Atg7^flox mice were obtained from Dr. Masaaki Komatsu of Tokyo Metropolitan Institute of Medical Science (54) and crossed with CD19-cre knock-in mice (The Jackson Laboratory) to obtain B/Atg7^–/– mice. MRL and MRL-lpr mice were obtained from the Jackson Laboratory. Sex and age-matched wild type or B/Atg7^–/– mice in the C57BL/6 background at the age of 8-12 weeks were used at the start of all experiments except noted. For pristane injection, healthy sex and age-matched B/Atg7^–/– mice and wild type controls (8-12 weeks old) were randomly separated into groups for pristane or phosphate-buffered saline (PBS) injection (as controls). A single dose of pristane (0.5 ml i.p.) was injected. Sera were collected 6 months post-injection and levels of autoantibodies were measured by ELISA. Mice were sacrificed at 6 months after pristane injection. Spleens were harvested for Fluorescence-Activated Cell Sorting (FACS) and kidneys and lungs were collected for histology analysis. At least 5 mice per group were used for each experiment. At least 10 mice per group were used for pristane injection experiments. The experiments were performed according to federal and institutional guidelines and with the approval of Institutional Animal Care and Use Committees of Baylor College of Medicine and Houston Methodist Research Institute.

The following antibodies from BD Biosciences were used for flow cytometry: biotinylated antibodies to CD4 (553728), CD8 (553029), IgM (553436), CD11b (553309), CD138 (553713) and GR-1 (553125); PE-conjugated antibodies to B220 (553090), CD5 (553023), CD11b (557397), IgD (558597), CD138 (553714); APC-conjugated antibodies to CD21 (558658), IgM (550676), IgD (560868), CD11b (553312), CD138 (558626) and GR-1 (553129); FITC-conjugated antibodies to GL7 (553666), IgD (553439); PE-Cy7 conjugated anti-CD11b (552820); Pacific Blue anti-CD3e (558214); PE-Cy5-anti-CD4 (553050) and APC-Cy7-anti-CD8a (557654). From Biolegend: APC-anti-mouse IgG1 (406610), PerCPCy5.5-anti-mouse IgG1(406612), Pacific Blue anti-CD38 (102720), FITC-anti-mouse IgG (405305), APC-anti-mouse IgG (405308), APC-Cy7-anti-mouse IgG (405316). From eBioscience: PerCP-Cy5.5-anti-B220 (45-0452-82), PE-anti-IgM (12-5890-83), PE-anti-CD23 (12-0232-82), Biotin-anti-mouse-IgD (13-5993-85), Streptavidin-PE (12-4317-87), PE-Cy7-streptavidin (25-4317-82). From the Jackson Immunoresearch Laboratories: normal rabbit IgG (015-000-002) or mouse IgG (011-000-002). From Abgent: anti-LC3 (AP1802a) for immunocytochemistry. From Invitrogen: Anti-CoxIV (459600) for immunocytochemistry. From Southern Biotechnology, HRP conjugated anti-mouse IgG or IgM.

Spleens were treated with 0.4 mg/ml liberase (Roche) at room temperature for 10 min to make single cell suspension of splenocytes, followed by lysis of red blood cells with ACK lysis buffer. After blocking with 1 μg/ml anti-CD16/CD32, 10 μg/ml rat IgG and 10 μg/ml hamster IgG, the cells were then incubated with various antibodies conjugated to FITC, PE, APC, PerCP-Cy5.5, Pacific Blue (BD Biosciences) or PE-conjugated anti-PDCA-1 (Miltenyi Biotec) and analyzed by flow cytometry. Double-stranded DNA (dsDNA) were conjugated to Alexa fluor 488 using ULYSIS Nucleic Acid Labeling Kit (U21650, Invitrogen) according to manufacturer’s instructions. Autoantigen RNP/Sm (The Binding Site) was labeled with Alexa Fluor 488 using the Microscale Protein Labeling Kit (A30006, Invitrogen). To detect dsDNA-specific memory B cells and germinal center B cells, total spleen cells were stained with PE-conjugated antibodies to CD11b, IgM, IgD, GR1 and CD138 (DUMP), APC-anti-mouse IgG, PerCP-Cy5.5-anti-CD19 or B220, Pacific Blue anti-CD38 and Alexa Fluor 488-dsDNA. To detect RNP-specific memory B cells and GC B cells, total spleen cells were stained with PE-conjugated antibodies to CD11b, IgM, IgD, GR1 and CD138 (DUMP), APC-anti-mouse IgG, PerCP-Cy5.5-anti-CD19 or B220, Pacific Blue anti-CD38 and Alexa Fluor 488-RNP/Sm. DUMP^-B220⁺IgG⁺dsDNA⁺CD38⁺ and DUMP^-B220⁺IgG⁺RNP/Sm⁺CD38⁺ memory B cells, DUMP^-B220⁺IgG⁺dsDNA⁺CD38^– and DUMP^-B220⁺IgG⁺RNP/Sm⁺CD38^– germinal center B cells were analyzed by flow cytometry.

CD19⁺IgM^lowIgD⁺CD23⁺IgG^- naïve mature B cells, CD19⁺DUMP^-IgG⁺CD38⁺dsDNA⁺ and CD19⁺DUMP^-IgG⁺CD38⁺RNP/Sm⁺ memory B cells were sorted from 3-4 months old lpr mice. RNA extracted from the cells was used to prepare cDNA with the High Capacity cDNA Reverse Transcription Kit (Life Technologies). Real-time PCR was performed using Taqman Universal PCR Master Mix with specific primers for autophagy genes or 18S rRNA from the TaqMan Gene Expression Assay Kit (AB Applied Biosystem) in the ABI PRISM 7000 Sequence Detection System. The assay IDs for the primers of the analyzed genes are: Mm00504340_m1 (Atg5), Mm00512209_m1 (Atg7), Mm00437238_m1 (ULK1), Mm00458725_g1 (MAP1LC3A), Mm00782868_sH (MAP1LC3B), and Mm00553733_m1 (Atg14). Relative gene expression was normalized to 18S rRNA.

To detect autoantibodies against autoantigens, 96-well plates coated with Sm, or RNP (The Binding Site) were incubated with serially diluted sera at 37°C for 2h. The plates were then washed and incubated with HRP-conjugated goat anti-mouse IgG (Southern Biotechnology, Birmingham, AL) at 37°C for 1h, followed by development with TMB peroxidase EIA substrate (Bio-Rad, Hercules, CA). The reaction was stopped with 1N H₂SO₄ and the optical density at 450 nm was measured using an ELISA reader. A mixture of sera from MRL-lpr mice was used to establish standard curves in each plate and antibody levels were shown as relative titers.

MultiScreen 96-well Filtration plates (Millipore) were coated with 2 μg/ml RNP+Sm (The Binding Site). Sorted RNP/Sm+ or nonspecific memory B cells (100, 1000, 10000/well) were activated with LPS (10ug/ml) for 3 days, then added to the plates and incubated at 37 °C for 5 h. The cells were lysed with H₂O and the wells were probed with HRP-conjugated goat anti-mouse IgG (Southern Biotechnology), followed by development with 3-amino-9-ethylcarbzole (Sigma).

Wild type mice which were injected with pristane 6 months earlier were used as donors of wild-type memory B cells. Sex and age-matched 8-12 week-old B/Atg7 ^–/– and wild type mice were used as recipients of the memory B cells. DUMP^-B220⁺IgG⁺CD38⁺RNP/Sm⁺ memory B cells were sorted from pooled spleens of pristane injected donor mice and adoptively transferred into wild type and B/Atg7^-/- recipient mice retrooribitally (10,000 cells/mouse). Sorted naïve cells (2x10⁵/mouse) from untreated mice were co-injected as filler cells to minimize cell loss during injection. One day after transfer recipient mice were injected with a single dose of 0.5 ml pristane i.p. Sera were collected 2 months post-injection. For some experiments, recipient mice were immunized with 20 μg RNP/Sm precipitated with 100 μl Inject Alum (Thermo Scientific) intraperitoneally (20 μg/mice, i.p.) and sera were collected 3 days later. Levels of autoantibodies in the sera of recipients were measured by ELISA.

To detect immune complex deposits in the kidney, frozen sections of the kidneys were stained with FITC-conjugated goat anti-mouse IgG (Sigma) or FITC-conjugated goat anti-mouse complement C3 (MP Biochemical, 55500) and analyzed under a fluorescent microscope. Scores of glomerular IgG and complement C3 deposition were assigned based on the intensity of IgG/C3 deposition (range 0-3) with 0 represents no deposition and 3 representing intense depositions. Sections of kidney and lung were also stained with hematoxylin and eosin (H&E). The degree of glomerulonephritis was graded using a glomerulonephritis activity score (range 0–24) developed for the assessment of lupus nephritis in humans (55). Glomerular cells in 10 glomeruli per section were counted. We also measured anti-nuclear antibodies (ANAs) in pristane-treated mice by staining of Hep2 cells according to our described protocol (56). Hep-2 cells on slides (Medical and Biological Laboratories) were incubated with serially diluted sera followed by staining with FITC-conjugated anti-mouse IgG (Sigma). The staining was visualized under a BX-51 fluorescence microscope (Olympus). LC3 staining were performed according to our previously described protocol (30, 31). Briefly, sorted RNP-specific memory B cells from MRL-lpr mice or pristane-treated wild type mice, and naïve B cells from untreated wild type mice were added to slides by cytospin. The cells were fixed, incubated with a rabbit antibody to processed LC3 (Abgent) and followed by staining with Alexa Fluor-conjugated secondary antibodies (Molecular Probes). The nucleus was counter-stained with DAPI. The cells were then analyzed using a SoftWorx Image deconvolution microscope (Applied Precision).

Data were presented as the mean ± SEM, and P values were determined by two-tailed Student’s t-test using GraphPad Prism software and are included in the figure legends. The comparison of survival curves between and WT and KO was performed by Log-rank (Mantel-Cox) test using Prism. Significant statistic differences (P<0.05 or P<0.01 or P<0.001 or P<0.0001) are indicated.

Autoimmune Fas-deficient lpr mice develop significant lymphoproliferation and increased autoantibodies against DNA and nuclear antigens (7, 57–60). These mice also exhibit nearly 10-fold increase in the percentage of IgG1⁺ memory B cells (16, 17). We therefore determined whether memory B cells specific for autoantigens could be detected in lpr mice. Anti-dsDNA and anti-RNP/Sm autoantibodies are associated with lupus in both humans and mice (61, 62). We generated fluorochrome-conjugated dsDNA and RNP/Sm in order to stain B cells specific for these autoantigens. Interestingly, we detected dsDNA-specific CD19⁺IgG⁺IgD^–IgM^–CD38⁺ memory B cells in MRL-lpr mice ( Figure 1A ). In contrast, these dsDNA-specific autoantigen-specific memory B cells were absent in age-matched control MRL mice ( Figure 1A ). We also detected RNP/Sm-specific memory B cells in MRL-lpr mice but not in control MRL mice ( Figure 1A ). Moreover, these CD19⁺IgG⁺IgD^–IgM^–CD38⁺RNP/Sm⁺ memory B cells could be activated by LPS in vitro to differentiate into a larger number of anti-RNP/Sm antibody secreting cells (ASCs) than RNP/Sm-negative memory B cells could, suggesting the CD19⁺IgG⁺IgD^–IgM^–CD38⁺RNP/Sm⁺ memory B cells identified by flow cytometry are bona fide RNP/Sm-specific memory B cells ( Figure 1D ). We found that B cells specific for dsDNA and RNP/Sm are predominantly the CD19⁺IgG⁺IgD^–IgM^–CD38⁺ memory B cells, but not the CD19⁺IgG⁺IgD^–IgM^–CD38^– germinal center (GC) B cells ( Figures 1A–C ). This indicates that lpr mice contain abundant memory B cells but not germinal center B cells specific for autoantigens.

We have previously observed increased autophagy in memory B cells (30). We therefore stained memory B cells with dsDNA and RNP/Sm and sorted these autoreactive memory B cells. We then examined whether these autoreactive memory B cells displayed increased autophagy gene expression. As shown by quantitative RT-PCR (qRT-PCR), autoreactive memory B cells also expressed increased levels of Ulk1 (Atg1) and Atg14 critical for autophagy initiation, as well as Atg5, Atg7, Map1lc3a and Map1lc3b that required for autophagosome maturation (63–65) ( Figure 1E ). These data suggest that autoreactive memory B cells from lpr mice display active autophagy.

The conversion from LC3-I to LC3-II isoforms is indicative of active autophagy (26, 66). We therefore performed Western blot analysis to detect LC3. In comparison with naïve B cells, memory B cells from wild type and lpr mice displayed increased levels of LC3-II ( Figure 1E ). Moreover, memory B cells from lpr mice displayed significantly increased LC3-II compared to wild type controls ( Figure 1F ). These data suggest that memory B cells from lpr mice display active autophagy.

It has been shown that autophagy is elevated in SLE, and may be essential for humoral autoimmune manifestations (38, 39, 49). We therefore investigated whether autophagy might be important for the protection of autoreactive memory B cells and the production of autoantibodies using a pristane-induced lupus model. We crossed CD19-cre mice with Atg7^flox mice (54) to generate B cell-specific deletion of Atg7 (B/Atg7^–/–) on the C57BL/6 background. B/Atg7^–/– and wild type mice were then administrated with pristane. Pulmonary hemorrhage resembling that seen in SLE has been reported to occur earlier in pristane-treated mice and causes mortality in a portion of mice on the C57BL/6 background (52, 53, 67). We observed that treatment with pristane caused 18% death in wild type mice within 4 weeks of pristane injection ( Figure 2A ). Interestingly, B/Atg7^–/– mice were resistant to pristane-induced death ( Figure 2A ). Presumably the mortality associated with pristane-induced pulmonary vasculitis (52, 53, 67) is prevented due to the loss of autophagy in B cells (also see Figure 3B ).

It has been reported that pristane-induced lupus induces production of a different spectrum of SLE associated autoantibodies in C57/BL6 and BALB/c mice (52, 53). C57/BL6 mice could produce anti-RNP and anti-Sm (57, 62) but not anti-dsDNA autoantibodies after pristane treatment (52, 53). We next measured lupus-associated anti-dsDNA, anti-RNP and anti-Sm autoantibody levels by ELISA in sera from these mice six months after pristane administration. Consistent with previous reports, we did not detect significant induction of anti-dsDNA autoantibodies in wild type or B/Atg7^–/– mice on the C57/BL6 background (data not shown). We detected significant induction of anti-RNP and anti-Sm in wild type mice after pristane treatment ( Figures 2B, C ). However, the induction of anti-RNP and anti-Sm autoantibodies was dramatically reduced by 10-fold in B/Atg7^–/– mice ( Figures 2B, C ).

We also measured anti-nuclear antibodies (ANAs) in pristane-treated mice by staining of Hep2 cells. As a positive control, we detected distinct nuclear staining with sera from B6/lpr mice ( Figure 2D ). We detected nuclear staining of Hep2 cells with sera from 90% pristane-treated wild type mice ( Figure 2D ). Among them, 1/10 sera gave strong nuclear staining (1:640 dilution); 4/10 sera gave medium nuclear staining (1:160 dilution); and 4/10 sera gave low nuclear staining (1:40 or 1:80 dilution). Majority of the staining pattern is speckled ( Figure 2D ), which are typically produced by elevated level of anti-RNP/Sm autoantibodies. Sera from one mouse gave strong cytoplasmic staining (data not shown). In contrast, no significant staining was observed using sera from 8 of the 10 B/Atg7^–/– mice ( Figure 2D ). Only 1 of the 10 sera from B/Atg7^–/– mice generated detectable but much weaker speckled nuclear staining in Hep2 cells (1:40 dilution), whereas weak cytoplasmic staining was found in sera from 1 mouse. These results indicate that pristane induces the production of ANAs in wild type mice but loss of autophagy in B cells significantly reduced ANA production in B/Atg7^–/– mice.

To determine whether deficiency in autophagy affects other manifestations of autoimmunity, we examined kidneys from mice treated with pristane. Pristane treatment has been shown to induce the development of glomerulonephritis in mice (52, 53, 68). Consistently, the sizes of glomeruli in the kidney sections from pristane-treated wild type were increased compared to untreated controls ( Figure 3A ). Moreover, increased number of intraglomerular mesangial cells were found in these wild type mice treated with pristane ( Figure 3A ). In contrast, pristane treatment did not induce the increases in cellularity or sizes of glomeruli in the kidney of B/Atg7^–/– mice. Glomerulonephritis activity score was 8.3 ± 2.8 versus 2.4 ± 1.8 for wild type and B/Atg7^–/– mice, respectively. These results suggest that autophagy deficiency in B cells suppress pristane-induced glomerulonephritis. We also examined lungs from mice treated with pristane for signs of lupus. Pulmonary hemorrhage has been shown in SLE patients and mice induced with pristane-induced lupus (69). We observed extensive perivascular and peribronchial infiltration in wild type but not autophagy-deficient mice at 6 months after pristane injection ( Figure 3B ). These results suggest that autophagy deficiency in B cells suppress pristane-induced autoimmune manifestations of lupus.

Autoreactive B cells can lead to over-production of antibodies and the deposition of antibody-immune complexes in the kidney, which can be the cause of glomerulonephritis. We therefore examined immune complexes in the kidneys of pristane-treated mice. As expected, we observed significant IgG deposition in the glomeruli of kidneys in wild type mice treated with pristane ( Figure 3C ). In contrast, IgG deposition was significantly reduced in B/Atg7^–/– mice (glomerulonephritis deposition score was 2.3 ± 0.5 versus 0.8 ± 0.6 for wild type and B/Atg7^–/– mice, respectively) ( Figure 3C ). Consistent with this finding, complement C3 staining were also dramatically reduced in B/Atg7^–/– mice when compared with wild type mice (glomerulonephritis deposition score was 2.2 ± 0.6 versus 0.7± 0.3) ( Figure 3D ), suggesting decreased complement activation in the absence of autophagy in B cells. These data support the conclusion that autophagy deficiency in B cells prevents autoimmune manifestations in the pristane-induced lupus model.

We next determined whether autophagy deficiency in B cells affects the composition and activation of different cell types in the immune system. We did not detect significant changes in the numbers and composition of T cells in B/Atg7^–/– mice compared to the wild type after the pristane treatment ( Figures 4A, B ). The number of B cells and the composition of mature B cells, as well as the transitional T1 and T2 cells, marginal zone (MZ) and follicular (FO) B cells (70) were also not significantly changed between wild type and B/Atg7^–/– mice at six months after pristane administration ( Figures 4A, C, D ). CD11c⁺CD11b⁺ dendritic cells and CD11c^–CD11b^high macrophages, which can phagocytose pristane, were increased after pristane injection ( Figure 4E ). However, their percentages were comparable between wild type and B/Atg7^–/– mice ( Figure 4E ).

We next determined whether autoreactive memory B cells were present after treatment with pristane. Switched memory-like B cells (CD19⁺CD138^–IgM^–IgD^–) were reported to be significantly increased in the pristane-treated BALB/c mice (71). However, whether autoantigen-specific memory B cells are expanded after pristane treatment have not been characterized. We also detected increased number of total IgG⁺ memory B cells (CD19⁺DUMP^–CD38⁺IgG⁺) in wild type mouse after pristane treatment ( Figure 4F ). Those IgG⁺ memory B cells were significantly reduced in B/Atg7^–/– mice ( Figure 4F ). Moreover, we found that RNP/Sm-specific memory B cells (CD19⁺DUMP^–CD38⁺ IgG⁺RNP/Sm⁺) in wild type mice could be detected by flow cytometry after injection with pristane ( Figure 4F ). These autoreactive B cells mainly display the DUMP^–IgG⁺CD38⁺ memory but not the DUMP^–IgG⁺CD38^–GC B cell phenotypes ( Figure 4F ). Moreover, such autoreactive memory B cells were significantly reduced in B/Atg7^–/– mice ( Figure 4F ). These results suggest that autophagy is important for the development of both autoreactive memory B cells and nonautoreactive memory B cells. We also examined the IgM^–IgD^–CD19^loCD138⁺ plasma cells (72) in the spleens of pristane-treated mice. We observed a slight decrease in splenic plasma cells in pristane treated B-Atg7^–/– mice, but it did not reach statistical significance ( Figure 4G ). It is possible that a continuous replenishment of new short-lived plasma cells induced by pristane treatment compensates the loss of those cells in the absence of autophagy.

We found that RNP⁺ memory B cells from pristane-treated wild type mice upregulated autophagy as shown by the increased LC3 punctate staining, to a similar extent as RNP⁺ memory B cells harvested from MRL-lpr lupus mice did ( Figure 5A ). To confirm the RNP⁺ memory B cells that we detected by flow cytometry ( Figure 4F ) could lead to autoantibody production, we sorted RNP/Sm-specific memory B cells from pristane-treated wild type mice and transferred them into wild type or B/Atg7^–/– recipient mice. We then challenged recipients of memory B cells and nonrecipient controls with RNP/Sm in Alum. Three days later, sera were collected to measure the production of RNP/Sm autoantibody. We detected production of RNP/Sm autoantibody in both wild type and B/Atg7^–/– recipient mice, suggesting CD19⁺DUMP^–CD38⁺IgG⁺RNP/Sm⁺ autoreactive memory B cells are functional RNP/Sm-specific memory B cells and can be rapidly activated to develop into autoantibody-producing cells ( Figure 5B ).

Next, we asked whether RNP/Sm-specific memory B cells could restore pristane-induced autoantibody production in B/Atg7^–/– mice. RNP/Sm-specific memory B cells from pristane-treated wild type mice were sorted, and adoptively transferred into wild type or B/Atg7^–/– recipient mice in parallel experiments. After 24 h, both the recipient mice and control mice without receiving cell transfer were challenged with pristane. Atg7^–/– RNP/Sm-specific memory B cells from B/Atg7^–/– mice were not used for transfer because not sufficient number of cells could be obtained. Two months after pristane administration, autoantibodies against RNP/Sm could be detected in some wild type mice but not B/Atg7^–/– mice without receiving cell transfer ( Figure 5C ). However, adoptive transfer of wild type RNP/Sm-specific memory B cells resulted in significant production of RNP and Sm-specific antibodies in both wild type and B/Atg7^–/– recipient mice two months post pristane treatment ( Figure 5C ). Flow cytometry analysis of memory B cells indicate those transferred RNP/Sm-specific memory B cells were expanded in the spleens of B/Atg7^–/– recipient mice compared with non-recipient controls ( Figure 5D ). These results suggest that autoreactive memory B cells can restore autoantibody production in B/Atg7^–/– mice. These results suggest that autophagy plays an important role in the protection of autoreactive memory B cells to maintain autoantibody production in lupus.