ER Proteostasis Regulators Reduce Amyloidogenic Light Chain Secretion Through an On-Target, ATF6-Independent Mechanism

The plasma cell secretion and toxic aggregation of amyloidogenic immunoglobulin light chains (LCs) causes proteotoxicity in Light Chain Amyloidosis (AL). We recently identified endoplasmic reticulum (ER) proteostasis regulators such as compound 147 that reduce secretion and aggregation of LCs implicated in AL (Plate, Cooley et al., 2016). Compound 147 promotes adaptive ER proteostasis remodeling through a mechanism involving covalent modification of multiple protein disulfide isomerases (PDIs) and subsequent activation of the ATF6 unfolded protein response (UPR) -associated transcriptional signaling pathway (Paxman, Plate et al., 2018). Here, we show that the 147-dependent reduction in amyloidogenic LC secretion from AL patient plasma cells is independent of ATF6 activation, but instead requires on-target PDI modification. Our results reveal pharmacologic targeting of PDIs as a potential strategy to ameliorate AL-associated proteotoxicity and demonstrate that 147 can influence ER proteostasis through multiple on-target mechanisms including ATF6 activation and PDI modification. IMPACT STATEMENT This study demonstrates the broad potential for endoplasmic reticulum proteostasis regulator compounds such as 147 to influence secretory proteostasis of disease-associated proteins through multiple on target mechanisms.


SUMMARY 23
The plasma cell secretion and toxic aggregation of amyloidogenic immunoglobulin light chains (LCs) causes 24 proteotoxicity in Light Chain Amyloidosis (AL). We recently identified endoplasmic reticulum (ER) proteostasis 25 regulators such as compound 147 that reduce secretion and aggregation of LCs implicated in AL (Plate, 26 Cooley et al., 2016). Compound 147 promotes adaptive ER proteostasis remodeling through a mechanism 27 involving covalent modification of multiple protein disulfide isomerases (PDIs) and subsequent activation of the 28 ATF6 unfolded protein response (UPR) -associated transcriptional signaling pathway (Paxman, Plate et al., 29 2018). Here, we show that the 147-dependent reduction in amyloidogenic LC secretion from AL patient plasma 30 cells is independent of ATF6 activation, but instead requires on-target PDI modification. Our results reveal 31 pharmacologic targeting of PDIs as a potential strategy to ameliorate AL-associated proteotoxicity and 32 demonstrate that 147 can influence ER proteostasis through multiple on-target mechanisms including ATF6 33 activation and PDI modification. Thus, new strategies are required to alleviate the LC proteotoxicity on distal tissues and allow 58 chemotherapeutic access to AL patients with severe organ involvement. 59 One strategy to reduce the secretion and toxic aggregation of amyloidogenic proteins is through the 60 adaptive remodeling of the ER proteostasis network comprising ER chaperones (e.g., BiP), folding enzymes 61 (e.g., protein disulfide isomerases, PDIs), and degradation factors ( Enhancing ER proteostasis through mechanisms such as activation of the unfolded protein response 71 (UPR)-associated transcription factor ATF6 can selectively reduce the secretion and toxic aggregation of 72 Here, we show that compound 147 reduces the secretion of the destabilized, amyloidogenic λ6a LC 97 ALLC from AL patient-derived plasma cells. However, this 147-dependent reduction in ALLC secretion is 98 refractory to co-treatments with ATF6 inhibitors, demonstrating that this compound reduces ALLC secretion 99 through an ATF6-independent mechanism. Instead, our results indicate that 147 reduces ALLC plasma cell 00 secretion through an on-target mechanism involving metabolic activation and covalent modification of multiple 01 PDIs. Consistent with this, other PDI inhibitors that covalently modify a similar subset of PDIs also reduce 02 ALLC plasma cell secretion through a mechanism analogous to that observed for 147. These results indicate 03 that pharmacologic targeting of PDIs using ER proteostasis regulators like 147 represents a potential strategy 04 to mitigate the plasma cell secretion of amyloidogenic LCs implicated in AL. Importantly, we further show that 05 147 is compatible with chemotherapeutics used to ablate AL plasma cells (e.g., bortezomib), indicating that 06 these two approaches could be used in combination to reduce both the LC-associated organ toxicity and 07 underlying plasma cell malignancy observed in AL pathogenesis. Ultimately, our results demonstrate that, 08 apart from promoting ATF6-mediated ER proteostasis remodeling, compound 147 can also correct pathologic 09 imbalances in ER proteostasis through the covalent targeting of PDIs. This highlights the broad potential for 10 this compound to influence ER proteostasis for structurally diverse proteins implicated in many types of protein 11 misfolding disease. 12 13 14

RESULTS 15
Compound 147 reduces ALLC secretion from AL patient plasma cells 16 We previously showed that 147 reduces secretion of the amyloidogenic λ6a LC ALLC from AL patient-derived 17 ALMC-2 cells by 50% (Plate et al., 2016). We confirmed this 147-dependent reduction in ALLC secretion by 18 both immunoblotting and ELISA ( Figure 1A and Figure 1 -figure supplement 1A). However, treatment with 19 147 did not significantly influence the viability of ALMC-2 cells, indicating that this reduced secretion cannot be 20 attributed to cell death (Figure 1 -figure supplement 1A). Compound 147 also did not significantly reduce 21  associated with toxic LC aggregation. Using a cycloheximide (CHX) chase assay, we showed that 147 reduces 31 the fraction of ALLC secreted by 30% ( Figure 1D). However, this reduction in ALLC secretion did not 32 correspond to a significant reduction in total ALLC over the 6 h time course of this experiment, indicating that 33 147 does not significantly increase ALLC degradation ( Figure 1E). This result is consistent with previous 34 results, demonstrating that both stress-independent ATF6 activation and treatment with 147 reduces ALLC 35 secretion through a mechanism involving its increased intracellular retention in complexes bound to ER 36

reduces intracellular ALLC in ALMC-2 cells. 39
Although 147 did not significantly increase ALLC degradation in our CHX experiment, we observed 30-50% 40 reductions in intracellular ALLC in 147-treated ALMC-2 cells by ELISA and immunoblotting (Figure 2A  The activation of ATF6 by 147 can be inhibited by co-administration of resveratrol or β-mercaptoethanol 99 (BME), which block different steps of the compound activation mechanism ( Figure 4A) . 00 Thus, we sought to define how treatment with resveratrol or BME influences 147-dependent reduction in ALLC 01 secretion from ALMC-2 cells. Interestingly, co-treatment with BME or resveratrol blocked 147-dependent 02 reductions in ALLC secretion (Figure 4D,E). Neither BME nor resveratrol impacted ALMC-2 cell viability in the 03 absence or presence of 147 (Figure 4 -figure supplement 1C,D). Similar results were observed for ALMC-2 04 cells treated with other active analogs of 147 (Figure 4 -figure supplement 1E). Importantly, co-treatment 05 with resveratrol or BME also blocked the 147-dependent reductions in intracellular ALLC measured by ELISA 06 (Figure 4 -figure supplement 1F). Resveratrol also inhibited the translation attenuation in 147-treated 07 ALMC-2 cells measured by [ 35 S] metabolic labeling (Figure 4 -figure supplement 1G). These results further 08 demonstrate that 147 reduces ALLC secretion from ALMC-2 cells through the same mechanism required for 09 ATF6 activation, involving both compound metabolic activation and covalent protein modification ( Figure 4A). 10 11

Compound 147 alters interactions between ALLC and ER PDIs. 12
The above results indicate that 147-dependent reduction in ALLC secretion from ALMC-2 cells requires 13 covalent modification of proteins. We previously identified proteins modified by 147 using an alkyne-modified 14 analog (147-alkyne; This suggests that 147 could reduce ALLC secretion through an ATF6-independent mechanism involving 30 covalent modification of these PDIs.

Pharmacologic PDI inhibition reduces ALLC secretion from ALMC-2 cells. 51
Interestingly, both RB-11-ca and KSC-34 reduced ALLC secretion from ALMC-2 cells, as measured by ELISA 52 ( Figure 6A). RB-11-ca reduced ALLC secretion by 35%, whereas KSC-34 reduced ALLC secretion by 25%. 53 Importantly, this reduction in secretion did not correspond to reductions in ALMC-2 cell viability ( RB-11-ca reduced ALLC interactions with PDIA1 and increased interactions with PDIA4 ( Figure 6E,F). 61 However, neither compound significantly influenced ALLC interactions with PDIA6. Again, these changes are 62 analogous to those observed with 147 ( Figure 5A,B). However, the reduction in ALLC interactions with PDIA1 63 was greater than that observed for 147, likely reflecting the increased labeling of PDIA1 by RB-11-ca ( Figure  64 5C,D). These results indicate that pharmacologic targeting PDIs with other PDI inhibitors reduces ALLC 65 secretion from ALMC-2 cells through a mechanism analogous to that observed for 147. This supports a model 66 whereby 147 reduces ALLC secretion through an on-target ATF6-independent mechanism involving covalent 67 modification of multiple PDIs. 68 69  (Figure 7A,B). While 147 treatment alone showed a modest ~15% 81 reduction in ALMC-2 viability (as reported previously (Plate et al., 2016)), this compound did not significantly 82 influence bortezomib-induced toxicity in these cells (Figure 7A,B). Furthermore, 147 did not increase or 83 decrease activation of the pro-apoptotic caspases 3/7 in ALMC-2 cells co-treated with subtoxic or toxic doses 84 of bortezomib, respectively ( Figure 7C). These results indicate that 147 does not interfere with bortezomib-85 induced plasma cell cytotoxicity and that these two compounds have the potential to be used in combination to 86 mitigate both the amyloid pathology and plasma cell malignancy associated with AL pathogenesis. The ELISA for ALLC and fully-assembled IgGs in conditioned media prepared on ALMC-2 or KAS-6/1 cells 84 was performed using an identical approach to that reported in (Plate et al., 2016). Briefly, ALMC-2 or KAS-6/1 85 plasma cells were plated 100,000 cells/well in 150 µL of media in 96-well MultiScreen HTS filtration plates 86 (EMD Millipore). Five replicates were treated with DMSO or compounds at the indicated concentrations and 87 incubated for 18 hr. Media was removed by filtration using a QIAvac 96 vacuum manifold (Qiagen) and wells 88 were washed two times with 150 µL media. Wells were then incubated with 150 µL of fresh media for 2 hr and 89 the conditioned media was harvested into a 96-well plate using the vacuum manifold. Whole lysates were 90 obtained by adding RIPA buffer to the cells in the filtration plates. Free LC and IgG concentrations were 91 determined by ELISA in 96-well plates (Immulon 4HBX, Thermo Fisher). Wells were coated overnight at 37 ºC 92 with sheep polyclonal free λ LC antibody (Bethyl Laboratories, A80-127A) at a 1:500 dilution or human IgG-93 heavy and light chain antibody (Bethyl Laboratories, A80-118A) at a 1:2000 dilution in 50 mM sodium 94 carbonate (pH 9.6). In between all incubation steps, the plates were rinsed extensively with Tris-buffered saline 95 containing 0.05% Tween-20 (TBST). Plates were blocked with 5% non-fat dry milk in TBST for 1 hr at 37ºC. 96

147-dependent reductions in ALLC secretion are compatible with bortezomib-induced ablation of AL
Media analytes were diluted between 5 -200 fold in 5% non-fat dry milk in TBST and 100 µL of each sample 97 was added to individual wells. Light chain or IgG standards ranging from 3 -300 ng/mL were prepared from 98 purified human Bence Jones λ light chain or human reference serum (Bethyl Laboratories, P80-127 and RS10-99 110). Plates were incubated at 37 ºC for 1.5 hr while shaking. Finally, HRP-conjugated goat anti-human λ light 00 chain antibody (Bethyl Laboratories, A80-116P) was added at a 1:5,000 dilution or HRP-conjugate IgG-Fc 01 fragment cross-adsorbed antibody (Bethyl Laboratories, A80-304P, 1:30,000 dilution) was added in 5% non-fat 02 dry milk in TBST, followed by a 1.5 hr incubation of the plates at 37 ºC. The detection was carried out with 2,2'-03 azinobis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS, 0.18 mg/mL) and 0.03% hydrogen peroxide in 04 100 mM sodium citrate pH 4.0. Detection solution (100 µL) was added to each well and the plates were 05 incubated at room temperature. The absorbance was recorded at 405 nm and the values for the LC standards 06 were fitted to a 4-parameter logistic function. Light chain or IgG concentrations were averaged from at least 3 07 independent replicates under each treatment and then normalized to vehicle conditions. 08 09 Immunoblotting for ALLC 10 Conditioned media was collected from ALMC-2 cells as previously described (Plate et al., 2016). Media was 11 denatured with Laemmli buffer + 100 mM DTT and boiled for 5 min prior to being separated by SDS-PAGE. 12 Conditioned media for non-reducing gels were prepared as above in the absence of DTT. Cell lysates were 13 prepared as previously described (Plate et al., 2016) in RIPA buffer with fresh protease inhibitor cocktail 14 (Roche). Total protein concentration in cellular lysates was normalized using the Bio-Rad protein assay. 15 Lysates were then denatured with Laemmli buffer + 100 mM DTT and boiled for 5 min before being separated 16 by SDS-PAGE. Proteins were then transferred onto nitrocellulose membranes (Bio-Rad) for immunoblotting 17 and blocked with 5% milk in Tris-buffered saline, 0.5 % Tween-20 (TBST). Membranes were then incubated 18 overnight at 4°C with primary antibodies. Membranes were washed in TBST, incubated with IR-Dye 19 conjugated secondary antibodies and analyzed using Odyssey Infrared Imaging System (LI-COR Biosciences). 20 Quantification was carried out with LI-COR Image Studio software. 21 To measure intracellular ALLC aggregation, ALMC-2 cells were lysed in RIPA buffer as described 22 above. Following centrifugation at 20,000 g for 10 min, we collected the soluble supernatant fraction (SN). We 23 then resuspended the pellet in RIPA + 2% SDS and DTT and sonicated to release aggregate protein. SN and 24 resolubilized pellets were then separated by SDS-PAGE and probed by immunoblotting as above. 25 26 ALLC Immunopurification. 27 ALLC was co-purified with ER proteostasis factors using an identical strategy to that previously reported (Plate 28 et al., 2016). Briefly, ALMC-2 cells were treated for 30 min a room temperature with PBS containing the 29 reversible crosslinker dithiobis (succinimidiyl propionate) (DSP, Thermo Scientific, 500 µM). This crosslinking 30 reaction was quenched by addition of 100 mM Tris pH 7.5 for 15 min. Lysates were then prepared in RIPA 31 buffer, as above. Total protein concentration in cellular lysates was normalized using Bio-Rad protein assay. 32 Cell lysates were then subjected to preclearing with Sepharose 4B beads (Sigma) at 4 °C for 1 h with agitation. 33 Lysates were then incubated overnight at 4 °C with Protein A beads coupled to λ LC antibody, as described 34 below. After four washes in RIPA buffer, crosslinks were cleaved and proteins were eluted by boiling in 35 Laemmli buffer with 100 mM DTT. Eluates were then separated by SDS-PAGE and immunoblotted as 36 described above. Quantifications of proteins co-purified with ALLC were normalized to the recovered ALLC, as 37 PBS and lysates were harvested using RIPA buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1 % SDS, 1% Triton 71 X-100, 0.5% deoxycholate) and protease inhibitor cocktail (Roche). ALLC was immunopurified using human 72 lambda LC coupled to Protein A Sepharose beads as described above. Lysates were incubated with the beads 73 overnight at 4 degrees. Beads were washed 5 times with RIPA buffer. ALLC was then isolated by boiling in 74 Laemmli buffer + 100 mM DTT and separated on SDS-PAGE. For whole cell lysates, cells were lysed in RIPA 75 buffer. Laemmli buffer + 100 mM DTT was then added to the lysates and separated by SDS-PAGE. All gels 76 were dried, exposed to phosphorimager plates (GE Healthcare), and imaged with a Typhoon imager. Band 77 intensities were quantified by densitometry in ImageQuant. 78 79

26
F. Normalized quantification of PDIA1 and PDIA4 from immunoblots as shown in Figure 6E. The recovery 27 of each PDI was normalized to the recovery of ALLC under each condition, allowing accurate 28 evaluation of the interaction between these two proteins. Error bars show SEM for n>2 independent 29 experiments. *p< 0.05 from a paired t-test.