The protease cathepsin K can debulk the cancer glycocalyx

Abstract

The cancer glycocalyx is characterized by the overexpression of large glycoconjugates, including mucins, proteoglycans, and polysialic acid, which collectively increase glycocalyx thickness and stiffness to promote tumor survival, metastasis, and progression. Previous work demonstrated that enzymatic degradation of cancer cell-surface mucins reduced tumor burden and metastasis in mouse models of breast cancer, thus validating glycocalyx remodeling as a therapeutic strategy. However, this work relied on an engineered bacterial mucin–selective protease, or mucinase, which does not degrade other bulky glycoconjugates and raises immunogenicity concerns because of its bacterial origin. A human enzyme that can degrade cell-surface mucins and other bulky glycoconjugates would address both limitations. We screened all 15 human cathepsins for their ability to degrade purified and cell-surface mucins, since multiple cathepsins have previously been shown to degrade mucins within regions of dense glycosylation. We found that cathepsin K (CTSK) uniquely degrades cell-surface mucins, proteoglycans, and polysialylated glycoproteins, and we demonstrated that CTSK reduces total glycocalyx thickness. These findings establish CTSK as a promising starting point for the development of a glycocalyx-debulking enzyme for cancer therapeutics.

Overexpression of large glycoconjugates, such as mucins, proteoglycans, and polysialic acid (polySia), is a common feature of the cancer glycocalyx (1). While these glycoconjugates vary in structure, as a collective, they enhance the thickness and stiffness of the glycocalyx, which, in turn, promotes cancer cell survival, metastasis, and disease progression (1). For example, primary tumors from patients with metastatic cancer exhibited increased expression of bulky mucins, proteoglycans, and ST8Sia2, a glycosyltransferase that builds polySia, compared with those with localized tumor growth (2). In murine breast cancer models, mucin overexpression led to increased lung metastasis (3). In addition, glycocalyx thickening has been shown to be a downstream consequence of oncogenic signaling (4). Mechanistically, glycocalyx thickening has been linked to enhanced integrin clustering, focal adhesion assembly, and downstream signaling events that promote cancer progression (2, 5).

The structures of these cancer-associated glycoconjugates underscore their effects on the physical properties of the glycocalyx. Mucin glycoproteins are characterized by a high density of O-glycosylated serine and threonine residues and extend from the cell surface as rigid, bottlebrush-like structures (6, 7). Proteoglycans are decorated with glycosaminoglycans (GAGs), such as heparan and chondroitin sulfate (CS), which are long polyanionic structures that also contribute to glycocalyx thickness (8). Similarly, polySia, a terminal modification of certain N- and O-glycans (9), has been shown to enhance glycocalyx bulkiness (10). While each of these glycoconjugate subtypes is individually associated with certain cancers, there is a strong correlation between a more general bulky glycocalyx transcriptional signature and metastatic disease (2). Thus, we are interested in developing therapeutic approaches that reduce glycocalyx bulk.

We recently described a strategy for selective cleavage of mucins from cancer cell surfaces using a cancer-targeting nanobody fused to an engineered version of the bacterial mucin–specific protease StcE (11). In preclinical models of breast cancer, this reagent reduced primary tumor burden and lung metastasis, thereby validating a glycocalyx remodeling approach to cancer treatment. However, the translation of this approach to a human therapeutic is limited by two factors. First, StcE is a bacterial enzyme that would likely be immunogenic in humans. And second, mucins are not the only contributors to the bulky cancer glycocalyx. An ideal glycocalyx-debulking enzyme would be of human origin and have broader activity on large cell-surface glycoconjugates beyond the mucin family.

Toward this end, we recently identified cathepsin D (CTSD) as a human protease capable of degrading purified mucins (12). Other cathepsins have also been implicated in mucin degradation, including cathepsin G (CTSG), cathepsin B, and cathepsin S (CTSS) (13, 14, 15). These cathepsins will cleave a protein backbone within regions of dense glycosylation but are not mucin selective. Congruent with these observations are structural studies indicating an open binding cleft, suggesting room to accommodate glycosylation (16). Based on these data, we hypothesized that the cathepsins, which span multiple mechanistic classes, might be a fruitful place to look for enzymes that act more broadly on bulky glycoconjugates.

Here, we evaluated the cleavage activity of all 15 human cathepsins against both purified and cell-surface mucins. We identified cathepsin K (CTSK) as the only cathepsin capable of degrading cell-surface mucins and debulking the cancer glycocalyx by cleaving proteoglycans and glycoprotein carriers of polySia. These results point to human CTSK as a promising starting point for the development of a therapeutic glycocalyx debulker.

Results

A subset of cathepsins degrade mucins in vitro

To begin, all 15 human cathepsins were commercially obtained (Tables 1 and S1), validated against the OMNICathepsin fluorogenic substrate (Fig. S1), and tested for their mucin degradation activity in vitro. Fluorophore-labeled MUC16—which was purified from the ovarian cancer cell line OVCAR-3 as previously described (12)—was treated with each cathepsin at pH 5, 6, and 7, and MUC16 degradation was evaluated via in-gel fluorescence (Figs. 1A and S2). The mucinases StcE and eStcE—an enzymatically detuned StcE mutant (11)—were included as positive controls of complete and partial MUC16 degradation, respectively. We selected purified MUC16 for our initial screening assay because in our prior work with CTSD, we demonstrated that purified MUC16 was resistant to standard mammalian proteases, and only proteolysis by StcE and CTSD led to observable degradation via in-gel fluorescence (12).

Table 1.

Peptidase family and endogenous tissue expression information for all 15 cathepsins tested

Name	Peptidase family	Tissue expression
Serine proteases
Cathepsin A	S10	Broadly expressed
Cathepsin G	S1	Bone marrow and neutrophils
Aspartic proteases
Cathepsin D	A1	Broadly expressed
Cathepsin E	A1	Immune cells
Cysteine proteases
Cathepsin B	C1	Broadly expressed
Cathepsin C	C1	Broadly expressed
Cathepsin F	C1	Broadly expressed
Cathepsin H	C1	Broadly expressed
Cathepsin K	C1	Osteoclasts and epithelial cells
Cathepsin L	C1	Broadly expressed
Cathepsin O	C1	Broadly expressed
Cathepsin S	C1	Antigen-presenting cells
Cathepsin V	C1	Thymus and testis
Cathepsin W	C1	Immune cells (NK and cytotoxic T cells)
Cathepsin Z	C1	Broadly expressed

Figure 1.

Multiple cathepsins degrade purified MUC16 in vitro.A, schematic describing the purification and fluorophore labeling of MUC16 for in-gel fluorescence assays (see the Experimental procedures section). B, digestion of IR800-labeled purified MUC16 with 100 nM cathepsins, StcE, or eStcE at pH 5, 6, and 7, quantified by in-gel fluorescence. Cathepsins have been labeled by their letter, for example, “A” refers to cathepsin A. Soluble MUC16 is over 3 million Da (62), which may explain why the IR800-MUC16 sits at the top of the gel, and any proteolyzed fragments result in lower molecular weight smears. Enzymes that degrade MUC16 are highlighted in pink. eStcE, enzymatically detuned StcE mutant; MUC, mucin; StcE, mucin-selective protease from Escherichia coli serotype O157:H7.

CTSD, CTSG, and CTSS degraded MUC16, consistent with our work with CTSD and reports of other mucins being degraded by CTSG and CTSS (13, 14, 15) (Fig. 1B). Cathepsins E (CTSE), F, K, L, V, and W also degraded MUC16, which, to our knowledge, is the first report of these enzymes proteolyzing an intact mucin, suggesting a broader role for cathepsins in mucin degradation. We also tested the cathepsins for proteolytic cleavage of the recombinant mucins, P-selectin glycoprotein ligand-1 (PSGL-1) and MUC1 (Fig. S3). Similar to purified MUC16, cathepsins D, E, G, K, L, and S showed proteolytic activity against recombinant PSGL-1 or recombinant MUC1, with CTSG and CTSK being active against both. For all substrates, the greatest mucin-degrading activity for each cathepsin was observed at pH 5, which is consistent with cathepsins being primarily lysosomal. Importantly, CTSG, CTSK, and CTSS retained activity against purified MUC16 even at pH 7, indicating promise for these cathepsins in extralysosomal spaces, such as the cell surface (Fig. 1B, right panel).

CTSK degrades mucins on cell surfaces across multiple cell lines

Next, we asked if the cathepsins would retain their mucin-degrading activities in the more complex cell surface environment, which would be necessary for debulking the cancer glycocalyx. K562 cells were treated with each cathepsin at pH 5, 6, or 7, and degradation of cell-surface mucins was measured by flow cytometry, as evaluated by a reduction in staining with antibodies against the mucins MUC1 and CD43 and the pan-mucin probe StcE^E447D (17) (Fig. 2A). Among all 15 cathepsins, only CTSK degraded K562 cell-surface mucins without inducing cellular toxicity, as observed by a reduction in MUC1 and StcE^E447D staining (Figs. 2, B–E and S4–S6). This activity was the strongest at pH 5 (Fig. S4) and partially preserved even at pH 7 (Fig. S6). Cathepsin W treatment led to not only a reduction in mucin staining at only pH 5 but also significantly increased cellular death. Initially, CTSE seemed to consistently lead to reduced cellular viability and mucin staining across the pH ranges. However, this was found to be due to the low pH of the CTSE enzyme buffer following activation (Table S1), and after readjusting the buffer pH, CTSE enzymatic activity showed no effect on viability or mucin staining (Fig. S7).

Figure 2.

Only cathepsin K degrades cell-surface mucins on K562s.A, schematic describing the flow cytometry assay used to evaluate the degradation of cell-surface mucins on K562 cells by cathepsins. B, cell viability of cells following enzymatic treatment. Normalized staining for (C) MUC1, (D) CD43, and (E) total mucins via StcE^E447D following enzymatic treatments for 3 h at pH 6 in Hanks' balanced salt solution (HBSS). Staining was normalized such that PBS-treated and fluorescence minus-one controls are defined as 100% and 0% staining within each replicate (n = 3–4 biologically independent replicates). Cathepsins have been labeled with their letter, for example, “A” refers to cathepsin A. CTSE was excluded because of the low pH of the CTSE activation buffer causing cellular toxicity (Fig. S7). See Figs.S4–S6 for representative flow cytometry histograms, assays performed at pH 5 and 7, and bar graphs with raw median fluorescence intensity (MFI) values. Data are shown as mean ± SD from three to four biologically independent replicates, and each dot represents a single replicate. p Values determined via one-way ANOVA corrected via Dunnett’s multiple comparison test, with a single pooled variance. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗ ≤ 0.001, and ∗∗∗∗p ≤ 0.0001. CTSE, cathepsin E; StcE^E447D, inactive point mutant of StcE used as a pan-mucin probe.

CTSK degraded cell-surface mucins less than the bacterial mucinase StcE but similarly to the less active engineered eStcE. We previously demonstrated that engineering StcE to have decreased activity (eStcE) was necessary to achieve cell-selective mucin degradation with targeting, and the more active StcE degraded mucins on off-target cells even when directly fused to the targeting moiety (11). Thus, the observed cell-surface activity of CTSK is promising for the desired cancer cell–targeted glycocalyx debulking function. Further optimization of this cell surface assay revealed a treatment condition—reduced cell density (2 million cells/ml as compared with 5 million cells/ml) and enzymatic treatments in pH 6.75 Hanks' balanced salt solution (HBSS)—in which all tested cell-surface mucins were degraded by CTSK (Fig. S8). These conditions were used for all remaining cell activity assays using the same commercial source of CTSK, which was found to be nearly 100% pure and active CTSK without any detectable contamination of other proteins (Fig. S9).

We tested CTSK with three additional cell lines—MCF10A, OVCAR-3, and H82—to better understand the generalizability of CTSK activity against cell-surface mucins (Figs. 3 and S10, A and D). These cell lines express different cell-surface mucins, as indicated by different banding patterns in the total mucin Western blot of the cell lysates (Fig. S10E). We used an engineered MCF10A cell line that was previously developed to have titratable levels of MUC1 via a doxycycline promoter (clone 1E7 from Park et al. (18)) to investigate if CTSK activity is retained in an extra-dense glycocalyx (Fig. 3B). CTSK significantly reduced levels of MUC1 on MCF10A and doxycycline-induced MCF10A (MCF10A^MUC1) cells, with greater relative cleavage observed with increased MUC1 expression (Fig. 3C). Interestingly, this activity was not observed via total mucin staining (StcE^E447D), despite clear evidence that MUC1 expression was driving levels of this staining (Fig. 3, B and D). The selected MUC1 antibody (clone 16A) binds within the proline–threonine–serine (PTS) domain of MUC1 against a repeated glycosylated peptide (19), whereas prior data support StcE^E447D binding more membrane proximally and externally from the PTS domain (20). These data are consistent with CTSK degrading MUC1 more N-terminally than the membrane-proximal StcE^E447D-binding site, which would leave StcE^E447D staining unchanged despite clear mucin degradation (Fig. S11). Furthermore, the recombinant MUC1 used in the in vitro digestions—which contains the regions N terminal to the PTS domains and only a small portion of the PTS domain—was minimally cleaved by CTSK (Fig. S3C). Together, these data support CTSK-degrading MUC1 within the PTS domains.

Figure 3.

CTSK degrades cell-surface mucins across multiple cell lines.A, schematic describing the flow cytometry assay for detecting the degradation of cell-surface mucins on H82, OVCAR3, MCF10A, and MCF10^MUC1 cells. B, median fluorescence intensity (MFI) of MUC1 and total mucins via StcE^E447D. C, normalized MFI of MUC1 staining for MCF10 ± MUC1 following enzymatic treatments for 1 h a 37 °C at pH 6.75 in Hanks' balanced salt solution (HBSS). D, normalized StcE^E447D staining for cell lines following enzymatic treatment for 1 h a 37 °C at pH 6.75. See Fig. S10, A–D for representative histograms. Data are shown as mean ± SD from two to four biologically independent replicates, and each dot represents a single replicate. P Values determined via (C) two-way ANOVA or (D) one-way ANOVA, both corrected via Tukey’s multiple comparison test, with a single pooled variance for each cell line. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗ ≤ 0.001, and ∗∗∗∗p ≤ 0.0001. CTSK, cathepsin K; MUC, mucin; StcE^E447D, inactive point mutant of StcE used as a pan-mucin probe.

CTSK treatment also significantly reduced total mucin staining on OVCAR-3 and H82 cells (Fig. 3D). None of the observed activity was present following treatment with heat-inactivated CTSK, indicating that the mucin degradation was indeed because of CTSK proteolytic activity. Overall, these data confirm that CTSK can robustly degrade cell-surface mucins across all tested cell lines.

CTSK cleaves within densely O-glycosylated mucin domains but is not mucin selective

The work presented thus far demonstrates that CTSK robustly degrades purified, recombinant, and cell-surface mucins. Because the dense glycosylation of mucins protects them from proteolytic cleavage by workhorse proteases like trypsin, we evaluated if CTSK can degrade mucins within regions of dense O-glycosylation, as we previously observed with CTSD (12). We performed glycoproteomics on peptides generated from CTSK treatment of the following substrates: recombinant PSGL-1, MUC1, and CD43, purified MUC16, the O-glycosylated nonmucin fetuin, and the non-O-glycosylated anti–human epidermal growth factor receptor 2 (HER2) nanobody. We generated cleavage motifs using only O-glycosylated peptides (Fig. 4A, left), nonmodified peptides (Fig. 4A, middle), and both O-glycosylated and nonmodified peptides (Fig. 4A, right). Across all peptides generated, CTSK had a nonspecific cleavage motif with minimal amino acid preference, except for proline or leucine in the P2 position. This is well documented for CTSK, and the P2 proline preference is unique compared with other closely related papain-like cysteine cathepsins.

Figure 4.

Cathepsin K (CTSK) tolerates glycans near the cleavage site.A, cleavage motif of CTSK was generated from mass spectrometry analysis of (left) glycopeptides, (center) nonmodified peptides, and (right) modified and nonmodified peptides generated from CTSK digestion of purified and recombinant mucins and nonmucin glycoproteins, followed by trypsin digestion (see the Experimental procedures section). The bar graphs on top of the glycopeptide cleavage motif indicate the frequency of O-glycosylation at each threonine and serine residue at that position. B, top, visualization of CTSK cleavage sites in recombinant MUC1 residues 24–47. The sialylated core-1 glycan at specific resides indicates that glycans were seen at those sites. This specific glycan was seen often in the dataset, but its depiction here is only intended to indicate glycosites, not to represent the diversity of all glycans detected at each glycosite in the dataset. Purple diamond, sialic acid; yellow circle, galactose; yellow square, N-acetylgalactosamine; and yellow circle, glycosylation site. Colored bars represent individual detected peptide sequences from CTSK cleavage only, with any shared peptides with chymotrypsin removed. Bottom, annotated spectrum for the indicated MUC1 O-glycopeptide. C, top, visualization of CTSK cleavage sites in recombinant P-selectin glycoprotein ligand-1 (PSGL-1) residues 148 to 197, represented as in (B), but this time from the tryptic + CTSK dataset, with all tryptic cleavage sites removed. Bottom, annotated spectrum for the indicated PSGL-1 O-glycopeptide. MUC, mucin.

The bars above the O-glycopeptide cleavage motif indicate the frequency of mucin-type O-glycosylation observed at each threonine or serine at that position (Fig. 4A, left). Importantly, glycans were tolerated near the cleavage site, and peptides were generated within densely glycosylated mucin domains, indicating a tolerance for mucin-type O-glycosylation (Fig. 4, B and C). However, CTSK was found to less frequently cleave mucins directly next to O-glycosites, as indicated by a low frequency of glycosylated threonines and serines at the P2–P1′ position, and deglycosylation of recombinant PSGL-1 and purified MUC16 increased proteolytic cleavage by CTSK (Fig. S12). Together, these data indicate that CTSK is not mucin selective but can degrade within regions of dense O-glycosylation—expanding its activity to proteins traditionally considered resistant to proteolysis by human enzymes.

CTSK degrades other bulky glycocalyx components and reduces overall glycocalyx thickness

Given the tolerance of CTSK for bulky glycans near the cleavage site, we next asked if CTSK could degrade other highly glycosylated proteins on the cell surface to further debulk the glycocalyx. As previously mentioned, in addition to mucins, proteoglycans with extended GAG chains and polySia-modified proteins are associated with aggressive cancer phenotypes and are known to contribute to glycocalyx bulk (1). These glycoconjugates are structurally distinct from mucins and serve as novel substrates to evaluate CTSK’s broader potential for cancer glycocalyx debulking (Fig. 5).

Figure 5.

Cathepsin K (CTSK) sheds bulky glycoconjugates across multiple cell lines. Schematic describing components of a bulky glycocalyx evaluated for CTSK activity: mucins, heparan sulfate, chondroitin sulfate, and polysialic acid. Monosaccharides are denoted according to the symbol nomenclature for glycans, as shown in the top right box (63). Sulfation of glycans is indicated above or below the glycans. CTSK cleaves the protein backbone (as indicated by the wider black line), whereas polySia-specific endosialidase (EndoNA), chondroitinase ABC, and heparinase I cleave the glycosidic linkages of polysialic acid, chondroitin sulfate, and heparan sulfate, respectively, degrading the glycans chains but not the protein backbone.

GAGs and polySia are found on many proteins, and while these extended glycan chains are not substrates of the protease CTSK, degradation of the protein carriers would shed them from the cell surface (Fig. 5). We thus evaluated the loss of heparan sulfate (HS), CS, and polySia from H82, OVCAR-3, MCF10A, and MCF10A^MUC1 cells following CTSK treatment (Figs. 6 and S13). For each assay, we evaluated the degradation of a bulky glycoconjugate on only the cell lines with high residency of that glycoconjugate (Fig. S14).

Figure 6.

Cathepsin K (CTSK) sheds bulky glycan polymers across multiple cell lines.A, schematic describing the flow cytometry assay for detecting shedding of glycan polymers from H82, OVCAR3, MCF10A, and MCF10^MUC1 cells following enzymatic treatments with CTSK, heat-inactivated CTSK (HI CTSK), StcE, heparinase, chondroitinase, and the polySia-specific endosialidase (EndoNA). Cells were stained for (B) heparan sulfate using fibroblast growth factor 2 (FGF2), which is a probe for heparan sulfate, (C) polysialic acid using anti-polySia antibody (clone 735), (D) chondroitin sulfate using anti–chondroitin sulfate antibody (clone CS-56), and (E) viability following enzymatic treatment of cells for 1 h at 37 °C at pH 6.75. Staining was (B) normalized median fluorescence intensity (MFI) to 100% buffer control and 0% secondary only or (C and D) quantified as percent positive staining because of the broad and non-normal distribution of the cell populations. See Fig. S13 for normalized staining values and representative histograms. F, change in glycocalyx thickness of YSCCC, MCF10A, and MCF10^MUC1 cells following enzymatic treatment relative to buffer control was measured using scanning angle interference microscopy. Each data point is the average of 20 to 50 individual cell measurements performed on a single day and represents an independent biological replicate. See Fig. S15 for individual cell measurements. Data are shown as mean ± SD from two to five biologically independent replicates, and each dot represents a single replicate. p Values determined via one-way or two-way ANOVA corrected via Tukey’s multiple comparison test, with a single pooled variance for each cell line. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗ ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Following CTSK treatment, HS staining was significantly reduced across all four cell lines (Fig. 6B). CTSK treatment strongly reduced polySia staining on H82 cells but not OVCAR-3 cells, indicating that the polySia carrier proteins on H82 cells are more amenable to CTSK cleavage than those on OVCAR-3 cells (Fig. 6C). Furthermore, CTSK treatment strongly shed CS from H82 cells (Fig. 6D). Heparinase ABC, chondroitinase I, and a polySia-specific endosialidase (EndoNA) (21) were used as controls for complete degradation of HS, CS, and polySia, respectively. Treatment with heat-inactivated CTSK did not induce shedding of CS, HS, or PolySia from any cell line.

Cellular toxicity was not observed with any enzymatic treatment, but CTSK consistently improved the viability of H82 cells (Fig. 6E). Interestingly, this improvement in viability was not observed following any of the other enzymatic treatments, indicating that this effect is driven by cleavage of other unevaluated cell-surface proteins or requires removal of multiple of the evaluated targets. In addition, StcE treatment led to a reduction in HS and CS staining consistent with previous reports of mucin domain–containing glycoproteins being observed to be modified with GAGs. However, CTSK treatment induced more shedding of GAGs and PolySia than StcE treatment, indicating that CTSK can debulk the glycocalyx via degradation of bulky glycoconjugates beyond just mucins.

We have so far evaluated changes in glycocalyx bulk via residency of individual bulky glycoconjugates, rather than changes in total glycocalyx thickness. We thus turned to a previously reported method for directly evaluating the thickness of the glycocalyx, scanning angle interference microscopy (SAIM) (22, 23). CTSK treatment of YSCCC cells—a cholangiocarcinoma cell line that expresses high levels of MUC1—reduced glycocalyx thickness, though not as drastically as StcE treatment (Fig. S15A). Quantification of glycocalyx thickness measurements for YSCCC and MCF10A ± MUC1 cells confirmed that CTSK treatment significantly reduced glycocalyx thickness, particularly under high MUC1 density conditions (Figs. 6F and S15, B–E).

These data indicate that CTSK is a human protease capable of debulking the cancer glycocalyx, as evidenced by a reduction in total glycocalyx thickness, cell-surface mucin cleavage, and GAG and polySia shedding across multiple cell lines.

Discussion

The work presented here is, to the best of our knowledge, the first report demonstrating that a human protease can debulk the cancer glycocalyx by degrading bulky transmembrane mucins, proteoglycans, and glycoproteins modified with polySia. Furthermore, the majority of cathepsins, spanning all cathepsin peptidase families, exhibited proteolytic cleavage against purified MUC16 in vitro, including previously unreported activity by capthepsins E, F, K, L V, and W. Of note, we have previously reported that CTSD knockout mice accumulate undegraded mucins in liver macrophages, indicating an endogenous function of CTSD in mucin catabolism (12). These data motivate further studies on the specificity, redundancy, and function of each of these cathepsins in mucin catabolism in health and disease.

Of all the tested cathepsins, only CTSK exhibited robust activity against cell-surface mucins. Furthermore, the demonstrated proteoglycan cleavage by CTSK is consistent with prior reports of CTSK degradation of aggrecan, a CS-modified proteoglycan that is a major component of cartilage (24). Collectively, these data suggest that CTSK could be a promising start to engineer a human glycocalyx-debulking therapeutic enzyme. Further work is needed to better understand why only CTSK exhibited proteolytic activity against cell-surface mucins, as compared with all the other tested cathepsins with promising in vitro activity.

Cathepsins are typically localized to endosomal and lysosomal compartments, but CTSK is released by osteoclasts to degrade collagen and elastin in the extracellular matrix during bone remodeling (25) and thus has well-characterized activity outside the cell. Furthermore, both mucins and collagen contain a high density of prolines, and mucins have an extended bottlebrush-like secondary structure that bears similarities to the polyproline helix of collagen (26, 27, 28). CTSK’s unique tolerance for proline at the P2 position and ability to cleave collagen at multiple sites within its native triple helix may underlie its robust activity against cell-surface mucins.

We had also hypothesized that the previously reported CS-mediated allosteric regulation of CTSK may rationalize its proteolytic activity against mucins on the cell surface (29, 30, 31). However, pretreatment of cells with chondroitinase prior to CTSK treatment did not affect mucin degradation (Fig. S16), suggesting that CS-mediated stabilization of CTSK was not driving mucin degradation in our cell line models.

The potential of CTSK for therapeutically debulking the cancer glycocalyx may appear to counter the large body of data linking CTSK overexpression with cancer progression (32). One mechanism underlying this connection is that degradation of the non–cell-surface tumor extracellular matrix (ECM) can release trapped growth factors as well as promote cancer cell migration, invasion, and malignancy (25, 32, 33, 34, 35, 36). Furthermore, CTSK can facilitate bone metastasis of breast and prostate cancer by promoting cancer cell invasion and bone remodeling (37, 38, 39, 40). CTSK inhibitors have thus been evaluated as cancer therapeutics in preclinical and clinical studies (41, 42, 43, 44, 45), but an intriguing possibility is the notion of engineering CTSK to target it away from the ECM and toward the cancer glycocalyx. Our previous work with the targeted bacterial mucinase demonstrated cell type–specific mucin degradation on the targeted tumor cells in vivo and not on intratumoral immune cells (11). This work suggests it may be possible to target CTSK to prevent CTSK-mediated ECM degradation. Development of a targeted CTSK would require engineering of the enzyme, as CTSK is not mucin specific and does degrade some other cell-surface receptors (Figs. 4 and S17), and a targeting moiety would be essential. However, our work presented here indicates the promise of CTSK as a starting point for a human enzyme that can therapeutically debulk the cancer glycocalyx.

Experimental procedures

Statistics and reproducibility

Biological replicates refer to experiments performed independently on different days. All statistical methods were performed using Prism 10 (GraphPad Software, Inc). Statistical tests are reported in each figure caption.

Cell lines

H82 (American Type Culture Collection [ATCC] HTB-175), HeLa (ATCC CCL-2), and K562 (ATCC CCL-243) cells were ordered from ATCC. MCF10A cells (1E7 clone) were generated as previously reported (18). OVCAR-3 cells were a gift from the Pedram laboratory. All cells were routinely tested to be negative for mycoplasma.

Cell culture

Cells were maintained at 37 °C and 5% CO₂. MCF10A ± MUC1 cells were cultured in Dulbecco's modified Eagle's medium:F12 supplemented with 5% New Zealand horse serum (Thermo Fisher Scientific), 20 ng/ml epidermal growth factor (Peprotech), 0.5 μg/ml hydrocortisone (MilliporeSigma), 100 ng/ml cholera toxin (MilliporeSigma), 10 μg/ml insulin (MilliporeSigma) and 1% penicillin/streptomycin (P/S). K562 and H82 cells were cultured in RPMI supplemented with 10% heat-inactivated fetal bovine serum (FBS; Thermo Fisher Scientific) and 1% P/S. OVCAR-3 cells were cultured in RPMI supplemented with 10% heat-inactivated FBS, 0.01 mg/ml bovine insulin (Sigma–Aldrich), and 1% P/S. Cells were counted using a Countess II FL automated cell counter (Thermo Fisher Scientific) following the manufacturer’s recommendations.

OMNICathepsin fluorogenic assay

All cathepsins were dissolved/aliquoted and stored per the manufacturer's recommendations, avoiding repeated freeze and thaws. CTSE was activated as in Table S1. All other cathepsins were used as supplied by the manufacturers. All cathepsins were diluted to 1 μM in PBS, other than CTSK. The reactions consisted of 3.85 pmol of each enzyme (StcE and eStcE were prepared as previously published (11) in 20 μM OMNICathepsin fluorogenic substrate (Enzo; BML-P139), 100 mM sodium acetate, 5 mM EDTA, pH 5.5. Reactions were initiated via the addition of 100 μl of OMNICathepsin in buffer to 3.85 μl cathepsins (5 μl of 0.77 μM CTSK). The standard curve was generated via serial dilutions of AMC standard (Enzo; BML-KI107) in 100 mM sodium acetate, 5 mM EDTA, pH 5.5. Reactions proceeded for 30 min at 37 °C, and the fluorescence signal (excitation: 365 nm; emission: 530 nm) was measured using a SpectraMax i3x microplate reader.

Purification and labeling of IR800 MUC16

MUC16 was purified from OVCAR-3 cell supernatant as previously reported (12). Following purification via size-exclusion chromatography using a HiPrep Sephacryl 16/60 S-500 HR column (Cytiva Life Sciences) in PBS, FPLC fractions were filtered via Acrodisc 0.2 μm 13 mm Supor membrane filters (Cytiva; 4602). Concentrations of protein within each FPLC fraction were determined via Pierce Quantitative Peptide Assay (Thermo Fisher Scientific) according to the manufacturer's recommendations. FPLC fractions were determined to be pure for MUC16 via anti-MUC16 Western blot. FPLC fractions (10–50 μM protein) were subjected to SDS-PAGE using 3% to 8% Criterion XT Tris-Acetate protein gels (Bio-Rad) and run in XT-Tricine (Bio-Rad) at 150 V for 80 min. The gel was transferred to a 0.2-μm nitrocellulose membrane using the Trans-Blot Turbo transfer system (Bio-Rad) at 2.5 A for 30 min. Total protein was quantified using REVERT stain (LI-COR Biosciences). The membrane was blocked in 5% (w/v) nonfat dry milk in PBS with 0.1% v/v Tween-20 (PBST) at room temperature for 1 h and incubated overnight with 2.7 μg/ml anti-MUC16 clone x75 (Novus; NB600-1468) in blocking buffer at 4 °C. Blots were washed three times for 5 min with PBST at room temperature. IRDye 800CW Goat anti-mouse IgG Secondary Antibody (LI-COR Biosciences) was used according to the manufacturer’s recommendations. Blots were imaged using an Odyssey CLx near-infrared fluorescence imaging system (LI-COR Biosciences). FPLC fractions with high MUC16 staining and low total protein staining were combined, and half of the purified MUC16 was frozen at −20 °C. IRDye 800 CW NHS Ester (LI-COR Biosciences) in dimethyl sulfoxide was added to the remaining purified MUC16 (27:1 dye:MUC16M ratio) and incubated at room temperature for 2 h, protected from light, turning end over end. The free dye was removed using three successive 40 K 0.5 ml Zeba Spin Desalting Columns (Thermo Fisher Scientific). IR800 MUC16 was aliquoted and frozen at −20 °C.

IR800 MUC16 in vitro digestions

CTSE was activated as in Table S1. All other cathepsins were used as supplied by the manufacturers. All cathepsins were diluted to 1 μM in PBS, other than for CTSK. IR800 MUC16 (300 nM) and cathepsins/mucinases (150 nM) were incubated for 16 h at 37 °C in a total of 8.34 μl of 1:1 PBS:100 nM sodium acetate with a final pH of 5, 6, or 7. Buffer pHs were calculated via a pH curve of adding 1 N HCl to 1:1 PBS:100 nM sodium acetate. 4X NuPAGE LDS Sample Buffer (Thermo Fisher Scientific) with 250 mM DTT (Thermo Fisher Scientific) was added to a final concentration of 1X LDS to each sample, and samples were boiled at 95 °C for 5 min. Samples were loaded onto a 3% to 8% CriterionXT Tris–Acetate protein gel (Bio-Rad) and run in XT-Tricine (Bio-Rad) at 150 V for 100 min. Gels were imaged using an Odyssey CLx near-infrared fluorescence imaging system.

Recombinant mucin in vitro digestions

Recombinant PSGL-1 (1 μg; Sino Biological; 13863-H08H) or MUC1 (Sino Biological; 12123-H05H) was incubated with each cathepsin at a 1:50 enzyme:substrate molar ratio for 16 h at 37 °C in 1:1 PBS:100 nM sodium acetate with a final pH of 5, 6, or 7 (buffer pHed as above). Samples were subject to SDS-PAGE using 4% to 12% Criterion XT Bis–Tris protein gel (Bio-Rad) and run in XT-MOPS (Bio-Rad) at 180 V for 1 h. Total protein was visualized with AcquaStain protein gel stain (Bulldog-Bio).

K562 cell-surface mucin degradation with all cathepsins

K562 cells were collected and resuspended at 5.56 × 10⁶ cells/ml in HBSS (pH 5, 6, or 7). Buffer pHs were calculated via a pH curve of adding 1 N HCl to 90% HBSS and 10% PBS without cells. K562 cells (90 μl) were mixed with 10 μl of 1 μM cathepsins (13 μl of CTSK), mucinases, or PBS per well in a 96-well V-bottom tissue culture plate and incubated for 3 h at 37 °C and 5% CO₂. Cells were washed twice with cold fluorescence-activated cell sorting (FACS) buffer (0.5% bovine serum albumin in PBS) and then stained for 30 min on ice protected from light with 2.5 μg/ml FITC anti-CD43 (clone MEM-59; BioLegend), 10 μg/ml PE/Cy anti-MUC1 (clone 16A; BioLegend), and 5 μg/ml AF647 StcEE447D (expressed, labeled, and verified in house as previously reported (17)) in cold FACS buffers supplemented with 0.1% benzonase (Sigma–Aldrich). Cells were washed twice with cold FACS buffer with 2 mM EDTA and stained with 1 μM Sytox Blue Dead Cell Stain (Thermo Fisher Scientific) in cold FACS buffer with 2 mM EDTA for 5 to 30 min. Cells were resuspended in 200 μl cold FACS buffer with 2 mM EDTA, and a minimum of 30,000 live, single cells up to 150 μl of sample were analyzed using a MACSQuant Analyzer 10 flow cytometer (Miltenyi Biotec). Samples were compensated using single-stained controls and gated in FlowJo, version 10.0. The median fluorescence intensity for each mucin stain (only including samples with at least 5000 single, live cells, given the limitations of the instrument) was normalized within each biological replicate with fluorescence minus one, and PBS-treated controls were used to define 0% and 100% cell-surface staining, respectively.

K562 flow cytometry experiments with CTSE

CTSE was activated as in Table S1 and diluted to 1 μM in PBS for 100 nM treatments and 0.5 μM in PBS for 50 nM treatments. For heat inactivation, 0.5 to 1 μM activated CTSE was heated at 80 °C for 20 min and then returned to ice until ready to use. CTSE buffer was prepared by mixing equal parts of 20 mM MES, 150 mM NaCl, pH 6.5, with 0.1 M sodium acetate, 0.5 M NaCl, pH 3.5. K562 cells growing in log phase were collected and resuspended at 2.22 × 10⁶ cells/ml in HBSS or RPMI (supplemented with 10% heat-inactivated FBS and 1% P/S) with pH 6, 6.75, or 7.5. Buffer pHs were calculated via the pH curve of adding 1 N HCl to 90% HBSS or complete RPMI and 10% PBS without cells. K562 cells (90 μl) were mixed with 10 μl of 0.5 to 1 μM (heat inactivated) CTSE, 10 μl PBS, or 10 μl CTSE buffer per well in a 96-well V-bottom tissue culture plate and incubated for 3 h at 37 °C and 5% CO₂. Cells were washed and stained with 2.5 μg/mL FITC anti-CD43 (clone MEM-59), 10 μg/ml PE/Cy anti-MUC1 (clone 16A), and 5 μg/mL AF647 StcE^E447D and Sytox Blue as above. Cells were resuspended in 200 μl cold FACS buffer with 2 mM EDTA, and a minimum of 15,000 live, single cells up to 150 μl of sample were analyzed using a MACSQuant Analyzer 10 flow cytometer (Miltenyi Biotec). Samples were compensated using single-stained controls and gated in FlowJo version 10.0. The median fluorescence intensity for each mucin stain was normalized within each biological replicate with Sytox Blue only and PBS-treated controls used to define 0% and 100% cell-surface staining, respectively.

Top-down mass spectrometry of commercial CTSK

Protein samples for top–down mass spectrometry (MS) were prepared using a modified methanol–chloroform–water precipitation and resolubilization method (46, 47). Briefly, 300 μl of chilled LC–MS-grade water (4 °C) was added to 100 μl of protein solution, followed by 400 μl of cold methanol (−20 °C). The protein solution was vortexed for 30 s, followed by the addition of 100 μl of cold chloroform (−20 °C) and an additional 30 s of vortexing to yield a biphasic mixture. The sample was then centrifuged for 10 min at 18,000g at 4 °C to clarify the mixture, after which the top layer of the solution was discarded without disturbing the protein pellet. Then, 400 μl of cold methanol (−20 °C) was added to the sample, gently vortexed, and centrifuged for 10 min at 18,000g at 4 °C. After the supernatant was discarded, the protein pellet was resolubilized with 3 μl of 80% formic acid (−20 °C) and diluted to 1% formic acid with 80:20 water:acetonitrile. Analysis was performed using an Agilent 1260 Infinity II high-performance liquid chromatography system coupled to an Agilent 6230 ToF LC–MS system (Agilent Technologies). The samples were injected onto an Agilent PLRP-S column (2.1 × 50 mm, 5 μm particle size, 1000 Å pore size) using a gradient of 10% to 90% mobile phase B (0–5 min, 10% B; 5–15 min, 20%–65% B; 15–18 min, 65%–90% B, 18–22 min, 90% B; 22–25 min, 10% B; mobile phase A set to 0.2% formic acid in water; mobile phase B set to 0.2% formic acid in acetonitrile). The flow rate was set to 200 μL min/min with a column temperature of 60 °C. Mass spectra were taken at a scan rate of 1 Hz over a 200 to 3200 m/z scan range with the time of flight set to extended dynamic range. A dual Agilent jet stream high-sensitivity ion source was used for nebulization, and the mass spectrometer was configured with the following instrument parameters: nebulizer (40  psi), sheath gas flow (12 L/min), sheath gas temperature (400 °C), gas temperature (275 °C), drying gas (12 L/min), VCap (3000 V), nozzle voltage (2000 V), fragmentor (250 V), skimmer (65 V), and Oct 1 RF Vpp (750 V). Mass spectra were output from the MassHunter (Agilent Technologies) software and analyzed using MASH Native (48) and UniDec (49).

Flow cytometry of mucins, HER2, integrins, and FOLR1 cleavage, and glycan shedding on H82, MCF10A ± MUC1, OVCAR-3, HER2+ K562, and HeLa cells

MCF10A cells were treated with 1 μg/ml doxycycline hyclate (Santa Cruz Biotechnology) in culture media for 24 h at 37 °C to induce expression of GFP-MUC1ΔCT (referred to as MCF10A^MUC1 cells). CTSK buffer (50 mM sodium acetate, 50 mM NaCl, 0.5 mM EDTA, and 5 mM DTT) was prepared, filtered, and stored at 4 °C. CTSK was heat inactivated at 95 °C for 20 min and then kept at room temperature until ready to use. MCF10A, HeLa, and OVCAR-3 cells growing in log phase were lifted with 0.25% trypsin–EDTA. K562 cells growing in log phase were collected, and all cell lines were resuspended at 2.22 × 10⁶ cells/ml in HBSS (pH 6.75). Buffer pHs were calculated via a pH curve of adding 1 N HCl to 90% HBSS and 10% CTSK buffer without cells. For each treatment in a 96-well ultra–low attachment flat bottom plate (Corning), 90 μl of cells were mixed with 10 μl of one of the following: CTSK buffer, CTSK directly from the manufacturer, heat-inactivated CTSK, StcE diluted to 770 nM in CTSK buffer, Bacteroides Heparinase I (New England Biolabs) diluted to 1200 U/ml in CTSK buffer, GFP-EndoNA (expressed from addgene plasmid #174724 as previously reported (21)) diluted to 100 μg/ml in CTSK buffer, or Chondroitinase ABC from Proteus vulgaris (Sigma–Aldrich) diluted to ≥1 U/ml in CTSK buffer. Cells were incubated for 1 h at 37 °C with 5% CO₂. Cells were transferred to a 96-well V-bottom plate and washed twice with cold FACS buffer. Cells were stained with primary antibodies (Table S2) for 30 min on ice protected from light in cold FACS buffer supplemented with 0.1% benzonase (Sigma–Aldrich). Cells were washed twice with cold FACS buffer and stained with secondary antibodies (Table S2) for 30 min on ice protected from light in cold FACS buffer. Cells were washed twice with cold FACS buffer supplemented with 2 mM EDTA and resuspended in 200 μl of SYTOX viability stain (Table S2) in cold FACS buffer supplemented with 2 mM EDTA. A minimum of 20,000 live, single cells up to 180 μl of sample were analyzed using a MACSQuant Analyzer 10 flow cytometer (Miltenyi Biotec). Samples were compensated using single-stained controls and gated in FlowJo version 10.0. The median fluorescence intensity for each stain (only including samples with at least 5000 single, live cells, given the limitations of the instrument) was normalized within each biological replicate, with buffer-treated cells and Sytox-only controls used to define 0% and 100% cell-surface staining, respectively. For each stain, within each biological replicate, a positive staining gate was drawn (only including samples with at least 2500 single, live cells, given the limitations of the instrument) such that 80% of the buffer-treated cells were positively stained.

O-glycoproteomics and cleavage specificity

Recombinant PSGL-1 (10 μg; Sino Biological; 13863-H08H), recombinant MUC1 (Sino Biological; 12123-H05H), recombinant CD43 (R&D Systems; 9680-CD-050), fetuin (Promega; V4961), and purified MUC16 (purified from OVCAR-3 as above) were mixed with CTSK or PBS control at a 1:50 CTSK:substrate molar ratio in 1:1 PBS:100 mM sodium acetate with a final pH of 5 (buffer pHed as above). Samples were incubated at 37 °C for 16 h. Samples were processed and analyzed by MS as previously reported (50). Loading and gradient elution were performed using the same conditions as described above, and an Orbitrap Fusion was used for MS analysis. Again, precursors were ionized using an EASY-Spray ionization source (Thermo Fisher Scientific) held at +2.2 kV compared with ground, the column was held at 40 °C, and the inlet capillary temperature was held at 275 °C. Survey scans of peptide precursors were collected in the Orbitrap from 500 to 1800 m/z with an automatic gain control (AGC) target of 400,000, a maximum injection time of 50 ms, an RF lens at 60%, and a resolution of 120,000 at 200 m/z. Monoisotopic precursor selection was enabled for peptide isotopic distributions, precursors of z = 2 to 10 were selected for data-dependent MS/MS scans for 3 s of cycle time, and dynamic exclusion was set to 45 s with a ±10 ppm window set around the precursor monoisotope. Precursor priorities were set to favor the highest charge state and the lowest m/z precursor ions, and an isolation window of 2 m/z was used to select precursor ions with the quadrupole.

All methods used were product dependent, where the presence of at least two of nine oxonium ions (126.055, 138.0549, 144.0655, 168.0654, 186.076, 204.0865, 274.0921, 292.1027, and 366.1395) in a “scouting” higher-energy collisional dissociation (HCD) MS/MS scan triggered acquisition of a second MS/MS can help improve the specificity of triggering on glycopeptide precursor ions (51). The “scout HCD” scan had an automated scan range determination and a first mass of 100 Th, a normalized collision energy of 36, an AGC target value of 50,000, a maximum injection time of 60 ms, and a resolution of 30,000 at 200 m/z. If at least two of the nine listed oxonium ions were present in the scout HCD scan within a ±10 ppm tolerance and were among the 20 most intense peaks, a second MS/MS scan was triggered. All triggered electron-transfer and higher-energy collision dissociation (EThcD) MS/MS scans used a fixed scan range of 120 to 4000 m/z (this extended mass range has been shown to improve glycopeptide characterization (52)), an AGC target of 100,000 ions, a maximum injection time of 200 ms, and a resolution of 30,000 at 200 m/z. Calibrated charge-dependent parameters for calculating reagent AGC targets and ion–ion reaction times were enabled for EThcD scans, and supplemental activation of 25 normalized collision energy was used (53).

Data analysis

All raw data were searched using O-Pair Search implemented in MetaMorpheus, which is available at https://github.com/smith-chem-wisc/MetaMorpheus (54). All files were searched in one batch so that the false discovery rate calculations were performed on the entire dataset. The “Glyco Search” option was selected, where the O-glycopeptide search feature was enabled and the Oglycan.gdb glycan database was selected, representing 12 common O-glycans: HexNAc(1), HexNAc(1)Hex(1), HexNAc(1)NeuAc(1), HexNAc(2)Hex(1), HexNAc(1)Hex(1)NeuAc(1), HexNAc(1)Hex(1)NeuAc(2), HexNAc(2)Hex(2)NeuAc(1), HexNAc(2)Hex(2)NeuAc(2), HexNAc(2)Hex(2), HexNAc(2)Hex(1)NeuAc(1), HexNAc(2)Hex(2)Fuc(1)NeuAc(1), and HexNAc(2)Hex(2)Fuc(1)NeuAc(2). The “Keep top N candidates” feature was set to 50, and the data type was set as HCD with child scan dissociation set as EThcD. The “Maximum OGlycan Allowed” setting was set to 5, where this number represents both the maximum number of O-glycan modifications that could occur on a glycopeptide candidate and the number of times each O-glycan could occur per peptide. Under Search Parameters, both “Use Provided Precursor” and “Deconvolute Precursors” were checked. Peak trimming was not enabled, and the top N peaks and minimum ratio were set to 1000 and 0.01, respectively. In silico digestion parameters were set to generate decoy proteins using reversed sequences, and the initiator methionine feature was set to “Variable.” The maximum modification isoforms allowed were 1024, and nonspecific digestion was enabled for peptides ranging from 5 to 25 residues. Precursor and product mass tolerances were 10 and 20 ppm, respectively, and the minimum score allowed was 3. Modifications were set as Carbamidomethyl on C as fixed and Oxidation on M and Deamidation on N as variable. Note, O-Pair Search returns a single identification representing two spectra, both an HCD and an EThcD spectrum. Identifications are made using the HCD spectrum, and the associated EThcD spectrum is used to localize O-glycosites.

The oglyco.psmtsv and single.psmtsv result files were used for all data processings as the O-glycopeptide and standard peptide identifications, respectively. For both O-glycopeptide and standard peptides, identifications were filtered to include only target matches (T) and identifications with a q value <0.01. O-glycopeptide identifications were filtered to include only level 1 or level 1b identifications, as described in the O-Pair Search reference, which includes only identifications with confident and unambiguous O-glycosite localization. To get cleavage motifs, identifications that were in common between the “trypsin-only” files were removed to retain identifications that resulted only from CTSK-derived cleavage. Note, this means that some semitryptic peptides remain from the sequential digestion data, where the nontryptic terminus is attributed to CTSK cleavage. The percentage of O-glycosylated serine and threonine residues was determined by counting the number of modified residues at a given position relative to the total number of residues. Serine and threonine data were summed, so this is an aggregate value for both residues. Calculations were performed in Microsoft Excel or in Visual Studio 2017 using C#, and data were graphed using OriginPro 2018. WebLogo was used for Logo plot generation (https://weblogo.berkeley.edu/logo.cgi) (55). To make this plot, residues +5 and −5 from the cleavage site were used. Spectral annotation was aided by the Interactive Peptide Spectral Annotator (http://www.interactivepeptide-spectralannotator.com) (56). Data are available at the MassIVE Repository with accession MSV000098227 (57). Reviewer login information is MSV000098227_reviewer with password CTSK.

CTSK digestion of deglycosylated mucins

Purified recombinant PSGL-1 (Sino Biological; 13863-H08H) was labeled with IRDye 800CW NHS Ester (LI-COR Biosciences) following the manufacturer’s recommendations. IR800 recombinant PSGL-1 and IR800 purified MUC16 were deglycosylated using protein deglycosylation mix II (New England Biolabs). Samples were incubated at room temperature for 30 min and then at 37 °C for 16 h. Substrates and enzymes were combined and incubated for 16 h at 37 °C in a total of 12 μl 1:1 mucin/enzyme:100 nM sodium acetate with a final pH of 5. Buffer pHs were calculated via a pH curve of adding 1 N HCl to 1:1 PBS:100 nM sodium acetate. Samples were subject to SDS-PAGE as above (see in vitro digestions).

Scanning angle interference microscopy

Silicon wafers with approximately 2000 nm of thermal oxide (Addison Engineering) were diced into 7 mm × 7 mm chips, and the oxide layer thickness of each chip was measured using a FilMetrics F50-EXR. Prior to use, the chips were cleaned with a 1:1 mixture of methanol and HCl for 15 min, followed by a 1-min treatment in a plasma cleaner (Harrick Plasma, PDC-001). The cleaned chips were then incubated with 4% (v/v) (3-mercaptopropyl)trimethoxysilane in absolute ethanol for 30 min at room temperature. After washing with absolute ethanol, they were further incubated with 4 mM 4-maleimidobutyric acid N-hydroxysuccinimide ester in absolute ethanol for 30 min and subsequently rinsed with PBS. The functionalized chips were then incubated overnight at 4 °C with 50 μg/ml human plasma fibronectin in PBS for conjugation. 1E7 cells, a previously validated MCF10A-derived model with doxycycline-inducible MUC1 expression, and YSCCC cells, a human cholangiocarcinoma cell line with high endogenous MUC1 expression, were seeded onto the chips at a density of 2 × 10⁴ cells/cm² in full culture medium (18). 1E7 cells were treated with 1 μg/ml doxycycline for 24 h to induce MUC1 expression. To enable SAIM imaging, both cell lines were further engineered to express the Bzip construct at the cell membrane, facilitating selective membrane staining through engineered leucine zipper interactions, as previously described (22, 58, 59). Cells were then treated with either 1:10 diluted CTSK enzyme in CTSK buffer (50 mM sodium acetate, pH 5.5, containing 50 mM sodium chloride, 0.5 mM EDTA, and 5 mM DTT) or 100 nM StcE enzyme (in full cell culture medium) for 1 h at 37 °C. After washing the cells twice with full cell culture medium, they were labeled with Azip dye conjugated to AF488 or AF647 in phenol red–free culture medium for 10 min at 37 °C. Following labeling, samples were rinsed three times with phenol red–free culture medium, inverted into a 35 mm glass-bottom imaging dish containing phenol red–free culture medium, and imaged at 37 °C. Details regarding the SAIM optical setup, sample preparation, image acquisition, and data analysis have been described previously (18, 60, 61).

Western blots

H82, K562 cells were spun down at 300g for 5 min, OVCAR3, MCF10A, and MCF10A^MUC1 cells were lifted with TrypLE, then spun down at 300g for 5 min. Cell pellets were washed twice with cold PBS and lysed in radioimmunoprecipitation assay buffer (Thermo Scientific, 89901) supplemented with 100× EDTA-free protease inhibitor cocktail (Thermo Scientific, 78425) on ice for 45 min with gentle rocking. The crude lysates were spun down at 18,000g for 10 min at 4 °C, with only the clarified portion collected. Clarified lysate protein concentration was then determined by bicinchoninic acid protein assay. Each cell lysate (30 μg) was loaded onto a 4% to 12% Criterion XT Bis–Tris precast gel and run in XT-MOPS at 180 V for 90 min. The gel was transferred to a 0.2-μm nitrocellulose membrane using the Trans-Blot Turbo Transfer System (Bio-Rad) at 2.5 A for 15 min with ice-cold buffer. The transferred membrane was blocked with 1x Carbo-free Blocking Solution (Vector Laboratories), supplemented with 0.1% Tween-20, for 1 h at room temperature. Mucin and MUC1 staining was performed with 15 μg/ml biotin-StcE^E447D and 2 μg/ml anti-MUC1 antibody (Novus Bio, clone 967) in PBST for 1 h at room temperature, followed by 3 × 10 min wash with PBS-T. Then, the membrane was incubated with IRDye 800CW Streptavidin and 680RD Goat anti-Mouse IgG (LI-COR Biosciences) according to the manufacturer's recommendations, with 3 × 10 min wash with PBS-T before imaging.

Data availability

Glycoproteomics data are available at the MassIVE Repository with accession MSV000098227. All other data are contained in this article.

Supporting information

This article contains supporting information (19, 20).

Conflict of interest

A patent application relating to the use of targeted cathepsin therapeutics has been filed by Stanford University (docket number STAN-2144WO). C. R. B. is a cofounder of Lycia Therapeutics, Palleon Pharmaceuticals, Enable Biosciences, Redwood Biosciences (a subsidiary of Catalent), Grace Science LLC, InterVenn Bio, TwoStep Therapeutics, Euler Biologics, ResNovas Therapeutics, and Firefly Bio. All other authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Robert Haltiwanger