α-Integrin expression and function modulates presentation of cell surface calreticulin

Calreticulin presentation on the cell surface is an important hallmark of immunogenic cell death (ICD), serving as the prophagocytic signal for macrophages. Cell adhesion is a physiologically relevant stimulus previously shown to increase calreticulin interaction with α-integrins via the juxtamembrane, cytosolic GFFKR motif. This study assessed whether integrin function can regulate surface calreticulin levels in ICD. We generated calreticulin-null T-lymphoblasts and confirmed the loss of surface calreticulin expression on cells treated with doxorubicin, an ICD inducer. Reconstituted expression with full-length calreticulin targeted to the endoplasmic reticulum (ER) successfully rescued doxorubicin-induced surface calreticulin. Reconstitution with a truncation mutant calreticulin targeted to the cytosol led to constitutively high surface calreticulin that was not further elevated by doxorubicin, suggesting calreticulin released from the stressed ER transits the cytosol before its translocation to the cell surface. When stimulated to engage integrin substrates, doxorubicin-treated wild-type T-lymphoblasts exhibited decreased surface calreticulin compared with cells under non-adherent conditions. The inhibitory effect on surface calreticulin was recapitulated for cells in suspension treated with a β1-integrin-activating antibody, 9EG7. Similarly, cells expressing a truncated α-integrin cytosolic tail, bearing only the juxtamembrane GFFKR calreticulin-binding motif, exhibited low surface calreticulin with doxorubicin treatment under non-adherent conditions. Using partial permeabilization techniques to distinguish between cytosolic and ER staining, we found that ICD inducers promoted the accumulation of cytosolic calreticulin with negligible change in total calreticulin, suggesting that integrin-mediated inhibition of surface calreticulin was due to reduced cytosolic to surface translocation. T-lymphoblasts co-treated with an ICD inducer and 9EG7 exhibited reduced phagocytosis by macrophages when compared with treatment with only ICD inducer. This study reveals a previously uncharacterized function of integrins as negative regulators of ICD by suppressing presentation of cell surface calreticulin.

Immunogenic cell death (ICD), or immunogenic apoptosis, is a form of cell death promoted by anthracycline-based chemotherapy, radiotherapy, and photodynamic therapy. 1,2 Unlike nonimmunogenic or tolerogenic apoptosis, ICD of tumor cells promotes their engulfment by professional phagocytes, leading to the activation of T cells and subsequent antitumor immune responses. 3 Such antitumor immune response could be beneficial for the treatment of metastatic diseases, where immune-based recognition of a primary treated tumor leads to targeting of tumors at a secondary site, a phenomenon known as the abscopal effect. 1,4 The key difference between nonimmunogenic and immunogenic apoptosis is the presentation and release of damage-associated molecular patterns (DAMPs) to the extracellular cell surface and milieu. 5 DAMPs have various mechanisms of action that include serving as a chemoattractant and the phagocytic 'eat me' signal for macrophages and dendritic cells to initiate the response to ICD. [6][7][8] Calreticulin (CRT) is a 46 kDa Ca 2+ -binding protein and molecular chaperone that is highly enriched within the lumen of the endoplasmic reticulum (ER). 9 The N-terminal domain carries a cleavable 17-amino-acid signal peptide that targets its insertion into the ER lumen. 10 Further enrichment of CRT within the ER is facilitated by a C-terminal KDEL ER-retention motif. The functional attributes of CRT outside of the ER remain a subject of intense interest. 11 When decorated on the surface of cells undergoing ICD, surface CRT is a DAMP that elicits the phagocytic response. 1,6,7 Compared with the ER, cytosolic CRT is of low abundance, 10 but its presence is supported by studies highlighting its intracellular transport 10,12 and requirements for the C-domain for the cytosolic localization. 13 More recently, somatic frameshift mutations resulting in CRT with a novel C terminus lacking the KDEL motif were described for a large subset of patients with myeloproliferative neoplasms. 14,15 Integrins are αβ-heterodimeric transmembrane cell adhesion receptors involved in numerous cellular functions including survival, growth, differentiation and apoptosis. 16 As such, integrin function has a pivotal role in the development of chemotherapeutic resistance and cancer relapse. 17,18 A physiological switch for integrin-mediated signaling is cell adhesion, which is facilitated by the activated, or high-affinity ligand binding, state of the integrins. Activated integrins undergo conformational changes that modulate the binding of effector proteins to the αand β-integrin cytosolic tails. 19 The tails of α-integrins share few sequence similarities, with the exception of the conserved membrane-proximal GFFKR motif that forms the inner membrane clasp with its β-subunit counterpart. 20 Several studies, including ours, have shown that integrin-mediated adhesion promotes binding of CRT, likely from the relatively low abundant cytosolic pool, to α-integrin tails in a manner requiring the GFFKR motif. [21][22][23][24] In this study, we investigated if α-integrin function may impact upon cell surface presentation of CRT using T-acute lymphoblastic leukemia (T-ALL) cells, which are amenable to cell adhesion studies and cytotoxic drug treatments to induce ICD. We present evidence suggesting that adhesion-mediated integrin-CRT interaction occurring in the cytosol constitute a mechanism reducing CRT translocation to the cell surface, which has implications in antitumor immunomodulatory activity.

Results
Doxorubicin induces surface CRT presentation in Jurkat T-lymphoblasts. To study ICD drug-mediated stimulation of surface CRT presentation in T-ALL cells, we used CRISPR-Cas9-mediated genome editing 25 to silence CRT expression (CRT null (CRT − / − )) in Jurkat T-lymphoblasts. Western blot analysis and sequencing of the CALR loci confirmed the loss of CRT expression resulting from a frameshift insertion and a predicted early translation termination (Figures 1a and b). To assess the effects of anthracycline on cell surface CRT, cells were treated with doxorubicin, and surface CRT assayed by flow cytometry on live, non-permeabilized cells. Doxorubicintreated wild-type (WT) cells exhibited~2-fold increase, whereas CRT − / − cells exhibited no change in surface CRT levels (Figures 1c and d).
Despite lacking CRT expression, CRT − / − cells exhibited a significant level of surface CRT when compared with the staining control. As this α-CRT-reactive signal was present in the serum-supplemented media (Figure 1d and Supplementary S1), we adapted and cultured cells in a serum-free synthetic replacement media. In a final culture state consisting of 0.3% fetal bovine serum (FBS) and 9.7% Cell-Ess, CRT − / − cells no longer exhibited any significant staining for surface CRT, with or without doxorubicin treatment (Figures 1e and f). Doxorubicin-mediated stimulation of surface CRT in WT cells remained equally robust (~2-fold increase) in 10 or 0.3% FBScultured conditions, indicating that most, if not all, of the surface CRT stimulated by anthracycline and detected with the α-CRT antibody is endogenous to the cell. These results indicate that anthracycline treatment of a T-leukemic cell line increased CRT trafficking to the surface from endogenous stores, and that this phenomenon is abrogated in cells lacking CRT expression.
Doxorubicin-mediated increase in surface CRT requires the ER-resident form. The N-terminal 17-amino-acid signal sequence (ss) of CRT is required for synthesis and insertion of the CRT propeptide into the ER lumen, whereas CRT expressed without ss is enriched in the cytosol. 10 To decipher the source of surface CRT in live cells, we used green fluorescent protein (GFP)-tagged CRT that was targeted either to the ER (ssGFP-CRT) or the cytosol (GFP-CRT) (Figure 2a). To validate their expected localization and targeting, CRT − / − cells transfected to express ssGFP-CRT or GFP-CRT was costained for protein disulfide isomerase (PDI), an ER-resident marker. For comparison, we stained for endogenous CRT and PDI in WT and CRT − / − cells. As expected, PDI exhibited colocalized staining with endogenous CRT in WT cells, and with ssGFP-CRT expressed in CRT − / − cells (Figure 2b). In contrast, PDI colocalized poorly with GFP-CRT in CRT − / − cells (Figure 2b), indicating that ssGFP-CRT targeted to the ER, whereas GFP-CRT remained diffuse in the cytosol.
Next, we analyzed surface CRT levels on live CRT − / − cells reconstituted with the ER-targeted ssGFP-CRT or the cytosolic-targeted GFP-CRT by flow cytometry. Expression of ssGFP-CRT restored the ability for doxorubicin-mediated stimulation of surface CRT in CRT − / − cells at~2-fold levels, comparable to WT cells (Figure 2c). In contrast, CRT − / − cells expressing GFP-CRT exhibited high surface CRT levels that was not further increased upon doxorubicin treatment ( Figure 2c). To confirm the functionality of the GFP-tagged CRT, we used GFP antibodies in place of CRT antibodies to detect the fusion GFP-CRT when presented on the surface of live, non-permeabilized cells. Non-transfected GFP-negative cells showed only background signals, whereas surface staining with GFP antibodies for CRT − / − cells expressing ssGFP-CRT or GFP-CRT were similar to that observed with CRT antibodies (Figures 2c and d), indicating that the fusion proteins were translocated to the cell surface and behave comparably to endogenous CRT. As a further control, we show that GFP was not present on the surface of CRT − / − cells expressing GFP alone (inset in Figure 2d).
These results show that ss-mediated targeting of CRT to the ER facilitated the increased surface CRT that was stimulated by doxorubicin. When targeted directly to the cytosol, high levels of CRTwere present at the surface in a manner that was not responsive to doxorubicin. Taken together, the data suggest a mechanism whereby doxorubicin-induced ER stress leads to CRT release from the ER lumen into the cytosol, following which CRT presentation on the surface is directly dependent on its increasing cytosolic concentration.
CRT is thought to be functionally pleiotropic with multiple cellular localizations. We assessed, and found that loss of CRT in T-lymphoblasts did not alter doxorubicin-treated extracellular release of ATP, a DAMP (Supplementary Table 1). 26 Similarly, loss of CRT and expression of the cytosolic GFP-CRT in T-lymphoblasts did not affect the expression of major histocompatibility complex (MHC) class I antigens (Supplementary S2). 27 In murine cells, anthracycline-stimulated surface translocation of ER-resident protein 57 (ERp57) is dependent on CRT. 28,29 We assessed ICD-induced surface ERp57 and found no significant difference between WT and CRT − / − lymphoblasts, or CRT − / − cells expressing GFP-CRT (Supplementary S3).
Expression and function of α-integrin reduces cell surface CRT. We, and others, reported that α-integrins exhibit increased interaction with CRT upon cell adhesion to integrin substrates. 21,23,24 This interaction is dependent on the GFFKR peptide motif found at the juxtamembrane cytosolic tail of α-integrins; thus, we speculated that α-integrin function may modulate surface CRT presentation. To address this, we assessed surface CRT levels for cells plated on integrin substrates. We used a Jurkat-derivative line lacking α4-integrins (α4 − / − ), and α4 − / − stably reconstituted with full-length α4 (α4 wt ), or with a cytoplasmic tail-truncated variant (α4δ) that terminates with the KxGFFKR cytosolic motif ( Figure 3a). Previously, we showed that α4δ-expressing cells do not adhere to an α4β1-specific substrate, yet exhibit increased α4δ binding to CRT in an adhesion-independent manner. 23 When plated on glutathione S-transferase (GST)-coated wells, a non-integrin engaging control substrate, doxorubicintreated α4δ cells showed significantly lower surface CRT compared with α4 − / − and α4 wt cells (Figure 3b). When plated on CS1, a fibronectin-derived fragment that specifically engages α4β1-integrins, 30 doxorubicin-treated α4 wt cells had lower surface CRT compared with α4 − / − cells (Figure 3b). When plated on Fn9.11, an RGD-containing fibronectin fragment that can engage multiple integrins (including α5β1integrins expressed in all three cell lines), 23 doxorubicintreated α4 − / − and α4 wt cells had lower surface CRT compared with the same cells plated on GST. Importantly, α4δ cells exhibited low surface CRT under both adherent and nonadherent conditions, and this low level was comparable to α4 − / − and α4 wt cells under adherent conditions ( Figure 3b).
As α4δ binds CRT in a constitutive manner, 23 we postulated that the juxtamembrane GFFKR motif may sequester CRT in the cytosol and reduce CRT translocation to the cell surface. In this manner, we predicted that cells with more α4δ expression will have less surface CRT. Thus, we gated a polyclonal α4δexpressing cell line for high, medium and low α4δ expression, and showed that surface CRT levels were inversely correlated with α4δ levels, both with and without doxorubicin treatment (Figure 3c). To determine if the juxtamembrane CRT-binding motif, GFFKR, is sufficient to inhibit surface CRT, we also assessed cells expressing the Tac carrier epitope fused to KLGFFKR (Tacδ) or the scrambled KLRFGFK (Tacδ scr ) (Figure 3a). 23 Similar to α4δ cells, doxorubicin-treated Tacδ cells had significantly reduced surface CRTwhen compared with control  (Figure 3d). Furthermore, expression of the cytosolic GFP-CRT resulted in increased surface CRT detected in control Tacδ scr cells, but not in Tacδ cells (Figure 3e). Unlike α4δ or α4 wt , Tacδ is a monomer that cannot heterodimerize with integrin-β1; thus, the reduced surface CRTobserved for Tacδ cells is directly attributed to the GFFKR motif. These results demonstrated that doxorubicinmediated surface CRT presentation was reduced for cells engaging integrin substrates. In contrast, cells expressing the juxtamembrane GFFKR motif (as α4δ or Tacδ) exhibited constitutively low levels of surface CRT in a manner that bypassed the requirement for cell adhesion.
Cell adhesion-mediated reduction in surface CRT involves β1-integrins. As β1-integrin heterodimerizes with multiple α-integrins, loss of the β1-subunit yields corresponding decreases in the expression of multiple integrins. We assessed and found decreased expression of α3-, α4and α5-integrins in β1 − / − cells compared with WT ( Figure 4a). We also confirmed that β1 − / − cells failed to adhere to the α4β1specific substrate CS1 (Figure 4b). Then, we assessed the functional contribution of β1-integrin on surface CRT presentation upon substrate adhesion. When plated on GST under non-adherent conditions, both β1 − / − and WT cells responded to doxorubicin treatment with comparable increases in surface CRT (Figure 4c). When plated on CS1 substrate to engage α4β1-integrins, WT cells exhibited a marked decrease, whereas β1 − / − cells exhibited no reduction in surface CRT compared with non-adherent conditions (Figure 4c). Thus, the evidence supports a modulatory role for integrin-mediated cell adhesion in decreasing surface CRT presentation for doxorubicin-treated cells.
Antibody-induced β1-integrin activation inhibits surface CRT presentation. Cell adhesion is facilitated by activated integrins that take on the high-affinity ligand binding state and our results suggest that this state correlates with reduced surface CRT. Therefore, we tested if antibodies known to promote β1-integrin activation, such as 9EG7, 31 could also downregulate surface CRT in the absence of cell adhesion.
We found that 9EG7 binds to suspension Jurkat cells in a concentration-dependent manner (Supplementary S4). Then, we assayed surface CRT for cells treated with 9EG7 and/or doxorubicin to induce surface CRT. Remarkably, increasing concentrations of 9EG7 reduced surface CRT on WT cells to the low levels observed on CRT − / − cells (Figure 5a). Importantly, incubation of 9EG7 with β1 − / − cells had no inhibitory consequence on doxorubicin-induced surface CRT, confirming the specificity of 9EG7 for β1-integrins. WT cells treated with the β1-specific but non-activating TS2/16 antibody also showed no inhibitory consequence on surface CRT. Notably, the 9EG7-mediated decrease of surface CRT in WT cells was observed for both doxorubicin-treated and nontreated conditions (Figure 5a), suggesting the generality of an integrin activation-mediated effect that did not require anthracycline induction of ER stress.
To gain insight on how integrin function can prevent surface presentation of CRT upon treatment with ICD inducers, we conducted integrin immunoprecipitation for doxorubicin-treated or untreated cells, and with and without 9EG7, as before. Doxorubicin effectively promoted an increased level of CRT that immunoprecipitated with α4-integrin, and this association was increased further for 9EG7-treated cells (Figure 5b). Furthermore, CRT − / − cells expressing either the ER-targeted ssGFP-CRT or cytosol-targeted GFP-CRT treated with 9EG7, but not doxorubicin, resulted in similarly and significantly reduced levels of surface CRT and GFP detected (Figures 5c and d). This confirmed that 9EG7mediated integrin activation can inhibit surface translocation of CRT from a predominantly cytosolic locale.

Cell adhesion and 9EG7 reduces surface CRT presentation in various T-ALL.
To determine if the suppression of surface CRT by stimulated integrin function occurred in other T-lymphoblasts, we repeated key assays using the human T-ALL cell lines THP-6, SUP-T1 and DND-41, as well as a primary human T-ALL, BD-67. As before, cells were treated with doxorubicin to stimulate surface CRT, plated on fibronectin-coated dishes to assess adhesion effects and/or treated in suspension with 9EG7 antibodies to assess integrin activation effects. In agreement with the Jurkat cell observations, doxorubicin-induced surface CRT was significantly reduced for cells adherent on fibronectin and for cells treated with 9EG7 when compared with their respective controls (Figures 6a-d).
CRT released from the ER by drug treatment is bound by integrins in the cytosolic compartment. As 9EG7 cotreatment inhibited surface CRT and increased CRT-integrin interaction ( Figure 5), we postulated that drug treatment promoted release of ER-resident CRT into the cytosol where it is able to interact with the integrin cytosolic domain. Doxorubicin is fluorescent and incompatible for multicolor immunofluorescence imaging, thus we used oxaliplatin, a non-fluorescent and well-characterized ICD-inducing agent. 32 First, we confirmed that surface CRT is induced in oxaliplatin-treated Jurkat cells, and that 9EG7 similarly inhibited surface CRT exposure (Figure 7a).
To enable differential immunostaining of cytosolic-and ERresident CRT, we used partial permeabilization techniques using digitonin, versus full permeabilization using TX-100. 10 WT cells show robust staining for PDI and CRT when fully permeabilized with Triton X-100, but not when treated with a digitonin concentration that enables permeabilization of the plasma membrane but not the ER (Supplementary S5). We observed a low but significant level of digitonin-permeabilized CRT in oxaliplatin-treated WT cells, suggesting staining of CRT other than the ER pool. To quantify the results, we repeated the assay by flow cytometry, as this allows assessment of total fluorescence that is not limited by optical sectioning (Figure 7b). Cells were treated with oxaliplatin, 9EG7 or both. Regardless of treatment conditions, TX-100-permeabilized WT cells exhibited comparable staining for CRT, indicating no significant changes of total CRT by the various treatments. In contrast, CRT staining in digitonin-permeabilized and oxaliplatintreated WT cells was significantly elevated over oxaliplatinuntreated cells, regardless of 9EG7 co-treatment (Figure 7b). In a similar manner, cells expressing Tacδ exhibited no increase in surface CRT upon oxaliplatin treatment, even though cytosolic CRT was significantly increased (Figures 7c  and d).
Taken together, our results show that cells treated with the integrin activator 9EG7, or expressing the minimal α-integrin GFFKR, have markedly reduced surface CRT when challenged with an ICD inducer, even though CRT was elevated in the extra-ER, cytosolic compartment. This cytosolic pool of CRT is also observed as an increased interaction with α4-integrin. As α-integrins interact with CRT via the cytosolic GFFKR motif, our results support a model where activated integrins bind cytosolic CRT and prevent its translocation to the cell surface.
9EG7-mediated reduction of surface CRT decreases phagocytic engulfment. To determine if the integrinmediated inhibition of surface CRT mediated by ICD inducers have a measurable outcome on target cell engulfment by professional phagocytes, we performed a phagocytosis assay using macrophages. Jurkat target cells were treated with oxaliplatin, 9EG7 and combinations thereof. To counter Jurkat cells' high levels of cluster of differentiation 47 (CD47) expression, 33 a known inhibitory receptor for macrophagemediated phagocytosis, 6,34 cells were also coincubated with the CD47-neutralizing antibody, B6H12. The treated Jurkat cells were coincubated with macrophages, and the labeled cell mixture was analyzed by flow cytometry to delineate and quantitate single-positive (Jurkat or macrophage) and double-positive (Jurkat/s engulfed by macrophage) In agreement with similar assays conducted by others, 35,36 we found that CD47 neutralization with B6H12 antibody was necessary to reveal a three-to fourfold higher rate of phagocytosis ( Figure 8). Under these conditions, macrophagemediated phagocytosis was significantly increased for oxaliplatin-treated compared with untreated Jurkat cells. Importantly, co-treatment with 9EG7 antibody resulted in reduced phagocytosis of cells treated with or without oxaliplatin. Taken together, our results show that the integrin-activating 9EG7 antibody can suppress surface presentation of CRT in T-lymphoblasts to levels that reduced its ability to be engulfed by macrophages.

Discussion
Our study is the first to describe a regulatory role for α-integrin function in cell surface CRT presentation. By generating and reconstituting CRT − / − T-lymphoblasts with GFP-tagged CRT that was either expressed in the cytosol or enriched within the ER lumen, afforded the opportunity to probe the subcellular source for surface CRT resulting from an ICD inducer. The increased surface CRT presentation following doxorubicin treatment of cells expressing the ER-targeted ssGFP-CRT is consistent with earlier reports, suggesting that the path to the cell surface requires induction of ER stress and CRT release from its highly enriched localization within the ER lumen. 8,37,38 This increase was detectable with either CRT or GFP antibodies, indicating that the translocation involved the intact fusion protein. When expressed within the cytosol without transiting the ER, GFP-CRT presented at high levels on the cell surface in a manner that could not be stimulated further with doxorubicin, thus effectively bypassing the requirement for ER stress. Several hypotheses have been put forth on the pathway for CRT transit from the ER to the cell surface. [39][40][41][42] Our findings support a mechanism that involves CRT release from the stressed ER to the cytosol via the ubiquitination pathway, [43][44][45][46][47] following which CRT translocates to the surface.
At the plasma membrane, α-integrins appear to regulate surface CRT presentation. The C-terminal cytosolic tail encoded by all α-integrins has a conserved GFFKR peptide motif required for heterodimer stabilization with β-integrin. 48,49 Structural studies has revealed that the αand β-integrin tails become physically separated upon integrin activation, 50 facilitating binding of proteins to the tails. We, and others, have reported that CRT interacts with integrins via this motif in an adhesion-dependent manner. 21,23,24 We adopted several strategies to test the assumption that increased integrin activation leads to increased intracellular CRT-integrin interaction, thus reducing 'free' cytosolic CRTable to translocate to the extracellular surface. As a physiologically relevant stimulus, we show that T-lymphoblast adhesion to an integrin substrate significantly reduces surface CRT in cells treated with ICD inducers. Similarly, suspension cells treated with the β1-integrin-activating antibody, 9EG7, elicited a greater suppressive effect on surface CRT when compared with adhesion, in most cases reducing surface CRT on WT cells to the levels observed for CRT − / − cells. Combined with the observations that 9EG7 treatment, or expression of the juxtamembrane-anchored cytosolic GFFKR motif, suppressed the appearance of surface CRT for CRT − / − cells expressing the cytosol-targeted GFP-CRT, we suggest that a cytosolic pool of CRT exists in T-lymphoblasts and that activated integrins interact primarily with cytosolic CRT.
Previously, we showed that expression of a transmembrane-anchored α-integrin tail consisting of only the GFFKR motif (as α4δ or Tacδ) resulted in increased interaction with CRT in an adhesion-independent manner. 23 In this study, we show that α4δ and Tacδ cells exhibit constitutively lower levels of surface CRT, again in an adhesion-independent manner. Furthermore, we correlated higher α4δ expression with lower surface CRT, suggesting that the truncation may have enabled binding of CRT to the now more accessible juxtamembrane GFFKR motif. A greater level of CRT associated with α4-integrins when cells were treated with an ICD inducer, likely resulting from a marked increase in cytosolic CRT released from the ER, now able to interact with α-integrin tails. We suggest the increased CRT-integrin interaction effectively sequestered CRT within the cytosol and prevented its translocation to the extracellular surface. Beyond the GFFKR-conserved motif, α-integrin tails share little sequence homology; 20 thus, an interesting question for future studies would be to compare surface CRT levels for cells specifically engaging various α-integrins.
As a DAMP expressed on the surface of cells undergoing ICD, surface CRT is crucial for the clearance of tumor cells by the host's immune system mediated by professional phagocytes. 3,6,7 Therefore, cell physiological events contributing to reduced surface CRT expression may lead to reduced antitumor responses resulting from ICD-based chemotherapy. The tumor microenvironment such as the bone marrow stroma may provide these protective stimuli in the form of adhesion substrates and stimulatory chemokines that promote integrin activation. 18 Thus, effective chemotherapy using ICD inducers may benefit from coadministration with integrin function blocking therapeutic antibodies.
CRT − / − cells was generated by the CRISPR-Cas9 methodology 25 using 5′-CGAGCCTGCCGTCTACTTCA-3′ CALR guide DNA and plasmid pX330. Following nucleoporation and sorting, CRT − / − clones were identified by fluorescence immunostaining for CRT and confirmed by sequencing the targeted genomic loci. Clone hCRT1-3 was used to generate the data shown, with the major phenotypes reproduced in three independent clones.
Primary T-ALL (BD-67) was obtained from the BC Children's Hospital Biobank (Vancouver, BC, Canada) with ethics approval from the BC Women's and Children's Hospital institutional review board (H12-03216). Relapsed bone marrow aspirate was injected via tail vein into NOD-scid/IL-2Rγ-null (NSG) mice (Jackson Laboratory, Bar Harbor, ME, USA). Mice were monitored for human leukemia engraftment by flow cytometric analysis of peripheral blood. Those with high leukemia burden were killed and their spleens (~80% CD45 + lymphoblasts) immediately sourced for primary T-ALL cells.
Flow cytometry. FACSCanto, LSRFortessa and Accuri C6 was used for analytical work and FACSAria (BD, Mississauga, ON, Canada) for cell sorting. Postacquisition analysis was carried out using FlowJo (Tree Star, Ashland, OR, USA).
Partial permeabilization and immunostaining for CRT. Following treatments, cells were fixed in 3.7% formaldehyde/PBS for 15 min and divided into two groups treated with 0.1% Triton X-100/PBS or 25 μg/ml digitonin/PBS (Acros, Waltham, MA, USA) for 5 min. Following washes, cells were immunostained for CRT and analyzed by flow cytometry. We optimized digitonin concentrations based on the ability to maintain ER integrity that precluded staining for PDI, an ER marker. 10 Fluorescence microscopy. Cells plated onto fibronectin-coated coverslips was fixed in 3.7% formaldehyde/PBS, permeabilized in 0.1% Triton X-100/PBS and immunostained accordingly. Images were acquired on an Olympus IX81 (Olympus Canada, Richmond Hill, ON, Canada) microscope equipped with a 60 × NA1.35 oilimmersion objective. Postacquisition processing was conducted using ImageJ (https://imagej.nih.gov/ij/). Some cells were digitonin permeabilized as described.