimulate B cells, induce Ig class switching, and promote plasma cell (PC) survival (29, 30, 35–38), APRIL may be an important (co-) contributor to SLE pathogenesis and, thereby, represents an appropriate therapeutic target in SLE.
To date, there have been no studies in either murine or human SLE that have solely targeted APRIL. That is, the benefit garnered from antagonizing APRIL per se remains uncertain. This point is not merely academic, since the combined antagonism of BAFF and APRIL may be more immunosuppressive than the antagonism of BAFF alone. Indeed, BR3-Ig (which antagonizes BAFF but not APRIL) and TACI-Ig (which antagonizes both BAFF and APRIL) displayed identical clinical efficacy in a head-to-head study despite TACI-Ig having a greater inhibitory effect on humoral immunity than did BR3-Ig (23).
To directly assess the contribution of APRIL to the development of SLE, we generated NZM.April–/– mice. Counterintuitively, T cells and B cells were globally expanded, and serologic autoimmunity and renal immunopathology tended to be greater, in NZM.April–/– mice than in WT mice. Moreover, development of renal immunopathology was identically mild in NZM.Baff–/–. April–/– mice as in NZM.Baff–/– mice despite the greater reduction of bone marrow (BM) PCs in the former. Collectively, these findings demonstrate that APRIL is dispensable for full-blown SLE in NZM mice, and they raise the concern that the combined antagonism of APRIL and BAFF may lead to greater immunosuppression beyond that of antagonism of BAFF alone but without a concomitant gain in therapeutic efficacy.
All mice were maintained in specific pathogen–free quarters at the University of Southern California. The experiments were approved by the Institutional Animal Care and Use Committee.
Female mice from 4 congenic NZM strains were studied: NZM WT, NZM.Baff–/–, NZM.April–/–, and NZM.Baff–/–.April–/–. The generation of NZM.Baff–/– mice has been previously described (25). To generate NZM.April–/– mice, the April–/– genotype from April–/– mice (mixed B6/129 background) (39) was introgressed into NZM mice via a marker-assisted selection protocol using markers to include those regions identified as susceptibility loci in NZM mice (25). To detect the disrupted April gene fragment (containing a neo insert), genomic DNA extracted from mouse tail clippings was polymerase chain reaction (PCR)–amplified for 30 cycles at 95°C for 60 seconds, 60°C for 30 seconds, and 72°C for 90 seconds. The primer sequences were as follows: for April, primer 1 was 5′-CAG-TCC-TGC-ATC-TTG-TTC-CA-3′ and primer 2 was 5′-GCA-GAT-AAA-TTC-CAG-TGT-CCC-3′; for neo, the primer was 5′-CTC-CCA-CTC-ATG-ATC-TAT-AAG-ATC-C-3′.
All 3 primers were added to a single reaction mixture. Band size for the intact April gene fragment is 712 bp, and band size for the disrupted April gene fragment (neo) is 464 bp. Thus, mice bearing the April+/+ genotype would yield only a 712-bp band, mice bearing the April–/– genotype would yield only a 464-bp band, and mice bearing the April+/– genotype would yield both the 712-bp and the 464-bp bands.
NZM.Baff–/–.April–/– mice were generated by intercrossing NZM.Baff–/– and NZM.April–/– mice and screening by PCR for the Baff–/–.April–/– genotype.
For determination of T cell and B cell subsets, murine spleen mononuclear cells were stained with combinations of fluorochrome-conjugated mAb specific for murine CD3, CD4, CD8, CD44, CD62L, CD45R (B220), CD19, CD21, CD23, or CD69 (BD PharMingen or eBioscience) and analyzed by flow cytometry (40). For PC determination, spleen cells or BM cells were surface-stained with a combination of fluorochrome-conjugated mAb specific for murine CD4, CD8, Gr1, F4/80, and B220 and intracellular-stained with fluorochrome-conjugated mAb specific for murine Ig κ and λ. PCs were taken to be the CD4–CD8–Gr1–F4/80–B220–Igκ/λ+ cells.
Serum levels of total IgG, IgG antichromatin, and IgG anti–double-stranded DNA (anti-dsDNA) antibodies were determined by enzyme-linked immunosorbent assay (ELISA) (26). Autoantibody optical density (OD) values were normalized to the mean OD of serum from 5-month-old MRL/lpr mice, the latter being arbitrarily assigned a value of 100 units/ml.
Serum BAFF was measured by ELISA (41). Assay plates were coated with a mouse BR3:human Fc fusion protein (Alexis Biochemicals) as the capture reagent, and biotinylated anti-mouse BAFF mAb 16D7 (Human Genome Sciences) followed by streptavidin–horseradish peroxidase was used as the detector. Results were interpolated from a mouse BAFF standard curve that was run on each plate.
Sections of formalin-fixed kidneys from WT or NZM.April–/– mice were stained with hematoxylin and eosin (H&E), periodic acid–Schiff (PAS), Masson's trichrome, and Jones' silver methenamine and were assessed by light microscopy for glomerular activity (hypercellularity, necrotizing lesions, karyorrhexis, cellular crescents, hyaline deposits), tubulointerstitial activity (interstitial cellular infiltration, tubular cell necrosis), chronic glomerular pathology (glomerulosclerosis, fibrous crescents), and chronic tubulointerstitial pathology (tubular atrophy, interstitial fibrosis). Each of the 4 categories was subjectively scored on a 0–3 scale, for a maximum composite score of 12. Kidneys from NZM.Baff–/– or NZM.Baff–/–.April–/– mice were evaluated after H&E staining.
Five-micron sections of snap-frozen kidneys were stained for total IgG deposition using fluorescein isothiocyanate–conjugated goat F(ab′)2 fragment anti-mouse IgG or C3 antibodies (MP Biomedicals) (26). Staining intensity was subjectively scored on a 0–4 scale.
Reagent strips for urinary protein (Albustix; Bayer) were dipped in mouse urine and were assigned a score (0–4 scale) by visual color comparison to the supplied standard color key. Severe proteinuria was defined as ≥3 on 2 consecutive examinations.
All analyses were performed using SigmaStat software (SPSS). Parametric testing between 2 matched or unmatched groups was performed by the paired or unpaired t-test, respectively. Parametric testing among 3 or more groups was performed by one-way analysis of variance (ANOVA). When the data were not normally distributed or the equal variance test was not satisfied, nonparametric testing was performed by the Mann-Whitney rank sum test between 2 groups and by Kruskal-Wallis one-way ANOVA on ranks among 3 or more groups.
In non–autoimmune-prone (mixed B6/129 background) mice, the numbers and percentages of total spleen B cells and T cells in April–/– mice were the same as those in WT mice (39, 42), although an increase in the percentage of memory (CD44highCD62Llow) CD4+ cells in the spleens of April–/– mice was reported by one group (39). Unexpectedly, NZM.April–/– mice underwent a global expansion of mononuclear cells in their spleens, affecting total B cells (CD19+), follicular B cells (CD23highCD21intermediate), marginal zone B cells (CD23lowCD21+), activated B cells (CD69+), total T cells (CD3+), total CD4+ cells, naive CD4+ cells (CD44lowCD62Lhigh), memory CD4+ cells (CD44high CD62Llow), and total CD8+ cells (Figures 1A–J). The percentages of these individual cell populations were not significantly different between WT mice and NZM.April–/– mice (data not shown), with the exception that the mean percentage of CD8+ cells was modestly lower in NZM.April–/– mice than in WT mice (19.7% versus 23.7%; P = 0.043).
Consistent with their global B cell (and T cell) expansion, NZM.April–/– mice harbored increased numbers of PCs in their spleens. In sharp contrast, BM PCs were markedly reduced in number (Figures 2A and B). This paucity of PCs in the BM of NZM.April–/– mice likely reflects the vital role of APRIL produced by BM stromal cells in plasmablast survival (38).
The dearth of BM PCs notwithstanding, serum IgG levels in NZM.April–/– mice tended to be higher, not lower, than those in WT mice. Similarly, serum IgG antichromatin, but not IgG anti-dsDNA, autoantibody levels were significantly higher in NZM.April–/– mice than in WT mice (Figures 2D–F). This disparity in serum IgG and IgG antichromatin levels between the two mouse cohorts cannot be attributed to a difference in circulating BAFF levels per se, since their serum BAFF levels were the same (Figure 2C).
At 3 months of age, the kidneys of NZM.April–/– mice and WT mice were histologically normal, with either no or minimal deposition of IgG or C3 in their glomeruli (data not shown).
By 5 months of age, similar moderate degrees of glomerular deposition of IgG and C3 had developed in WT and NZM.April–/– mice (Figure 3A, parts a–d). The deposits were predominantly in the mesangial areas, with minimal deposition along the glomerular capillary walls. Histologically, glomerular hypercellularity was readily appreciated in both mouse cohorts, with the hypercellularity actually being somewhat greater in NZM.April–/– mice than in WT mice. Glomerular size was increased in both, and cellular crescents were present. Moreover, mesangial matrix deposition (as detected by PAS staining) and collagen deposition in the glomeruli (as detected by silver staining) were readily detectable in both groups. In addition, some degree of interstitial inflammation with perivascular leukocyte infiltration was also observed in both groups (Figure 3A, parts e–l). Overall, the kidney scores tended to be higher in NZM.April–/– mice than in WT mice, although this difference did not achieve statistical significance (Figure 3B).
By 8 months of age, glomerular deposition of IgG, but not C3, had greatly increased in both mouse cohorts (Figure 3A, parts m–p). This was accompanied by marked renal histopathology, including the development of glomerular fibrotic crescents, glomerulosclerosis, tubular atrophy, increased interstitial inflammation and fibrosis, and striking degrees of perivascular leukocyte infiltration around both arterioles and venules (Figure 3A, parts q–x). As at 5 months of age, the kidney scores tended to be higher in NZM.April–/– mice than in WT mice, although this difference again did not achieve statistical significance (Figure 3B).
In both NZM.April–/– and WT mice, clinical disease (severe proteinuria) started to develop at 5 months of age, with 50% of the mice in each cohort being affected between the ages of 7 and 8 months (Figure 3C). Overall, the onset and incidence of clinical disease in these two cohorts were identical (Figure 3C). (Mortality could not be accurately assessed, since clinically sick mice were routinely euthanized upon development of ascites and lethargy.)
Although the development of clinical disease in NZM.April–/– mice was identical to that in WT mice, the increased numbers of T cells and B cells, the increased circulating levels of IgG antichromatin antibodies, and the trend toward more severe renal immunopathology collectively suggested that the life-long absence of APRIL might promote or accelerate, rather than inhibit or delay, the development of SLE. Since TACI-Ig (atacicept), which antagonizes both BAFF and APRIL, has been evaluated in murine SLE (10, 22, 23) and is presently undergoing clinical evaluation in human SLE (43), we wanted to determine whether elimination of both BAFF and APRIL might lead to a more severe SLE phenotype than that resulting from elimination of BAFF only.
Comparison of NZM.Baff–/–.April–/– mice to NZM.Baff–/– mice revealed that this was not the case. Both spleen and BM PC populations were reduced in NZM.Baff–/– mice relative to those in WT mice (P = 0.016 and P = 0.007, respectively) (Figures 4A and B). Of note, BM, but not spleen, PCs in NZM.Baff–/–. April–/– mice were further reduced relative to those in NZM.Baff–/– mice (P = 0.016 for BM PCs and P = 0.222 for spleen PCs) (Figures 4A and B), again consistent with an essential role in plasmablast survival for BM stromal cell–produced APRIL (38). Moreover, serum levels of total IgG tended to be lower in NZM.Baff–/–. April–/– mice than in NZM.Baff–/– mice (P = 0.089) (Figure 4C), and serum IgG antichromatin and anti-dsDNA antibody levels were unmistakably lower in NZM.Baff–/–.April–/– mice than in NZM.Baff–/– mice (P = 0.005 and P < 0.001, respectively) (Figures 4D and E). Of note, serum IgG anti-dsDNA antibody levels were very high in some NZM.Baff–/– mice, which is consistent with our previous report of elevated IgG anti-dsDNA antibody titers in NZM.Baff–/– mice as they age (25).
In any case, glomerular IgG deposition was very limited in both NZM.Baff–/– and NZM.Baff–/–.April–/– mice, and no glomerular C3 deposition could be detected. This is in striking contrast to the copious deposition of both IgG and C3 in the glomeruli of age-matched WT mice (Figure 5). Moreover, histologic evaluation demonstrated only occasional increases in glomerular cellularity in either NZM.Baff–/– mice or NZM.Baff–/–.April–/– mice without changes in glomerular size or increases in mesangial matrix. Perivascular leukocyte aggregates were rarely observed, and tubular atrophy or lesions were absent. Again, this is in striking contrast to the widespread pathologic changes in WT mice. Importantly, none of the NZM.Baff–/– or NZM.Baff–/–.April–/– mice developed clinical disease by the time they were euthanized (8 months of age).
The contribution of BAFF to SLE has been firmly established. In mice, constitutive overexpression of BAFF in hosts that are otherwise not autoimmune-prone leads to the development of SLE-like features (8–10), and treatment of SLE mice with a BAFF-specific antagonist ameliorates disease (21, 23). In humans, circulating BAFF levels are elevated in SLE patients and correlate with disease activity (14, 15, 17), and treatment with belimumab, a BAFF-specific antagonist, is efficacious in many SLE patients (27, 28).
In contrast, the connection between APRIL and SLE is tenuous. Unlike the SLE-like features that develop in BAFF-Tg mice, no such features develop in APRIL-Tg mice (44). Moreover, unlike the repeatedly observed increased circulating BAFF levels in human SLE (14, 15, 19, 45, 46), normal circulating levels of APRIL have been found in at least 2 independent studies of human SLE (34, 47). Finally, no studies prior to the present one had addressed the effect on SLE of selective elimination of APRIL. Although the combined antagonism of APRIL and BAFF is therapeutically successful in murine SLE (10, 22, 23) and is being evaluated in human SLE (43), the role of antagonism of APRIL per se cannot be unequivocally determined from such studies.
To directly assess the contribution of APRIL to SLE, we generated SLE-prone NZM mice that are deficient only in APRIL. Since in non–autoimmune-prone (B6/129 mixed background) April–/– mice, the numbers of B cells (and T cells) are, for the most part, identical to those in WT mice, as are the serum IgG levels (39, 42), we anticipated that these parameters in NZM.April–/– mice would be identical to those in their WT counterparts. Given the ability of APRIL to costimulate B cells, induce Ig class switching, and promote PC survival (29, 30, 35–38), we predicted that if there were any effects of APRIL deficiency on B cells and/or IgG levels in NZM.April–/– mice, those effects would be inhibitory. Unexpectedly, the numbers of B cells (and T cells) were globally increased, rather than decreased, in NZM.April–/– mice, as were circulating levels of at least one IgG autoantibody (antichromatin).
The factors that promote expansion of B cells and augment serologic autoimmunity in NZM.April–/– mice remain to be formally identified, but the absence of BAFF/APRIL heterotrimers in these mice may be highly relevant. Although BAFF/APRIL heterotrimers are biologically active in in vitro assays and circulate at elevated concentrations in SLE patients (33, 34), the in vivo activity of such heterotrimers is not known. Since BAFF/APRIL heterotrimers are less potent in vitro than are BAFF homotrimers (34), it is likely that BAFF/APRIL heterotrimers are also less potent in vivo. These heterotrimers can compete with BAFF homotrimers for the BAFF receptors, leading to functional down-modulation of BAFF activity. In the absence of APRIL, there would be no BAFF/APRIL heterotrimers. Thus, the natural heterotrimer-mediated down-modulation would be absent, leading to “uncontested” BAFF homotrimers and increased BAFF activity. That is, despite serum BAFF levels being identical in NZM.April–/– and WT mice, the net biologic activity of BAFF in the former might be greater than that in the latter. Such greater BAFF activity could well lead to increased numbers of B cells and circulating levels of autoantibodies in an APRIL-deficient SLE-prone host.
Moreover, BAFF has direct effects on T cells in addition to its direct effects on B cells. BAFF can costimulate the in vitro proliferation of, and cytokine production by, T cells by directly engaging them (48, 49). Thus, loss of BAFF/APRIL heterotrimers in NZM.April–/– mice with the (presumed) attendant increase in BAFF potency may have contributed to the expansion of T cells. Unfortunately, reagents that specifically neutralize BAFF/APRIL heterotrimers without also neutralizing BAFF and/or APRIL homotrimers do not yet exist, so our hypothesis cannot yet be formally tested, and the role of BAFF/APRIL heterotrimers per se in SLE remains an enigma.
In any case, NZM.April–/– mice developed renal immunopathology and clinical disease at least to the same degree and with the same time course as did WT NZM mice, demonstrating unequivocally that APRIL per se is dispensable to the development of full-blown SLE. This is in sharp contrast to the considerable protection from renal immunopathology and near-total protection from clinical disease in NZM.Baff–/– mice (25). Of note, the development of WT-like clinical disease in NZM.April–/– mice despite the T cell and B cell expansions and the increase in circulating IgG antichromatin antibody levels strongly suggests that cell numbers or serum autoantibody titers per se may offer only limited insights into the development of clinically relevant parameters.
The dispensability of APRIL to the development of full-blown SLE did not exclude a possible contributory role of APRIL in the development of SLE. That is, the role of APRIL in the development of SLE might only be appreciated when a more dominant contributor is lacking. Given the biologic relatedness of APRIL to BAFF, we reasoned that APRIL might play a more discernible role under BAFF-deficient conditions.
To test this, we compared NZM.Baff–/–.April–/– mice to NZM.Baff–/– mice. Rather than the levels of serum IgG antichromatin and IgG anti-dsDNA antibody being increased as might have been predicted from observations in NZM.April–/– mice, they were lower in NZM.Baff–/–.April–/– mice than in NZM.Baff–/– mice. Importantly, the superimposition of APRIL deficiency in a BAFF-deficient environment has no effect on BAFF/APRIL heterotrimers, since such heterotrimers are already not extant. Thus, the absence of APRIL in NZM.Baff–/–.April–/– mice aggravates, rather than ameliorates, the absence of BAFF and results in fewer BM PCs and reduced circulating levels of autoantibodies. This preservation of spleen PCs in NZM.Baff–/–.April–/– mice is reminiscent of the preservation of spleen PCs in NZB/NZW mice following neutralization of both BAFF and APRIL with TACI-Ig (22). Of note, spleen PCs were not preserved in NZM 2410 mice following treatment with TACI-Ig (23), indicating that the combined inhibition/elimination of BAFF and APRIL may have differential effects in different SLE models. In any case, renal immunopathology was identically minimal in NZM.Baff–/–.April–/– and NZM.Baff–/– mice.
With the caveats that findings in mice may not be fully translatable to humans and that studies in genetically-deficient hosts may not fully reflect outcomes in hosts treated with pharmacologic antagonists, our findings question the wisdom of antagonizing both BAFF and APRIL rather than antagonizing BAFF alone. Our cellular and serologic findings in NZM. Baff–/–.April–/– mice are remarkably similar to those previously reported in NZM 2410 mice treated with TACI-Ig to antagonize both BAFF and APRIL (23). Importantly, NZM 2410 mice treated in that study with BR3-Ig (to antagonize BAFF only) had the same favorable clinical and pathologic outcomes as did mice treated with TACI-Ig. That is, whether BAFF and APRIL were antagonized genetically or pharmacologically, the clinical and pathologic diseases were ameliorated to the same degree as when only BAFF was antagonized. Thus, antagonism of both BAFF and APRIL may simply increase the risk of toxicity (especially infections) without adding a clinical benefit. One clinical trial of atacicept (in combination with mycophenolate mofetil) in SLE was prematurely terminated because of an increase in serious infections (Clinical Trial.gov identifier NCT00573157), so circumspection is certainly warranted.
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. Dr. Stohl had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design.C. O. Jacob, Stohl.
Acquisition of data.C. O. Jacob, Guo, N. Jacob, Pawar, Putterman, Quinn, Migone.
Analysis and interpretation of data.C. O. Jacob, Guo, N. Jacob, Pawar, Putterman, Quinn, Cancro, Stohl.
Author Mignone is an employee of Human Genome Sciences.
The authors thank Dr. Raif Geha (Children's Hospital Boston, Boston, MA) for his kind gift of April–/– mice (B6/129 mixed background).