The carboxyl-terminal sequence of bim enables bax activation and killing of unprimed cells

The Bcl-2 family BH3 protein Bim promotes apoptosis at mitochondria by activating the pore-forming proteins Bax and Bak and by inhibiting the anti-apoptotic proteins Bcl-XL, Bcl-2 and Mcl-1. Bim binds to these proteins via its BH3 domain and to the mitochondrial membrane by a carboxyl-terminal sequence (CTS). In cells killed by Bim, the expression of a Bim mutant in which the CTS was deleted (BimL-dCTS) triggered apoptosis that correlated with inhibition of anti-apoptotic proteins being sufficient to permeabilize mitochondria isolated from the same cells. Detailed analysis of the molecular mechanism demonstrated that BimL-dCTS inhibited Bcl-XL but did not activate Bax. Examination of additional point mutants unexpectedly revealed that the CTS of Bim directly interacts with Bax, is required for physiological concentrations of Bim to activate Bax and that different residues in the CTS enable Bax activation and binding to membranes.


Introduction 48 49
Apoptosis is a highly conserved form of programmed cell death that can be 50 triggered by extrinsic or intrinsic signals. It plays a fundamental role in maintaining 51 homeostasis by eliminating old, excessive or dysfunctional cells in multi-cellular 52 organisms (Kerr, Wyllie, and Currie 1972). Defective regulation of apoptosis has been 53 found in many diseases (Favaloro et al. 2012) and is considered one of the hallmarks of 54 cancer (Hanahan and Weinberg 2011). 55 Bcl-2 family proteins play a decisive role in apoptosis initiated by intrinsic signaling 56 by regulating the integrity of the mitochondrial outer membrane (MOM). Commitment to 57 apoptosis is generally regarded as due to MOM permeabilization (MOMP) releasing 58 cytochrome c and pro-apoptotic factors from the intermembrane space into the 59 cytoplasm. These factors activate the executioner caspases that mediate cell death 60 (Chipuk, Bouchier-Hayes, and Green 2006). Direct interactions between Bcl-2 family 61 proteins govern both initiation and inhibition of MOMP (Kale,Osterlund,and Andrews 62 function as a regulator of anti-apoptotic proteins, as it binds and thereby inhibits by 98 mutual sequestration all known anti-apoptotic proteins (Chen et al. 2005;Shamas-Din, 99 Kale, et al. 2013). Until recently, It was unknown why Bim binds to Bcl-XL with sufficient 100 affinity to resist displacement by small molecule BH3-mimetics, while other BH3 101 proteins, such as Bad, are easily displaced (Aranovich et al. 2012). In addition to 102 interactions via the BH3-domain, residues within the Bim CTS bind to Bcl-XL, and 103 thereby increase the affinity of the interaction by "double-bolt locking" providing an 104 explanation for the observations with BH3 mimetic drugs (Liu et al. 2019). Here we 105 investigated whether the CTS of Bim also contributes to the functional and physical 106 interactions between Bim and Bax. 107 We demonstrate that both primary cells and cell lines have a range of apoptotic 108 responses to the expression of a truncated BimL protein lacking the CTS (BimL-dCTS), 109 while expression of full-length BimL was sufficient to kill all of these cells. To determine 110 the molecular mechanism that underlies this difference, the two pro-apoptotic functions 111 of Bim; activation of Bax and inhibition of Bcl-XL, were quantified using purified full-112 length BimL protein and cell free assays. Replacing the CTS of Bim with an alternative 113 tail-anchor that binds the protein to mitochondrial membranes did not fully restore Bax 114 activation function, demonstrating that sequences within the Bim CTS rather than 115 membrane binding contribute to Bax activation. Site-directed mutagenesis of the Bim 116 CTS also revealed residues important for binding to membranes that were not required 117 for Bax activation (e.g. I125). Furthermore, specific residues within the CTS were 118 identified that are required for BimL to efficiently activate Bax, but that are not required 119 for BimL to bind to and inhibit Bcl-XL. Evidence in cell free assays demonstrated that 120 BimL CTS residues L129 and I132 physically interact with Bax and are required to 121 activate it. These mutants were used to show that BimL residues L129 and I132 are also 122 required for BimL to efficiently kill cells resistant to BimL-dCTS, demonstrating that it is 123 necessary to activate Bax to kill these unprimed cells. Together, our data demonstrates 124 that the unusual sequence of the CTS of Bim separately controls both membrane 125 binding and Bax activation. 126 127

Results 128
The CTS of Bim variably contributes to the pro-apoptotic activity of Bim in different cell 129

lines. 130
Removing the CTS from Bim abrogates pro-apoptotic activity in HEK293 cells 131 (Weber et al. 2007). While this observation has generally been ascribed to loss of 132 binding of Bim to MOM our observation that the CTS is also involved in binding BimEL to 133 Bcl-XL (Liu et al. 2019) suggested that there may be other explanations for the loss of 134 pro-apoptotic activity for Bim when the CTS is removed. To determine the contribution of 135 the Bim CTS to pro-apoptotic activity, a BimL mutant was generated in which the 136 previously characterized membrane binding domain (carboxyl-terminal residues P121-137 H140) were deleted (BimL-dCTS) (Wilfling et al. 2012), (Liu et al. 2019). This mutant 138 was expressed in cells and the effectiveness of induction of cell death was compared to 139 expression of full-length BimL by confocal microscopy. To detect expression of the 140 constructs in live cells, they included an N-terminally fused Venus fluorescent protein 141 (indicated by a superscripted v in the name). Thus a construct in which Venus was fused 142 to BimL is referred to here as V BimL while the mutant lacking the CTS is V BimL-dCTS. 143 As an inactive control we used V BimL-4E a mutant in which four conserved hydrophobic 144 residues in the BH3 domain of BimL were replaced with glutamate, thereby preventing 145 binding to all other Bcl-2 family proteins (Chen et al. 2005), (Liu et al. 2019). 146 To assay pro-apoptotic activity, the constructs were expressed in primary cells and 147 cell lines and both expression and cell death were measured using confocal microscopy. 148 Apoptosis was assessed by detecting externalization of phosphatidylserine by Annexin V 149 staining in cells expressing detectable levels of V BimL or the V BimL mutants as 150 measured by Venus fluorescence. As expected, expression of V BimL induced apoptosis 151 in all cell types tested, while the negative control V BimL-4E did not ( Figure 1A). As 152 reported previously for Bim-dCTS, the fluorescent version ( V Bim-dCTS) failed to induce 153 cell death in HEK293 cells (Weber et al. 2007). In contrast, expression of V BimL-dCTS 154 induced apoptosis to levels similar to V BimL in HCT116 and MEF cells but had reduced 155 potency in BMK and CAMA-1 cells. Thus the CTS of Bim contributed variably to the pro-156 apoptotic activity of Bim in different cell lines despite having equal expression across all 157 cell types (Figure 1 -figure supplement 1). 158 To determine if this difference in response to V BimL-dCTS expression is a function 159 of the extent to which the apoptotic machinery is loaded in mitochondrial outer-160 membranes, mitochondria were purified from cells resistant (HEK293) and sensitive 161 (MEF) to V BimL-dCTS expression and assayed by BH3-profiling (Potter and Letai 2016) 162 to measure loading of anti-apoptotic proteins with BH3 proteins or active Bax/Bak. Unlike 163 BH3-profiling experiments conducted with BH3-peptides, in these experiments purified 164 full-length proteins were used. Thus, purified BimL, BimL-dCTS, Bad and Noxa proteins 165 were incubated with mitochondria from each of the cell lines and mitochondrial outer 166 membrane permeabilization (MOMP) was measured by separating supernatant and 167 pellet fractions for each reaction, and immunoblotting for cytochrome c released from the 168 intermembrane space as previously described (Pogmore et al. 2016). Immunoblots were 169 quantified and MOMP assessed as % cytochrome c released ( Figure 1B). As expected 170 from the data in Figure 1A, addition of recombinant BimL was sufficient to induce 171 cytochrome c release from mitochondria from both HEK293 and MEF cells. However, 172 addition of BimL-dCTS induced cytochrome c release only in the MEF mitochondria 173 confirming that resistance to BimL-dCTS in HEK293 cells is manifest at mitochondria. 174 One potential explanation for this difference is that the mitochondria in the cell 175 lines have different dependencies on multi-domain anti-apoptotic proteins for survival, a 176 phenomenon known as priming. If BimL-dCTS has lost one of the functions of Bim such 177 as activating Bax or Bak or inhibiting one of the multi-domain anti-apoptotic proteins Bcl-178 2, Bcl-XL and Mcl-1 it would be expected to have different activities on mitochondria with 179 different priming. Therefore, to better understand why BimL-dCTS can only permeabilize 180 MEF mitochondria and not mitochondria from HEK293 cells, we compared the sensitivity 181 of mitochondria from the two cell types to addition of BH3-proteins Bad and Noxa that 182 inhibit Bcl-2 and Bcl-XL or Mcl-1, respectively, but that do not activate Bax or Bak (Kale,183 Osterlund, and Andrews 2017). Incubation of full length Bad and/or Noxa with 184 mitochondria from HEK293 cells failed to induce cytochrome c release, while the 185 addition of Noxa was sufficient to permeabilize MEF mitochondria ( Figure 1B). This data 186 suggests that HEK293 cells do not depend on expression of Bcl-2, Bcl-XL or Mcl-1 187 sequestering active Bax, Bak or their BH3-activators while mitochondria from MEFs 188 depend on expression of Mcl-1 to prevent apoptosis (Lessene et al. 2013 Full-length BimL is required to kill cultures of primary cortical neurons 219 Our data with cell lines and their respective purified mitochondria suggests that 220 BimL-dCTS does not kill cells that do not depend on anti-apoptotic proteins for survival. 221 To test this in a more biologically relevant system, we cultured primary murine cortical 222 neurons and assayed their response to expression of the BimL mutants. For regulated 223 expression in primary cortical neurons the coding regions for V BimL, V BimL-4E, and 224 V BimL-dCTS were cloned into a tetracycline-responsive lentiviral vector, and introduced 225 into primary cortical neuron cultures through lentiviral infection. After culture for 8 days in 226 vitro, BimL expression was induced in the neurons by the addition of doxycycline. 227 Neuronal cell death was assayed using confocal microscopy after staining neurons with 228 propidium iodide (PI), a dye that exclusively stains the nuclei of dead cells. 229 Quantification of Venus-expressing neuronal cell bodies revealed that as expected 230 V BimL expression killed cultured primary neurons while V BimL-4E did not (Figure 2A-B). 231 However, the expression of V BimL-dCTS was largely ineffective to induce cell death in 232 cultured primary cortical neurons ( Figure 2B). Our data is consistent with previous 233 reports suggesting that primary murine cultures of hippocampal neurons become 234 resistant to induction of apoptosis by external stimuli over time in culture. This resistance 235 has been reported to be due to a difference in Bcl-2 family protein expression that 236 results in decreased mitochondrial 'priming', explaining why our cultures of primary 237 cortical neurons are resistant to V BimL-dCTS (Sarosiek et al. 2016). 238 To determine if resistance to induction of cell death by BimL-dCTS is due to 239 differential sensitivity of neuronal mitochondria to induction of MOMP by BimL and BimL-240 dCTS, mitochondria were isolated from embryonic day 15 (E15) mouse brains, the same 241 age used to culture primary cortical neurons. Brain mitochondria were used instead of 242 isolating mitochondria from neuronal cultures due to the low yield from primary cultured 243 neurons. Untreated mitochondria from day E15 brain released only low levels of 244 cytochrome c. As expected, addition of 0.1nM recombinant BimL was sufficient to elicit 245 MOMP as measured by cytochrome c release and detection in the supernatant. In 246 contrast, 100 times more BimL-dCTS (10nM) failed to induce MOMP ( Figure 2C). 247 Taken together our data suggest that BimL-dCTS kills cells in which the 248 mitochondria are sensitive to inhibition of anti-apoptotic proteins by sensitizers such as 249 Bad and Noxa. Thus, BimL-dCTS did not permeabilize mitochondria extracted from 250 HEK293 cells or E15 whole murine brains, and as a result, BimL-dCTS expression did 251 not kill HEK293 cells or primary cultures of cortical neurons. This finding suggests that 252 inhibition of anti-apoptotic proteins is not sufficient to kill these cells. Therefore, BimL-253 dCTS differs mechanistically from BimL as the latter kills both cell types resistant and 254 sensitive to BimL-dCTS. Compared to BimL, BimL-dCTS is missing the membrane 255 binding domain and therefore is not expected to localize at mitochondria (Liu et al. 256 2019), however, the relationship between Bim binding to membranes and Bim mediated 257 Bax activation has not been extensively studied. To determine how the molecular 258 mechanism of BimL-dCTS differs from BimL the activities of the proteins were analyzed 259 using cell free assays. 260 The Bim CTS mediates BimL binding to both Bax and membranes 277 To investigate the pro-apoptotic mechanism of BimL and BimL-dCTS without 278 interference from other cellular components, both were purified as full-length 279 recombinant proteins and assayed using liposomes and/or isolated mitochondria. To 280 measure direct-activation of Bax by Bim, either BimL or BimL-dCTS was incubated with 281 recombinant Bax and liposomes encapsulating the dye and quencher pair: ANTS (8-282 Aminonaphthalene-1,3,6-Trisulfonic Acid, Disodium Salt) and DPX (p-Xylene-Bis-283 Pyridinium Bromide). In this well-established assay , increasing 284 amounts of BimL activated Bax resulting in membrane permeabilization measured as an 285 increase in fluorescence due to the release and separation of encapsulated dye and 286 quencher ( Figure 3A). This result is consistent with previous observations that picomolar 287 concentrations of BimL induce Bax-mediated membrane permeabilization (Sarosiek et 288 al. 2013). In contrast, three orders of magnitude higher concentrations of BimL-dCTS 289 were required to induce Bax-mediated liposome permeabilization ( Figure 3A very small amounts of Bim were sufficient to trigger membrane permeabilization 300 because once activated, Bax recruits and activates additional Bax molecules (Tan et al. 301 2006). 302 To assess the impact of the Bim CTS on the interaction between Bim and Bax, 303 binding was measured using Förster resonance energy transfer (FRET). For these 304 experiments recombinant BimL proteins were labelled with the donor fluorophore 305 Alexa568, while Bax was labelled with the acceptor fluorophore Alexa647. 306 Unexpectedly, and unlike the BH3-only protein tBid (Lovell et al. 2008), BimL bound to 307 Bax even in the absence of membranes ( Figure  BimL-dCTS (4 nM) was incubated with the indicated amounts of Alexa647-labeled Bax 327 and FRET was measured from the decrease in Alexa568 fluorescence. 328 (D) Bim binding to Bax measured by FRET in samples containing mitochondrial-like 329 liposomes. FRET was measured as in (C) with 4nM Alexa568-labeled BimL or BimL-330 dCTS and the indicated amounts of Alexa647-labeled Bax. 331

__________________________________________________________________ 332
To confirm in our system that the labeled BimL proteins bind to membranes via the 333 CTS sequence, binding of Alexa568-labeled recombinant BimL and BimL-dCTS to DiD 334 labeled liposomes was measured by FRET ( Figure 4A). In these experiments DiD 335 serves as an acceptor for energy transfer from Alexa568 labeled BimL. The same 336 approach was used to quantify BimL binding to mitochondrial outer membranes with 337 mitochondria isolated from BAK -/mouse liver ( Figure 4B), which lack Bax and Bak 338 (Shamas-Din, Bindner, et al. 2013). In both cases, BimL spontaneously bound to 339 membranes with picomolar affinity, while stable binding of BimL-dCTS to liposomes and 340 mitochondria was not-detectable ( Figure 4A-B). Furthermore, BimL-dCTS again had no 341 relevant binding to Bax even in the presence of purified mitochondria ( Figure 4C). 342 Taken together, our data strongly suggest that the CTS of Bim is required for both 343 BimL to bind to membranes in vitro and for binding Bax with or without membranes. 344 Alternatively purified BimL-dCTS may be completely non-functional. To demonstrate that 345 Alexa568-labeled BimL (n=5) or BimL-dCTS (n=3) to the indicated amounts of DiD 362 labeled mouse liver mitochondria was assessed by measuring FRET. 363 (C) Deletion of the CTS prevented Bim binding to Bax at mitochondria. Bim binding to 364 Bax was measured by FRET in samples containing mouse liver mitochondria, 4nM 365 Alexa568-labeled BimL (grey) or BimL-dCTS(black) and the indicated amounts of 366 Alexa647-labeled Bax (n=3). 367

____________________________________________________________________ 368
The CTS is not required for BimL to inhibit Bcl-XL 369 In addition to direct Bax activation, Bim promotes apoptosis by binding to Bcl-XL 370 and displacing either activator BH3-proteins (Mode 1) or activated Bax or Bak (Mode 2) 371 (Llambi et al. 2011). In the ANTS/DPX liposome dye release assay, BimL-dCTS was 372 functionally comparable to the well-established Bcl-XL inhibitory BH3-protein Bad in 373 tBidmt1 ( To identify which residues in the Bim CTS mediate binding to membranes and/or 413 Bax we generated a series of point mutations. Sequence analysis using HeliQuest 414 software (Gautier et al. 2008) predicts that the Bim CTS forms an amphipathic α-helix 415 ( Figure 6A). Two arginine residues (R130&134) are predicted to be on the same 416 hydrophilic side of the helix, whereas hydrophobic residues (e.g. I125, L129, I132) face 417 the other side ( Figure 6A). To determine the functional importance of these residues, 418 Bim CTS mutants were created including: BimL-CTS2A in which R130 and R134 were 419   BimL-CTS2A had less effect on Bim binding to membranes ( Figure 6B). Despite the 429 dramatic changes in affinity for membranes among Bim CTS mutants, the mutations did 430 not abolish binding to Bax both in the presence and absence of membranes ( Figure 6B). 431 Indeed most of the mutants had Kd values for binding to Bax of less than 100 nM and to 432 our surprise many of them bound to Bax better in solution than on membranes. This data 433 further confirms that binding to membranes and Bax are independent functions of the 434 Bim CTS. In the case of BimL-I125E, a mutant that activates Bax to permeabilize 435 liposomes, the initial interaction with Bax must occur in solution as neither protein 436 spontaneously binds to membranes ( Figure 6B). 437 Unexpectedly, there was not a good correlation between BimL binding to 438 membranes and Bax activation. For example, while BimL bound to membranes with a 439 Kd of 31 pM, BimL-CTS2A and BimL-I125E bound to membranes very poorly (Kds of 440 ~600 and >1000 pM, respectively) yet both mutants triggered Bax mediated membrane 441 permeabilization, demonstrating that specific residues in the CTS rather than binding to 442 membranes enabled BimL to mediate Bax activation. Moreover, BimL binding to Bax 443 was also not sufficient to activate Bax efficiently. BimL-L129E and BimL-I132E are two 444 Bim mutants that do not bind membranes, retain reasonable affinities for Bax in the 445 presence of membranes (Kds ~100-200nM), but were unable to activate Bax ( Figure  446 6B). These results indicate that these two residues play a key function in Bax activation. 447 As expected, the negative control BimL-4E mutant does not bind to nor activate Bax 448 even though its CTS is intact and the protein binds membranes ( Figure 6B). This result 449 confirms the essential role of the BH3 domain and suggests that the Bim CTS provides a 450 secondary role in Bax binding rather than providing an independent high affinity binding   Our binding and mutagenesis data suggest that the Bim CTS binds to and 542 activates Bax in solution and on membranes. To detect this binding interaction we used 543 a photocrosslinking approach, in which a BimL protein was synthesized with a 544 photoreactive probe attached to a single lysine residue positioned in the CTS using an in 545 vitro translation system containing 5-azido-2-nitrobenzoyl-labled Lys-tRNA Lys that 546 incorporates the lysine analog (ANB-Lys) into the polypeptide when a lysine codon in 547 the BimL mRNA is encountered by the ribosome. The BimL synthesized in vitro was also 548 labeled by 35 S via methionine residues enabling detection of BimL monomers and 549 photoadducts by phosphor-imaging. 550 The radioactive, photoreactive BimL protein was incubated with a recombinant 551 His6-tagged Bax protein in the presence of mitochondria isolated from BAK -/mouse liver 552 lacking endogenous Bax and Bak to prevent competition and increase BimL-Bax protein 553 interactions. Mitochondrial proteins were then separated from the soluble ones by 554 centrifugation. Both soluble and mitochondrial fractions were photolyzed to activate the 555 ANB probe generating a nitrene that can react with atoms in close proximity (< 12 Å from 556 the C of the lysine residue). Thus, for photoadducts to form, the atoms of the bound 557 Bax molecule are likely to be located in or near the binding site for the Bim CTS. The 558 resulting photoadduct between the BimL and the His6-tagged Bax was enriched by 559 Ni2+-chelating agarose resin and separated from the unreacted BimL and Bax 560 monomers using SDS-PAGE. The 35 S-labeled BimL in the photoadduct with His6-tagged 561 Bax and BimL monomer bound to the Ni2+-beads specifically via the His6-tagged Bax or 562 nonspecifically were detected by phosphor-imaging. A BimL-Bax specific photoadduct 563 was detected when the ANB probe was located at four different positions in the Bim CTS 564 on both hydrophobic and hydrophilic surfaces of the potential -helix ( Figure 8A). These 565 photoadducts have the expected molecular weight for the Bim-Bax dimer, and were not 566 detected or greatly reduced when the ANB probe, the light, or the His6-tagged Bax 567 protein was omitted ( Figure 8A). Consistent with the FRET-detected BimL-Bax 568 interaction in both solution and membranes, the BimL-Bax photocrosslink occurred in 569 both soluble and mitochondrial fractions. Less photocrosslinking occurred in the 570 mitochondrial fraction likely due to the fact that in membranes homo-oligomerization of 571 activated Bax competes with hetero-dimerization between BimL and Bax. 572 As expected, BimL-Bax photocrosslinking was detected in both soluble and 573 mitochondrial fractions when the ANB probe was positioned in the Bim BH3 domain as a 574 positive control ( Figure 8B). Crosslinking with the Bim BH3 domain is consistent with the 575 canonical BH3 interaction well supported by experimental evidence including co-crystal 576 structures and NMR models ( (Walensky et al. 2008;Robin et al. 2015). Furthermore, 577 loss of photocrosslinking for BimL mutants with the BH3 4E mutation that abolished 578 binding to Bax demonstrates that direct binding between the proteins is required for 579 crosslinking to be detectable ( Figure 8C). Therefore, the crosslinking data suggests that 580 similar to the BH3 domain, the Bim CTS binds to Bax. To further demonstrate that the 581 CTS of Bim binds to Bax independent of both membrane binding and Bax activation the 582 experiment was repeated with BimL-L129E, a mutant that binds Bax without activating it 583 and that does not bind membranes ( Figure 6B. As shown in Figure 8C, the L129E 584 mutation in the CTS did not inhibit photocrosslinking of BimL to Bax in either the soluble 585 or mitochondrial fractions. Furthermore, this mutant also resulted in photocrosslinking to 586 Bcl-XL, consistent with data demonstrating that the Bim CTS also binds to this anti-587 apoptotic protein ( Figure 8C  labeled methionine residues, were synthesized in vitro, and incubated with His6-tagged 599 Bax protein (6H-Bax) in the presence of mitochondria lacking endogenous Bax since 600 Bak. The mitochondria were then separated from the soluble proteins by centrifugation 601 and both fractions were photolyzed. The resulting radioactive BimL/6H-Bax 602 photoadducts were enriched with Ni 2+ -beads, and analyzed by SDS-PAGE and 603 phosphor-imaging. BimL/6H-Bax dimer specific photoadducts were detected in both 604 mitochondrial and soluble fractions and indicated by arrowheads. They were of reduced 605 intensity or not detected in control incubations in which the ANB probe, light (h) or 6H-606 Bax protein was omitted, as indicated. The radioactive BimL monomers are indicated by 607 Together, our data suggests that specific residues within the Bim CTS are involved in 627 different aspects of BimL functioning to activate Bax. Residue I125 is required for Bim to 628 bind to mitochondria but is of lesser importance in activating Bax. In contrast, residues 629 L129 and I132 are not required for BimL to bind Bax but are important for it to efficiently 630 activate Bax. Finally BimL-dCTS functions only to bind and inhibit Bcl-XL. The defined 631 mechanism(s) of these mutants makes them useful for probing the differential sensitivity 632 of HEK293 and MEF cells to expression of V BimL-dCTS as seen in Figure 1. Expression 633 of the mutants in HEK293 cells by transient transfection revealed that similar to V BimL-634 dCTS, expression of either V BimL-L129E or V BimL-I132E was not sufficient to kill 635 HEK293 cells, despite expression of either mutant being sufficient to kill the primed MEF 636 cell line (Figure 9). In contrast, HEK293 cells were killed by expression of V BimL-I125E, 637 albeit to a lesser extent than by V BimL (Figure 9). This result is consistent with our 638 findings with purified proteins showing that the EC50 for liposome permeabilization by 639 BimL-I125E was 100 nM compared to ~ 1nM for BimL ( Figure 6B). The activity of V BimL-640 I125E also demonstrates that BimL binding to membranes is not required to kill HEK293 641 cells as BimL-I125E does not bind membranes ( Figure 6B). Together, this data suggests 642 that unlike MEF cells, only mutants of BimL that can efficiently activate Bax kill HEK293 643 cells. The indicated cell lines were transiently transfected with DNA to express V BimL, and the 650 indicated V BimL mutant proteins. Cells expressing Venus fusion proteins were stained 651 with the nuclear dye Draq5 and rhodamine labelled Annexin V, and apoptosis was 652 assessed by confocal microscopy as in Figure 1. The y-axis indicates Annexin V 653 Positivity (%), which was calculated based on the total number of Venus expressing cells 654 that also score positive for Annexin V rhodamine fluorescence. A minimum of 400 cells 655 were imaged for each condition. Individual points (open circles) represent the average 656 for each replicate, while the bar heights, relative to the y-axis, represent the average for 657 all three replicates. A one-way ANOVA was used within each cell line followed by a 658 Tukey's multiple comparisons test to compare the means of each transfection group. *p-659 values less than 0.05, **p-values less than .01, ns, non-significant p-values (>0.05).  Figure 6B. BH3-proteins that do not efficiently activate BAX, such as BimL-683 L129E or BimL-I132E, interact primarily with anti-apoptotic proteins (illustrated here as 684 Bcl-XL since it was possible to measure binding with purified proteins). The binding 685 measurements in Figure 6B allow prediction of the outcome of more subtle differences in 686 interactions for BimL and its mutants. For example, even though BimL-I125E activates 687 Bax the concentration required is around 100nM while the dissociation constant for Bcl-688 XL is less than 3nM ( Figure 6B) such that in cells BimL-I125E would preferentially bind 689 and inhibit Bcl-XL rather than activate Bax ( Figure 10B). While the CTS is necessary for 690 Bim to activate Bax at physiologically relevant concentrations, membrane binding 691 mediated by the CTS is not a prerequisite for interaction with Bax. Rather, binding to 692 membranes increases subsequent Bax activation possibly through facilitating Bax 693 conformational changes on the membrane ( Figure 7B; compare BimL-dCTS-MAO and 694 BimL-dCTS; and Figure 6B compare BimL, BimL-CTS2A and BimL-I125E). Thus it is 695 likely that in cells expressing endogenous Bim, binding to membranes contributes to the 696 efficiency with which the protein kills cells. Nevertheless, there exists the distinction in 697 mechanism between Bim and tBid, as tBid requires membrane binding and a 698 subsequent conformational change in order to bind and efficiently activate Bax (Lovell et 699 al. 2008), while BimL can do so in solution via dual interactions by the Bim BH3 and CTS 700 regions ( Figure 3C and 10). ( Figure 6B). BimL binds membranes and can activate Bax. BimL-I125E (yellow star) has 718 no detectable membrane binding activity but still binds to and activates Bax, albeit with 719 reduced activity compared to BimL ( Figure 6B). At physiologically relevant 720 concentrations BimL-L129E, BimL-I132E, and BimL-dCTS do not activate Bax. 721 However, BimL-L129E and BimL-I132E binding to Bax is not reduced enough to account 722 for the loss in Bax activation and membrane permeabilization suggesting these two 723 residues are involved in activating Bax. 724 (B) In primed cells, one or more pro-apoptotic proteins (activated Bax/Bak and/or a 725 Bax-activating BH3-protein) are sequestered by anti-apoptotic proteins at the MOM. For 726 simplicity only active Bax is shown. Depending on the amount of active pro-apoptotic 727 protein sequestered and the amount of free inactive Bax and or Bak in the cell, BimL 728 may initiate apoptosis primarily by inhibiting anti-apoptotic proteins or by activating Bax 729 and inhibiting anti-apoptotic proteins. The Bim CTS is not required for binding to and 730 inhibiting anti-apoptotic proteins as BimL-L129E, BimL-I132E, and BimL-dCTS bind to 731 anti-apoptotic proteins such as Bcl-XL and release both pro-apoptotic BH3-proteins and 732 Bax ( Figure 5 and Figure 6B), thus enabling killing of primed cells. Overall, our data suggests a model in which the unusual CTS of Bim is not only 774 required for binding to membranes but is directly involved in the activation of Bax. This 775 function is crucial for BimL in killing unprimed cells. The CTS also increases the affinity 776 of Bim for binding to Bcl-XL and Bcl-2 that is sufficient to induce apoptosis in primed 777 cells ( Figure 10). The very much higher affinity of Bim for Bcl-XL and Bcl-2 compared to 778 Bax also ensures that in cells with excess anti-apoptotic proteins Bim is effectively 779 sequestered and neutralized. In separate studies, we demonstrate that the additional experiments are shown as individual points, some points are not visible due to overlap. 816 The mutants analyzed are indicated to the right of the graphs. To permit accurate 817 estimation of the binding constants presented in Figure 6, data was collected to 818 saturation for all mutants (for some curves 1600nM or 3200nM acceptor concentrations 819 were required). For presentation purposes all curves were truncated at 1000 nM. 820 __________________________________________________________________ 821 822

Protein Purification 824
Wild type and single cysteine mutants of Bax, Bcl-XL, and cBid were purified as 825 described previously . cBid mutant 1 (cBidmt1) was purified with the 826 same protocol used for cBid . Bad was purified as described previously 827 (Lovell et al. 2008). 828 His-tagged Noxa was expressed in E. coli strain BL21DE3 (Life Tech, Carlsbad, CA). E. 829 coli cells were lysed by mechanical disruption with a French press. The cell lysate was 830 diluted in lysis buffer (10mM HEPES (7.2), 500nM NaCl, 5mM MgCl2, 0.5% CHAPS, 831 1mM DTT, 5% glycerol, 20mM Imidazole) and Noxa was purified by affinity 832 chromatography on a Nickel-NTA column (Qiagen, Valencia, CA). Noxa was eluted with 833 a buffer containing 10mM HEPES (7.2), 300mM NaCl, 0.3% CHAPS, 20% glycerol, 834 100mM imidazole, dialyzed against 10mM HEPES 7.2, 300mM NaCl, 10% glycerol, 835 flash-frozen and stored at -80 °C. 836 Purification of BimL and single cysteine mutants of BimL was carried out as previously 837 described (Liu et al. 2019). Briefly, cDNA encoding full-length wild-type murine BimL was 838 introduced into pBluescript II KS(+) vector (Stratagene, Santa Clara, CA). Sequences 839 encoding a polyhistidine tag followed by a TEV protease recognition site 840 (MHHHHHHGGSGGTGGSENLYFQGT) were added to create an in frame fusion to the 841 N-terminus of BimL. All of the purified BimL proteins used here retained this tag at the 842 amino-terminus. However, control experiments demonstrated equivalent activity of the 843 proteins before and after cleavage with TEV protease. Mutations as specified in the text 844 were introduced into this sequence using site-directed mutagenesis. 845 Bim was expressed in Arabinose Induced (AI) E. coli strain (Life Tech, Carlsbad, CA). E. 846 coli were lysed by mechanical disruption with a French press. Proteins were purified 847 from the cell lysate by affinity chromatography using a Nickel-NTA column (Qiagen, 848 Valencia, CA). A solution containing 20mM HEPES pH7.2, 10mM NaCl, 0.3% CHAPS, 849 300mM imidazole, 20% Glycerol was used to elute the proteins. The eluate was 850 adjusted to 150 mM NaCl and applied to a High Performance Phenyl Sepharose (HPPS) 851 column. Bim was eluted with a no salt buffer and dialyzed against 10mM HEPES pH7.0, 852 20% glycerol, flash-frozen and stored at -80 °C. 853

Protein labeling 854
Single cysteine mutants of Bax, Bcl-XL, cBid and Bad were labeled with the 855 indicated maleimide-linked fluorescent dyes as described previously (Pogmore et al. 856 2016;Lovell et al. 2008;Kale et al. 2014). Single cysteine mutants of Bim were labeled 857 with the same protocol as cBid with the exception that the labeling buffer also contained 858 4M urea. 859

Bim binding to membranes 860
Liposomes (100 nm diameter) with a lipid composition resembling MOM were 861 prepared as described previously . Mouse liver mitochondria were 862 isolated from Bak -/mice as previously described (Pogmore et al. 2016). Liposomes and 863 Heavy membranes enriched in mitochondria were isolated as described previously 888 (Pogmore et al. 2016;Brahmbhatt et al. 2016). Membrane fractions (1mg/ml) were 889 incubated with 500nM of the specified BH3 proteins (Bim, Bad and/or Noxa). For E15 890 brain mitochondria, 0.5mg/mL of membrane fractions were used and incubated with the 891 indicated amounts of BH3-only proteins for 30 min at 37 o C. Membranes were pelleted by 892 centrifugation at 13000g for 10 min and cytochrome c release was analyzed by 893 immunoblotting using a sheep anti-cytochrome c antibody (Capralogics). Mitochondria 894 from embryonic mouse brains for BH3 profiling experiments were prepared from ~20 895 mouse embryos, E15 in age, following the same protocol used for liver mitochondria 896 (Pogmore et al. 2016). 897

Photocrosslinking of Bim to Bax 915
The photocrosslinking method for studying interactions among the Bcl-2 family proteins 916 has been described in detail (Lin, Johnson, and Zhang 2018)

. Briefly. [ 35 S]Met-labeled 917
BimL proteins with a single ANB-Lys incorporated at specific locations were 918 synthesized using an in vitro translation system. 10 l of the resulting BimL proteins 919 were incubated at 37 o C for 1 h with 1 M of 6H-Bax or 6H-Bcl-XL protein and Bak -/-920 mouse liver mitochondria (0.5 mg/ml total protein) in a 21-l reaction adjusted by buffer