Inhibition of Carbonic Anhydrase using SLC-149: Support for a Non-catalytic Function of CAIX in Breast Cancer

Abstract

Carbonic anhydrase IX (CAIX) is considered a target for therapeutic intervention in solid tumors. In this study, the efficacy of the inhibitor, SLC-149 (4-(3-(2,4-difluorophenyl)-oxoimidazolidin-1-yl)benzenesulfonamide), is evaluated on CAIX and a CAIX-mimic. We show that SLC-149 is a better inhibitor than Acetazolamide against CAIX. Binding of SLC-149 thermally stabilizes CAIX-mimic at lower concentrations compared to CAII. Structural examinations of SLC-149 bound to CAIX-mimic and CAII explain binding preferences. In cell culture, SLC-149 is a more effective inhibitor of CAIX activity in a triple negative breast cancer cell line than previously studied sulfonamide inhibitors. SLC-149 is also a better inhibitor of activity in cells expressing CAIX versus CAXII. However, SLC-149 has little effect on cytotoxicity, and high concentrations are required to inhibit cell growth, migration, and invasion. These data support the hypothesis that CAIX activity, shown to be important in regulating extracellular pH, does not underlie its ability to control cell growth.

INTRODUCTION

The zinc metalloenzyme Carbonic Anhydrase IX (CAIX) is a homodimeric, type I transmembrane protein with its catalytic domain facing the extracellular environment. ^1-3 This isoform, like other members of the α-CA family, catalyzes the reversible hydration of CO₂ to HCO₃⁻. CAIX expression in normal tissue was originally thought to be limited to epithelial cells that line the gastrointestinal tract. ^{4, 5} The human protein atlas (proteinatlas.org) now shows antibody-validated expression in the bile ducts of the liver and glandular cells of the gallbladder. However, expression of CAIX is induced in many solid tumors, including breast cancer, as the gene that encodes for CAIX (CA9) is under the control of the hypoxia-inducible factor 1 (HIF1) transcription factor. ⁶ In fact, CAIX is a marker for hypoxic regions of breast tumors, ⁷ an indicator of poor patient prognosis ^8-10, associated with estrogen receptor (ER) negative tumor tissue, ^{8, 11} and contributes to the growth advantage of cancer cells. ^11-13 Like CAIX, CAXII is a transmembrane CA isoform with an extracellular catalytic domain. It is unique from CAIX as it lacks the proteoglycan-like (PG) domain and its expression is associated with estrogen/progesterone positive (ER⁺) breast cancers and is under regulation through estrogen action. ^{14, 15} In this setting, CAXII is a predictor of a more positive patient outcome, ^{7, 16} which suggests that targeting the activity of this CA isoform, in contrast to CAIX, may be detrimental.

Nearly 100 years ago, Warburg showed cancer cells switch their utilization of glucose from aerobic metabolism (which provides energy from oxidative phosphorylation in the mitochondria) to anaerobic glycolysis in the cytoplasm as their main source of energy (which forces higher glycolytic rates).^{17, 18} This is driven by hypoxia, which leads to increased production of lactic acid and thus acidification of the tumor microenvironment. Under these conditions, extracellular pH (pHe) stabilizes to ~6.8, which leads to the death of neighboring normal cells while cancer cells survive, colonize, and grow in the host tissue. ¹⁹ Ultimately, this glycolytic switch persists even in the presence of adequate oxygen, which suggests permanent changes to the genome. This condition, also discovered by Warburg, is called “aerobic glycolysis” and is considered a cancer hallmark. ²⁰ Current literature supports a role for CAIX in maintaining this reduced pHe in the microenvironment ^21-24 by exhibiting stability ²⁵ and high activity ^{21, 26} at low pH. This acidic environment provides a barrier against chemotherapeutic drugs and radiation therapy. ²⁷ Additionally, the acidic pHe leads to activation of integrin receptors ²⁸ and extracellular metalloproteases, ²⁹ both of which facilitate the migration of cancer cells to secondary sites.

CAIX is under consideration, by both academic and pharmaceutical entities, as a potential target for intervention in breast cancer (see ³⁰ for review). Girentuximab (G250) was one of the first CAIX targeted inhibitors to be tested in cancer patients in a clinical trial setting. ³¹ G250 is a monoclonal antibody that binds to the catalytic domain of CAIX and inhibits its activity. However, this monoclonal antibody did not achieve its primary endpoint and therefore failed to enter the clinic. ³² Another monoclonal antibody, M75, is used for CAIX detection via binding to the PG domain. ³³ Many small molecule inhibitors have also been designed to selectively target CAIX. For instance, NICOX S.A. is a CAIX and CAXII patented inhibitor. ³⁴ Other CAIX-specific inhibitors have been previously developed, such as imidazole-substituted benzene sulfonamides, and their application in cancer cells has been described by Mboge et al. ¹³

One of the more successful small molecular inhibitors SLC-0111, developed by SignalChem Lifesciences Corporation (SLC), is a ureido-substituted benzene sulfonamide that is in clinical trials for the treatment of CAIX expressing tumors (https://clinicaltrials.gov/ct2/show/NCT022158500). The Ki values of this compound for CAIX and CAII are 45 nM and 960 nM, respectively. ³⁵ Recently, SLC synthesized imidazole-related derivatives of SLC-0111 and plans to develop one of these inhibitors, SLC-149, as a more potent CAIX specific inhibitor. SLC-149 is a patented ³⁶ imidazole-substituted benzene sulfonamide (Figure 1), which is currently in the pre-clinical stage to establish treatment efficacy.

Figure 1: SLC-149, 4-(3-(2,4-difluorophenyl)-2-oxoimidazolidin-1-yl)benzenesulfonamide.

Herein, the effect of this inhibitor on the catalytic activity of a recombinant CAIX is evaluated, based on stopped flow kinetics data that compared the Ki values of SLC-149 for CAII and CAIX. These data revealed similar relationships between the IC₅₀ and Ki values of SLC-149 for CAIX and CAII. An esterase-based kinetics assay was also used to evaluate the specificity and inhibitory constants of SLC-149 against CAII and CAIX-mimic. The esterase data showed a similar trend in IC₅₀ values between CAII and CAIX-mimic with specificity toward CAIX-mimic. CAIX-mimic is an engineered CAIX mimetic constructed with a modified sequence of CAII to mimic the CAIX active site. This CAIX-mimic is widely used as a replacement to recombinant CAIX as it is more soluble in aqueous solutions. ^{37, 38} Using differential scanning calorimetry, it was shown that a ~100-fold higher concentration of SLC-149 was required to induce a melting temperature shift (Tm) in CA II relative to CAIX-mimic, which infers preferential binding of SLC-149 to CAIX. Crystallography and structural modeling were utilized to understand molecular interactions within the CA catalytic pocket and develop structure-activity relationships for CAII, CAIX, and CAXII, and infer the importance of residue 131. A triple negative breast cancer cell line that expresses CAIX on its cell surface and an estrogen receptor positive breast cancer cell line that expresses only CAXII were used to measure CAIX and CAXII activity in cell culture. This revealed that the Ki values were 100 times lower for CAIX than CAXII despite similar affinities for purified recombinant proteins. However, the ability of SLC-149 to block cell growth was minimal and at high concentrations in both of these lines. Additionally, there was no effect on migration or invasion in cells expressing CAIX. This adds to the evidence ^{13, 39-42} that a non-catalytic function of CAIX may regulate its ability to block cell and tumor growth despite its capacity to control pHe through its catalytic activity. ^21-24

RESULTS

Binding of SLC-149 to CA II and CAIX

The Ki values for SLC-149 were screened using stopped flow against all active human CA isoforms and compared to Acetazolamide (AZM, a standard clinically used CA drug) (Table 1). These data revealed that SLC-149 was a low to sub-nM inhibitor against CAII and CAIX, with a 10- and 5-fold concentrations. These binding concentrations were significantly lower than that of CAII, where between 0.1 and affinity over AZM, respectively. In addition, the screen showed SLC-149 had a ~4-fold selectivity of CAII over CAIX (Table 1). To evaluate the efficacy of SLC-149 for CAIX-mimic, esterase activity assays were conducted for comparison against CAII (Table 2). These data showed sub-micromolar inhibition of both CAIX-mimic (0.23 ± .01 μM) and CAII (0.78 ± .02 μM). It should be noted that the stopped flow data are a better indicator of inhibition of CA activity than the esterase assay. That said, the esterase activity did confirm that CAIX-mimic was both active and inhibited by SLC-149. Despite multiple attempts to express the CAXII catalytic domain, these efforts have failed and thus we lack esterase activity.

Table 1.

Inhibition constants of SLC-149 and acetazolamide bound to specific CA isoforms.

	Ki (nM)	Ki (nM)
CA isoform	SLC-149	ACZ
I	4.4	250
II	0.94	12.1
III	> 10,000	> 10,000
IV	8.6	74
VA	108.7	63
VB	24.6	54
VI	98.2	11
VII	69.2	0.25
IX	4.1	25.8
XII	4.8	5.7
XIII	21.1	17
XIV	194.4	41

These data represent the mean of 3 independent assays, measured by stopped flow kinetics. Errors were in the range of ± 5-10% of the reported values.

Table 2.

Inhibition constants of SLC-149 and acetazolamide bound to CA isoforms measured by esterase activity.

Isoform	SLC-149 IC50 (μM)	ACZ IC50 (μM)
CAII	0.78 ± .02	0.24 ± .01
CAIX-mimic	0.23 ± .01	1.28 ± .14

CAII is a full-length protein, while CAIX-mimic is a recombinant protein that contains only the catalytic domain with specific amino acid substitutions (see Methods). The IC₅₀ values, calculated from the effect of SLC-149 or acetazolamide (ACZ) on esterase activity, represent the average of three independent assays ± SEM.

In addition, differential scanning fluorimetry was used to measure the Tm shift for SLC-149, at concentrations ranging from 0.001, 0.01, 0.1, and 1 mM, for CAII and CAIX-mimic (Figure 2, A & B). These data demonstrate that SLC-149 bound and thermally stabilized CAIX-mimic between 0.001 and 0.01 mM concentrations. These binding concentrations were significantly lower than that of CAII, where between 0.1 and 1.0 mM was needed to induce the Tm melt shift (Figure 2, A & B). Indeed, this shows that the concentration of SLC-149 required to stabilize CAIX-mimic is ~100-fold lower than for CAII (Figure 2 C). This infers that SLC-149 has a greater binding affinity for CAIX-mimic than CAII. Yet, the change in Tm (Δ TM) upon binding of SLC-149 showed the greatest change in Tm stabilization for CAII (6°C) while increasing CAIX-mimic by ~4°C (Figure 2 D). This implies that SLC-149 bound with more interactions to CAII, forming a more thermally stable complex over CAIX-mimic.

Figure 2. Stability of human CAs in the presence of SLC-149 using Differential Scanning Fluorimetry.

Purified CAII (Panel A, gray) and CAIX-mimic (Panel B, cyan) were incubated with 0.001, 0.01, 0.1, and 1.0 mM concentrations of SLC-149 and heated over a range of temperatures (°C). Relative fluorescent units (signal) are plotted against temperature change for a single run of four samples. Panel C plots the melting temperatures against SLC-149 concentration for each CA (color-coded as in A & B). Panel D compares the change in melting temperatures (Δ Tm) for each CA exposed to specific concentrations of SLC-149 (color-coded as in A & B). Data for C and D represent the average ± S.E.M., n = 4.

To further explore the interactions of SLC-149 with CAII and CAIX-mimic, the crystal structure of the complexes for both were determined. The crystallographic data is shown in Supporting Table S1. Both CAII and CAIX-mimic SLC-149 complexes crystalized in the space group P2₁. The diffraction data sets redundancy ranged from 2.5-2.7 with completeness above 90%. The CAII and CAIX-mimic SLC-149 complex structures were determined to resolutions of 1.4 and 1.8 Å, respectively.

As observed for other ureido sulfonamide CA complex structures, ⁴³ SLC-149 was bound in both structures directly to the active site zinc (~2.0 Å), displacing the zinc-bound water (Figures 3 and 4). Also consistent with other sulfonamide-based compounds, hydrogen bonds were observed between the N1 (amino) of sulfonamide and O atom of T199 (2.7-2.9 Å) and the O atom of sulfonamide and N atom of T199 (2.8-3.0 Å) in both proteins (Figure 3). Additional hydrogen bonds and hydrophobic interactions were observed between SLC-149 and active site residues. The interacting surface area with SLC-149 was similar in the two structures; 517 and 520 Ų in CAII and CAIX-mimic, respectively.

Figure 3. Crystal structure of CAII and CAIX-mimic in complex with SLC-149.

Panel A represents CAII (grey) and Panel B represents CAIX-mimic (cyan). SLC-149 is depicted in salmon color. Hydrogen bonds (2.5-3.5 Å) are shown as black dotted lines. The zinc is shown as a magenta sphere and the solvent is depicted as red spheres. Hydrophobic and hydrophilic regions of the active site are colored orange and purple, respectively. Panels C and D show density maps of SLC-149 in the active sites of CAII and CAIX-mimic. CAII (PDB ID: 6NM0); CAIX-mimic (PDB ID: 6NLV).

Figure 4. Stick representation of compound SLC-149 within the active site of human CAs.

Shown are the catalytic sites of CAII (Panel A, grey) and CAIX-mimic (Panel B, cyan) bound to SLC-149 (colored in lime green for CAII and in purple for CAIX) with specific active site residues labeled. The active site zinc is shown as a magenta sphere. Residue 131 in the active site is featured in these models, which differs between CAII (Phenylalanine, F) and CAIX (Valine, V). Panel C compares the two structures, indicating the change in SLC-149 conformation as a result of residue 131

In the crystal structure of CAII (PDB: 6NM0), SLC-149 formed an additional bridging hydrogen bond between the O of the linker with Q92 (2.9 Å) through a water molecule (3.3 Å) (Figure 3A). Four residues (V121, F131, L198, and W209) were involved in intermolecular interactions with SLC-149. Of note in CAII, the tail of SLC-149 projected out, away from the hydrophobic pocket of the active site (Figure 5A). This was most likely caused by the bulky phenylalanine at position 131 that sterically blocked the tail (Figure 4A and 5A).

Figure 5. Crystal and model structures of compound SLC-149 bound to the active site of human CAs.

SLC-149 (salmon) was co-crystalized with CAII (Panel A, grey) and CAIX-mimic (Panel B, cyan). Zinc is indicated by a magenta sphere. Hydrophobic and hydrophilic regions of the active site are colored orange and purple, respectively. SLC-149 was modeled into CAXII (Panel C, wheat). Residue 131 is a Phenylalanine in CAII (A), a Valine in CAIX (B), and an Alanine in CAXII (C). Residue 132 is a Glycine in CAII (A), a Glycine in CAIX (B), and a Serine in CAXII (C). Residue 135 is a Valine in CAII (A), Valine in CAIX (B), and a Serine in CAXII (C). Residue 204 is a Leucine in CAII (A), an Alanine in CAIX-mimic (B), and an Asparagine in CAXII (C). The homology coordinates for CAXII are available in Supporting Information 2.

In the crystal structure of CAIX-mimic (PDB: 6NLV), the inhibitor also formed an additional hydrogen bond but now between the O of sulfonamide with T200 (2.9 Å) through water molecules (3.0 Å) and an F atom on the tail of SLC-149 with G132 (Figure 3B). The structure also exhibited more intermolecular forces and hydrophobic interactions with seven residues (V121, V131, G132, V135, V143, L198, and W209) compared to the CAII complex structure. This was a consequence of the tail of SLC-149 being oriented more towards the hydrophobic pocket of the active site (Figure 4B & 5B). The presence of a Valine at position 131 in CAIX (instead of Phenylalanine in CAII) allowed the SLC-149 tail to bind towards the hydrophobic pocket due to less steric hindrance.

In 2018, a similar study was conducted to analyze the binding of a series of ureido-substituted benzenesulfonamides (U-CH₃, U-F (SLC-0111), U-NO₂) to several CA isoforms. ¹³ SLC-149 binds in a similar fashion to U-F in CAII and most like U-NO₂ in CAIX-mimic (Supporting Figure S1). ¹³ The Ki of U-F in CAII is 960 nM and the Ki of SLC-149 in CAII is 0.94 nM. As such, SLC-149 appears to be a better inhibitor of CAII compared to similar previously synthesized inhibitors. The Ki of U-NO₂. in CAIX is 1 nM and the Ki of SLC-149 in CAIX is 4.1 nM. These analogous Ki values can be explained by the similar modes of binding between the two compounds. In addition to the crystallography structural studies, SLC-149 was also in silico docked to CAXII to provide structural rational for its inhibition of CAXII at nM affinity (Figure 5C). The modeling showed SLC-149 sulfonamide to bind the Zinc and inhibit enzyme activity in a similar manner as for CAII and CAIX. Intermolecular interactions occurred with residues Q92, V121, S132, L198, T200, and W209. With respect to the inhibitor tail, it was orientated towards the hydrophobic pocket of the active site due to Alanine at position 131 (compared to Phenylalanine in CAII). The Alanine provided less steric hindrance for SLC-149 binding, allowing an additional interaction of the F in the tail region to form a H-bond with S135 (3.1 Å). CAXII differs from CAII in four other amino acid residues within the active site that change the size and/or Van der Waal surface. Residue 132 is a Serine in CAXII (compared to a Glycine in CAII) and residue 135 is a Serine in CAXII (compared to a Valine in CAII). These two residues are polar rather than non-polar, causing additional steric hindrance that prevents SLC-149 from binding as tightly within CAXII. Residue 204 is an Asparagine in CAXII (compared to a Leucine in CAII). The polar nature of this residue may act to repel SLC-149. Overall, there are more bulky residues in the active site of CAXII that would lead to a lower binding affinity.

Effect of SLC-149 on CA Catalytic Activity in Breast Cancer Cells

To provide background for the MIMS assay, Figure 6 shows the catalytic progress curves of CA activity in UFH-001 cells (Panel A) and T47D cells (Panel B). Because UFH-001 cells express both CAII and CAIX (Supporting Figure S3A), the progress curves that follow CA activity are biphasic on a semi-log plot. The first phase (phase 1, 20-40 sec after addition of cells) reflects CAII activity. ⁴⁴ The second phase (phase 2, 100-400 sec in length) reflects CAIX activity. The slopes of these phases are used to calculate rate constants (k). Phase 1 represents the rapid diffusion of ¹³C¹⁸O₂ into cells where CAII catalytic cycles deplete ¹⁸O from H¹³C¹⁸O followed by efflux of ¹³CO₂ species from the cell. The concentration of CO₂ and the isotopic forms of CO₂ are measured in a continuous fashion by the mass spec. For ease of presentation, we have selected the values of atom fraction enrichment of ¹⁸O in ¹³CO₂ at 25 sec intervals. Phase 1 (activity of CAII) is best observed in Figure 6A when acetazolamide (ACZ) is present which inhibits only exofacial CA activity, as ACZ is impermeant during the time course of the experiment. ⁴⁴ The slope of Phase 2 in the presence of acetazolamide approaches the activity observed in the non-catalytic phase (the few time points before addition of cells). We attribute the inhibited activity as loss of CAIX activity, as UFH-001 cells lack CAXII expression (see Supporting Figure S3). When acetazolamide is absent, phase 1 is abbreviated through competition by CAIX for substrate thereby reducing the intracellular substrate concentration available to CAII. Under these conditions, phase 2 gives a better estimate of the total exofacial CAIX activity (compare phase 2 between the absence and presence of ACZ). In contrast to ACZ, the permeant CA inhibitor, ethoxzolamide (EZA), blocks all activity so that the rate constant is reduced to that of non-catalytic activity. Thus, this methodology allows the measurement of both CAII and CAIX activity. In T47D cells (Figure 6B), which lack intracellular CAII activity (Supporting Figure S3B), the progress curves are linear. Now, inhibition by ACZ and EZA is nearly equivalent. We interpret this as inhibition of CAXII activity because T47D cells lack CAIX but express CAXII (Supporting Figure S3). Similar data for an impermeant, pegylated sulfonamide (N-3500) (described by Delacruz et al. ⁴⁴) confirm the above conclusions.

Figure 6. Catalytic progress curves for CAIX in UFH-001 cells and CAXII in T47D cells.

UFH-001 and T47D cells were released from plates using cell release buffer that contains EDTA instead of trypsin to avoid loss of the ectodomain of the CAIX (in UFH-001 cells) and CAXII (in T47D cells). These cells under went extensive washing with medium that contained no bicarbonate, but buffered with Hepes (pH 7.4). After counting, 500,000 cells per mL were added to the reaction vessel, containing bicarbonate-free medium/Hepes at 16 °C in which was dissolved ¹⁸O-enriched ¹³CO₂/H¹³CO₃⁻ at 25mM total ¹³CO₂, at a point when the non-catalytic rate had been established (see arrow). A membrane inlet was immersed in the medium in the reaction vessel and used to detect the atom fraction of ¹⁸O in extracellular species of ¹³CO₂. In this experiment, we used a pegylated sulfonamide (N-3500) at 25 μM to block exofacial CA activity. A single representative experiment (from 3) is shown for each cell type.

In earlier studies, we have shown that a set of related ureido sulfonamide inhibitors affect different CA isoforms with significantly different efficacies in the context of their native cellular environments and breast cancer subtypes. ¹³ These experiments revealed that ureido sulfonamides were substantially more effective in blocking CAIX activity in receptor positive T47D cells despite similar Ki values for the triple negative UFH-001 cells than CAXII in the estrogen purified proteins. ^{45, 46} SLC-149 shows this same trend (Figure 7), although by comparison, SLC-149 is a much stronger inhibitor than those previously studied (see Supporting Figure S4B). In these experiments, the effect of SLC-149 on CA activity was measured in both normoxic and hypoxic cells, as breast cancer cells are often exposed to low oxygen conditions in vivo. The MCF10A cells, used as a control cell line, express CAIX only under hypoxic conditions. ⁴⁷ Thus, the activity of CAIX in normoxic MCF10A cells was very low, which means that the Ki value was difficult to measure. Activity was somewhat higher in hypoxic cells so the Ki value could be determined with somewhat more accuracy (6.1 ± 2.0 nM). The UFH-001 cells express CAIX under both normoxic and hypoxic conditions and at higher levels that MCF10A cells ⁴⁵ although the level and activity of CAIX increases (~2-fold) when the cells are exposed to hypoxia ^{11, 13} (see Supporting Figure S3A). The Ki values for SLC-149-induced inhibition of CAIX in cells exposed to normoxic or hypoxic conditions were in the same range (48.9 nM ± 0.05 vs 20.8 ± 0.05 nM, respectively). T47D cells express only CAXII and its expression is not affected by hypoxia as previously shown ^{11, 13, 47} (see also Supporting Figure S3B). The Ki values for SLC-149-induced inhibition of CAXII were essentially identical under normoxic and hypoxic conditions (5.4 ± 0.06 μM vs 5.2 ± 0.21 μM, respectively). Note that these values are two orders of magnitude higher than those calculated for inhibition of CAIX in UFH-001 cells.

Figure 7. Effect of SLC-149 on CA activity in breast cancer cell lines.

CA activity was measured in normoxic and hypoxic MCF10A cells, UFH-001 cells, and T47D cells in the presence of an extensive range of SLC-149 to determine Ki values using the MIMS assay (described above and in Methods). Average Ki values of compound SLC-149 in MCF10A, UFH-001 and T47D cells, under normoxic or hypoxic (16 h) conditions, are shown. Data represent the mean of 3 independent experiments ± SEM.

Impact of SLC-149 on breast cancer cell growth, cytotoxicity, migration, and invasion

Previous experiments have shown that the “anti-cancer” effect of ureido sulfonamide inhibitors required significantly higher concentrations of drug, compared to that needed for inhibition of activity in intact cells. However, the Ki values for SLC-149-induced inhibition of CAIX activity, particularly, were lower than observed for these ureido sulfonamides by 3- to 10-fold depending upon the inhibitor ¹³ (Supporting Figure S4B). We thus predicted that SLC-149 would be a better inhibitor of cell growth. Yet, growth of the MCF10A, UFH-001, and T47D cells was only affected by extremely high concentrations of SLC-149 (over 48 h of treatment) reducing the growth (measured using MTT assay) by only about 25% (Figure 8A-F and Supporting Figure S4C). Hypoxia did not further sensitize these cells to the inhibitor. UFH-001 cells were the only cell type that exhibited cytotoxicity and only after 48 h of exposure to SLC-149 (Figure 9E and Supporting Figure S4D). Finally, we tested the effect of SLC-149 on migration and invasion, but only in UFH-001 cells because neither MCF10A nor T47D cells display migration or invasion ability in culture. However, SLC149 did not significantly affect these phenotypic features, although the highest concentration tested was 100 μM (Supporting Figure S5).

Figure 8. Effect of SLC-149 on breast cancer cell growth.

Breast cancer cell lines grown under normal culture conditions for 24 h were exposed to SLC-149, under normoxic or hypoxic conditions, for 16 h [A) MCF10A, B) UFH-001 and C) T47D] or 48 h [D) MCF10A, E) UFH-001 and F) T47D]. MTT assay was performed at 16 h and 48 h, respectively. Data shown are an average of at least 3 independent experiments and are represented as the mean ± SEM.

Figure 9. Effect of SLC-149 on breast cancer cell viability.

The cytotoxic effects of compound SLC-149 were evaluated using the LDH release assay. Breast cancer cells were grown in 96-well plates and exposed to SLC-149, under normoxic or hypoxic conditions, for 16 h [A) MCF10A, B) UFH-001 and C) T47D] or 48 h [D) MCF10A, E) UFH-001 and F) T47D] ²². LDH release was assayed after treatment, results were evaluated, and data analyzed using Prism. Data shown are representative of 3 independent experiments ± SEM.

DISCUSSION AND CONCLUSIONS

SLC-149 exhibited 4-fold inhibition specificity toward CAIX-mimic over CAII based on calculated IC₅₀ values from the esterase activity assay (Table 2). These data were surprising as the Ki values determined by stopped flow analysis for CAII and CAIX showed the inverse relationship (Table 1). Within the limits of each of these very different experiments, the measured inhibition of SLC-149 was similar, only differing by a few-fold. While kinetic constants for CA, like Ki and Km, can be estimated with the pNPA esterase assay or by CO₂ exchange as in the stopped flow assay, these two methods will give different kinetic constants. However, within each method the values can be compared. Stopped flow assays showed little to no selectivity of SLC-149 for CAIX over CAII. That said, the Tm shift of CAIX-mimic (Figure 2) induced by SLC-149 binding was observed at a substantially 100-fold lower concentration of inhibitor compared to CAII. These differences in affinity may be attributed to variations in interactions within the active site, the smaller inhibitor bound surface area in CAII, and the F131V variation. In the CAIX-mimic, SLC-149 formed additional hydrogen bond interactions with G132 along with T200 via water molecules while CAII interacted with only Q92 through water molecules. The hydrogen bond with G132 allowed for the different conformation in CAIX-mimic (Figure 4C). The inhibitor was further stabilized in the CAIX-mimic by intermolecular forces and hydrophobic interactions with G132, V135, and V143.

Variation from F131V contributed to a difference in orientation in which the SLC-149 tail projected out of the active site in CAII to accommodate F131 steric hindrance while it moved towards the hydrophobic pocket in CAIX-mimic, thus increasing interactions with the hydrophobic region. It was expected that the loss of interactions and F131V variation would increase affinity in the CAIX-mimic versus CA II, which was observed by differential scanning fluorimetry.

As mentioned, possible differences in the binding affinities for SLC149 among these isoforms could occur because of differences in residue type at position 131. In CAIX this residue is a Valine and in CAXII an Alanine, both of which are less bulky and hydrophobic than Phenylalanine in CA II. Also, residue 135 is a Serine in CAXII and analogous to that position in CAII, but in CAIX this is a Valine, which could also affect the binding affinity for SLC149.

Furthermore, studies that examine anti-cancer activity in cell culture still fall short as SLC-149 had little effect on growth, migration, or invasion of UFH-001 cells, although we have shown that other sulfonamides in this same class do block these features at concentrations that are about 100-fold higher than Ki values for blocking CA activity in intact cells. ¹³ Yet, studies that employed ureido sulfonamides, for example SLC-0111/U104/U-F which do block cell growth in UFH-001 cells, have shown success in blocking tumor growth ¹² and metastasis ⁴⁵ in animal models. In these animal studies, a rough calculation of the final concentration of sulfonamide in the circulation is about 1 mM, which is where we see inhibition of cell growth in cell culture. However, when we knock out CAIX in UFH-001 cells, the inhibitors have the same effect on cell growth. This suggests that this group of sulfonamides have non-specific effects at least in cell culture and that these effects are independent of CAIX catalytic activity. ¹³ SLC-149 does not have this effect, which suggests that it is more specific for targeting activity than other ureido sulfonamides, however it does not have the desired anti-cancer activity. That said, this might be the perfect drug for combination therapy, one that can create a pH imbalance in the tumor microenvironment, and one which exhibits growth inhibition. There is precedence for this as SLC-0111 has been shown to work in combination therapy to sensitize cancer cells to other chemotherapy drugs. ⁴⁸

In summary, our data show that SLC-149 binds to and inhibits CAIX in silico, in vitro, and in cell culture of breast cancer cells. Yet, inhibition of catalytic activity does not appear to impair cell growth, migration, or invasion, which suggests that a non-catalytic function of CAIX may regulate its ability to block cell and tumor growth despite its capacity to control pHe through its catalytic activity.

EXPERIMENTAL METHODS

Synthesis of 4-(3-(2,4-difluorophenyl)-2-oxoimidazolidin-1-yl)benzenesulfonamide (SLC-149)

1. Sulfanilamide (10.00 g, 58.1 mmol) was suspended in acetone (30 mL), chloroacetyl chloride (4.49 mL, 55.3 mmol) was added drop-wise at room temperature (RT). The reaction mixture was stirred at 95 °C for 1 h and then cooled to RT. Ice-water was added and the resulting mixture was stirred and filtered. The solid collected was washed with ice-water and re-crystallized from ethanol. The solid was collected by filtration and dried in vacuo to afford chloro-4-sulphamoylacetanilide as a white solid (8.84 g). The filtrate was concentrated to yield more product as a white solid (3.42 g). The total yield was 89%. ¹H NMR (300 MHz, DMSO-d₆) δ 10.61 (s, 1H), 7.81-7.71 (m, 4H), 7.26 (br s, 2H), 4.29 (s, 2H).

2. The chloro-4-sulphamoylacetanilide (0.70 g, 2.81 mmol) and 2,4-difluoroaniline (2.17 mL, 11.24 mmol) were heated in a seal tube at 140 °C overnight. Then 20 mL of dichloromethane was added followed by the addition of 10 mL brine. The mixture was extracted 3 times with dichloromethane, and then dried with anhydrous sodium sulfate. After removal of the solvent, the residue was recrystallized from ethanol to afford 0.41 g of 2-((2,4-difluorophenyl)-amino)-N-(4-sulfamoylphenyl)acetamide as a white solid in 42% yield. ¹H NMR (400 MHz, DMSO-d₆) δ 10.34 (s, 1H), 7.85-7.70 (m, 4H), 7.22 (br s, 2H), 7.20-7.05 (m, 1H), 6.60-6.90 (m, 1H), 6.70-6.55 (m, 1H), 5.70 (m, 1H), 3.97 (d, J = 6.3 Hz, 2H).

3. 2-((2-fluorophenyl)amino)-N-(4-sulfamoylphenyl)acet-amide (0.40 g, 1.17 mmol) was suspended in tetrahydrofuran (1.17 mL), then borane dimethyl sulfide complex (2M in tetrahydrofuran, 1.32 mL, 2.64 mmol) was added dropwise at 0 °C. The mixture was refluxed for 1 h and then cooled to 0 °C. Methanol (10 mL) was added at 0 °C, and the resulting mixture was stirred at 0 °C for 10 min. Hydrogen chloride (gas) was bubbled in to change the pH <2. The mixture was then refluxed for 30 min and cooled to RT . The residue was treated with aqueous sodium hydroxide to pH >12, extracted with ethyl acetate (3 x 80 mL), and dried over magnesium sulfate. After filtration and removal of the solvent, the residue was treated with dichloromethane and filtered. The solid collected was dried in vacuo to yield 4-((2-((2,4-difluorophenyl)amino)ethyl)amino)benzene-sulfonamide as a white solid in 91% yield (0.35 g). ¹H NMR (400 MHz, DMSO-d₆) δ 7.53 (d, J = 8.7 Hz, 2H), 7.15-7.05 (m, 1H), 6.92 (br s, 2H), 6.90-6.82 (m, 1H), 6.81-6.70 (m, 1H), 6.65 (d, J = 8.7 Hz, 2H), 6.47 (br s, 1H), 5.36 (br s, 1H), 3.40-3.20 (m, 4H).

4. To a solution of 4-((2-((2,4-difluorophenyl)amino)ethyl)-amino)benzene-sulfonamide (0.34 g, 1.04 mmol) was added dropwise triphosgene (0.15 g, 0.52 mmol) solution in tetrahydrofuran. The resulting mixture was stirred at RT for 1 h. The precipitate obtained was collected and washed with tetrahydrofuran to afford 4-(3-(2-fluorophenyl)-2-oxoimidazolidin-1-yl)benzenesulfonamide as a white solid in 19% yield (70 mg). ¹H NMR (300 MHz, DMSO-d₆) δ 7.78 (d, J = 9.6 Hz, 2H), 7.74 (d, J = 9.6 Hz, 2H), 7.58 (td, J = 8.8, 6.4 Hz, 1H), 7.38 (m, 1H), 7.23 (br s, 2H), 7.15 (m, 1H), 4.05 (m, 2H), 3.95 (m, 2H). MS (ES+) m/z 354.2 (M + 1). The HPLC trace of SLC-149, which shows 98% purity, is available in Supporting Information 3.

Protein expression and purification.

CAII (full length), CAIX-mimic (catalytic domain), cDNA were transformed into E. coli BL21 (DE3) as previously described ⁴³. Protein expression was induced at 37 °C using isopropyl β-D thiogalactoside (IPTG) (final concentration of 0.1μg/mL) and a final concentration of 1 mM zinc sulfate (ZnSO₄). After 3 h, cells were collected by centrifugation and harvested. Protein purification was performed by affinity chromatography using a p-(aminomethyl) benzenesulfonamide agarose affinity column (Sigma) and eluted with 0.4 M sodium azide and 50 mM Tris-HCl, pH 7.8. The eluted proteins were dialyzed against 50 mM Tris-HCl pH 7.8 and concentrated using Amicon ultra-filtration tubes with a 10 kDa molecular weight cut-off. The purity of the proteins was assessed by SDS-PAGE stained with Coomassie blue. The concentration was determined by UV spectroscopy at 280 nm using a molar extinction coefficient of 54,800 cm⁻¹mol⁻¹. The CAIX-mimic was engineered by site-directed mutagenesis of wildtype CAII to resemble the active site of wildtype CAIX. Mutations include A65S, N67Q, E69T, I91L, F131V, K170E and L204A resulting in 93% sequence similarity between the catalytic domains of the CAIX-mimic and CAIX (Supplemental Table 1) ^{37, 38}.

Esterase and stopped flow analysis.

In addition to catalyzing the reversible hydration of CO₂, CAs exhibit esterase activity in their catalytic pocket which can be used as a surrogate for the hydration reaction, albeit at lower rates of catalysis. This activity was measured using p-nitrophenyl acetate (pNPA) as the substrate. This is a colorimetric assay which detects a color change when the acetic acid group is cleaved by CA activity ⁴⁹. The concentration of CA used in this assay was 0.02 mg/mL (667 nM for CAII, 667 nM and for CAIX-mimic). A concentration of pNPA at 3 μM was used to calculate IC₅₀ values over a range of SLC-149 concentrations. Ethoxzolamide, a potent inhibitor of CA activity, was used as a control for complete inhibition relative to that for SLC-149. An average of three replicates are reported for the IC₅₀ values. Stopped flow spectroscopy was used to determine the Ki values of SLC-149 and acetazolamide according to a method originally described by Khalifah ⁵⁰ and more recently by Angeli et al. ⁵¹

Differential scanning fluorimetry (DSF).

DSF was used to assay the conformational stability of CAII and CAIX-mimic by changes in melting temperature (Tm) in the presence of increasing concentrations of SLC-149 (from Signalchem Lifesciences, Richmond, British Columbia, CA). Samples of purified CAII and CAIX-mimic (0.25 mg/mL) were incubated with specific concentrations of SLC-149 (0.001-1.0 mM). Prior to data collection, samples were incubated on ice with the fluorescent dye, Sypro-Orange at 0.01% (Invitrogen) for 30 min. Melting curve assays were conducted in a quantitative PCR instrument (Corbett Research) with the temperature increasing from 30 to 99 °C, increasing at a rate of 0.1 °C/6 sec. Solutions containing only buffer were used to assess background for data processing. The Tm was defined as the maximal value of the first derivative (dRFU/dT; change in fluorescence/change in temperature) of the signal that is produced in terms of relative fluorescent units (RFU).

Crystallization and inhibitor soaking.

CAII and CAIX-mimic proteins were crystallized using the hanging drop vapor diffusion method in precipitant solution (1.6 M Sodium citrate, 50 mM Tris-HCl at pH 7.8). Each 5 μL crystallization drop contained CAII or CAIX-mimic protein with precipitant solution. The stock concentration used to obtain both CAII and CAIX-mimic crystals was 10 mg/mL, resulting in a final concentration of 5 mg/mL/drop. Crystals were observed after 1 week. At this point, compound SLC-149 was soaked into the crystals. Briefly, a 60 mM stock solution of SLC-149 was prepared in 100% DMSO. The compound was further diluted to a final concentration of 20 mM before adding to the crystallization drop. Crystals were soaked for 24-48 h with the inhibitor. Prior to data collection, crystals were cryopreserved and stored in liquid nitrogen.

X-ray crystallography.

X-ray diffraction was collected at Cornell High Energy Synchrotron Source (CHESS) on beamline F1 using a 24-pole wiggle with X-ray energy of 12.68 keV and a wavelength of 0.9177 Å or SSRL on 14-1 beamline. At CHESS, the data sets were collected using an ADSC Quantum 270 CCD detector at a crystal-to-detector distance ranging from 200-250 mm with a 1° oscillation angle and an exposure time of 1-3 sec per image. A total of 360 images were collected. For the SSRL data sets, the crystal-to-detector distance ranged from 200 mm, 0.15° oscillation angle, and exposure time of 1.5 sec per image for a total of 1200 images. The data was indexed, integrated, and scaled using X-ray detector software (XDS) to a monoclinic P2₁ space group. ⁵² Molecular replacement was performed for initial phasing in PHENIX, ⁵³ using previously published CAII (PDB: 3KS3) and CAIX-mimic (PDB: 3IAI) structures as a search model with zinc and solvent removed. ⁵⁴ PHENIX was also used for further structure refinement and generation of ligand restraints. Crystallographic object-oriented toolkit (COOT) ⁵⁵ and LigPlus were used to generate ligand-protein complexes and to determine bond lengths and angles. The inhibitor-bound surface area was calculated using PDB e-PISA (http://www.ebi.ac.uk/pdbe/prot_int/pistart.html) based on the work of Krissinel and Henrick. ⁵⁶ Figures were generated using PyMol (version 1.5.0.4 Schrodinger, LLC).

Cell Culture.

The MCF10A cells (a gift from Dr. Brian Law, University of Florida) were plated at a density of 20,000 cells/mL in 10 cm dishes and cultivated in DMEM/Ham’s F12 medium supplemented with 5% horse serum (Sigma Aldrich), 10 μg/mL insulin, 20 ng/mL epidermal growth factor (EGF) (Upstate Biochem), and 100 ng/mL dexamethasone (Sigma). These cells served as a control breast cancer line because they are immortalized but not transformed. They express CAIX only under hypoxic conditions and do not express intracellular CAII nor cell surface CAXII. ⁴⁷ T47D cells (a gift from Dr. Brian Law, University of Florida) were plated in 10 cm dishes at 20,000 cells/mL and maintained in McCoys 5A medium supplemented with 10% FBS and 10 μg/mL insulin. These cells exhibit the luminal (estrogen-receptor positive) phenotype, and express only CAXII at the cell surface and lack intracellular CAII. ⁴⁷ UFH-001 cells were plated in 10 cm dishes at a density of 10,000 cells/mL and maintained in DMEM supplemented with 10% FBS. UFH-001 cells are a newly characterized line that exhibits the triple negative breast cancer phenotype ⁵⁷ and are available for purchase (Sigma, SCC210). These cells express intracellular CAII and cell surface CAIX, but not CAXII. All cell lines were maintained at 37 °C in humidified air with 5% CO₂. Experiments were conducted when cells achieved ~70% confluence, unless otherwise specified. For hypoxic treatment, cells were placed in humidified Billups Rothenberg Metabolic Chambers and exposed to 1% O₂, 5% CO₂ and balanced N₂ for designated times at 37 °C. All cell lines were authenticated by STR analysis (Bio-Synthesis Inc. and IDEXX BioResearch) and tested for mycoplasma.

Membrane Inlet for Mass Spectrometry.

CA catalysis was measured by the exchange of ¹⁸O from species of ¹³CO₂ into water determined by membrane inlet mass spectrometry (MIMS). ^{58, 59} The application of this method in demonstrating exofacial CA activity in breast cancer cells has been described previously ²¹ and more recently in Chen et al. ¹¹ and Mboge et al. ¹³ Briefly, cells were collected from culture plates using cell release buffer (Gibco), washed with bicarbonate-free DMEM buffered containing 25 mM Hepes (pH 7.4) and counted. Cells (5.0 × 10⁵ cells/mL, unless otherwise stated) were added to a reaction vessel containing 2 mL of Hepes-buffered, bicarbonate-free DMEM at 16°C in which ¹⁸O-enriched ¹³CO² /H¹³CO₃ at 25 mM total ¹³CO₂ species were dissolved. We chose 16 °C for the reaction temperature (in contrast to the traditional 25 °C) for two reasons. First, the lower temperature slows the enzyme-mediated reactions to better contrast the intracellular and extracellular CA activities. Second, we sought to prevent endocytotic events during the course of the experiment, which is accomplished at temperatures slightly below the phase transition for lipids in the plasma membrane. The membrane inlet was immersed in the medium in the reaction vessel and used to detect the atom fraction of ¹⁸O in extracellular ¹³CO₂. This activity was measured after addition of cells (previously exposed to normoxic or hypoxic conditions) in medium with specific CA inhibitors, including SLC-149. This exchange is a specific measure of CA activity during the time frame of the assay (~ 6 min). Ki values were calculated using the following equation:

$$(\mathrm{V}\text{-}\text{unc})=(\text{Vo}-\text{unc})∕(1+\mathrm{I}∕\text{Ki}),$$

where (V) is the rate at a specific inhibitor concentration, (unc) is the uncatalyzed rate, (Vo) is rate in the absence of inhibitor, and (I) is inhibitor concentration. Additional details are included in the text and Figure 6.

Cell Growth Assay.

Cell growth in the presence or absence of compound was analyzed using 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT assays). Briefly, cells were seeded in 96-well plates and treated with different concentrations of compound after 24 h. The cells were then incubated under normoxic or hypoxic conditions under the different treatment conditions for 48 h or 96 h. The medium was replaced every two days. MTT reagent (10% v/v of 5 mg/mL) was added to the cells after each time point and the cells were incubated for an additional 4 h. The medium was removed without disturbing the cells and DMSO was added to each well. The DMSO-dissolved cells were incubated while shaking for 15 min in the dark. Absorbance was measured at 570 nm using an Epoch microplate reader (Biotek, Winooski, VT). IC₅₀ values (inhibitor concentration where the response is reduced by half) were calculated using Prism 7.0c for Macs. Non-linear regression analysis was performed on % dose response curves with log plots, interpolating the value at 50% from the least squares fit of the growth curves.

Lactate Dehydrogenase Release Assay.

Cytotoxicity was estimated using the release of lactate dehydrogenase (LDH) activity (Sigma-Aldrich, St. Louis, MO). Cells were plated in 96-well plates. SLC-149 was added the next day within a specific range of concentrations. After 48 h, medium was collected from each well to measure the amount of released LDH activity. The amount of LDH released from each sample was measured at RT at 450 nm using a BioTek Epoch microplate reader. LDH activity is reported as nmol/min/mL of medium.

Cell Migration and Invasion Assays.

Serum starved cells were plated at a density of 50,000 cells in 300 μL per insert in 24-well cell migration and invasion plates (Cell BioLabs, San Diego, CA). The cells were allowed to migrate (24 h) or invade (48 h) from the insert containing serum free medium, in the presence of specific concentrations of SLC-149, or without inhibitor as a control, towards the well that contained medium with 10% FBS. Fixing and staining of cells terminated the assay. Images were then collected and analyzed. All cell lines were maintained at 37 °C in humidified air with 5% CO₂ for the duration of the experiment.

Statistical Analysis.

Data are presented as the mean ± SEM unless otherwise specified. Differences between the treatment groups, when applicable, were analyzed using Student’s t-test in GraphPad Prism, version 8.3.1: p-values < 0.05 was considered statistically significant.

ABBREVIATIONS USED

CA

Carbonic Anhydrase

hCA

Human Carbonic Anhydrase

CAIX-SV

Carbonic Anhydrase IX-Surface Variant

HIF1

Hypoxia-Inducible Factor 1

ER

Estrogen Receptor

pHe

Extracellular pH

DSF

Differential Scanning Fluorimetry

MTT

3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide

IPTG

Isopropyl β-D thiogalactoside

pNPA

p-nitrophenyl acetate

T_m

Melting Temperature

RFU

Relative Fluorescence Units

CHESS

Cornell High Energy Synchrotron Source

XDS

X-ray Detector Software

COOT

Crystallographic Object-oriented Toolkit

EGF

Epidermal Growth Factor

TNBC

Triple Negative Breast Cancer

MIMS

Membrane Inlet Mass Spectrometry

LDH

Lactate Dehydrogenase

RT

room temperature