Intracellular trafficking of furin enhances cellular intoxication by recombinant immunotoxins based on Pseudomonas exotoxin A Open Access

Furin (EC 3.4.21.75) is a ubiquitously expressed mammalian protease that belongs to the proprotein convertase subtilisin/kexin (PCSK) family of serine endoproteases (reviewed in Seidah and Prat, 2012). Within the secretory pathway, furin proteolytically activates numerous proprotein substrates that include endogenous pro-hormones and growth factors as well as pathogen-associated viral glycoproteins and bacterial toxins. Furin is important for maintaining cellular homeostasis and has an essential role in embryogenesis. It is also implicated in diseases ranging from Alzheimer's and cancer to bird flu, Ebola, and anthrax (reviewed in Thomas, 2002). Recently, furin has been implicated in viral infection by SARS-CoV-2, and there is evidence that a novel furin cleavage site on the viral spike protein may be responsible for its high infectivity and overall virulence (Coutard et al., 2020; Hoffmann et al., 2020; Johnson et al., 2021).

Studies describing the role of furin in both normal cellular activity and disease have provided important molecular insights (reviewed in Thomas, 2002; Jaaks and Bernasconi, 2017; Braun and Sauter, 2019). Evaluation of furin knockout mice showed an embryonic lethal phenotype (Roebroek et al., 1998), but furin knockdown by siRNA was not lethal to cultured cells (Tay et al., 2012). Furin expression varies widely among the cell lines evaluated in the Human Protein Atlas (www.proteinatlas.org; Karlsson et al., 2021), but naturally furin-deficient cell lines, such as the CHO-derived RPE.40 (Moehring and Moehring, 1983; Watson et al., 1991) and colon carcinoma line LoVo (Takahashi et al., 1995, 1993), are rare. The limited selection of furin-deficient cell lines and variation among existing lines has limited our capability to study the role of furin in cultured cells.

One exogenous furin substrate is Pseudomonas exotoxin A (PE). PE is a protein virulence factor secreted by Pseudomonas aeruginosa but can also be conjugated to antibody fragments to make a targeted cell killing agent called a recombinant immunotoxin (RIT) (reviewed in Weldon and Pastan, 2011). PE-based RITs can be used therapeutically for the treatment of cancers; several are in clinical trials and one (moxetumomab pasudotox; Kreitman and Pastan, 2020) has been FDA-approved for the treatment of hairy cell leukemia. Furin plays an important, but incompletely understood, role during the intoxication of PE and PE-based RITs. Prior research has established that furin proteolytically activates the toxin during its intracellular trafficking route (Chiron et al., 1994; Gordon et al., 1995; Gu et al., 1996; Inocencio et al., 1994). In addition, furin-deficient LoVo cells show poor sensitivity to RITs (Chiron et al., 1997), furin inhibitors diminish RIT activity (Weldon et al., 2015), and mutations to the furin cleavage site of RITs can influence their cellular toxicity (Weldon et al., 2015).

In addition to furin-mediated cleavage, circumstantial evidence suggests that furin may act as a transporter for PE and PE-based RITs. The furin cleavage site on PE is a remarkably poor substrate for furin, demonstrating the most unfavorable cleavage characteristics of all substrate sequences evaluated in two separate studies (Izidoro et al., 2009; Remacle et al., 2008). Our own unpublished experience with in vitro cleavage assays confirms this. Mutations around the furin cleavage site of RITs can enhance both furin cleavage and RIT cytotoxicity, but there was no correlation between the two parameters (Weldon et al., 2015). Evidence also suggests that furin can act as a non-proteolytic chaperone, as demonstrated in the case of matrix metalloproteinase-28 (MMP-28) transport from the TGN to the cell surface, where MMP-28 is released without being cleaved (Pavlaki et al., 2011). Taken together, these points suggest that the role of furin is more complex than a cleavage activation step in the intoxication pathway.

In this study, our objective was to investigate the role of furin in the intoxication pathway of PE-based RITs. Specifically, we sought to understand the relationship between the cleavage, cytotoxicity, and intracellular localization of RITs and furin. We hypothesized that in addition to cleavage, furin can also function in directing RITs to the necessary subcellular compartments required for productive intoxication, such as from endosomes to the Golgi. This role of furin is separate from its traditional proteolytic activity and informs the development RITs and other antibody conjugates that utilize a furin-cleavable linker to separate cargo from the antibody. Consequently, the composition of the furin site may influence the intracellular trafficking and the location of cleavage for RITs and other furin-cleavable conjugates.

To explore the role of furin in PE RIT intoxication, we have generated a furin-deficient HEK293 cell line, ΔFur293. HEK293 cells are well suited for understanding the role of furin in cellular processes; they have excellent growth characteristics, are widely studied, and express clearly detectable levels of furin. We employed a variant HEK293 line, HEK293 FRT, that contains a single Flp recombinase recognition target site (FRT) stably integrated into its genome. Using a CRISPR/Cas9 system, we genetically modified HEK293 FRT cells to eliminate endogenous furin expression. Results were confirmed by sequencing, western blots, cytotoxicity assays, intracellular cleavage assays, and complementation with transgenic furin. The ΔFur293 line will be a convenient and valuable tool for studies evaluating the role of furin in a variety of natural and disease-related processes.

We then employed the ΔFur293 cell line to investigate the influence of furin on the activity of PE-based RITs. We stably introduced mutant forms of furin with either altered intracellular trafficking patterns or decreased catalytic activity into ΔFur293 cells. The results of RIT cytotoxicity assays support the importance of furin in the intoxication pathway of PE-based RITs and indicate that both furin cleavage and trafficking play important roles during RIT intoxication.

Using CRISPR-Cas9 based gene editing we generated a furin-deficient HEK293 FRT cell line without detectable furin expression by western blot (Fig. 1). The HEK293 FRT FURIN KO cell line was shortened to ΔFur293 for convenience. The gene sequence of the two furin alleles (Fig. S1) and STR profiling (Table S2) of the chosen clone were determined to confirm its identity.

To evaluate the effect of furin on the efficacy of PE-based recombinant immunotoxins (PE RITs), the cytotoxicity of HB21-LR (anti-transferrin receptor single-chain variable fragment PE24 RIT; Zhu and Weldon, 2019) was evaluated against HEK293 FRT and ΔFur293 cells (Fig. 2). Each cell line was evaluated in the presence and absence of a proprotein convertase inhibitor [PPCI; 4-(guanidinomethyl)phenylacetyl-Arg-Val-Arg-4-amidinobenzylamide], a cell permeable reversible inhibitor of furin (K_i=16 pM) that has been shown to have minimal toxicity in tissue culture at concentrations of up to 50 μM (Becker et al., 2012). Preliminary analysis of PPCI indicated that a 1 μM concentration was adequate to achieve the maximum influence on RIT cytotoxicity (Fig. S2). A representative set of cytotoxicity assays is shown in Fig. S3.

Analysis showed that the EC₅₀ of HB21-LR against HEK293 FRT cells significantly increased by approximately 8-fold (P=0.0030) in the presence of PPCI. No difference in HB21-LR cytotoxicity against ΔFur293 cells with and without PPCI was found. When compared to untreated HEK293 FRT cells, ΔFur293 cells in the presence and absence of PPCI showed significant increases in EC₅₀ of approximately 13-fold (P=0.0005) and 10-fold (P=0.0044). Comparison of PPCI-treated HEK293 cells with ΔFur293 cells showed no statistically significant difference in EC₅₀. Interestingly, a small (1.5-fold) but significant difference was observed between the EC₅₀ of HEK293 FRT and ΔFur293 cells when both lines were in the presence of PPCI (P=0.0007). Overall, these data show that furin deletion results in dramatic decrease in HB21-LR toxicity.

As an additional control, we evaluated HB21-LR against the human colon carcinoma cell line LoVo, which lacks active furin (Takahashi et al., 1995, 1993). As with ΔFur293 cells, no significant difference was observed with or without PPCI (Fig. S4).

We next assessed the effect of restoring furin to ΔFur293 cells by stably transfecting them with the wild-type furin gene (Fur) inserted into the genomic FRT site (ΔFur293/Fur cells). Results from cytotoxicity assays are shown in Fig. 3. As a control, we also evaluated ΔFur293 cells stably transfected with the gene for chloramphenicol acetyltransferase (CAT) at the FRT site (ΔFur293/CAT).

We then evaluated the cleavage efficiency of HB21-LR in each cell line over an 8 h time course (Fig. 4). Lysates from treated cells were examined for full length and cleaved HB21-LR by western blot at various time intervals (Fig. 4A). Each band was evaluated by densitometry and normalized against the actin loading control. The cleaved fraction of total protein was plotted against time for each of the cell lines (Fig. 4B). In HEK293 FRT cells, the percentage of total toxin cleaved within the cell peaked at 50%, while in ΔFur293 cells the percentage of cleaved toxin peaked at less than 10%. The ΔFur293/Fur cells demonstrated enhanced cleavage of the toxin, reaching a peak level of over 90% RIT cleaved at the 4 h time point. We were unable to harvest a sufficient quantity of intact ΔFur293/Fur cells to analyze at the 8 h time point due to cell death from their high sensitivity to HB21-LR.

We next generated three furin mutants designed to evaluate the role of furin in RIT intoxication: two mutants with altered intracellular trafficking patterns (S773A/S775A and S773D/S775D) and one mutant with diminished catalytic activity (N295A). The trafficking mutants target two serine residues (positions 773 and 775) in the acidic cluster/TGN sorting signal within the cytoplasmic tail of furin. The phosphorylation state of these serine residues controls the trafficking of furin between the TGN and the cell surface (Jones et al., 1995). The catalytic activity of furin cannot be completely eliminated from the mature protein because of its requirement for autocatalytic processing (Anderson et al., 2002). Reduced activity, however, can be obtained by mutation of an asparagine at position 295, which corresponds to the position of the oxyanion hole in its catalytic mechanism (Roebroek et al., 1994).

The three mutants were stably transfected into the FRT site of ΔFur293 cells (ΔFur293/Fur^ADA, ΔFur293/Fur^DDD, ΔFur293/Fur^Ala-295). Initial cytotoxicity analysis of these mutants showed significant differences between clones expressing the same furin variant (data not shown), which suggested that clones might exhibit variable expression levels of the furin transgene. To test this, we compared furin expression from two clones of each mutant, as well two clones of wild-type furin stably transfected into both ΔFur293 and HEK293 FRT cells. Fig. S5B shows a western blot of cell lysates from each line. The results were quantified by densitometry (Fig. S5A) and used to normalize the results of cytotoxicity assays. This normalization eliminated discrepancies between different clones expressing the same mutant (Fig. 5).

The S773A/S775A alanine trafficking mutant of furin (Fur^ADA) causes it to become trapped in early endosomes (Jones et al., 1995; Schapiro et al., 2004). We observed no difference between HB21-LR cleavage in ΔFur293/Fur and ΔFur293/Fur^ADA cells (Figs 4 and 6), and an 8-fold resistance to HB21-LR cytotoxicity (Fig. 5). The S773D/S775D aspartic acid trafficking mutant of furin (Fur^DDD) is depleted from early endosomes and preferentially retrieved to the TGN or cell surface, exhibiting a more disperse localization in the cell (Jones et al., 1995). Cleavage was decreased by approximately 20% when compared to ΔFur293/Fur cells (Figs 4 and 6). Cytotoxicity was decreased nearly 3-fold compared to ΔFur293/Fur cells and increased 2.6-fold compared to ΔFur293/Fur^ADA cells (Fig. 5). The N295A furin oxyanion hole mutant (Fur^Ala-295) has decreased catalytic activity compared to wild-type furin (Roebroek et al., 1994) while having an identical trafficking pattern (Creemers et al., 1995). The ΔFur293/Fur^Ala-295 cells are approximately 40% less efficient at cleaving HB21-LR when compared to ΔFur293/Fur cells (Figs 4 and 6). This resulted in an approximately 8-fold decrease in cytotoxicity relative to ΔFur293/Fur (Fig. 5).

To determine if a stable interaction between furin and PE-based RITs can occur, we analyzed the formation of a stable PE-furin complex by co-immunoprecipitation. Soluble furin with a C-terminal myc/FLAG tag and either HB21 (PE38 RIT) or HB21-LR (PE24 RIT) were incubated together and precipitated using either anti-myc monoclonal or anti-PE polyclonal antibodies. We found that furin could be detected following capture of the complex by the anti-PE antibody (Fig. S6A). The reverse experiment using the anti-myc antibody to capture the complex, however, did not definitively show precipitation of the RIT (Fig. S6B).

Here we present a furin deficient derivative of the HEK293 FRT cell line, HEK293 FRT FURIN KO, abbreviated as ΔFur293. Successful disruption of the furin gene by CRISPR/Cas9 was confirmed by western blot analysis (Fig. 1) and sequencing (Fig. S1). Compared to the human furin gene (Ensemble Transcript ID: ENST00000268171.8) one allele showed a 105 bp deletion (nucleotides 7168-7272) and a second allele showed a 106 bp deletion (nucleotides 7168-7273). The deletion spans most of the signal peptide domain (residues 1-26) and the amino terminal portion of the prodomain (residues 27-107). The 106 bp deletion also introduces a frameshift mutation beginning at position 42, leading to a nonsense product.

STR profile analysis (Table S2) confirmed the origin and identity of ΔFur293. HEK293 FRT and ΔFur293 were identical to each other at 20 of the 22 loci and showed a partial match at two loci (D19S433 and D8S1179). Although variation from the reference profile was observed, the variation was consistent with existing variation in the reference sequence (Bairoch, 2018), reports in the literature (Bylund et al., 2004), and variation among STR profiles from biological resource centers.

Cytotoxicity analysis of HEK293 FRT and ΔFur293 cells in the presence and absence of the furin inhibitor PPCI (Fig. 2) showed that cells in the absence of furin or in the presence of a furin inhibitor are approximately 10-fold more resistant to HB21-LR. No cytotoxicity difference was found when comparing ΔFur293 cells to HEK293 FRT cells with PPCI. These data conclusively demonstrate that furin enhances toxin sensitivity. A small (1.5-fold) but significant difference was observed when both PPCI-treated HEK293 FRT and PPCI-treated ΔFur293 cells were evaluated for cytotoxicity. This observation suggests that cells may retain a small quantity of active furin in the presence of PPCI and support the use of ΔFur293 over PPCI for future studies.

Insertion of the wild-type furin gene into ΔFur293 cells (ΔFur293/Fur) restored and enhanced RIT sensitivity (Fig. 3). This effect had not previously been observed but is consistent with our expectation that increased furin expression enhances RIT cytotoxicity. A comparison of furin expression levels by western blot indicated that ΔFur293/Fur expresses significantly more furin than HEK293 FRT (Fig. S5). Control experiments with transgenic chloramphenicol acetyltransferase showed no difference in RIT sensitivity from the parental line.

We next performed a RIT intracellular cleavage analysis of the HEK293 FRT, ΔFur293, and ΔFur293/Fur cell lines treated with HB21-LR (Fig. 4). Compared to the parent line, ΔFur293 cells show greatly diminished cleavage and ΔFur293/Fur cells show enhanced cleavage. This is consistent with the cytotoxicity data and western blot evaluation of expression levels (Fig. S5). We further observed that ΔFur293 cells exhibited a small percentage of cleaved toxin. The presence of cleavage products in these cells may be due to other PCSKs with similar cleavage site specificities. It is also possible that cleavage sites for lysosomal proteases could result in similarly sized protein fragments that would be indistinguishable by western blot analysis (Weldon et al., 2009). Further research will be needed to explore the influence of other proteases on RIT cleavage and cytotoxicity.

We then evaluated how complementation of furin by mutants would influence RIT cleavage and cytotoxicity. The results of our analysis of agree with previous work indicating that furin cleavage is important to RIT cytotoxicity but is not the sole means by which furin influences the efficacy of PE-based RITs (Weldon et al., 2015). Two trafficking mutants with alterations in the cytoplasmic tail at phosphorylation signal sequences Ser⁷⁷³ and Ser⁷⁷⁵ (Fur^ADA and Fur^DDD) and a mutant with diminished catalytic activity (Fur^Ala-295) were also evaluated.

The S773A/S775A alanine trafficking mutant of furin (Fur^ADA) causes it to become trapped in early endosomes (Jones et al., 1995; Schapiro et al., 2004). We expected no change in cleavage because PE RITs are most likely cleaved by furin in early endosomes. Although we observed no difference between RIT cleavage in ΔFur293/Fur and ΔFur293/Fur^ADA cells (Figs 4 and 6), we did observe an 8-fold resistance to RITs in cells expressing Fur^ADA (Fig. 5). These results suggest that the decrease in cytotoxicity is due to furin being trapped in early endosomes and are consistent with the hypothesis that furin has a transport role during RIT intoxication.

The S773D/S775D aspartic acid trafficking mutant of furin (Fur^DDD) is depleted from early endosomes and preferentially retrieved to the TGN or cell surface, exhibiting a more disperse localization in the cell (Jones et al., 1995). We expected that RIT cleavage would be decreased in cells expressing this furin variant because of the constitutive signal directing furin out of early endosomes. Consistent with our expectations, cleavage was decreased by approximately 20% when compared to both ΔFur293/Fur and ΔFur293/Fur^ADA cells (Fig. 6). Cytotoxicity, however, was decreased nearly 3-fold compared to the wild-type furin transgene and increased 2.6-fold compared to the alanine trafficking mutant (Fig. 5). This intermediate cytotoxicity phenotype indicates that the altered localization pattern of the aspartic acid trafficking mutant provides a relative improvement in cytotoxicity compared to ΔFur293/Fur^ADA. The result is consistent with the furin acting in a transport role during RIT intoxication, as the movement of furin out of early endosomes to the TGN is enhanced.

The N295A furin oxyanion hole mutant (ΔFur293/Fur^Ala-295) has decreased activity compared to wild-type furin (Roebroek et al., 1994) while having an identical trafficking pattern (Creemers et al., 1995). Our results confirmed a decrease in RIT cleavage efficiency. The ΔFur293/Fur^Ala-295 cells are approximately 40% less efficient at cleaving the RIT when compared to ΔFur293/Fur cells (Figs 4 and 6). This resulted in an approximately 8-fold decrease in cytotoxicity relative to ΔFur293/Fur (Fig. 5). Comparing ΔFur293/Fur^Ala-295 cells to ΔFur293/Fur^ADA cells demonstrated a decrease in cleavage efficiency but no difference in cytotoxicity. This indicates that interfering with either the trafficking or catalytic activity of furin have similar inhibitory effects on RIT intoxication.

We conclude that altering the trafficking pattern or blocking the activity of furin can affect RIT cytotoxicity. Our results are consistent with the hypothesis that furin acts in a transport capacity to direct PE and PE-based RITs out of endosomes and into the TGN for retrograde trafficking to the ER. Although these experiments have tested only a single cell line (HEK293) with a single RIT (HB21-LR), the results are consistent with previous evidence from other RITs on other cell lines (Weldon et al., 2015) and we believe that our results will be broadly applicable to the use of furin-cleavable linkers. Future experiments evaluating the colocalization of furin and RITs by immunofluorescence will explore this hypothesis further.

Single guide RNAs (sgRNAs) to the furin gene were designed using the Broad Institute sgRNA design tool (CRISPick; Doench et al., 2016; Sanson et al., 2018). Three sgRNAs that scored highly and targeted locations early in the furin gene were selected: GCGCATCCCTGGAGGCCCAG (score 0.735013), GAAGGTCTTCACCAACACGT (score 0.606152), and GCTACCACCCATAGCAACCA (score 0.689511).

All oligonucleotides used in this study obtained from Integrated DNA Technologies, Inc. (IDT, Coralville, IA, USA) and are detailed in Table S1. Oligonucleotides corresponding to the selected furin sgRNAs were ligated into the pSpCas9(BB)-2A-Puro (PX459) V2.0 plasmid, a gift from Dr Feng Zhang [Broad Institute of Massachusetts Institute of Technology (MIT) and Harvard, Cambridge, Massachusetts, USA, Addgene plasmid #62988; http://n2t.net/addgene:62988; RRID: Addgene_62988; Ran et al., 2013]. Human furin cDNA (product #RC204279) was obtained from OriGene (Rockville, MD, USA) and inserted into the pcDNA5/FRT plasmid for stable transfection into HEK293 FRT cells. Mutations in the furin gene was performed by site-directed mutagenesis. The HB21-LR expression plasmid pHB21-LR was generated by mutation of the original pHB21 plasmid from the Pastan Lab. Plasmids were transformed into NEB5α competent E. coli (NEB, Ipswich, MA, USA) and DNA was harvested using the Nucleospin plasmid isolation kit (Macherey Nagel, Bethlehem, PA, USA). All constructs were confirmed by gel electrophoresis and sequencing (Macrogen USA, Rockville, MD, USA).

All restriction enzymes were purchased from NEB and PCR amplification was performed with Q5 High-Fidelity PCR (NEB) following the manufacturer's recommendations. DNA ligations were performed using the Blunt/TA Ligase Master Mix (NEB) followed by transformation into NEB5α competent E. coli cells.

HEK293 FRT cells were a kind gift from Dr Ira Pastan (NCI/NIH). They contain a single stably integrated Flp recombinase recognition target (FRT) site at a transcriptionally active genomic locus. LoVo cells (CCL-229) were obtained from ATCC (Manassas, VA, USA). All cells were grown at 37°C with 5% CO₂ in high glucose Dulbecco's Modified Eagle Medium (DMEM) with sodium pyruvate (Corning) supplemented with 10% FBS (GeneMate) and 2 mM Glutagro L-glutamine supplement (Corning). HEK293 FRT cells were cultured in zeocin (100 µg/ml, Invitrogen) to maintain the integrated FRT site. Cells were passaged once they reached >80% confluency.

Transfections were performed using Turbofection 8.0 (Origene, Rockville, MD, USA) following the manufacturer's recommendations. Individual sgRNA plasmids and combinations of plasmids were transfected. Puromycin was added to each well at a concentration of 3 μg/ml following initial passaging. The transfection utilizing all three sgRNA plasmids showed the most marked reduction in furin levels as determined by western blot analysis and was chosen for clonal selection by serial dilution in a 96-well plate according to protocol (Ryan, n.d.; https://www.corning.com/catalog/cls/documents/protocols/Single_cell_cloning_protocol.pdf).

Cells at >80% confluency were harvested and resuspended at a density of 5×10⁶ in 1 ml of cold RIPA buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS) with a broad-spectrum protease inhibitor (Thermo Fisher Scientific, A32953). The mixture was incubated on ice for 10 min, sonicated, and stored at −20°C. Protein concentrations were determined using a bicinchoninic acid assay (BCA) protein assay kit (Thermo Fisher Scientific, 23227).

Genomic DNA was extracted from the selected monoclonal population (Wizard^® Genomic DNA Purification Kit, Promega, Madison, WI, USA) following the manufacturer's recommended protocol. DNA was PCR amplified (primers hFUR Exon 1-3 PCR For and hFUR Exon 1-3 PCR Rev; see Table S1) and purified (Monarch PCR & DNA cleanup kit, NEB). The resulting PCR product was sequenced using the hFur Exon 1 Seq primer (Table S1). Initial sequencing showed that exon 2 of the furin gene had a deletion of approximately 105 nucleotides. However, the sequencing chromatogram indicated heterozygous alleles of the furin gene and prevented an exact sequence.

To specify the genomic sequence of ΔFur293, overlap extension PCR was performed on the genomic DNA to clone exon 2 of the furin gene into the pUC57 plasmid (Bryksin and Matsumura, 2010, 2013). Briefly, genomic DNA from ΔFur293 cells was PCR amplified (primers hFUR Exon 1-3 PCR For and hFUR Exon 1-3 PCR Rev; see Table S1). This PCR product was purified, and a 2 ng/μl solution was prepared. A second PCR reaction was performed using the product from reaction 1 and primers that contained overhangs complementary to portions of pUC57-kan (pUC57 Furin Ex 2 OvEx C+A and pUC57 Furin Ex 2 OvEx D+B; see Table S1). The resulting product from reaction 2 was purified and used as a primer for PCR amplification of the pUC57 plasmid. The product from reaction 3 was DpnI treated, purified, and transformed into NEB5α. Multiple colonies were grown for DNA extraction. In total, ten plasmids were sequenced using the M13 R primer, showing two distinct alleles differing by a single nucleotide (Fig. S1).

To confirm the identity and establish a baseline STR profile for our cell lines, we evaluated 22 loci from ΔFur293 and its parent cell line HEK293 FRT. Genomic DNA was extracted from cultured cells using the Wizard^® Genomic DNA Purification Kit and amplified using the Investigator 24plex QS Kit (Qiagen, Hilden, Germany) on a Veriti 96-well thermal cycler (Applied Biosystems, Foster City, CA, USA). Each sample was separated on a 3500 Genetic Analyzer (Applied Biosystems). The resulting electropherograms were analyzed with GeneMapper ID-X software (Applied Biosystems) version 1.5. We compared our results to each other and to a reference profile for HEK293 (RRID:CVCL_0045) in the Cellosaurus database (Bairoch, 2018).

The recombinant immunotoxin HB21-LR (Zhu and Weldon, 2019), derived from the single-chain anti-transferrin receptor RIT HB21 (Batra et al., 1991) using the ‘LR’ modifications (Weldon et al., 2009), was expressed and purified following the published protocol (Pastan et al., 2004), except for the following modifications. A starting quantity of 40 mg was used instead of 100 mg. All reagents were scaled proportionally, but concentrations remained equivalent. The column chromatography employed a single anion-exchange step instead of the two-step protocol. In place of the Q Sepharose and Mono Q columns as described (Pastan et al., 2004), only a HiTrap Q HP column (Cytiva, Marlborough, MA, USA) was used. The expression plasmid was a kind gift from Dr Ira Pastan (Laboratory of Molecular Biology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA).

Cells (10⁶) were seeded in 60×15 mm dishes with 4 ml fresh medium and incubated for approximately 24 h. After reaching 60-80% confluency, HB21-LR was added directly to the 4 ml of media to a final concentration of 0.1 μg/ml. Cells were incubated at 37°C for various time intervals, after which the media was aspirated and the dish was washed consecutively with 1 ml cold DPBS, 1 ml cold stripping buffer (1 mg/ml BSA in 0.2 M glycine, pH 2.5), and 1 ml cold DPBS. Ice-cold RIPA buffer (100 μl) with protease inhibitor cocktail (Thermo Fisher Scientific) was added to the dish, cells were dislodged using a cell scraper, and the suspended cells were transferred to a microcentrifuge tube. This process was repeated with an additional 100 μl of ice-cold RIPA buffer to collect any cells remaining on the dish. The RIPA solution was then used to prepare total cell lysates as previously described.

Cell lysates containing 20 μg of total protein were analyzed by western blot. The specific primary and secondary antibodies and their corresponding dilutions are indicated in the appropriate figure legends. For furin expression level analysis, mouse anti-furin antibody (mAb, Santa Cruz Biotechnology, sc-133141), mouse anti-β-actin (mAb, Thermo Fisher Scientific, BA3R), and goat anti-mouse IgG alkaline phosphatase conjugated antibody (mAb, Santa Cruz Biotechnology: sc-2058) were used. For internalization and cleavage assay analysis, rabbit anti-PE (pAb, NCI/NIH), rabbit anti-β-actin (mAb, Boster Bio: M01263), and mouse anti-rabbit IgG alkaline phosphatase conjugated antibody (mAb, Santa Cruz Biotechnology: sc-2358) were used. Antibodies were visualized chromogenically by incubating the membrane in an alkaline phosphatase buffer (130 mM Tris, pH 9, 150 mM NaCl, 100 µM MgCl₂) and 2% BCIP/NBT (Roche, 11681451001).

Densitometry was performed using Bio-Rad's Image Lab Software (Version 6.01). For expression level analysis, the ratio of the target intensity to that of the loading control was calculated for each sample and normalized to the cell line with the lowest expression. For internalization and cleavage analysis, the ratio of the intensity between total internalized toxin and cleaved toxin was calculated.

Viability of cell lines treated with immunotoxins was measured using the Cell Counting Kit-8 assay (Dojindo Molecular Technologies: CK04) as previously described (Weldon et al., 2015). PPCI [4-(guanidinomethyl)phenylacetyl-Arg-Val-Arg-4-amidinobenzylamide] was purchased from Sigma-Aldrich (#537076). Comparisons between two groups were analyzed using a paired, two-tailed t-test assuming a Gaussian distribution. Comparisons of three or more groups were analyzed using a one-way ANOVA with a Geisser-Greenhouse correction and a Tukey-Kramer multiple comparisons test using GraphPad Prism.

Protein A/G MagBeads (55 µl) (Genscript, L00277) were prepared following the manufacturer's recommendation and incubated at 4°C with mild agitation in wash buffer (20 mM Na2HPO4 [pH 7.0] and 0.15 M NaCl) with 5% dry milk powder for 2 h to block the beads and prevent non-specific binding. After 2 h, a magnetic separation rack was used to separate the beads from the supernatant. The supernatant was discarded, and the beads were resuspended in 110 µl of wash buffer with 1% dry milk powder. Samples were prepared in 0.2 ml PCR tubes using PBS supplemented with 2 mM CaCl2 to a final volume of 250 µl. A 4 nM concentration of furin and an 8 nM concentration of HB21 or HB21-LR was employed. A total of 0.25 µg of rabbit anti-PE (pAB, kindly provided by Dr Ira Pastan, NCI/NIH), mouse anti-c-myc (mAb, Genscript, A00704), and mouse anti-FLAG (mAb, Genscript, A00187) was used. The appropriate components for each sample were combined and incubated at 4°C with gentle agitation.

After approximately 2 h, 10 µl of blocked magnetics beads were added to each sample and incubated for 3 to 4 h at 4°C with gentle agitation. After incubating, samples were placed on a magnetic stand for 5-10 min and the supernatant was carefully removed using a gel loading tip. The samples were removed from the magnetic stand and the beads were washed with 200 µl of PBS supplemented with 2 mM CaCl2 and 1% dry milk powder. The supernatant was removed as described and the wash step was repeated four additional times. After removing the supernatant from the final wash, the samples were placed back on the magnetic stand for 5 min and any remaining supernatant was removed. The samples were then resuspended in 30 µl of 1X LDS Buffer (Thermo Fisher Scientific, NP0007), heated to 95°C for 5-10 min, cooled to room temperature, centrifuged to pellet the beads, and placed on the magnetic stand for 1-2 min. The supernatant was transferred into a new tube using a gel loading tip and analyzed by western blot.

The authors would like to thank Ira Pastan and the Laboratory of Molecular Biology, CCR, NCI, NIH for their support and encouragement of this work. We are also grateful for helpful comments and suggestions from Margaret Regan, Matthew Bonett, Eber Guzman-Cruz, and Dayshia Kerney.