PDZ Proteins SCRIB and DLG1 Regulate Myeloma Cell Surface CD86 Expression, Growth, and Survival

Multiple myeloma, a disease of antibody-producing plasma cells, is the second most common hematologic malignancy in the United States (1). In 2020, approximately 32,000 patients were diagnosed with myeloma and 13,000 deaths occurred (2). Although therapeutic agents such as proteasome inhibitors, immunomodulatory drugs, and targeted antibody treatments have increased median survival rates, efficacy of these agents is still limited, and the majority of patients become drug resistant and relapse (3–5). While normal plasma cells are highly reliant on interactions with bone marrow stromal cells for survival, myeloma cells may become extramedullary in advanced stages and can survive and proliferate outside of the bone marrow microenvironment (6–9). Thus the mechanisms that allow multiple myeloma cells to become independent of the bone marrow microenvironment are central to understanding how the disease progresses to this terminal treatment refractory stage.

Some clues underlying myeloma cell progression can be found within the bone marrow microenvironment where the prosurvival interactions of myeloma cells with stromal cells and the extracellular matrix have been previously studied (6, 10–12). One such stromal–myeloma cell interaction involves the binding of myeloma cell receptor CD28 to CD80/CD86 of a dendritic cell (13). This CD80/CD86–CD28 interaction is primarily known in context of T-cell costimulatory response. During this process, the T-cell receptor is first activated by MHC peptide complex of an antigen-presenting cell (APC), and CD80/86 expressed on APCs binds to T-cell CD28 to provide costimulation to maximally activate T-cell proliferation, and survival (14–18). Because most myeloma cells coexpress CD28 and CD86, we hypothesized a similar role for the proteins in myeloma cells. Specifically, because the CD86–CD28 interaction promotes survival in T cells, this suggests a possible mechanism by which myeloma cells survive independent of stromal cell signals. This pathway could represent a therapeutic target in myeloma, especially as the FDA has already approved inhibitors of the CD86–CD28 interaction (CTLA4-Ig) for treatment of graft-host rejection and autoimmune disorders (19–22).

CD86 has been primarily characterized as a ligand of CD28, and numerous CD28 signaling motifs and pathways have been identified (23–26). However, recent work on B-cell function in mice (27–33) and dendritic cells (34) demonstrates that CD86 also signals upon ligation to initiate specific responses. Consistent with this, our lab has recently shown that CD86 is necessary for myeloma cell survival and drug resistance (13, 35). The cytoplasmic region of CD86 is important for conferring these effects as well as inducing molecular changes in myeloma cells such as upregulation of IRF4, ITGB1, and ITGB7 (35). This suggests that proper surface expression of CD86 and its downstream signaling is important in myeloma, While a specific polylysine motif in the cytoplasmic tail has been shown to associate CD86 with the cytoskeleton to maintain surface expression in APCs (16), other motifs in the cytoplasmic tail are not known. CD86 contains numerous sites of N-linked glycosylation in its extracellular domain that can direct trafficking to the plasma membrane. However, additional means of regulating CD86 surface expression have remained understudied. Therefore, we set out to further elucidate mechanisms of proper CD86 trafficking in myeloma via its cytoplasmic tail.

Human embryonic kidney 293T (HEK293T) cell line was purchased from ATCC. Cells were cultured and seeded in 6-well plates containing 2 mL DMEM (Corning) supplemented with 10% FBS (Gemini) at 37°C in an incubator with 5% CO₂. Myeloma cell lines were cultured as described previously (36). MM.1s were provided by Dr. Steven Rosen, KMS18 were purchased from the Japanese Cell Resource Bank, and RPMI8226 were purchased from ATCC. Cells were cultured in RPMI1640 media supplemented with 10% FBS at 37°C in an incubator with 5% CO₂. Cell lines were Mycoplasma tested using PCR-based Mycoplasma testing service from ATCC. RPMI8226 and HEK293T cells were most recently tested for Mycoplasma in January, 2022 while KMS18 and MM.1s were most recently tested in July, 2021. Cells were passaged for 4–6 weeks and experiments were performed between 1 and 6 weeks following thaw. Identity of myeloma cells was validated by short tandem repeat. This was repeated for the KMS18 and RPMI8226 cells after introduction of the Cas9 vector to assure these cells were of the correct origin.

HEK293T cells were seeded on coverslips on a 6-well plate. The next day, they were transfected using Lipofectamine-2000 (Invitrogen) according to manufacturer's instructions after reaching a confluency of 70%–90%. Forty-eight hours after transfection, cells were either lysed for protein extraction or fixed for immunofluorescence. Cells were transfected with CD80 (Genscript) and CD86 constructs cloned into pcDNA3.1+ plasmid as described previously (35).

Cell pellets were lysed in RIPA buffer with protease and phosphatase inhibitors as described previously (9). Lysates were quantified using the bicinchoninic acid assay (Thermo Fisher Scientific), and lysates were run in SDS-PAGE gels, then blotted as described previously (9). Primary antibodies used included: Mouse monoclonal α-CD86 (R&D), mouse monoclonal α-β-actin (Sigma), Streptavidin horseradish peroxidase (HRP; Millipore) and rabbit polyclonal α-HA (Abcam). The following secondary antibodies were used: anti-mouse IgG-HRP and anti-rabbit IgG-HRP (Santa Cruz Biotechnology).

Cells were grown on glass coverslips (Thermo Fisher Scientific) coated with 5 μg/cm² fibronectin unless otherwise specified (Millipore). Twenty-four hours after transfection, cells were fixed with PHEMO buffer (68 mmol/L PIPES, 25 mmol/L HEPES, 15 mmol/L EGTA-Na₂, 3 mmol/L MgCl_2•6H₂O, 10% DMSO, pH 6.8) supplemented with 3.7% formaldehyde (Thermo Fisher Scientific), 0.05% glutaraldehyde (Thermo Fisher Scientific) and 0.2% Triton X-100 (Bio-Rad) for permeabilization when indicated. CD86 was labeled by staining with CD86-APC antibody (Caprico Biosciences). CD80 was labeled by staining with CD80-FITC antibody (BD Biosciences). Mannose-6 phosphate receptor was labeled overnight at 4°C using a rabbit polyclonal antibody α-Mannose-6 phosphate receptor provided as a gift from Dr. Paul Luzio (Cambridge Institute for Medical Research; ref. 37). Other antibodies used were Rabbit α SYNTENIN (Abcam), Rabbit α SCRIB (Cell Signaling Technology) and Rabbit α DLG1 (Thermo Fisher Scientific), Mouse α early endosomal antigen-1 (EEA1; BD Biosciences). Goat α rabbit Alexa Fluor 488 (Invitrogen) or Goat α Mouse Alexa Fluor 488 secondary antibody was used to stain cells for 1 hour at room temperature. Coverslips were then mounted on microslides (Thermo Fisher Scientific) using Prolong Gold containing 300 nmol/L 4′-6-diamidino-2-phenylindole dilactate (DAPI; Invitrogen). Cells were imaged using a Leica TCS SP8 inverted confocal microscope (63× oil HC PL APO, NA 1.4). Mander's correlation coefficient was analyzed for five cells in three independent assays (15 cells total) and determined using Colocalization tool on FIJI ImageJ.

For BioID analysis, 50 million cells of MM.1s BirHA or MM.1s CD86-BirHA were cultured in 150 cm² flasks. Subsequent BioID was performed as described previously (38). Gene ontology was performed using String.db. PDZ domain–containing proteins were identified using the HUGO nomenclature online database.

SCRIB/CD86 and DLG1/CD86 interactions were detected in situ using Duolink II secondary antibodies and detection kits (Sigma-Aldrich, #DUO92002, #DUO92004, and #DUO92008) according to the manufacturer's instructions. Briefly, proximity ligation assay (PLA) probes and primary antibodies: Rabbit α HA (Abcam), Mouse α SYNTENIN (Abnova) Mouse α SCRIB (GeneTex), and Mouse α SAP97/DLG1 (Santa Cruz Biotechnology) were applied to fixed cells. Then, Duolink secondary antibodies were added. Polymerase and amplification buffer were added to amplify a positive signal (red dot) and detected by confocal microscopy. DAPI was used for counterstaining of the nucleus.

RNA was extracted and qRT-PCR was performed as described previously using Applied Biosystems High Capacity cDNA Reverse Transcription Kit (Life Technologies; ref. 36). Resulting cDNA was amplified using the TaqMan Gene Expression Master Mix (Life Technologies) on the CFX96 Real-Time PCR System following the manufacturer's protocol (Bio-Rad). Probes used were Glyceraldehyde-3-phosphate dehydrogenase, SYNTENIN, CD86, SCRIB, DLG1, and IRF4 (Applied Biosystems).

Cell-surface expression CD86-APC (Caprico Biosciences) and integrin β7-PE (BD Biosciences) were measured via flow cytometry. A total of 100,000 live cells were collected, washed with 1× PBS, and stained with appropriate antibodies in 50 μL FACS staining buffer. After incubation of 15 minutes at 4°C in the dark, cells were washed in 1× PBS and resuspended in 400 μL FACS staining buffer. Samples were run on a FACS Symphony A3 Flow Cytometer (BD Biosciences) and analyzed using FlowJo software.

Inducible Cas9-expressing cells were generated using the pCW-Cas9 plasmid through lentiviral infection (39). Virus was produced by transfecting HEK293T cells with pCW-Cas9 (gift from Eric Lander and David Sabatini: Addgene plasmid # 50661) and packaging plasmids DR8 and VSVg using Lipofectamine 2000 (40). Lentiviral infection and selection of cells were performed as described previously (39)

CD86, SYNTENIN, SCRIB, and DLG1 single-guide RNA (sgRNA) clones were generated by designing specific sgRNAs using the online CRISPOR tool (41) and cloning them into pLX-sgRNA (gift from Eric Lander and David Sabatini: Addgene plasmid # 50662) using the following primers: GTCGAGTGTGCTACTCAACTCGTTTTAGAGCTAGAAATAGCAA (forward) and GAGTTGAGTAGCACACTCGACGGTGTTTCGTCCTTTCC (reverse). The sgRNA sequences are provided in Supplementary Table S1. Virus generation and infection of inducible Cas9 cells were as described previously (39). Twenty-four hours after infection, 10 μg/mL blasticidin was added to select cells for 7 days.

KMS18 and RPMI8226 Cas9 inducible cells were infected with sgRNA containing lentiviral particles as described above. The following day, live cells were counted using 100 μL of cells and 100 μL of FITC-conjugated counting beads (Thermo Fisher Scientific) and run on a FACS Symphony A3 Flow Cytometer (BD Biosciences). A total of 100,000 live cells were then seeded on 24-well plates in 2 mL of complete media. Doxycycline hyclate (1 μg /mL; Sigma) was then added to cells and counts were repeated using FITC-conjugated counting beads as per the manufacturer's instructions for 7 days.

Cell death was measured by annexin V-fluorescein isothiocyanate (BioVision) and propidium iodide (PI) staining as described previously (36).

Statistical significance was assessed using two-tailed Student t test using GraphPad Prism.

We previously generated RPMI8226 myeloma cell lines which stably overexpress full-length CD86 (CD86FL) or a “tail-less” mutant of CD86 (CD86TL; ref. 35). CD86TL expresses the extracellular and transmembrane domain and seven amino acids of the cytosolic domain (Fig. 1A). We used vectors coding for CD86FL and CD86TL and transiently transfected them into HEK293T, a cell line which does not endogenously express CD86. We then stained fixed cells with a CD86 antibody to determine CD86 localization 2 days after transfection (Fig. 1B). While CD86FL can be observed at the perimeter of the CD86FL cells, CD86TL formed a more diffuse staining pattern with several puncta inside the cells. To verify that this phenotype was not due to a discrepancy in transfection efficiency, we took lysates of these cells and performed Western blot analysis. There was no decrease in CD86 total protein expression and in fact, the CD86TL expression was significantly higher than CD86FL (Fig. 1C). We also determined that the stable RPMI8226-CD86FL myeloma cells also produced a uniform stain around the perimeter while the RPMI8226-CD86TL myeloma cell lines had more punctate staining (Fig. 1D). To determine what stage CD86 trafficking to the surface is inhibited, we costained the HEK293T cells with Mannose-6 phosphate receptor (M6PR), a Golgi marker, as well as CD86 (Fig. 1E). We then analyzed colocalization of these proteins using a Mander's correlation coefficient and observed a significant increase in colocalization between M6PR and CD86 in the CD86TL compared with CD86FL-transfected HEK293T (Fig. 1F). Because CD86 is glycosylated, we would expect it to be exported from the Golgi under normal conditions following its synthesis and posttranslational modifications. The presence of CD86 in the Golgi apparatus could indicate a defect in anterograde trafficking or increased retrograde trafficking. Therefore, we used a pulse-chase experiment utilizing cell surface biotin labeling to measure rates of CD86 delivery to the plasma membrane following trypsin cleavage. We observed that CD86FL was able to repopulate the cell surface after 1 hour while CD86TL exhibited delayed recovery (Supplementary Fig. S1A). To determine whether truncation of the CD86 cytoplasmic tail results in increased retrograde trafficking, we also costained with CD86 and an endosomal marker, EEA1 and saw no significant difference in colocalization with CD86 in CD86FL and CD86TL transfected lines (Supplementary Fig. S1B). Together, these data show that the cytoplasmic tail has a role in conferring proper transport of CD86 from the Golgi apparatus to the membrane.

The cytoplasmic region of CD86 is composed of a 61 amino acid long tail. Because we detected a trafficking defect in the CD86TL-transfected HEK293T, we defined which areas of the tail are important for CD86 trafficking. Homology alignment of human CD86 to other higher mammalian species identified numerous conserved and potentially important areas of the cytoplasmic tail (35). Full-length CD86 is 329 amino acids in length, and there may be multiple motifs working in conjunction along the protein to regulate trafficking. Therefore, instead of mutating specific motifs, we developed CD86 truncation mutants that were 282 (CD86∆282) 298 (CD86∆298), and 315 (CD86∆315) amino acids long, respectively (Fig. 2A). We then transfected the truncation mutants into HEK293T cells and observed CD86 trafficking 2 days following transfection (Fig. 2B). While the truncation mutants trafficked CD86 more effectively than CD86TL, none were able to fully phenocopy CD86FL. To determine what point the trafficking is inhibited, we costained M6PR and CD86 in pcDNA3.1 vector control, CD86FL-, CD86∆282-, CD86∆298-, and CD86∆315-transfected cells (Fig. 2C). When we quantified colocalization using a Mander's correlation coefficient, we observed that the truncation mutants each had significantly higher colocalization of M6PR and CD86 compared with CD86FL (Fig. 2D). Furthermore, there is less colocalization the longer the tail, suggesting that there are several regions of the tail that are important for proper CD86 trafficking.

Because full-length CD86 contains 329 total amino acids, and the CD86∆315 mutant improperly traffics, we identified 14 amino acids at the C-terminus of CD86 to further dissect. Among higher order mammalian species, CD86 contains a conserved class I Psd95/DLG1/zo-1 (PDZ)-binding motif at its C-terminus (Fig. 3A). A class I PDZ-binding motif is composed of a S/T-X-ϕ motif where X represents any amino acid and ϕ represents a hydrophobic residue (42), and human CD86 C-terminus meets these criteria with TCF at its C-terminus. This motif is commonly found at the C-terminus of receptor proteins and is recognized by adaptor proteins that contain a PDZ domain. PDZ proteins can then perform numerous signaling or scaffolding functions specific to cell and context (42).

To determine whether the PDZ-binding motif has a role in cellular trafficking of CD86, we developed a truncation mutant which lacks the final three amino acids of CD86, effectively truncating the PDZ-binding motif (CD86∆326). We then transfected this construct into HEK293T and costained with M6PR and CD86 2 days after transfection (Fig. 3B). After quantification of colocalization using a Mander's correlation coefficient, we observed significantly more colocalization of M6PR and CD86 in the CD86∆326 HEK293T compared with CD86FL, illustrating that the PDZ-binding motif is important for proper CD86 trafficking to the cell surface (Fig. 3C).

Like CD86, CD80 is another receptor protein classically recognized to be involved in the T-cell costimulatory response (14). Unlike CD86, CD80 is not expressed in myeloma cells, so the two proteins may not have redundant roles. Interestingly human CD80 does not contain a PDZ-binding motif, thereby suggesting a further difference in trafficking between the two proteins in immune cells (Supplementary Fig. S2A). We developed a “tail-less” mutant of CD80 to assay whether CD80 cytoplasmic tail also confers a proper trafficking phenotype (Supplementary Fig. S2B). When we transfected HEK293T with CD80FL and CD80TL we observed no notable difference in trafficking between the two proteins, illustrating a different mechanism of transport between the two proteins (Supplementary Fig. S2C).

To determine possible interacting partners for the CD86 cytoplasmic tail, we utilized the BioID proximity-based labeling assay. This method utilizes a constitutively active biotin ligase (BirA) that is engineered to biotinylate proteins within 10 nm of it (38). We developed a construct with BirA containing an HA tag (BirHA) cloned into the CD86 tail prior to the PDZ domain (CD86-BirHA; Supplementary Fig. S3A). We transfected soluble BirHA or CD86-BirHA into the myeloma cell line, MM.1s, resulting in stable cell lines expressing either protein (Fig. 4A; Supplementary Fig. S3B). We then used streptavidin-conjugated beads to pull down proximal proteins in either cell line (Supplementary Fig. S3C). On-bead trypsin digestion was performed and unique peptides were analyzed using mass spectrometry.

Proteins were then stratified on the basis of peptide-spectrum match (PSM) score. This assigns a numerical value that expresses the likelihood that fragmentation of a peptide is contained in the experimental spectrum (43). Proteins with a PSM score of greater than 5 met the determined threshold for a positive hit. Proteins were then deemed to be enriched in the CD86 cytoplasmic tail if a threefold higher PSM score in MM.1s CD86-BirHA was observed in comparison with MM.1s BirHA. In total, there were 225 proteins that were associated with the CD86 cytoplasmic tail (Supplementary Table S2). Using the online database, string.db, we performed gene ontology analyses for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis and Reactome pathways (Fig. 4B) and found that the top hits for each pathway were endocytosis in KEGG (FDR = 9.34 × 10⁻⁶) and membrane trafficking in reactome pathways (FDR = 1.13 × 10⁻⁵). The Reactome pathways analysis also revealed additional functions such as receptor tyrosine kinase signaling, Rho GTPases, glycosylation, intra-Golgi and retrograde Golgi-ER traffic, and COPI-mediated anterograde transport which providing additional information about the biological processes controlled by CD86-binding partners.

We next determined how many of the stratified hits were PDZ domain–containing proteins. Using the HUGO nomenclature online database, we identified 152 known PDZ domain-containing proteins and cross-referenced them with our 225 enriched proteins. Ten proteins with PDZ domains were found among our hits: SLC9A31, SCRIB, SYNTENIN, PDLIM, PARD3, GORASP2, CASK, AHNAK, DLG1, and ERBB2IP (Fig. 4C). Using the CoMMpass dataset, a longitudinal study which follows patient gene expression throughout treatment, we determined that all 10 PDZ domain proteins are expressed to varying degrees in patients with primary myeloma (Supplementary Fig. S4A). To stratify proteins to focus, we analyzed patient subtype expression. Both CD86 and CD28 have increased expression in the MAF myeloma subtype (7, 35). We identified SCRIB, AHNAK, and PARD3 as genes with the highest expression in the MAF subtype and DLG1 having the lowest expression in this subtype (Supplementary Fig. S4B). Among the three highest expressing proteins, we decided to focus on SCRIB as it is prognostic of poor patient progression-free and overall survival (Supplementary Fig. S4C). We also wanted to test a PDZ protein that has median relative expression in the MAF subtype and selected SYNTENIN (encoded by SDCBP) as it has been shown to regulate surface levels CD138. CD138 or SYNDECAN is a marker of myeloma cells with a role in myeloma cell adhesion, and we therefore were interested in the potential role of SYNTENIN in myeloma (44). Furthermore, using the CoMMpass dataset, we observed a positive correlation between SDCBP and CD86 mRNA expression in patients (Supplementary Fig. S4D). Taken together, these proteins may have implications in CD86 expression as well as myeloma cell survival and bone marrow localization, and warranted further exploration.

SCRIB and DLG1 are apical-basal polarity proteins that have been shown to regulate surface levels of numerous proteins (45, 46) and have been shown to form complexes with one another to regulate polarity axes of cells (47). Using a PLA, we verified that SCRIB and DLG1 interact with CD86 while SYNTENIN does not (Fig. 4D). We also costained CD86-overexpressing RPMI8226 cell lines with CD86 and either SYNTENIN, SCRIB, DLG1, or a secondary antibody-only control (Fig. 4E). We compared colocalization of CD86 with either a nonspecific secondary antibody-only control, SYNTENIN, SCRIB, or DLG1. While there was no significant difference between SYNTENIN colocalization and the secondary-only control, there was a significant increase in SCRIB and DLG1 colocalization compared with SYNTENIN (Fig. 4F). Identification of SCRIB and DLG1 colocalization with CD86 led us to further explore their function in multiple myeloma.

We next wanted to elucidate the role of SCRIB and DLG1 in CD86 trafficking to the cell surface. While they have been involved in numerous cancers, their roles in multiple myeloma have been unexplored (48, 49). We utilized myeloma cell lines, KMS18 and RPMI8226, which were engineered to stably express a doxycycline-inducible Cas9 enzyme. sgRNA targeting either CD86 (sgCD86), SCRIB (sgSCRIB), and DLG1 (sgDLG1) were generated on the basis of specificity and efficiency using the online CRISPOR tool (41). As a control, a single-guide vector targeting the viral incorporation site, AAVS1, was used (plx-sgRNA; ref. 40). We used lentiviral particles to infect myeloma cells with sgRNA and then selected for cells containing the vectors for 7 days using blasticidin. Following selection, cells were plated and doxycycline was added to activate the Cas9 enzyme expression (Fig. 5A). We have previously determined a role for CD86 in myeloma cell survival and hypothesized that SCRIB and DLG1 may be involved as well. We therefore did not establish stable cell lines and instead performed a transient assay measuring cell counts, cell death, and CD86 surface expression for 7 days.

We first verified that CD86 surface expression was decreased with the addition of CD86 guides. Indeed, we observed reduced CD86 expression in two guides targeting CD86 as soon as 3 days after doxycycline addition (Fig. 5B). At day 3, we also observed decreased CD86 surface expression with three SCRIB guides in KMS18 (Fig. 5C and D, and G; Supplementary Fig. S5A). Similarly, we observed decreased CD86 surface expression in the sgDLG1 KMS18 cells after 3 days (Fig. 5E–G; Supplementary Fig. S5A). While surface levels were decreased, we did not observe any difference in total CD86 protein levels (Supplementary Fig. S5B). We repeated this experiment using two guides against SYNTENIN (sgSYNTENIN) with the hypothesis that CD86 surface expression would not be affected as SYNTENIN does not colocalize with CD86. Indeed, we observed no change in CD86 surface expression (Supplementary Fig. S6A and S6B). These data suggest that SCRIB and DLG1 affect CD86 surface expression in myeloma, and we wanted to further explore what effect SCRIB knockout (KO) and DLG1 KO would have on myeloma cells.

To determine whether KO of CD86, SYNTENIN, SCRIB, or DLG1 resulted in less proliferation of myeloma cells, we used FITC-conjugated beads and flow cytometry to count live cells 0 to 7 days following doxycycline addition. We conducted these experiments in KMS18 and RPMI8226 myeloma cell lines and found that KO of CD86, SCRIB, and DLG1 (Fig. 6A–C) resulted in significantly diminished cell growth compared with plx-sgRNA control cells. We hypothesized that this decrease in cell count could be largely due to an increase in cell death. We have previously shown using short hairpin RNA knockdown that loss of CD86 results in significantly higher cell death in both KMS18 and RPMI8226 (35). Indeed, we saw that KO of CD86 via CRISPR-Cas9 resulted in increased cell death in both lines (Fig. 6D). SCRIB and DLG1 gene editing resulted in an increase in cell death in both cell lines (Fig. 6E and F). We also observed decreased cell proliferation and increased cell death in SYNTENIN KO lines despite the lack of change in CD86 surface expression (Supplementary Fig. S6C and S6D). This suggests a role for SYNTENIN in myeloma viability that is independent of control of CD86 trafficking.

Interestingly, the majority of cell death occurred during days 1–3 in all KO lines. This is particularly evident in RPMI8226 lines where approximately 80% of all KO cells die by day 3 (Fig. 6D–F; Supplementary Fig. S7A and S7B). RPMI8226 are largely dependent on CD86 for survival (35), and the majority of CD86 KO cells die by day 3 (Fig. 6D). In the RPMI8226 SCRIB KO and DLG1 KO, a “CD86 low” population fails to grow out, and the few cells that do survive past day 3 have normal protein expression of CD86 (Supplementary Fig. S7B). Furthermore, in the KMS18 SCRIB KO and DLG1 KO, the “CD86 low” populations do not grow out past day 3, and CD86 resurfaces in days 4–7 (Supplementary Fig. S5C). This points to a population of cells that did not efficiently edit CD86 as the population that was able to effectively grow out.

To further investigate the role that SCRIB and DLG1 had in CD86-mediated cell survival, we costained RPMI8226-CD86FL cells with CD86 and either SCRIB or DLG1 (Fig. 7A). Because CD86 signals via cell–cell contact, we hypothesized that there would be a difference in colocalization of CD86 and SCRIB or DLG1 at points of cell contact compared with noncontact sites. Surprisingly, we observed a decrease in SCRIB/DLG1 colocalization with CD86 at sites of contact compared with noncontact sites (Fig. 7B). We also noticed that these contact sites had higher expression of CD86 in the RPMI8226-CD86FL cell lines (Figs. 4E and 7C). We hypothesized that binding of CD86 to CD28 on myeloma cells stabilizes CD86 expression and allows SCRIB and DLG1 to leave. Consistent with this possibility, CD86FL-transfected HEK293T, a cell line which does not express CD28, displayed no difference in CD86 expression in contact sites compared with noncontact sites (Fig. 7C and D). To verify whether the cell death in the SCRIB and DLG1 KO lines was due in part to the lack of a CD86-intrinsic signal we determined whether loss of the PDZ domain proteins shared other characteristics with the loss of CD86. We have previously shown that loss of CD86 in myeloma cell lines leads to decreased expression of IRF4 and Integrin β7 (35), two proteins which have been implicated in myeloma cell survival (50, 51). We quantified mRNA expression of IRF4 in our CD86, SCRIB, and DLG1 RPMI8226 KO lines and observed a significant decrease with every sgCD86 and sgSCRIB line and two of three DLG1 sgRNAs (Fig. 7E). In addition, integrin β7, decreased in all CD86, SCRIB, and DLG1 sgRNAs 2 days following doxycycline addition (Fig. 7F). This suggests that the decrease in CD86 surface expression observed with the loss of SCRIB or DLG1 results in a decrease in CD86-mediated growth and survival signaling. To verify that loss of CD86 signaling is important for SCRIB KO- and DLG1 KO-induced cell death, we overexpressed CD86FL in SCRIB KO and DLG1 KO RPMI8226 cells and observed that dominant CD86 expression rescues cell death induced by SCRIB/DLG1 ablation (Fig. 7G; Supplementary Fig. S7C).

Interaction between CD28 and CD86 on myeloma cells facilitates myeloma cell survival in a bone marrow–independent niche. We previously demonstrated a role for the CD86 cytoplasmic tail in drug resistance and induction of molecular changes in myeloma cells. This study illustrates how the CD86 cytoplasmic tail mediates effective trafficking of the protein to the cell surface. We took advantage of a cell line lacking endogenous CD86 to monitor trafficking of full-length CD86 or a tail-less mutant of CD86 and observed a clear difference in efficiency of CD86 to traffic to the plasma membrane. We have shown that lacking the tail results in accumulation of CD86 at the Golgi apparatus, and KEGG pathway analysis has revealed numerous proteins involved in endocytosis in proximity to CD86 cytoplasmic tail. While retrograde trafficking may play a role in CD86 surface expression, the lack of difference in EEA1 colocalization between CD86FL- and CD86TL-transfected HEK293T suggest that the tail primarily affects anterograde trafficking.

The CD86 cytoplasmic tail is 61 amino acids long and makes up almost 20% of the protein sequence. It follows that there may be numerous regions that can be possible binding regions for intracellular proteins to influence trafficking. Because the length of the tail inversely corresponds with the amount of CD86 that colocalizes with M6PR, we hypothesize that a variety of proteins may be binding to the tail at different locations to influence CD86 localization both inside and outside of the Golgi. In addition, KMS18 cytoplasmic tail contains a SNP resulting in a A304T change whereas MM.1s and RPMI8226 have an allele that is conserved among higher order mammalian species. This polymorphism is associated with increased cancer risk and allograft rejection, illustrating the importance for further study of various regions of the CD86 cytoplasmic tail (52–54).

The discovery of a PDZ-binding motif at the C-terminus provides clues as to how the protein is being regulated. PDZ domain proteins have been largely studied and have numerous context-dependent roles in a variety of cell types. The difference in trafficking effectiveness between full-length CD86 and CD86 lacking the PDZ-binding motif in HEK293T shows the importance of this motif for proper surface transport. The existence of the PDZ-binding motif on CD86 but not CD80 and lack of importance of CD80 tail in trafficking to the surface may also provide clues in antigen presenting cells toward the preferential expression of either CD80 or CD86 during T-cell costimulation. While the CD80 cytoplasmic tail does not seem to be necessary for trafficking to the surface, it has previously been shown to influence CD80 localization at the cell surface during T-cell costimulation (15, 55). Interestingly, ICAM-1, another receptor protein that is involved in both myeloma cell adhesion (56) and immunologic signaling (57) contains a PDZ-binding motif. SYNDECAN-1 (CD138) is yet another PDZ-binding motif-containing protein whose surface levels have previously been shown to be regulated by SYNTENIN (44). CD138 is a heparan sulfate proteoglycan important for myeloma cell adhesion, and this may be a means by which SYNTENIN affects myeloma cell growth and survival (58). The presence of a three-amino acid motif on proteins that are implicated in myeloma growth and survival represents a specific target for potential therapeutic treatment.

We mapped the interaction network of CD86 cytoplasmic tail using the BioID method followed by LC/MS. Among our enriched proteins, 10 of them contained PDZ domains. Of these 10 PDZ domain proteins, half of them (CASK, DLG1, PARD3, PDLIM, SCRIB) are involved in cell polarity. All of these five proteins except PDLIM have been shown to interact with one another (59–61). A recent study found that SCRIB can regulate CD86 surface levels in activated APCs (62). Our findings expand this research to malignant plasma cells, identify a role for DLG1 in CD86 regulation as well, and determine a novel outcome of this process in regulating myeloma growth and survival.

One notable enriched protein in the interactome without a PDZ domain was SLC3A2. This protein makes up the heavy chain of CD98, which is a determinant of immunomodulatory imide drug (IMiD) activity in multiple myeloma (63). IMiDs are cytostatic drugs that halt proliferation in myeloma. We have also previously found that CD98 light chain (SLC7A5) is significantly downregulated in both CD28- and CD86-silenced cells. Because our data suggest that CD86 may have a role in proliferation, the functional interaction between CD86 and CD98 in myeloma warrants further investigation.

The similar phenotype of decreased CD86 surface expression in SCRIB KO and DLG1 KO cells suggests that these proteins may be forming a complex with one another. Alternatively, they could be part of a larger complex which regulates CD86 membrane trafficking and localization. The classical function of these two proteins may underlie that not only the amount of CD86 surface expression but also its location is important for maintenance of cellular survival. For example, CD86 surface expression at points of cell–cell contact appears to be higher (Figs. 4E, 7C and D). Furthermore, we found a decrease in colocalization between CD86 and SCRIB/DLG1 in areas of cell contact (Fig. 7A and B). This suggests that SCRIB and DLG1 may be helping to traffic CD86 to areas of the cell surface where it can bind to CD28. Binding to CD28 then stabilizes CD86 surface expression and can allow SCRIB and DLG1 to leave the site of contact to perform other roles in the cell.

Our data also show that SCRIB and DLG1 are important for cell growth and viability in multiple myeloma cell lines, KMS18 and RPMI8226. The majority of cell death appears to take place within the first 4 days of doxycycline addition in both cell lines. Interestingly, in KMS18 we start to see expression of CD86 decrease by day 2 in our CD86 KO, SCRIB KO, and DLG1 KO. However, following day 4, the “CD86-low” population in the SCRIB KO and DLG1 KO do not grow out while a population of cells with normal expression of CD86 survives and proliferates. We see this to a greater effect in RPMI8226, a cell line that is more dependent on CD86. In this line, a CD86-low population is unable to grow out and roughly 80% of SCRIB/DLG1 KO cells die by day 3. The decrease of IRF4 and integrin β7 expression in the SCRIB/DLG1 KO cells and rescue of cell mortality via dominant CD86 expression points to the lack of CD86 signaling as a mechanism by which these cells die. Because we have also shown a role for CD28 in both normal and malignant plasma cell survival, we cannot rule out a role for diminished CD28 signaling as a result of lower CD86 surface expression contributing to cell death as well (26). The presence of a population of normally expressing CD86 cells is likely due to incomplete gene editing; however, it remains possible that activation of a compensatory mechanism to maintain CD86 surface expression in a small fraction of cells may occur. In addition, SCRIB and DLG1 have pleiotropic roles and may also be conferring myeloma cell survival via additional mechanisms.

Taken together, our data identify a means of CD86 transport and expression at the surface of myeloma cells. It illustrates that SCRIB and DLG1 are two polarity proteins expressed in myeloma that can transport CD86 to the membrane. Proper transport is reliant on a complete cytoplasmic tail with a PDZ-binding motif at its C-terminus. Ablation of SCRIB or DLG1 results in a decrease in CD86 surface expression, myeloma cell survival, and proliferation. These results elucidate a role for PDZ proteins in regulation of myeloma growth and may provide new insights for targeted therapeutic advances in multiple myeloma.

K.P. Lee reports non-financial support from Bristol Myer Squibb outside the submitted work. L.H. Boise reports grants and personal fees from AstraZeneca and personal fees from Abbvie outside the submitted work. No disclosures were reported by the other authors.

T. Moser-Katz: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft. C.M. Gavile: Resources, investigation, methodology, writing–review and editing. B.G. Barwick: Resources, data curation, formal analysis, writing–original draft. K.P. Lee: Conceptualization, resources, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing. L.H. Boise: Conceptualization, resources, supervision, funding acquisition, investigation, writing–original draft, project administration.

This work was supported by R01 CA121044 (K.P. Lee), R01 CA192844 (L.H. Boise), and Paula and Rodger Riney Foundation (L.H. Boise). B.G. Barwick was supported by Developmental Funds from Winship Cancer Institute of Emory University, post-doctoral fellowship PF-17-109-1-TBG from the American Cancer Society, a Research Fellow Award from the MMRF, and American Society of Hematology Scholar Award. Research reported in this publication was supported in part by the Pediatrics/Winship Flow Cytometry Core of Winship Cancer Institute of Emory University, Children's Healthcare of Atlanta and was also supported in part by the Emory Integrated Genomics Core (EIGC), Emory Integrated Proteomics Core, and Integrated Cellular Imaging shared resources of Winship Cancer Institute of Emory University and NIH/NCI under award number P30CA138292. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

We thank Dr. Vikas Gupta for generation of the KMS18 and RPMI8226 Cas9 cell lines. We also thank Drs. Rachel Turn and Richard Kahn for antibodies used for immunofluorescence and their assistance and Dr. Andrew Kowalczyk for assistance with the biotin pulse chase surface labeling experiment.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.