HDP-101 potently reduces viability of myeloma cells
To evaluate the activity of HDP-101, we studied a panel of BCMA-expressing myeloma cell lines (Fig. 1A) representing common molecular subtypes, including translocation (t)(11;14), t(4;14), t(14;16), gain and amplification of 1q, and different TP53 status and IL-6 dependency. HDP-101 potently reduced viability in all of these lines in a dose-dependent manner (Fig. 1B), as evidenced by median inhibitory concentrations (IC50) ranging from 40 pM to 38.6 nM (Table S2). As the IC50 values across the cell lines did not correlate with surface BCMA expression (R2 = 0.094), we examined the impact of the GSI RO4929097 (27), which increased surface BCMA levels (Fig. S1A). When the GSI was added to HDP-101, the combination enhanced the reduction in viability compared with either agent alone (Fig. S1B). One prominent feature of myeloma clinically is the benefit that neoplastic cells receive from their interactions with the marrow microenvironment, including adhesion-mediated drug resistance (31). To model this, we co-cultured myeloma cells with Green fluorescent protein (GFP)-labeled HS-5 stromal cells, exposed these co-cultures to HDP-101 and control compounds, and evaluated them for Annexin V and TO-PRO-3 staining by gating on GFP-positive and-negative cells. HS-5 cells were essentially unaffected by HDP-101 as the viable (Annexin V−/TO-PRO-3−) cell fraction was stable under all tested conditions (Fig. 1C). In contrast, HDP-101 reduced the proportion of viable MM1.S myeloma cells to 36.7% from the baseline of 74.0%, while this fraction was 37.1% in the presence of HS-5 cells (Fig. 1C). Notably, free α-amanitin, the unconjugated BCMA antibody, and an anti-digoxigenin/α-amanitin control ADC did not impact viability. Qualitatively similar findings were noted when this same experiment was performed with H929 cells (Fig. S2). One especially relevant product of the microenvironment is APRIL, since it binds BCMA and contributes to multiple aspects of the transformed phenotype (5). Importantly, when excess APRIL or soluble BCMA were added, while there was slight interference with the activity of HDP-101 at low ADC concentrations (Fig. S3A-B), at higher, physiologically relevant levels, HDP-101 overcame these soluble factors. Finally, acquired drug resistance is another prominent aspect of myeloma, and agents that overcome resistance hold promise for relapsed/refractory disease. There was no difference between the efficacy of HDP-101 in MM1.S versus MM1.R (RRID:CVCL_8794) dexamethasone-resistant cells (Fig. 1D, left upper panel), which showed greater potency in these studies compared to Fig. 1B since cells were incubated for 96 hours, demonstrating also a time-dependent effect. Similarly, the sensitivity of RPMI 8226 cells (RRID:CVCL_0014) resistant to melphalan (RRID:CVCL_J434) or doxorubicin (RRID:CVCL_J431) was similar to (Fig. 1D), or even slightly higher than their WT counterparts (Table S3). Moving to models of novel drug resistance, HDP-101 activity was not reduced in bortezomib-resistant RPMI 8226 or KAS-6/1, or in carfilzomib-resistant RPMI 8226, U266 (RRID:CVCL_0566), or KAS-6/1 cells (Fig. 1D, Table S3). Finally, resistance to the cell-autonomous effects of lenalidomide was studied, and no reduction in efficacy was seen (Fig. 1D, Table S3).
Del17p reduces POLR2A expression, increasing HDP-101 sensitivity
Del17p producing TP53 loss has been linked to POLR2A co-deletion and haploinsufficiency in colorectal and prostate carcinoma (21, 22). Therefore, we evaluated the Multiple Myeloma Research Foundation’s CoMMpassSM database, and first confirmed that POLR2A copy number was reduced in del17p patients compared with those without del17p (mean 0.0373 vs. -0.5688, respectively, p < 0.0001; Fig. 2A). This was associated with a significant reduction in POLR2A mRNA levels by RNASeq (16.72978 FPKM (fragments per kilobase of exon model per million reads mapped) vs. 12.88536 FPKM, respectively, p < 0.0001; Fig. 2B). Notably, no change was seen in BCMA copy number and mRNA expression with del17p (Fig. S4). To model this in vitro, we started with TP53 WT H929, MM1.S, and MOLP-8 cells and used genome editing to knockout (KO) TP53 (Fig. S5A). Next, POLR2A was subjected to knockdown (KD) using shRNAs (Fig. S5B), and we exposed these to HDP-101. Compared to WT cells, TP53 KO alone showed no impact on HDP-101 sensitivity in H929 and MM1.S cells, while there was a small degree of sensitization in MOLP-8 cells (Fig. 2C). In contrast, POLR2A KD alone had a more significant sensitizing effect and was comparable to that seen in the dual TP53/POLR2A KO/KD cells (Fig. 2C). Similarly, by Annexin V staining, HDP-101 induced evidence of decreased cell viability and increased apoptosis with increasing surface exposure of phosphatidyl-serine in the POLR2A KD and dual TP53/POLR2A KO/KD models (Fig. 2D). Finally, we also evaluated mitochondrial membrane potential changes, and the TP53 WT/POLR2A KD and TP53 KO/POLR2A KD cells both had substantially decreased potentials (Fig. 2E). In addition, cell cycle analysis was performed on the four H929-derived cell lines, and HDP-101 increased the apoptotic, sub-G0/G1 cell population to a greater extent in the two POLR2A KD models (Table S4). Together, the data support the statement that HDP-101 shows enhanced efficacy driven by reduced POLR2A levels, including in models of del17p myeloma, where sensitivity is increased compared to WT 17p controls.
HDP-101 induces ER stress and ICD while reducing anti-apoptotic modulators
To gain mechanistic insights into the impact of HDP-101, gene expression analysis was performed on TP53 WT H929 and MM1.S cells (Fig. 3A), and also on their TP53 KO counterparts (Fig. S6). Pairwise analysis (Fig. S7) revealed that HDP-101 down-regulated more genes than were up-regulated, consistent with its mechanism as a POLR2A inhibitor. Enriched genes with a false discovery rate of < 5% are presented in Table 1 as hallmark gene sets (32) identified from the Molecular Signatures Database, with a focus on gene sets up-regulated in all four models. Notably, the UPR, a target of several therapeutics effective against myeloma (33), was induced by HDP-101. As this gene set contained Activating transcription factors (ATF)-5 and ATF6, as well as X-box protein (XBP)-1, we performed Western blotting to determine if there was evidence of UPR activation at the protein level. Both ATF6, which represents one arm of the ER UPR, and ATF5, which regulates the mitochondrial UPR, were induced by HDP-101 (Fig. 3B), and this tended to occurr to a greater extent in the POLR2A KD cell models. Similarly, Inositol-requiring enzyme 1 (IRE1) was induced, which led to enhanced downstream expression of the short isoform of XBP1 (Fig. 3B). On the down-regulated side, and consistent with the known mechanism of action of α-amanitin (34), RNA polymerase II and III levels were decreased (Fig. 3C). Notably, HDP-101 also decreased expression levels of several anti-apoptotic proteins, including Myeloma cell leukemia 1 (MCL1) and BCL-xL, as well as the X-linked inhibitor of apoptosis (Fig. 3C). These occurred in association with increased markers of apoptosis, including of cleaved poly-(ADP-ribose) polymerase (PARP1) and cleaved Caspases 3 and 7 (Fig. 3B,C).
Table 1
Hallmark gene sets identified from the Molecular Signatures Database using gene expression data from myeloma cells exposed to HDP-101
Gene Sets Up-regulated |
HALLMARK_G2M_CHECKPOINT |
HALLMARK_E2F_TARGETS |
HALLMARK_IL2_STAT5_SIGNALING |
HALLMARK_UNFOLDED_PROTEIN_RESPONSE |
HALLMARK_HYPOXIA |
HALLMARK_MTORC1_SIGNALING |
HALLMARK_MYC_TARGETS_V2 |
HALLMARK_ANDROGEN_RESPONSE |
HALLMARK_MYC_TARGETS_V1 |
HALLMARK_GLYCOLYSIS |
Several ADCs have the ability to induce both direct cytotoxic effects as well as ICD as part of their mechanisms of action (2) and, given the link between UPR activation and ICD (35), we next sought to determine if HDP-101 could so as well. Starting with the MM1.S cell model, isotype antibody or HDP-101-treated cells were analyzed to detect externalization of Calreticulin (CRT) or High mobility group box 1 (HMGB1) by immunofluorescence. Compared to the isotype negative control, HDP-101 enhanced cell surface levels of both CRT and HMGB1 (Fig. 4A), consistent with induction of ICD as seen with bortezomib as a positive control. Next, we expanded our panel to include H929 and MM1.S cells that were TP53 WT, TP53 KO, POLR2A KD, or dual KO/KD, and looked at CRT and HMGB1, as well as externalization of Heat shock protein 70 (HSP70) and HSP90. Once again compared to the isotype negative control, HDP-101 induced externalization of all four markers of ICD in both cell lines, with a trend seen towards increasing ICD in the H929 model system (Fig. 4B). Finally, we further broadened our analysis to look at our larger panel of myeloma cell lines representing multiple molecular subtypes of myeloma. Compared to the isotype control, HDP-101 enhanced cell surface levels of Calreticulin, HMGB1, and HSP70 in all of the cell lines (Fig. S8), suggesting consistent activation of the UPR and ICD.
Activity of HDP-101 in vivo may engage NK cells and macrophages
We next tested the ability of HDP-101 to exert effects against primary samples derived from patients, and separated these initially into fractions enriched for CD138-positive tumor cells and CD138-negative tumor microenvironment (TME) cells. HDP-101 showed minimal effects on the proportion of TME cells that were in the early apoptotic (Annexin V+/TO-PRO-3−), late apoptotic (Annexin V+/TO-PRO-3+), or necrotic (Annexin V−/TO-PRO-3+) stages of cell death (Fig. 5A) compared to the isotype control. In contrast, HDP-101 induced a substantial decrease in the viable cell fraction (Annexin V−/TO-PRO-3−) and an increase predominantly in the late apoptotic fraction under the condition tested (Fig. 5A). To examine the activity of HDP-101 in vivo, we developed a systemic xenograft using luciferase (luc)-labeled MM.1S WT TP53 cells in NOD-SCID mice. Once engrafted, mice were randomized to produce three groups with equivalent disease burden to receive a single dose of vehicle, HDP-101 at 4 mg/kg, or the anti-digoxigenin/α-amanitin ADC control at 4 mg/kg. Compared to vehicle, the negative control/non-targeted α-amanitin-ADC did not substantially reduce disease burden (Fig. 5B,C). In contrast, HDP-101 first slowed disease progression, and this was followed by complete disease regression by day 32 (Fig. 5C). Similar patterns were seen in the MM1.S TP53 KO, POLR2A KD, or dual KO/KD models, with regression of disease and no evidence of relapse even after a single 2 mg/kg dose of HDP-101 (Fig. 5D). Amanitins are able to impact quiescent cells (24, 25, 26) that could be equivalent to myeloma-initiating cells, which have been hypothesized as potentially contributing to drug resistance and disease relapse after other anti-myeloma therapies (36, 37). Therefore, we considered whether elimination of such cells could have contributed to the lack of relapse after HDP-101, and used staining for Aldehyde dehydrogenase (ALDH) to identify them (36). ALDH-positive cell fractions of H929 and MM.1S cells proved to be more clonogenic than their ALDH-negative counterparts (Fig. S9A,B). Notably, HDP-101 showed similar activity against the two, with no significant difference in the IC50 (Fig. S9C).
As disease relapse was not see in NOD-SCID mice xenografted with MM1.S cells and treated with HDP-101 at 100 days, we rechallenged a pilot cohort of five of these mice with luc-labeled MM1.S cells. Whereas engraftment was virtually 100% at baseline, none of these mice developed detectable tumor after another 100 day follow-up period (Fig. 6A). While these mice are immune deficient, recent re-evaluation of this model has indicated that they retain aspects of the innate immune system, including NK cells and macrophages (38). To determine if these could be contributing to the lack of re-engraftment, we repeated this experiment except that, just prior to disease re-challenge, we randomized mice to receive a dose of either control rabbit antiserum or a rabbit anti-Asialo-GM1 antibody. While the control group showed a range of fluorescence values after the initial MM1.S re-injection (Fig. 6B) consistent with some baseline circulating tumor cells, this signal rapidly disappeared (Fig. 6C). In contrast, treatment with anti-Asialo-GM1 to deplete NK cells and macrophages promoted tumor engraftment and led to a significantly higher disease burden (Fig. 6B,C).
HDP-101 is more potent ADC than belantamab mafodotin
Belantamab mafodotin (BelaMaf), a BCMA-targeted ADC linked to monomethyl-auristatin F, was the first drug in its class to achieve a regulatory approval for myeloma. However, more recently, it was withdrawn as a randomized study of this agent did not show sufficient superiority compared to a standard of care. Given the encouraging activity of HDP-101, we compared it directly to BelaMaf in both our in vitro and in vivo model systems. In the former, HDP-101, an ADC with a drug:antibody ratio (DAR) of 2 (24), more potently reduced viability of all of the myeloma cell lines we tested at 48 hours than was the case for BelaMaf (Table S2), which has a DAR of 4. At 96 hours, HDP-101 showed both greater overall efficacy than BelaMaf in our H929- and MM1.S-based WT, TP53 KO, POLR2A KD, and dual KO/KD models, as well as preferential activity in the del17p dual knockout cells (Table 2). Next, we tested primary patient-derived samples, and found that HDP-101 induced a greater loss of viability than did BelaMaf at two different concentrations (Fig. 7A). Finally, we prepared a xenograft based on the del17p MM1.S dual TP53 KO/POLR2A KD model, and treated these with a single dose of isotype control, or either 0.1 or 0.5 mg/kg of BelaMaf or HDP-101. This dose was selected to produce sub-total myeloma cell killing by HDP-101 to allow for better comparisons between the activity of the two ADCs. At day 60 (Fig. 7B), control mice had substantial disease burden that was reduced only modestly by BelaMaf at either of the two doses (Fig. 7C). In contrast, a statistically significant tumor growth inhibition was seen at both doses of HDP-101 (Fig. 7C) which, as expected, was greater at the higher dose, where some mice had no measurable tumor (Fig. 7B).
Table 2
Median inhibitory concentrations (pM) for H929 and MM1.S-based myeloma cell line models treated with belantamab mafodotin or HDP-101 for 96 hours*
| HDP-101 | BelaMaf |
| H929 | MM1.S | H929 | MM1.S |
TP53 WT/POLR2A WT | 155 \(\pm\) 0.008 | 243 \(\pm\) 0.04 | 1,157 \(\pm\) 0.2 | 2,625 \(\pm\) 0.4 |
TP53 KO/POLR2A WT | 114 \(\pm\) 0.01 | 270 \(\pm\) 0.05 | 1,296 \(\pm\) 0.1 | 3,145 \(\pm\) 0.6 |
TP53 WT/POLR2A KD | 29 \(\pm\) 0.01 | 109 \(\pm\) 0.01 | 1,269 \(\pm\) 0.2 | 2,686 \(\pm\) 0.5 |
TP53 KO/POLR2A KD | 13 \(\pm\) 0.005 | 75 \(\pm\) 0.01 | 922 \(\pm\) 0.08 | 4,096 \(\pm\) 1.1 |
Abbreviations: BelaMaf, belantamab mafodotin; KD, knock down; KO knock out; POLR2A, RNA polymerase II subunit alpha; TP53, p53 tumor suppressor; WT, wild type |