BCL-2 as a Therapeutic Target in Human Tubulointerstitial Inflammation

Abstract

Objective

In lupus nephritis (LuN), tubulointerstitial inflammation (TII) is associated with in situ adaptive immune cell networks that amplify local tissue damage. As patients with severe TII often fail conventional therapy and develop renal failure, understanding these in situ mechanisms might reveal new therapeutic targets. We hypothesized that in TII, dysregulated apoptotic regulators maintain local adaptive immunity and drive inflammation.

Methods

We developed novel computational approaches that, when applied to multicolor confocal images, quantified apoptotic regulator protein expression in selected lymphocyte subsets. This approach was validated using laser capture microdissection (LCM) coupled to qPCR. Furthermore, we explored the consequences of dysregulated apoptotic mediator expression in a murine model of LuN.

Results

Analyses of renal biopsies from LuN and mixed cellular allograft rejection patients revealed that BCL-2 was frequently expressed in infiltrating lymphocytes while expression of MCL-1 was low. In contrast, the reciprocal pattern of expression was observed in tonsil germinal centers. These results were consistent with RNA expression data obtained using LCM and qPCR. BCL-2 was also highly expressed in tubulointerstitial infiltrates of NZB/W F₁ mice. Furthermore, treatment of NZB/W F₁ mice with ABT-199, a selective oral inhibitor of BCL-2, prolonged survival and prevented proteinuria and development of TII in a prevention model. Interestingly, glomerular immune complexes were partially ameliorated by ABT-199 and serum anti-dsDNA antibody titers were unaffected.

Conclusion

These data demonstrate BCL-2 as an attractive therapeutic target in LuN manifesting TII.

Systemic lupus erythematous (SLE) is the prototypical systemic autoimmune disease. The central pathogenic mechanism is thought to be a fundamental failure of lymphocytic tolerance and subsequent selection of pathogenic autoreactive populations. One mechanism of tolerance that fails is clonal deletion by programmed cell death or apoptosis (1).

Dysregulation of the pro- and anti-apoptotic gene products that regulate programmed cell death can lead to survival of autoreactive B and T cells, and autoimmunity (2, 3). Transgenic mice over-expressing the anti-apoptotic molecule, B cell lymphoma 2 (BCL-2) in B lymphocytes develop a lupus-like illness with anti-nuclear autoantibodies and glomerulonephritis (GN) (4). In humans, BCL-2 proteins have been found to be overexpressed in peripheral blood, especially in circulating T cells (5–7). However, peripheral blood lymphocytes do not necessarily provide a reliable sampling of populations mediating end-organ damage (8, 9).

The disparity between peripheral and in situ immunity is evident in lupus nephritis (LuN) which is the most common severe manifestation of SLE (10, 11). Tubulointerstitial inflammation (TII) is frequently found in LuN, and its presence and severity predict renal failure more than glomerular inflammation (12). TII in LuN is characterized by in situ antigen-driven clonal expansion of B cells suggesting local propagation of adaptive autoimmune responses (13). This is quite different from GN which is thought to arise from deposition of circulating immune complexes and subsequent inflammation. A few studies have examined BCL-2 expression in LuN. However, they have focused on GN and have been largely inconclusive (14–16). These studies and histologically-based studies in general have been limited because tools to objectively and quantitatively assess the distribution and prevalence of protein expression within specific cell populations have been lacking.

Herein, using novel quantitative image-analysis tools, we demonstrate that BCL-2 is specifically up-regulated in T and B cells infiltrating the tubulointerstitium in human LuN but not in glomeruli. This pattern of expression is in contrast to that observed in secondary lymphoid organs where BCL-2 is down-regulated upon response to antigen (17, 18). A similar pattern of high BCL-2 expression was observed in mixed cellular renal allograft rejection (MR) suggesting that BCL-2 dysregulation might be a general feature of inflammation. Finally, treatment of NZB/W F₁ mice with ABT-199, a specific inhibitor of BCL-2, protected against nephritis primarily by inhibiting TII. These studies suggest that BCL-2 inhibition would be clinically beneficial in LuN.

Materials and Methods

Human Studies

Clinical characteristics of patients with renal biopsies are provided in Supplemental Materials and Methods.

Two-dimensional confocal microscopy and image processing

Three-µm thick fresh frozen sections were stained with immunofluorescent (IF) antibodies against BCL-2 (mouse, DAKO, Carpinteria, CA), MCL-1 (mouse, Thermo Scientific, Rockford, IL), BIM (rabbit, Cell Signaling, Danvers, MA), CD20 (mouse, DAKO and rabbit, Abcam, Cambridge, MA), CD4 (rat, Novus Biologicals, Littleton, CO) and 4’,6-diamidino-2-phenylindole (DAPI) (Invitrogen, Carlsbad, CA), and then fluorescently labeled with species-specific secondary anti-IgG antibodies (Invitrogen).

Images were obtained with a TCS SP2 Leica laser scanning confocal microscope (Leica Microsystems, Buffalo Grove, IL) as described (19). ImageJ (http:/imagej.nih.gov/ij/) was used for background subtraction, fluorescence threshold calibration, despeckling and exclusion of small particles (performed by KK).

Super-resolution imaging was obtained by using Leica’s Ground State Depletion followed by Individual Molecule return. Slides were extensively washed and mounted on depression slides with b-Mercaptoehtylamine in PBS. Twinsil (Picodent, Wipperfürth, Germany) was used to seal the edges. Images were then acquired sequentially at 160X magnification with dye laser turned to 488 nm then 642 nm with the dye pump at 50 – 100%. 10 – 20% of events were acquired at a threshold of 10 photons for both channels until no new events were seen.

Computer-assisted analysis (Automated Cell Phenotyping, ACP

Digital high-power field images (HPFs) were analyzed using automated two-dimensional image analysis software as described in the Results and summarized in Supplemental Figure 1 (19, 20). This software identified nuclei (DAPI staining) and specific membrane stains. These were converted into binary signals based on signal intensity-threshold, and then assigned specific cell types based on proximity between membrane stains and nuclei of 1 pixel or less. It then counted the number of specific cell nuclei per HPF. The average number of HPFs per patient was 4.8 ± 3.2 (minimum of 2 in one sample).

Laser capture microdissection and quantitative PCR of BCL-2, MCL-1 and BIM

Frozen tissue sections (8 µm) from LuN, tonsil and normal kidney samples were immunostained using anti-CD20 antibody (mouse, DAKO) directly conjugated to Alexa Fluor (Zenon Alexa Fluor 594 Mouse IgG_2a kit, Life Technologies, Grand Island, NY). Laser capture microdissection (LCM) was then performed as described (19) using the Arcturus Pixcell II (Molecular Devices, Sunnyvale, CA). An average of 40 – 60 cells/LCM cap were subjected to RNA isolation, cDNA synthesis and qPCR using indicated primers (Supplemental Materials and Methods).

Animal Studies

Mice

Female NZB/W F₁ mice approximately 26 weeks of age were obtained from Jackson Laboratory (Bar Harbor, ME) and acclimated for one week.

Efficacy and toxicity studies in mice

Mice were randomly assigned to treatment groups: vehicle controls for ABT-199 and mycophenolate mofetil (MMF), ABT-199 at varying doses and MMF (21). Urine was collected at weekly intervals and tested for protein levels using Albustix reagent strips (Siemens 2191, Pittsburgh, PA) with proteinuria defined as urine protein above 300 mg/dL on at least 2 consecutive tests or prior to death. Mice exhibiting proteinuria prior to treatment were excluded. Treatments were administered daily via oral gavage starting at 26 weeks of age. Retro-orbital blood samples were analyzed using a Cell-Dyn 3700 Hematology Analyzer (Abbott, Lake Forest, IL). Plasma anti-dsDNA antibody levels were measured by QUANTA Lite dsDNA ELISA assay (Inova Diagnostics, San Diego, CA).

For flow cytometry, bone marrow and spleens from mice were collected, processed and analyzed as described (22) using indicated antibodies (Supplemental Materials and Methods). The antibodies were directly coupled to fluorescein isothicyanate, phycoerythrin, phycoerythrin and cyanine 5, allophycocyanin, or peridinin chlorophylla and cyanine 5.5, and the results were analyzed with FlowJo (version 8.5, Treestar Inc).

Immunofluorescent and double-/single-label immunohistochemistry microscopy

As vehicle mice became moribund, kidney tissues were collected and fixed in 10% neutral buffered formalin. In addition, after 24 weeks of treatment, representative mice from each group were euthanized. Kidneys were collected, bisected and either snap frozen in Tissue-Tek O.C.T. media (Sakura Finetek, Torrance, CA) or fixed in 10% neutral-buffered formalin and then paraffin-embedded.

Double- and single-label immunohistochemistry (IHC) was performed as essentially as described (13) with a primary antibody to BCL-2 (goat, R&D Systems, Minneapolis, MN), CD45R/B220 (rat, BD Biosciences, San Diego, CA), CD138 (rat, R&D Systems), CD3 (rabbit, Thermo Scientific) or IBA1 (rabbit, Wako, Japan), followed by secondary antibodies (Vector, Burlingame, CA), Leica Bond polymer for HRP or AP (Leica Biosystems, Buffalo Grove, IL), and Deep Space Black chromogen (Biocare Medical, Concord, CA). Following hematoxylin counterstaining, cells were imaged using a Zeiss Axioskop 2 microscope (Thornwood, NY).

IgG IF was performed with anti-mouse IgG Alexa 488 secondary antibody (chicken, Thermo Fisher Scientific) or IgY (IgG) -488 (chicken, Jackson ImmunoResearch) and semiquantitatively scored as previously described (23).

Study Approval

The human studies were approved by the University of Chicago Medical Center Institutional Review Board. All animal experiments were in compliance with AbbVie’s Institutional Animal Care and Use Committee (IACUC) and the National Institute of Health (NIH) Guide for Care and Use of Laboratory Animals guidelines in a facility accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care.

Results

Quantifying complex cell phenotypes in situ

It is difficult to assess the frequency of specific cells expressing particular cytosolic or surface markers because robust methods are lacking to segment the cytoplasmic boundary of one cell from another. Recently, we developed computational approaches (Cell Distance Mapping, CDM) to understand the spatial relationships between different lymphocyte subpopulations in confocal laser scanning multicolor microscopy (CLSM) images of human tissue (19, 20). Using CDM, cell subsets were identified and positioned within tissue by assigning membrane stain signatures back to their corresponding nuclei.

This initial version of CDM used simple computational approaches in which nuclear signatures were uniformly segmented using Otsu adaptive thresholding coupled to top-hat morphological opening operations. However, for dense cellular infiltrates, this approach resulted in under-splitting of cell nuclei (19). Therefore, we developed additional tools for defining and splitting nuclear signatures (Supplemental Figure 1A and B). Briefly, nuclei were detected using DAPI and images acquired using CLSM (Step 1). The DAPI image was pre-processed to reduce background noise and enhance contrast (Step 2). Sobel edge detection was then used to identify nuclei outlines (Step 3a). Subsequently, circular Hough transform and watershed thresholding were applied to identify nuclei outlines missed by edge detection analysis (Step 3b). Finally, post-processing refined nuclear segmentation (Step 4). An example of this approach is provided in Figure 1A. In the first row, is a raw DAPI image of lupus tubulointerstitial infiltrates (Step 1). This image was then converted to a binary image and processed (Step 2), and segmented using edge detection, circular transform and watershed thresholding (Steps 3–4) (second row).

Figure 1. Automated cell phenotyping of BCL-2 expression in infiltrating CD20 and CD4 lymphocytes.

(A) Multichannel confocal immunofluorescent staining of human lupus nephritis tissue for BCL-2, CD20, CD4 and DAPI (nuclei) (scale bar: 20 µm). The raw images of specific membrane markers (first row) were converted to binary signals through ACP (second row), where the nuclear signals were expressed as outlines. The membrane markers were then assigned to corresponding nuclei for specific cell definition (third row). White asterisks (*) denote examples of cells that stained for both CD20 and CD4. Lastly, specific nuclear assignments were combined to generate BCL-2⁺CD20⁺ (magenta) and BCL⁺CD4⁺ (yellow) nuclei (cells) (fourth row). From these data, the BCL-2 frequency per HPF was automatically calculated. (B) Example of cells staining for both CD20 and CD4 (scale bar: 2 µm). The first row represents a cell expressing both CD20 and CD4 as imaged by conventional confocal microscopy. The second row represents the super-resolution confocal microscopy image of the same cell demonstrating intertwining of discrete CD20⁺ and CD4⁺ membranes. ACP = automatic cell counting and HPF = digital high-power field image\

To define expression distributions of specific proteins within individual cells, segmented nuclei (second row, column 1) were converted to outlines that were then applied to corresponding confocal images for specific markers, in this example CD20, BCL-2 and CD4 respectively (second row, columns 2 – 4). Expression of specific proteins was then assigned to particular nuclei to define cells that expressed CD20, BCL-2 or CD4 (third row). Finally, nuclear assignments were combined to define which CD20⁺ and CD4⁺ nuclei (cells) co-expressed BCL-2 (fourth row). This technique, which we refer to as Automated Cell Phenotyping (ACP), provides rich data sets on the relative distribution of proteins in different cell populations in an unbiased and high-throughput manner.

As can be seen, some cells appear to stain with both CD20 and CD4 (examples denoted with an “*” in Figure 1A). This apparent double staining does not reflect a failure of nuclear segmentation as higher magnification examples revealed single cells co-staining for CD20 and CD4 (Figure 1B, first row). Therefore, we imaged the cell in Figure 1B (first row) with super-resolution confocal microscopy (Figure 1B, second row) which uses ground state depletion to provide up to 20 nm resolution in the x- and y- axis. As demonstrated, apparent double staining by CLSM represented close approximation of discrete CD20⁺ and CD4⁺ membrane regions. Therefore, apparent double positive cells likely represent cross-sections of T:B cognate pairs (19).

BCL-2 is up-regulated in human tubulointerstitial inflammation

We next used ACP to define expression distributions for apoptosis regulators in human LuN compared to those observed in secondary lymphoid organs. Multi-color confocal imaging of human tonsils revealed that BCL-2 was commonly expressed in lymphocytes outside GCs and infrequently in those activated cells residing in the GCs (Supplemental Figure 2) (17, 18, 24, 25). In contrast, infiltrating B and T lymphocytes in lupus TII commonly express BCL-2 (Figures 1 and 2A). Interestingly, expression of BCL-2 in glomeruli was much less than that observed in the tubulointerstitium. B and T lymphocytes were also infrequent in glomeruli and these did not express detectable BCL-2 (Figure 2A and Supplemental Figure 3).

Figure 2. Dysregulation of BCL-2 in LuN and MR TII.

(A) and (B) Multichannel confocal immunofluorescent staining of human LuN tissue (A) and MR (B) for BCL-2, CD20, CD4 and DAPI (nuclear) (scale bar: 20 µm). The white line in (A) represents the border between tubulointerstitium (marked “TI”) and glomerulus (marked “G”). (C) Frequency of BCL-2 expressing CD20⁺ and CD4⁺ lymphocytes in tonsil FO, GC, LuN and MR. Each data point represents the frequency of BCL-2⁺ cells per biopsy. Central tendency shown is the median value, with error bars representing the interquartile range. P values were generated by two-tailed Mann-Whitney U test. *: P < 0.05, ***: P < 0.0001. LuN = lupus nephritis, MR = renal allograph mixed cellular rejection, TII = tubulointerstitial inflammation, G = glomerulus, FO = primary follicle and GC = germinal center

We next examined renal biopsies from patients manifesting MR. Surprisingly, BCL-2 was frequently expressed in the B and T lymphocytes infiltrating the allograft (Figure 2B). As in LuN, B and T cells primarily infiltrated the tubulointerstitium and were infrequent in glomeruli (data not shown).

We next utilized ACP to quantify the expression of BCL-2 in LuN and MR, compared to normal expression patterns observed inside and outside GCs. Tissue samples from patients with LuN (n = 22) and MR (n = 10), and tonsillectomy control samples (n = 10) were stained with antibodies specific for BCL-2, CD20, CD4 as well as DAPI and visualized by CLSM. For tonsil samples, digital high-power field images (HPFs) were manually segregated into GC (negative control) and primary follicle (FO, positive control) regions of interest. For LuN and MR, all HPFs with evident B and T cells were included in the analysis. It was found that the median percentages of CD20⁺ and CD4⁺ cells that co-expressed BCL-2 in LuN cases were 35.4 % (IQR 24.9 % – 43.2 %) and 22.5 % (IQR 19.5 % – 33.7 %) out of total number of CD20⁺ and CD4⁺ cells, respectively. These findings were striking when compared to tonsil GCs where there was scarcity of BCL-2 positive cells (median frequency = 3.7 % with IQR 0.6 % – 5.5 %, P < 0.0001) (Figure 2C). The frequency of BCL-2 positive cells in LuN cases was comparable to the one in FOs for CD20⁺ cells, whereas it was slightly lower for CD4⁺ cells (P < 0.05).

The distribution of BCL-2 frequency was broad in LuN, yet none of the BCL-2 frequency values in either CD20⁺ or CD4⁺ cells overlapped with those in GCs. The lymphocyte frequency of BCL-2 expression in LuN did not correlate with ISN/RPS class, activity or chronicity indices, TII scores or number of CD20⁺ or CD4⁺ cells per HPF (data not shown). The study sample description is included in our Supplemental Materials and Methods.

The MR samples also showed significantly higher percentages of BCL-2 positive cells compared to tonsil GC controls (P < 0.0001), suggesting that enhanced BCL-2 expression may be a general feature of TII. The frequency of BCL-2 positive cells in the MR samples was even higher than the one found in FOs for CD20⁺ cells (P < 0.05), whereas there was no difference in BCL-2 positivity between the MR and FO samples for CD4⁺ cells.

MCL-1 is down-regulated in tubulointerstitial inflammation

In contrast to BCL-2, the anti-apoptotic molecule MCL-1 increases in lymphocytes upon encountering antigen and is required for normal GC responses (26, 27). Indeed, MCL-1 expression was readily detectable only in those CD20⁺ and CD4⁺ cells within the tonsillar GCs (Supplemental Figure 4A). In contrast, the expression of the pro-apoptotic molecule BIM (28), was similar in the FO and GC regions (Supplemental Figure 4B).

We utilized ACP to quantify the expression of MCL-1 and BIM in different tonsil regions and compare it to the expression patterns in LuN. Seven LuN cases and seven tonsil cases were stained for MCL-1 (Supplemental Figure 4C) while nine LuN and 10 tonsil controls were stained for BIM in conjunction with CD20, CD4 and DAPI (Supplemental Figure 4D). Application of ACP revealed that the median frequency of MCL-1 expressing CD20⁺ and CD4⁺ cells was much lower in LuN compared to GC controls (Figure 3A, P < 0.001) and was comparable to expression frequencies observed in follicular lymphocytes. There was little variability in MCL-1 expression in LuN, with no overlap between the LuN and GC data sets. In contrast, there was little difference in BIM expression in CD20⁺ cells in FOs, GCs or LuN (Figure 3B). The frequency of BIM⁺ cells was lower for CD4⁺ lymphocytes in LuN (P < 0.0001), although the absolute difference in prevalence was modest.

Figure 3. Low MCL-1 and constitutive BIM expression in tubulointerstitial inflammation.

(A) and (B) MCL-1 (A) and BIM (B) frequency comparison across tonsil FO, GC and LuN in CD20⁺ and CD4⁺ cells. Each data point represents the total frequency of MCL-1⁺ or BIM⁺ cells per subject. Central tendency shown is the median value. Error bars represent the interquartile range. P values were generated by two-tailed Mann-Whitney U test. (C) LCM/qPCR analysis showing gene expression in CD20⁺ lymphocytes in tonsil FO and GC and LuN tissue for BCL-2MCL-1 and BIM. The fold changes are expressed on a log scale and compared through two-tailed Mann-Whitney U test. *: P < 0.05, **: P < 0.001, ***: P < 0.0001 and NS: not significant. LuN = lupus nephritis, TII = tubulointerstitial inflammation, LCM = laser capture microdissection, ACP = automatic cell counting, FO = primary follicle and GC = germinal center.

Dysregulated BCL-2 and MCL-1 mRNA expression in human LuN

We next used LCM to sample mRNA from CD20⁺ cell rich areas in tonsil FOs and GCs, LuN, and normal kidney tissues, and then performed qPCR to assess the expression of BCL-2, MCL-1 and BIM. As shown in Figure 3C, BCL-2 was indeed up-regulated in CD20⁺ lymphocytes in tonsil FOs compared to those in tonsil GCs (P < 0.05). Consistent with our previous ACP data, higher levels of BCL-2 expression were found in LuN compared to tonsil GCs (P < 0.05) whereas BCL-2 expression in LuN was slightly lower than that found in FOs (P < 0.05). MCL-1 expression showed the opposite pattern as it was down-regulated in both tonsil FOs and LuN cases compared to tonsil GCs (P < 0.05). There was no difference in MCL-1 expression between LuN cases and FOs. BIM was constitutively expressed. Thus, our LCM/qPCR analyses of mRNA expression confirmed observed patterns of protein expression.

Selective BCL-2 inhibition prevents lupus nephritis in NZB/W F₁ mice

The above findings suggest that the lymphocytes infiltrating the kidney in LuN might be sensitive to BCL-2 inhibition. To begin to address this possibility, we used NZB/W F₁ mice which develop a spontaneous lupus-like disease including GN and TII with lymphocyte aggregates (29). Kidneys from 39 week-old mice were harvested, and paraffin-embedded 5-µm sections were stained for histology, and with antibodies to BCL-2, B220 and CD3 for double-labeled IHC. As described (29), perivascular B and T cell infiltrates were a usual feature of the TII in these mice (Figure 4A). Similar to what was observed in humans, these cells co-expressed BCL-2.

Figure 4. Inhibiting BCL-2 in NZB/W F₁ lupus mice diminished perivascular and interstitial inflammation and prolonged survival.

(A) Double-labeled immunohistochemical staining of NZB/W F₁ mouse LuN tissue demonstrating perivascular infiltrates of BCL-2 expressing B (B220⁺) and T (CD3⁺) lymphocytes (scale bar: 20 µm). v: vessel. (B) and (C) Graphs showing proteinuria (B) and survival rates (C) after ABT-199 treatment for 24 weeks at indicated dosages compared to vehicle-treated mice. MMF was used as a positive control. Log-rank Mantel-Cox survival analysis was used for analysis. *: P < 0.05, **: P < 0.001 and ***: P < 0.0001. LuN = lupus nephritis, hema = hematoxylin and MMF = mycophenolate mofetil.

We next examined whether BCL-2 inhibition would attenuate nephritis in NZB/W F₁ mice. Beginning at 26 weeks of age, cohorts of NZB/W F₁ mice were treated daily with the oral BCL-2 specific inhibitor ABT-199 at doses ranging from 1.2 mg/kg/d to 100mg/kg/d, or ABT-199 vehicle or MMF at 100 mg/kd/d. Cohorts were then followed for up to an additional 24 weeks. Untreated NZB/W F₁ mice started developing proteinuria at 32 weeks (Figure 4B) with essentially all mice developing proteinuria by 45 weeks. Mortality closely followed the onset of proteinuria (Figure 4C) with a median survival of 43 weeks. In contrast, mice treated with increasing doses of ABT-199 were progressively protected from proteinuria and had improved survival. The 3.7 mg/kg/d dose attenuated proteinuria (P < 0.05) while almost all mice receiving 33 or 100 mg/kg dose were protected from developing proteinuria. Treatment with MMF also protected against proteinuria (P < 0.0001 vs. vehicle control group) similarly to the 33 and 100 mg/kg/d doses of ABT-199. In terms of survival, a significant response was observed even with the 1.2 mg/kg dosing of ABT-199 (P < 0.05 vs. vehicle control group) while all mice survived to 46 weeks if they received either the 33 or 100 mg/kg/d dose (P < 0.0001). Similar survival was observed with MMF given at 100 mg/kg/d. These data demonstrate that both the 33 and 100 mg/kg/d doses of ABT-199 were similarly efficacious as MMF.

Treatment with ABT-199 also greatly attenuated the histological features of nephritis. Sections from ABT-199-vehicle control mice at 50 weeks of age (24 weeks of treatment) demonstrated GN and in the tubulointerstitium, perivascular infiltrates, dilated tubules and tubular casts (Figure 5A, top left panel). These pathological changes were greatly attenuated in mice receiving ABT-199 at 33 or 100 mg/kg/d dose (Figure 5A and data not shown). Semi-quantitative scoring of H&E slides (Figure 5B) in mice treated with either 33 mg/kg/d or 100 mg/kg/d of ABT-199 revealed diminished GN scores (1.6 versus 4.0 for vehicle control; P < 0.01), attenuated tubular damage (0.04 versus 2.25; P < 0.01), and diminished perivascular infiltrates (0.14 versus 2.63; P < 0.01), similar to those treated with MMF. These data indicate that ABT-199 doses of 33 mg/kg/d or higher are as effective as MMF in resolving the histological features of nephritis in NZB/W F₁ mice.

Figure 5. Inhibiting BCL-2 in NZB/W F₁ lupus mice attenuated TII.

(A) Histological comparison of kidneys from ABT-199-vehicle-, ABT-199- (33 mg/kg/d) and MMF-treated NZB/W F₁ mice. ^: tubular casts, *: glomerular disease, #: cell infiltrates. (B) Semi-quantitative scoring of H&E slides of vehicle-(n = 8), ABT199-(n = 7) and MMF-treated (n = 3) mice compared by two-tailed Mann-Whitney U test. *: P < 0.05. TII = tubulointerstitial inflammation, GN = glomerulonephritis, Vehicle-A = vehicle for ABT-199 and MMF = mycophenolate mofetil.

The effects of ABT-199 on morphologic glomerular damage were less pronounced, compared to tubular damage and perivascular infiltrates (Figure 5B). While histologic evidence of tubular damage and perivascular infiltrates was nearly abolished, ABT-199 reduced the glomerular disease score by only approximately two-fold. Similar trends were observed with MMF. These observations suggest that ABT-199 might differentially affect the glomerular and tubulointerstitial compartments in LuN.

Renal samples from mice treated with either vehicle or ABT-199 (33 mg/kg/d) were further analyzed by single-color IHC for B220, CD138, CD3 or IBA1 (a pan-macrophage marker) (Supplemental Figure 5). While cells positive for each of these markers were readily apparent within the tubulointerstitial and perivascular infiltrates of vehicle mice, they were almost absent in mice treated with ABT-199. These data demonstrate that ABT-199 greatly diminishes the frequencies of adaptive and innate immune cells infiltrating the kidney in NZB/W F₁ mice.

While ABT-199 also decreased glomerular immune complex deposits, the effect was only modest compared to the complete resolution of TII in treated mice as shown above (Figure 6A-B). Immune complex deposition in glomeruli is often associated with peripheral titers of anti-dsDNA antibodies (30). Indeed, treatment with 33 mg/kg of ABT-199 did not diminish anti-dsDNA antibody titers in NZB/W F₁ mice (Figure 6C). While there are other mechanisms for anti-dsDNA antibody production (31), they can also arise from long-lived plasma cells (32) that are dependent upon MCL-1 for survival (33). Consistent with the selectivity of ABT-199 for BCL-2, bone marrow resident plasma cells were not affected by treatment with ABT-199 (Figure 6D). In contrast, treatment with ABT-199 diminished B220⁺IgM⁺ mature B cells in the bone marrow by 54% while the precursor populations, B220^lowIgM⁺ immature B cells, were normal and earlier progenitors (B220⁺IgM⁻) populations were increased. These data are consistent with the known roles of BCL-2 in late but not early B lymphopoiesis (34).

Figure 6. Inhibiting BCL-2 modestly diminished glomerular IgG deposition and did not alter serum anti-dsDNA antibody titers.

(A) Renal IgG deposition after ABT-199 and MMF treatment compared to respective vehicle treatment. (B) Semi-quantitative scoring of IgG deposition in glomeruli, compared vehicle, and ABT199 and MMF treated mice by two-tailed Mann-Whitney U test. *: P < 0.05 (C) Graphs showing anti-dsDNA antibody titers in both vehicle and ABT-199 treated mice. There was no statistically significant difference at each time point measured (two-tailed Student’s t-test). (D) FACS analysis of different B cell subtypes in bone marrow of treated mice. Two-tailed Mann-Whitney U test was used for comparison. *: P < 0.05. (E) Graphs showing dose-dependent peripheral lymphocyte number reduction with ABT-199 treatment with platelet sparing. One-way analysis of variance was used. *: P < 0.05. MMF = mycophenolate mofetil and FACS = flow cytometry

ABT-199 diminished peripheral blood lymphocyte counts in a dose dependent manner (Figure 6E). A dose of 1.2 mg/kg/day decreased lymphocytes by 30% while the 33 mg/kg/d dose decreased lymphocytes by 69%. Treatment with the 100 mg/kg/d dose did not lead to further substantial lymphocyte depletion (72%). In contrast, platelets were preserved at all tested doses of ABT-199 (Figure 6E). Analysis of lymphocyte subsets in the spleen (Supplemental Figure 6) revealed depletion of both CD4⁺ and CD8⁺ T cells as well as CD19⁺ B cells. Within the B cell lineage, Transitional 2 (T2), mature and GC B cells were reduced. In contrast, T1 cells, marginal zone B cells and B1 cells were relatively preserved.

Discussion

In secondary lymphoid organs, lymphocyte activation is normally associated with the coordinated down-regulation of BCL-2 and the induction of MCL-1, with the latter being required for GC responses (17, 18, 26, 27). In contrast, our data reveal that activated infiltrating lymphocytes in human LuN retain a pattern of BCL-2 and MCL-1 expression similar to that observed in naïve lymphocytes. A similar pattern of differential expression was observed in MR suggesting that high BCL-2 expression in infiltrating lymphocytes might be a general feature of inflammation rather than a specific feature of autoimmunity. Finally, in NZB/W F₁ mice, renal infiltrating inflammatory cell populations were very sensitive to BCL-2 inhibition. These data suggest that inhibiting BCL-2 will decrease in situ immunity in LuN and possibly other inflammatory renal diseases.

Through the development and application of automated and high-throughput methods of cell counting within human tissue, we were able to objectively and quantitatively measure the expression and frequency of defined subsets of infiltrating cells. ACP allowed us to accurately compare expression frequencies in clinical biopsies between patients and disease states to identify underlying commonalities in both autoimmune and alloimmune renal disease. A commonality, the dysregulation of BCL-2, was then explored in a relevant murine model to demonstrate that BCL-2 function was critical for renal inflammation. These findings demonstrate a general approach in which quantitative analysis of specific regulatory pathways operative in situ in human disease identifies putative disease effector mechanisms that can then be tested in relevant murine models.

The application of ACP to human renal biopsies allowed us to identify regional variance in the expression of BCL-2. Surprisingly, BCL-2 expressing lymphocytes were usually found in the tubulointerstitium and not in glomeruli. This is consistent with previous observations that LuN TII is associated with the features of in situ adaptive immune selection including intrarenal B cell clonal expansion, ongoing somatic hypermutation and cognate T cell help (13, 19). MR also has similar features of in situ adaptive immunity (35). This is different from LuN GN which is a manifestation of systemic autoimmunity in which antibodies to ubiquitous self-antigens, especially dsDNA, deposit within glomeruli where they induce inflammation, damage and scarring (13, 36, 37). Interestingly, treatment with ABT-199 had less effect on GN than TII. Furthermore, circulating anti-dsDNA antibodies were not affected by treatment with ABT-199. These data highlight the pathogenic differences between lupus GN and TII and suggest that targeted therapies might preferentially attenuate specific manifestations of LuN.

Dysregulation of BCL-2 in TII could arise from several mechanisms. Local cytokines could enhance BCL-2 expression. In human lupus, IL-2, IL-4, IL-7 and IL-15 have been shown to induce BCL-2 production in peripheral mononuclear cells (38). Furthermore, STAT2, 3, 5, 6 and c-Myb can directly bind and induce Bcl-2 (39). Another possibility is that lymphocytes infiltrating the tubulointerstitium have more of a memory than a GC phenotype. However, such cells are reported to express high levels of both BCL-2 and MCL-1 which we did not observe (27, 40).

MCL-1 is an anti-apoptotic protein that is broadly expressed in peripheral lymphocytes where it is required for both maintaining peripheral lymphocyte populations (41) and GC B cells (26, 27). However, in LuN, the fraction of lymphocytes expressing MCL-1 was lower than GCs and comparable to that found in follicular B cells. This apparent failure to upregulate MCL-1 likely contributes to exquisite sensitivity of in situ lymphocytes to BCL-2 inhibition.

A previous study has shown that ABT-737, a duel inhibitor of BCL-2 and BCL-X_L, significantly reduced disease severity in IFN-α induced NZB/W F₁ mice (23). However, concomitant platelet depletion was observed due to the requirement for BCL-X_L in platelet survival (42). The high specificity of ABT-199 for BCL-2 circumvented platelet toxicity while preserving efficacy in a murine model of SLE. Furthermore, the specificity of ABT-199 allowed us to demonstrate that specific cell subsets and disease manifestations were dependent upon BCL-2. It is important to note that ABT-199 led to dose-dependent peripheral lymphopenia of approximately 70% in mice treated with either 33 mg/kg/d or 100 mg/kg/d. While lymphocyte depletion could be a consequential mechanism of action, ABT-199 likely has additional mechanisms as TII was completely resolved. For example, BCL-2 antagonists deplete plasmacytoid dendritic cells and diminish interferon-α production (43).

One limitation of the study was that a preventive murine model was used to show the efficacy of BCL-2 blockade. Another study using a classical treatment model is needed to further support its clinical use.

In LuN (44), up to about 40 – 60 % of patients develop end-stage renal disease over 10 years despite conventional therapy (45). Clearly, better treatments for LuN are needed. Our studies suggest an essential role for BCL-2 in maintaining pathogenic B and T lymphocyte populations in renal inflammation. However, additional studies are required to test whether inhibiting BCL-2 will improve renal survival in patients with LuN and other inflammatory renal diseases.