Conjugation to Native and Nonnative Triscatecholate Siderophores Enhances Delivery and Antibacterial Activity of a β-lactam to Gram-negative Bacterial Pathogens

Abstract

Developing new antibiotics and delivery strategies is of critical importance for treating infections caused by Gram-negative bacterial pathogens. Hijacking bacterial iron uptake machinery, such as that of the siderophore enterobactin (Ent), represents one promising approach towards these goals. Here, we report a novel Ent-inspired siderophore–antibiotic conjugate (SAC) employing an alternative siderophore moiety as the delivery vector and demonstrate the potency of our SACs harboring the β-lactam antibiotic ampicillin (Amp) against multiple pathogenic Gram-negative bacterial strains. We establish the ability of N,N′,N″-(nitrilotri-1,2-ethanediyl)tris[2,3-dihydroxybenzamide] (TRENCAM, hereafter TC), a synthetic mimic of Ent, to facilitate drug delivery across the outer membrane (OM) of Gram-negative pathogens. Conjugation of Amp to a new monofunctionalized TC scaffold affords TC-Amp, which displays markedly enhanced antibacterial activity against the gastrointestinal pathogen Salmonella enterica serovar Typhimurium (STm) compared to unmodified Amp. Bacterial uptake, antibiotic susceptibility, and microscopy studies with STm show that the TC moiety facilitates TC-Amp uptake by the OM receptors FepA and IroN and the Amp warhead inhibits penicillin-binding proteins. Moreover, TC-Amp achieves targeted activity, selectively killing STm in the presence of commensal Lactobacilli. Remarkably, we uncover that TC-Amp and its Ent-based predecessor Ent-Amp achieve enhanced antibacterial activity against diverse Gram-negative ESKAPE pathogens that express Ent uptake machinery, including strains that possess intrinsic β-lactam resistance. TC-Amp and Ent-Amp exhibit potency comparable to the FDA-approved SAC cefiderocol against Gram-negative pathogens. These results demonstrate the effective and potent application of native and appropriately designed nonnative siderophores as vectors for drug delivery across the OM of multiple Gram-negative bacterial pathogens.

Graphical Abstract

Introduction

Gram-negative bacterial pathogens pose a serious threat to human health, in part due to the impermeability of the outer membrane (OM) to many therapeutics. Clinically relevant bacteria of concern, such as the ESKAPE pathogens and the World Health Organization priority pathogens,¹ are primarily Gram-negative bacteria, including Pseudomonas aeruginosa (Pa), Acinetobacter baumannii (Ab), Escherichia coli (Ec), Salmonella spp., and Klebsiella pneumoniae (Kp). Our current arsenal of drugs targeting Gram-negative bacteria has been depleted in the face of antimicrobial resistance, especially through mechanisms related to decreasing intracellular drug accumulation, such as porin modification/downregulation and expression of efflux pumps.^2–4 Furthermore, the wide use of broad-spectrum antibiotics can accelerate the development of antimicrobial resistance, while simultaneously increasing the risk for secondary infections and alterations of the gut microbiota (i.e., dysbiosis).^{3, 5} It is thus important to implement molecular designs that address OM impermeability and resistance mechanisms of Gram-negative pathogens as well as selectivity in the next generation of antibiotics. Employing vectors for drug delivery that target endogenous and essential bacterial receptors, such as those involved in Fe acquisition, is one promising strategy to address these needs.^6–7

Fe is an essential nutrient for the vast majority of bacterial species; however, its bioavailability is exceedingly low, due to the relative aqueous insolubility of Fe(III) and the sequestration of Fe in mammalian hosts by proteins such as transferrin and ferritin.⁸ Fe can be further restricted by the host during infection as part of nutritional immunity.⁹ In order to acquire sufficient Fe to proliferate, bacteria biosynthesize and secrete siderophores into the extracellular space, which chelate Fe(III) with high affinity and are then transported by specific cognate receptors into the cell for Fe release and utilization.¹⁰ Enterobactin (Ent 1, Figure 1a) is an archetypal triscatecholamide siderophore produced by Enterobacteriaceae that has been well characterized for its coordination chemistry and biological activity.^{8, 11} Closely related to Ent are the salmochelins, C-glucosylated Ent congeners produced and utilized by strains harboring the pathogen-associated iroA gene cluster, such as pathogenic Ec, Kp, and Salmonella.^12–13 TonB dependent receptors (TBDRs) are OM proteins that utilize the inner membrane (IM) proton gradient to power active transport of the substrate, and are responsible for the recognition and transport of Ent and the salmochelins through the OM. Here we focus on the OM receptors present in Salmonella enterica serovar Typhimurium (STm) as a model system (Figure 1b). This gastrointestinal pathogen employs Ent and salmochelins for Fe acquisition in the mammalian host and expresses three OM receptors for Fe(III)-bound catecholate siderophores.¹⁴ FepA transports Ent, whereas IroN, encoded within the iroA gene cluster, imports the salmochelin diglucosylated Ent (DGE) as well as Ent.^{13, 15} STm also expresses Cir, which has been implicated in uptake of Ent degradation products, namely linear 2,3-dihydroxybenzoylserine fragments (DHBS)_n (n = 1, 2, 3).^{14, 16} Following OM transport by TBDRs, Fe(III)-bound Ent and salmochelins are chaperoned by the periplasmic binding protein FepB to the IM permease FepCDG, an ATP-binding cassette transporter. Once in the cytoplasm, the central trilactone ring is hydrolyzed by the esterases Fes or IroD to yield DHBS fragments, and subsequent Fe reduction, putatively by the reductase YqjH, facilitates release of the nutrient Fe(II).^{8, 17–18}

Figure 1.

(a) Structures of siderophores Ent 1 (native) and TC 2 (synthetic mimic); (b) cartoon overview of Ent OM TBDRs in STm and their preferred siderophores; (c) Structures of siderophore–antibiotic conjugates evaluated in this study, Ent-Amp 3 and TC-Amp 4; β-lactam cargo shown in blue.

Siderophore–antibiotic conjugates (SACs) are designed to hijack the active transport capacity and molecular recognition of Fe-siderophore acquisition systems, and thus have the potential to overcome the OM impermeability of Gram-negative bacteria and narrow the antibiotic spectrum of their drug cargos.^{7–8, 19–20} A recent clinical implementation of this strategy is highlighted by the FDA-approved antibiotic cefiderocol (Cfdc), a catechol-modified cephalosporin (Figure S1)²¹ applauded for its ability to evade resistance strategies involving efflux pump upregulation or modifications to porins.^22–23 Several bidentate siderophore-modified (mono-catechol or hydroxypyridone) β-lactams have reached preclinical or clinical trials over the years, but no others have progressed to market.⁷ In a few notable cases, conditions of Fe limitation and effects of native siderophore production were found to be important but overlooked factors in evaluating the efficacy of the drug and the development of adaptive resistance.^24–27 In contrast to these pharmaceuticals employing simplified, bidentate moieties imitating siderophore functionalities, several of the most effective SACs reported in the literature have incorporated or closely mimicked the structure of naturally occurring siderophores,²⁸ including our established platform for native Ent-based conjugates. Our strategy of monofunctionalized Ent conjugates maintains the high affinity, hexadentate Fe chelating scaffold and retains the interaction with Ent uptake machinery.²⁹ This approach has provided receptor-dependent delivery of several different drug cargos to the periplasm (β-lactams ampicillin, amoxicillin, cephalexin, and meropenem) of Ec^29–31 and Salmonella³² and the cytoplasm (ciprofloxacin and cisplatin) of Ec.^33–34 Ent- and DGE-β-lactam conjugates, such as Ent-Amp (3, Figure 1c), displayed enhanced antibacterial activity compared to the parent antibiotics, and the DGE-β-lactams and Ent-Cipro achieved selectivity towards strains harboring the iroA gene cluster.^{30, 33} Our explorations of antibacterial activity towards species beyond Ec and Salmonella have been limited; however, we hypothesize that our conjugates can be transported by other bacterial pathogens that import Ent, such as the ESKAPE pathogens Ab and Pa that use Ent as a xenosiderophore.^35–36

Ent is a highly symmetric molecule that lacks a native functionalization handle. Consequently, many efforts by others employed synthetic mimics of Ent in order to facilitate installation of conjugation handles as well as improve conjugate stability and streamline the synthesis.^{20, 37–40} Early screening studies of conjugates employing biscatechol and triscatechol analogs showed potential for delivering β-lactam drug cargos, but received little follow-up.^41–45 In recent years, rationally designed SACs employing triscatecholamide Ent mimics where the trilactone core of Ent has been substituted by an alkyl or mesitylene backbone have also shown promise in delivering drug cargos with periplasmic targets.^{37–39, 46} Nevertheless, it is unclear how these mimics directly compare to the native siderophore scaffold for their uptake and activity. We also consider the instability of the trilactone ring of Ent, both to chemical hydrolysis and potential nonspecific enzymatic breakdown (e.g., host esterases). Implementing a nonhydrolyzable backbone in place of the trilactone ring has the potential to improve the hydrolytic stability of the conjugate, which may be advantageous for future applications in preclinical models.

To evaluate the role of the native Ent scaffold and to introduce alternative options for the siderophore moiety of our conjugates, we expanded our repertoire of SACs to include a synthetic Ent mimic with a nonnative, nonhydrolyzable backbone. We turned to the synthetic siderophore N,N′,N″-(nitrilotri-1,2-ethanediyl)tris[2,3-dihydroxybenzamide] (TRENCAM, hereafter TC 2, Figure 1a), a triscatecholamide with a tris(2-aminoethyl)amine (TREN) backbone that was designed to mimic Ent.^47–48 TC binds Fe(III) tightly (pM = 27.8, vs. pM = 35.5 for Ent)^{47, 49} and is recognized and transported by FepA.^50–51 Leveraging the wealth of fundamental characterization and prior studies on TC, we selected it as a promising candidate for our SAC strategy. Based on its close structural mimicry of Ent^52–54 and its known interactions with Ent-binding proteins,^{50–51, 55–59} we hypothesized that TC can effectively serve as a vector for selectively transporting drug cargos into Gram-negative bacteria. We reasoned that evaluating a TC-based conjugate, particularly in comparison to an analogous Ent-based conjugate, would provide an excellent approach to elucidate the uptake and antibacterial activity of a conjugate based on a nonnative scaffold.

Herein we report the design and synthesis of a monofunctionalized TC scaffold and a TC-ampicillin conjugate (TC-Amp 4, Figure 1c) and demonstrate that TC-Amp provides efficient antibiotic delivery across the OM and enhances the antibacterial activity of Amp, showing minimum inhibitory concentration (MIC) values and time-kill kinetics comparable to those of Ent-Amp. Moreover, we establish that Ent-Amp and TC-Amp exhibit antibacterial activity against multiple Gram-negative ESKAPE pathogens that express Ent uptake machinery, including those with intrinsic β-lactam resistance and a cystic fibrosis clinical isolate of Pa. Finally, we show that the antibacterial potency of Ent-Amp and TC-Amp is generally comparable to that of Cfdc, at least under the conditions used in this work. Collectively, these studies advance our understanding of the ability of SACs to hijack Ent uptake machinery and expand the scope of both Ent- and TC-based conjugates for the delivery of antibiotic cargo across the OM of Gram-negative bacteria.

Results and Discussion

STm takes up ferric TC via Ent OM receptors.

To elucidate the potential of TC as a drug delivery vector, we investigated the capacity of the Ent uptake machinery expressed by STm to transport ferric TC. We aimed to evaluate the uptake efficiency of ferric TC compared to that of ferric Ent, and determine the primary receptors involved. To measure siderophore transport, we treated STm with ⁵⁷Fe(III)-loaded Ent or TC and monitored ⁵⁷Fe uptake by quantifying cell-associated ⁵⁷Fe using inductively coupled plasma mass spectrometry (ICP-MS). As a stable, low abundance isotope (2.12% abundance vs. 91.8% for ⁵⁶Fe), ⁵⁷Fe can effectively be used as a biological tracer; we and others have monitored the ratios of ⁵⁷Fe to other isotopes of Fe as an approach to track different Fe sources.^60–63 We chose to measure uptake in STm as a case study because this gastrointestinal pathogen produces and utilizes Ent and employs both FepA and IroN to transport it across the OM, and we recently demonstrated that Ent-based SACs are highly active against STm.³² Additionally, STm only produces the catecholate siderophores Ent and salmochelin. To avoid any convoluting effects of natively biosynthesized Ent, we generated STm strains with an entC knockout background. EntC is an isochorismate synthase that catalyzes an early step in the biosynthetic pathway to dihydroxybenzoic acid, a necessary precursor for Ent biosynthesis;¹¹ thus, the entC mutant strains are Ent (and therefore salmochelin) deficient.

STm cultures grown in an Fe-deficient modified M9 medium (<1 μM Fe, Table S5) were treated with varying concentrations of Ent or TC preloaded with 0.9 equivalents of ⁵⁷Fe (⁵⁷Fe-Ent or ⁵⁷Fe-TC) for 1 h. STm entC showed concentration-dependent uptake of both ⁵⁷Fe-Ent and ⁵⁷Fe-TC when treated at 10 μM or 1 μM, whereas 0.1 μM was not significantly different from the untreated control (Figure 2a). ⁵⁷Fe-TC treatments resulted in robust ⁵⁷Fe accumulation by STm (~70% that of ⁵⁷Fe-Ent). This finding is in line with prior reports that demonstrated the recognition of TC by Ec FepA using the radio tracer ⁵⁹Fe, and indicated Ent as a preferred substrate for Ec FepA compared to TC based on binding and transport efficiencies.^50–51 Initial recognition of substrates by FepA is thought to depend on the interaction of the catechol rings with the extracellular loops of the TBDR, while its ability to discriminate among triscatechol siderophores is speculated to depend on additional interactions within a secondary binding site.^{51, 64–67} In the case of TC, the loss of H-bonding groups in the TREN backbone compared to the trilactone ring has been proposed to account for its lower affinity and transport by FepA compared to Ent.⁵¹ Nevertheless, as our results support, the otherwise close structural mimicry of Ent by TC is sufficient for recognition and efficient uptake by the OM receptors for Ent.

Figure 2.

⁵⁷Fe-TC uptake by Ent transport machinery in STm. (a) Uptake of ⁵⁷Fe-Ent or ⁵⁷Fe-TC by STm entC; (b) uptake of ⁵⁷Fe-TC by STm entC or STm entC strains that lacked the Ent OM receptors FepA and/or IroN; the significance was compared to the parent entC strain. Assays performed in modified M9 medium, 1 h, 30 °C, n = 5. Statistical differences were calculated using two-tailed Student’s t test assuming unequal variances; * p < 0.05, ** p < 0.005, *** p < 0.0005.

Considering the uptake of ⁵⁷Fe-TC by STm entC, we measured uptake in the entC fepA, entC iroN, and entC fepA iroN mutant strains to evaluate the roles of FepA and IroN in transporting Fe-TC through the OM (Figure 2b). Strains expressing either FepA or IroN showed ⁵⁷Fe-TC uptake generally equivalent to the parent entC strain. In the strain lacking both Ent OM receptors, ⁵⁷Fe-TC uptake was significantly decreased for 10 μM treatment and effectively reduced to levels comparable to the untreated control for 1 μM treatment, supporting the key roles of FepA and IroN in the uptake of Fe-TC. ⁵⁷Fe-Ent exhibited a similar uptake profile among the OM receptor mutants (Figure S2). Overall, the significant and receptor-specific uptake of TC observed in these studies provided support for pursuing TC-based SACs.

Design and synthesis of TC-Amp.

To evaluate our hypothesis that TC can serve as an alternative vector for antibiotic delivery across the OM through FepA and IroN, we prepared a TC–ampicillin (TC-Amp) conjugate. Monofunctionalized TC intermediates and TC-Amp were designed and synthesized (Scheme 1) based on our Ent-β-lactam conjugates.^{29, 68} To achieve monofunctionalized TC scaffolds, we derivatized one catechol ring at the C5 position because this strategy, inspired by the structures of naturally occurring salmochelins and class IIb microcins,^{13, 68–69} has proven effective for cargo delivery by our Ent-based conjugates. Benzyl protected benzoic acid precursors 5 and 6 were synthesized as previously reported.⁶⁸ Then, a one-pot amide coupling was carried out with PyAOP as the coupling agent, combining 5, 6, and commercially available TREN 7 to assemble the TC scaffold. This reaction was performed using a 1.5:1:1 ratio of 5:6:7 based on prior optimization of monofunctionalized Ent synthesis to favor formation of the mono-substituted form.⁶⁸ The reaction produced a mixture of the tri-, di-, and mono-substituted analogs, and unfunctionalized benzyl protected TC, as expected. The mixture of products was separated by column chromatography using alumina rather than silica, improving resolution and avoiding retention of the tertiary amines, enabling isocratic elution. The desired mono-alkene product 8 was acquired as an off-white, iridescent foam in up to 43% yield. This intermediate was then oxidized in one step by OsO₄ and oxone in DMF to afford carboxylic acid 9 in 60% yield. In this step, careful pH adjustment was required after quenching the reaction with sodium sulfite prior to aqueous/organic extraction to avoid protonation of the tertiary amine by the generated sulfuric acid. The resulting carboxylic acid handle installed in the C5 position can facilitate a variety of possible chemistries to attach linkers and cargos, as we demonstrated for monofunctionalized Ent.⁶⁸ In this study, we chose to append a PEG₃ linker with a terminal azide to carboxylic acid 9 via amide coupling using HATU and HOAt to give 10 with 79% yield. Then, benzyl deprotection of the catechols was carried out employing BCl₃ to preserve the azide moiety, affording 11 in 44% yield following purification by reverse-phase preparative high pressure liquid chromatography (HPLC). Overall, this synthesis affords a monofunctionalized TC scaffold that is hydrolytically stable and possesses a flexible and stable linker and click chemistry capacity for modular cargo attachment.

Scheme 1: Synthesis of TC-Amp^a.

^a PyAOP = (7-azabenzotriazol-1-yloxy)tripyrrolidinophosphonium hexafluorophosphate, DIPEA = N,N-diisopropylethylamine, HATU = hexafluorophosphate azabenzotriazole tetramethyl uronium, HOAt = 1-hydroxy-7-azabenzotriazole, PMB = pentamethylbenzene

For the cargo, we chose ampicillin (Amp), a β-lactam amenable to modification without loss of antibacterial activity.²⁹ Because Amp targets penicillin-binding proteins, periplasmic proteins involved in cell wall biosynthesis, it requires transport across the OM of Gram-negative bacteria to exert its activity. Thus, employing Amp as the drug cargo enabled us to study siderophore-mediated OM transport. Alkyne-functionalized Amp 12 was prepared as previously described,²⁹ then conjugated to 11 via copper-catalyzed azide-alkyne cycloaddition. Cu(I) was generated by in situ reduction of CuSO₄ by sodium ascorbate (NaAsc) and stabilized by tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]-amine (TBTA) to prevent degradation of the β-lactam.²⁹ Purification was conducted by preparative HPLC with 0.05% TFA in the solvents to minimize β-lactam hydrolysis, yielding TC-Amp 4 in high purity (Figure S4) and 65% yield.

TC-Amp shows enhanced antibacterial activity against STm.

We tested the antibacterial activity of TC-Amp in parallel with Ent-Amp and the parent antibiotic Amp against STm. Side-by-side evaluation of these compounds provided an opportunity to investigate how the nonnative, nonhydrolyzable backbone of TC would affect activity, and examine whether differences between Ent and TC manifest in the comparison of the respective antibacterial activity of their conjugates. For these assays, we employed the Fe-deficient modified M9 medium used in the ⁵⁷Fe uptake studies, which provides Fe-limiting conditions and induces expression of siderophore uptake machinery.³³ We carried out a 10-fold dilution series, and administered SACs with apo-siderophore moieties, which presumably scavenge Fe from the growth medium.

TC-Amp showed enhanced antibacterial activity compared to Amp against STm, with almost complete inhibition of growth at 10⁻⁸ M, compared to the MIC of 10⁻⁵ M for Amp, representing a >100-fold increase in potency (Figure 3a). The activity of TC-Amp was comparable, if slightly attenuated, relative to that of Ent-Amp, which exhibited an MIC of 10⁻⁸ M, consistent with what we have previously observed for STm.³² The enhanced antibacterial activity of TC-Amp compared to Amp suggests that TC can act as an effective vector for the delivery of the β-lactam cargo into the periplasm of STm.

Figure 3.

TC-Amp exhibits enhanced antibacterial activity and accelerated time-kill kinetics against STm. (a) Antibacterial activity of TC-Amp against STm compared to that of Ent-Amp and Amp; performed in modified M9 medium, 20 h, 30 °C, n ≥ 3, mean ± STD. (b) Time-kill kinetic assays for STm treated with TC-Amp, Ent-Amp, or Amp, and 1% DMSO control; performed in modified M9 medium, 37 °C, n ≥ 3, mean ± STD.

TC-Amp displays accelerated killing of STm.

The active transport of SACs has the propensity to not only increase the potency of the warhead but also accelerate cell killing by the antibacterial cargo. To evaluate the impact of TC modification on the rate of cell killing, we carried out time-kill kinetics studies by treating STm (~3 × 10⁸ CFU/mL) with TC-Amp, Ent-Amp, and Amp in parallel. Treatment with 5 μM TC-Amp led to accelerated cell killing of STm relative to treatment with 50 μM Amp, and the time-kill kinetic profile of TC-Amp was indistinguishable from that of Ent-Amp (Figure 3b). These results are consistent with a model whereby introducing active transport by the TBDRs as the mode of uptake for the SACs versus the passive diffusion of Amp provides more rapid cell killing. This finding further supports the ability of TC to mediate delivery of the Amp drug cargo through the OM of STm in a manner comparable to that of Ent.

TC-Amp activity depends on Ent OM receptors.

To ascertain the involvement of Ent receptors in recognizing TC-Amp and transporting the conjugate into the periplasm, mutants in select TBDRs of STm were tested for their susceptibility to TC-Amp. Because we observed FepA- and IroN-dependent uptake of ⁵⁷Fe-TC (Figure 2), we anticipated both OM receptors would contribute to TC-Amp uptake. Other studies of catecholate-β-lactam conjugate activity in Ec have reported uptake dependence on the OM receptors for DHBS fragments Cir and Fiu, including triscatecholamide scaffolds with TREN or mesitylene backbones, as well as Cfdc.^{23, 38, 44–45} Although negligible uptake of ⁵⁷Fe-TC by the STm entC fepA iroN mutant was observed (Figure 2b), we wondered if uptake of the conjugate would differ from that of the unmodified siderophore (as seen by others³⁸), so we included Cir in the panel of TBDRs considered. We note that STm does not express Fiu, a catecholate siderophore transporter found in Ec. We prepared combinatorial knockouts of cirA, fepA, and iroN in STm, enabling us to compare the activity of TC-Amp in the seven resulting mutant strains (fepA, iroN, cirA, fepA iroN, cirA fepA, cirA iroN, and cirA fepA iroN) to the parental STm strain. Single mutants of fepA and iroN showed decreased susceptibility to TC-Amp, with nearly complete growth inhibition at 10⁻⁷ M, reflecting ~10-fold decrease in potency relative to TC-Amp against STm (Figure 4a). The fepA iroN double mutant showed further attenuated susceptibility, with an MIC of 10⁻⁶ M (almost 100-fold decrease compared to the parent strain). Taken together, these results demonstrate the dependence of TC-Amp antibacterial activity on uptake by both FepA and IroN, generally consistent with what we have observed for Ent-Amp.³² In the cirA mutant strains, we observed negligible effect of the single knockout of cirA on susceptibly to TC-Amp compared to the parent strain, and the cirA iroN and cirA fepA iroN strains showed similar susceptibility as the iroN and fepA iroN strains, respectively (Figure S5). These findings suggest that TC-Amp is not taken up by Cir in STm. By contrast, when STm and its cirA mutants were tested for their susceptibility to Cfdc, a 10-fold decrease in susceptibility was observed in the absence of Cir (Figure S6a), supporting its role in transporting this monocatechol-modified β-lactam antibiotic.

Figure 4.

TC-Amp activity is attenuated against STm Ent OM receptor mutants and in the presence of exogenous Ent. (a) Antibacterial activity of TC-Amp against STm and its isogenic mutants in fepA, iroN, and fepA iroN; performed in modified M9 medium, 20 h, 30 °C, n ≥ 3, mean ± STD; (b) TC-Amp and Ent-Amp antibacterial activity against STm in the presence of exogenous Ent, performed in 50% MHB + 100 μM DP, 20 h, 30 °C, n ≥ 3, mean ± STD.

Surprisingly, the cirA fepA mutant was sensitized to TC-Amp (Figure S5). A potential explanation is that deletion of cirA and fepA could lead to an overproduction of IroN, which may enhance the import of the conjugate and increase its potency. We also observed sensitization of the fepA and fepA iroN mutants to Cfdc (Figure S6b), which similarly could be due to an increased reliance on Cir in the absence of FepA and IroN, leading to enhanced import and activity. This type of phenomenon has also been observed by others for catechol-based conjugates tested in Ec OM receptor mutants.^{38, 70}

A few minor differences in the antibacterial activity profiles of TC-Amp and Ent-Amp do suggest slightly different effects on the receptor mutants: the STm fepA single mutant showed no change in susceptibility to Ent-Amp³² while a ~10-fold decrease in susceptibility was observed for TC-Amp, potentially indicating a greater reliance on FepA for the transport of TC-Amp. Additionally, TC-Amp inhibited growth of the fepA iroN and cirA fepA iroN strains at 10⁻⁶ M. For Ent-based conjugates, we routinely observed growth inhibition of Salmonella and Ec CFT073 fepA iroN strains at 10⁻⁵ M, attributed to Fe sequestration and starvation.^31–32 To evaluate any growth inhibitory effects of the TC moiety alone, we assayed the antibacterial activity of unmodified TC against STm and the fepA iroN mutant strain as a control (Figure S7). The parent strain exhibited no sensitivity to TC, whereas the fepA iroN mutant was inhibited at 10⁻⁶ M TC, mirroring the effect of TC-Amp. Thus, we speculate that TC-Amp exerts an Fe starvation effect on these receptor mutant strains at 10⁻⁶ M (Fe content in the medium is < 10⁻⁶ M). It is possible that we observe this effect at a lower concentration for TC-Amp than for Ent-based conjugates due to the stability of the TC moiety to hydrolysis.

TC-Amp competes with Ent for Ent OM receptors.

To further explore the interaction of Ent OM receptors with TC-Amp and Ent-Amp, we performed antibacterial activity assays of TC-Amp or Ent-Amp (0.1 μM) against STm in medium supplemented with exogenous Ent (0–10 μM) (Figure 4b). This approach enabled us to probe the competition for receptor recognition, because Ent uptake facilitates growth promotion whereas conjugate uptake inhibits growth. We employed 50% Mueller Hinton Broth (MHB; ~2 μM Fe, Table S5) supplemented with 2,2’-dipyridyl (DP) to induce Fe starvation as previously done for uptake competition studies.³⁰ TC-Amp exhibited a similar MIC against STm in 50% MHB + 100 μM DP as in modified M9 medium (Figure S8). When treated with TC-Amp, growth recovery of STm occurred when at least 0.5 μM Ent (5-fold excess) was administered. When treated with Ent-Amp, at least 2 μM Ent (20-fold excess) was required to recover STm growth, similar to what we observed for Ec CFT073 and UTI89.³⁰ These results illustrate the competition that occurs between TC- or Ent-based SACs and Ent at the OM receptors. Comparison of the profiles of TC-Amp and Ent-Amp indicated that TC-Amp activity was attenuated by the addition of exogenous Ent to a greater extent than Ent-Amp, suggesting that the ability of TC-Amp to compete with Ent for uptake by STm is lesser than that of Ent-Amp. This result is reminiscent of early competition studies of TC and Ent binding to FepA in Ec.⁵¹ Nevertheless, the potent antibacterial activity of TC-Amp against STm indicates that TC-Amp is recognized and transported into the periplasm even though endogenous Ent may be present in the culture. To probe the effects of natively produced Ent on the activity of the conjugate, we also tested the antibacterial activity of TC-Amp against STm entC in modified M9 medium (Figure S9). This strain was fully inhibited at the lowest concentration of TC-Amp tested, 10⁻⁹ M. The sensitization of the entC mutant relative to the parent strain suggests that some competition between endogenous Ent and TC-Amp for Fe and receptor recognition occurs in culture. Taken together, these findings further corroborate the recognition of TC-Amp by Ent OM receptors in STm, in both the presence and absence of Ent, and inform the ability of TC-Amp to compete with Ent for uptake by STm.

TC-Amp induces cell elongation and death in STm, consistent with the mode of action of Amp.

To provide further evidence for the mode of action of the Amp warhead following the import of TC-Amp by OM receptors, we carried out differential interference contrast (DIC) and fluorescence microscopy to examine the morphology and viability of treated cells (Figure 5). STm (~10⁹ CFU/mL) was treated with sub-MIC concentrations of TC-Amp or Amp estimated based on this higher cell density.³² When administered below the MIC, Amp primarily targets PBP3, a penicillin-binding protein that participates in cell division; consequently, Amp treatment induces bacterial cell filamentation in Ec.^71–72 We previously observed the same phenomenon for Amp treatment of STm.³² When treated with TC-Amp, STm cells exhibited cell elongation (~10 μm cell perimeter) compared to untreated cells (average of 4.8 μm cell perimeter). This morphology is comparable to the filamentation induced by sub-MIC Amp treatment and is consistent with the periplasmic delivery and activity of the Amp warhead. In contrast, the fepA iroN mutant strain showed normal cellular morphology upon TC-Amp treatment, in line with the dependence of conjugate uptake and activity on Ent OM receptors.

Figure 5.

TC-Amp induces cellular morphologies consistent with PBP3 inhibition by Amp. Representative DIC and fluorescence micrographs of STm and its fepA iroN mutant untreated or treated with sub-MIC concentrations of TC-Amp or Amp. Assays performed in modified M9 medium, 20 h, 30 °C, scale bar = 10 μm.

In parallel, we examined the viability of treated cells by employing LIVE/DEAD staining, which utilizes fluorescent nucleic acid stains SYTO 9 and propidium iodide to distinguish between cells with intact OM (alive, green) and damaged OM (dead, red).⁷³ As expected, the vast majority of untreated cells stained green. By contrast, Amp treatment and TC-Amp treatment of STm resulted in a mixture of green and red cells, indicating some live cells and some dead cells, the latter of which presumably underwent lysis due to penicillin-binding protein inhibition. TC-Amp-treated fepA iroN cells stained green and thus remained alive, consistent with reduced conjugate activity in the absence of the TBDRs. Overall, our observations of STm cell morphology and viability are consistent with our results for Ent-Amp,³² and further support the ability of TC to act as a vector for Amp delivery into the STm periplasm, where it can interact with the penicillin-binding proteins it targets.

TC modification enables selective killing of STm in the presence of commensal Lactobacilli.

Concomitant with the ability to enhance delivery of β-lactams into bacteria expressing Ent uptake machinery, siderophore modification has the potential to prevent uptake of these broad-spectrum antibiotics by bacteria that do not have Ent receptors, generating narrow spectrum antibacterial agents. Lactobacilli are Gram-positive commensal bacteria of the human gastrointestinal tract that may aid the host in combating STm infections.^74–79 Therefore, it is important that antibiotics administered do not also deplete the Lactobacilli population. Lactobacilli have been considered to require minimal metabolic iron and do not employ siderophores for iron acquisition.^80–83 We employed Lactobacillus rhamnosus (Lr) GG, a probiotic that is susceptible to Amp, to probe whether TC-Amp provides selective killing of STm in the presence of Lactobacilli. This strain was previously used in our studies of the selectivity of Ent- and salmochelin-based conjugates of Amp and amoxicillin toward uropathogenic Ec (UPEC).³⁰

Antibacterial activity assays showed that the growth of Lr GG monoculture was unaffected by treatment with up to 10 μM TC-Amp, whereas treatment with 10 μM Amp fully inhibited its growth (Figure S10). We carried out time-kill kinetics assays on cocultures of STm and Lr GG to evaluate the selectivity of TC-Amp toward STm in the presence of this commensal Lactobacilli. When the coculture (~3×10⁸ CFU/mL of each strain) was treated with 5 μM TC-Amp, STm was rapidly killed while Lr GG growth was unaffected (Figure 6). The same concentration of Amp exhibited slight growth inhibitory effects over the course of 4 h on both STm and Lr GG in coculture. Thus, the enhanced antibacterial activity of TC-Amp toward STm combined with the attenuated sensitivity of Lr GG to the TC-modified β-lactam permits conditions where TC-Amp can achieve selective killing of an enteropathogen in the presence of the commensal Lactobacilli, demonstrating the capacity of TC-modification to target drug delivery to the Ent-utilizing Gram-negative organism. Together with what we have previously shown for Ent-based conjugates and UPEC,³⁰ this result further highlights that siderophore conjugation can serve to narrow the antibiotic spectrum of the cargo. Whether Ent- and TC-based conjugates affect commensals that were recently discovered to take up Ent, such as Bacteroides thetaiotaomicron (which notably employs a distinct uptake mechanism involving a secreted Ent-binding lipoprotein),^84–85 requires investigation as well as what interplay may occur between commensal and pathogenic strains and SACs in the complex host environment.

Figure 6.

TC-Amp selectively kills STm in the presence of commensal Lactobacilli. Time-kill kinetics assay of STm + Lr GG cocultures treated with 5 μM TC-Amp, 5 μM Amp, and 1% DMSO control; performed in 1:1 MRS/MHB + 200 μM DP, 37 °C, n ≥ 3, mean ± STD.

TC-Amp is active towards other pathogenic Enterobacteriaceae.

To examine the antibacterial activity of TC-Amp beyond STm, we assessed its activity against other pathogenic strains within Enterobacteriaceae that are susceptible to Ent-based conjugates due to their expression of Ent uptake machinery. These pathogens include Salmonella enterica serovar Enteritidis (SEd), as well as UPEC strains CFT073 and UTI89, all of which express FepA and IroN.^{29, 32} We note that CFT073 also possesses an additional catecholate receptor Iha.⁸⁶ All three organisms were susceptible to Amp with an MIC of 10⁻⁵ M and displayed the expected susceptibility to Ent-Amp with MICs of 10⁻⁷–10⁻⁸ M.^31–32 TC-Amp displayed enhanced antibacterial activity against SEd with partial growth inhibition at 10⁻⁸ M, and an MIC of 10⁻⁷ M (Figure 7a), a profile similar to STm. TC-Amp also exhibited enhanced antibacterial activity against both UPEC strains, with partial growth inhibition at 10⁻⁷ M and MICs of 10⁻⁶ M (Figure 7b, 7c). Overall, these experiments extend the applicability of TC-Amp to enhance the activity of the Amp warhead against clinically relevant Ent-producing bacterial pathogens.

Figure 7.

TC-Amp exhibits enhanced antibacterial activity against additional Enterobacteriaceae. Antibacterial activity of TC-Amp compared to that of Ent-Amp and Amp against (a) SEd, (b) Ec CFT073, and (c) Ec UTI89. Assays performed in modified M9 medium, 20 h, 30 °C, n ≥ 3, mean ± STD.

Preservation of the Amp warhead by a β-lactamase inhibitor reveals enhanced activity of TC-Amp and Ent-Amp against Kp.

We also tested the activity of Ent-Amp and TC-Amp against another member of Enterobacteriaceae, Kp ATCC 13883, which biosynthesizes and utilizes Ent, and expresses FepA, Cir, and IroN.⁸⁷ We observed no susceptibility of this strain to Ent-Amp or TC-Amp up to 10⁻⁵ M (Figure S12a). This observation is consistent with the negligible activity of Ent-β-lactam conjugates against Kp ATCC 13883 in 50% MHB supplemented with DP.²⁹ Because Kp possesses a chromosomally encoded serine β-lactamase⁸⁸ and is intrinsically resistant to Amp, we posited that this inactivity was due to β-lactamase inactivation of the aminopenicillin warhead. Coadministration of the β-lactamase inhibitor sulbactam (SB) with Amp is required to achieve Amp activity against Kp, and Amp/SB combination therapy is approved for treating infections of Kp.⁸⁹ We observed that Kp ATCC 13883 was resistant to Amp up to 10⁻⁴ M, but cotreatment with SB (1:0.75 Amp/SB, based on the clinical formulation) partially inhibited growth at 10⁻⁵ M and exhibited an MIC of 10⁻⁴ M (SB showed no activity when administered alone) (Figure S12b). Thus, we hypothesized that cotreatment with SB could reveal enhanced antibacterial activity of Ent-Amp and TC-Amp against Kp. A relatively high concentration of SB would be required to inhibit the β-lactamase based on the Amp/SB MIC, while the active uptake of the conjugates would reduce the quantity of the Amp warhead necessary to inhibit growth. Thus, we tested a 10-fold dilution series of Ent-Amp or TC-Amp in medium supplemented with 10 μM SB, which is similar to the concentration for SB and other β-lactamase inhibitors published in Clinical and Laboratory Standards Institute breakpoint guidelines (4 μg/mL, ~17 μM for SB).⁹⁰ Under these conditions, we observed MIC values of 10⁻⁶ M for both conjugates against Kp ATCC 13883, corresponding to a 100-fold increase in potency compared to cotreatment with Amp and SB (Figure 8). These findings support the hypothesis that the β-lactamase produced by Kp is indeed responsible for the inactivity of Ent-Amp and TC-Amp. Moreover, the results demonstrate that in this Amp-resistant strain, siderophore conjugation can still enhance activity so long as the warhead is preserved. This result opens new doors regarding SAC activity against Kp, which is relatively understudied in the field, and illuminates the possibility of SAC utilization within combination therapies.^91–92 It also motivates the study of conjugates employing more advanced, β-lactamase resistant warheads, such as meropenem,³¹ to target β-lactamase-producing strains.

Figure 8.

TC-Amp and Ent-Amp show enhanced activity against Kp when co-administered with SB. Antibacterial activity TC-Amp and Ent-Amp with 10 μM SB compared to Amp/SB (1:0.75) against Kp ATCC 13883. Assays performed in modified M9 medium, 20 h, 30 °C, n ≥ 3, mean ± STD.

TC-Amp and Ent-Amp are active against other Gram-negative pathogens that use Ent as a xenosiderophore.

We looked beyond Enterobacteriaceae and evaluated two additional Gram-negative bacterial ESKAPE pathogens, Ab and Pa. Both pathogens employ Ent as a xenosiderophore; these species do not produce Ent but express OM Ent receptors (FepA in Ab and PfeA in Pa).^{16, 35–36} Ab and Pa are both intrinsically resistant to many β-lactams due to a combination of (i) OM impermeability based on porin structure and prevalence, (ii) efflux pump expression, and (iii) β-lactamase activity.^93–97 SACs have the potential to circumvent the former two resistance mechanisms based on their active transport through the OM and resulting increased cellular accumulation. We found that Ab ATCC 17978, Pa PAO1, and Pa JSRI-1 grew sufficiently well in modified M9 medium, so antibacterial activity assays for these organisms were carried out using this medium. We note that we previously observed little sensitivity of Ab ATCC 17961 and Pa PAO1 to Ent-based conjugates in 50% MHB supplemented with DP,^29–30 but we revisited these organisms in the current study as the growth medium can affect the activity of SACs.³³ Additionally, Ab 17978 has since been characterized for Ent uptake.³⁶

Ab ATCC 17978 is a well-studied clinical isolate and required 10⁻⁴ M Amp to completely inhibit growth. When treated with the SACs, growth of this strain was significantly inhibited at sub-micromolar concentrations of TC-Amp or Ent-Amp, with the MIC for each conjugate at 10⁻⁶ M (Figure 9a). Based on the antibacterial activity plots, TC-Amp appears slightly more active than Ent-Amp. To determine the role of the Ent receptor in Ab in taking up Ent-Amp and TC-Amp, the Ab 17978 fepA mutant was examined (Figure S13a, S13b). This strain showed reduced susceptibility to Ent-Amp and TC-Amp compared to the parent strain, consistent with the involvement of FepA in transporting Ent-Amp and TC-Amp and facilitating their enhanced antibacterial activities against Ab.

Figure 9.

TC-Amp and Ent-Amp exhibit enhanced activity against Ab and Pa strains. Antibacterial activity of TC-Amp and Ent-Amp compared to Amp against (a) Ab 17978, (b) Pa PAO1, and (c) Pa JSRI-1. Assays performed in modified M9 medium, 20 h, 30 °C, n ≥ 3, mean ± STD.

Pa PAO1 is a laboratory strain that is commonly used for genetic manipulation, whereas Pa JSRI-1 is a clinical isolate from the lung of a cystic fibrosis patient.⁹⁸ Both Pa strains were insensitive to Amp up to 10⁻⁴ M. When treated with the SACs, growth of Pa PAO1 was >80% inhibited by either Ent-Amp or TC-Amp at 10⁻⁶ M (Figure 9b). Pa JSRI-1 was partially inhibited at 10⁻⁷ M Ent-Amp and 10⁻⁶ M TC-Amp, and both conjugates inhibited growth ~90% at 10⁻⁵ M (Figure 9c). It is remarkable that a Pa isolate adapted to the human lung shows similar susceptibility to the SACs as the laboratory strain. This finding provides support for future testing of these SACs in preclinical models.

In contrast to our findings for Kp, enhanced activity of the Amp conjugates against Ab and Pa was observed without the need for a β-lactamase inhibitor. This result illustrates that the different resistance mechanisms bacteria may rely on to achieve resistance to an antibiotic can manifest in differential susceptibility to an SAC employing that warhead. The ability of TC-Amp and Ent-Amp to inhibit these Amp-resistant strains provides further support for the strategy of siderophore conjugation to overcome permeability and accumulation related drug resistance and to hijack xenosiderophore uptake machinery. Other examples of triscatechol-based conjugates have shown promise against Ab and Pa,^37–39 and our findings further highlight the feasibility of using Ent-inspired SACs against these pathogens.

Ent-Amp and TC-Amp activity is comparable to Cfdc.

The monocatechol-β-lactam Cfdc (Figure S1) is indicated for treatment of complicated urinary tract infection and bacterial pneumonia caused by many Gram-negative organisms including Ec, Kp, Pa, and Ab, and has been tested for its in vitro potency against large panels of bacterial pathogens.²³ To further benchmark and contextualize the antibacterial activity of our triscatecholamide conjugates, we performed a comparative analysis of Cfdc, Ent-Amp, and TC-Amp activity against eight Gram-negative bacterial pathogens in modified M9 medium (Table 1). Under these conditions, Cfdc treatment afforded MIC values that were generally consistent with those previously reported for Pa, Ab, and Kp strains,^{23, 99} and ≥100-fold lower than the MIC values of Amp. Overall, the Ent- and TC-based SACs performed comparably to Cfdc. For almost all strains tested, Ent-Amp and TC-Amp achieved potency equivalent to or within 10-fold of Cfdc. Remarkably, Ent-Amp and TC-Amp were found to be more potent than Cfdc against both Salmonella serovars. In the absence of a β-lactamase inhibitor, Cdfc is >100-fold more potent against Kp than the conjugates, whereas equivalent potency is observed when the conjugates are co-administered with SB. We speculate that Cfdc is not a substrate for the penicillinase expressed by Kp ATCC 13883; its highly functionalized cephalosporin core was designed to evade β-lactamases, and it retains activity against bacteria harboring various classes of β-lactamases, including several Kp strains.^{21, 23} Thus, we reason that comparison of Ent-Amp/TC-Amp + SB to Cfdc is justified. Our findings motivate further investigation of these SACs in more physiologically relevant models, as well as future study of their performance in vivo relative to the clinically implemented Cfdc.

Table 1.

Antibacterial activity of compounds in this study^a

		Concentration (M)a
Strain	Amp	Ent-Amp	TC-Amp	Cfdc
STm	10−5	10−8	~10−8	10−7
SEd	10−5	10−9	10−7	10−7
Ec CFT073	10−5	~10−8	10−6	10−8
Ec UTI89	10−5	10−7	10−6	10−8
Ab 17978	10−4	10−6	10−6	10−7
Pa PAO1	>10−4	~10−6	~10−6	~10−7
Pa JSRI-1	>10−4	~10−5	~10−5	10−6
Kp 13883	>10−4	>10−5	>10−5	10−6
+ SB	10−4	10−6	10−6	n.d.

^aConcentration (M) that resulted in >90% reduction in OD₆₀₀ value; ~ indicates concentration with 80–90% reduction in OD₆₀₀ observed; n.d., not determined. For values in μM and μg/mL, see Table S6.

Conclusion and Outlook

In this work, we demonstrated the ability of TC, a triscatecholamide siderophore analog with a nonnative backbone, to facilitate the delivery of β-lactam drug cargo through the OM of Gram-negative bacteria via Ent uptake machinery. We established the synthesis of a monofunctionalized TC scaffold with a conjugation handle installed in the C5 position of one catechol moiety, which provides hydrolytic stability and allows for modular cargo attachment by amide coupling or click chemistry. Conjugation of the antibiotic cargo Amp to TC afforded enhanced antibacterial activity and accelerated killing of STm while preserving the mode of action of the Amp warhead, which binds to penicillin-binding proteins in the periplasm following import of the conjugate by Ent OM receptors. TC conjugation also targeted the broad-spectrum antibiotic Amp to Gram-negative species with Ent OM receptors and away from non-Ent-utilizing commensal Lactobacilli. Overall, the activity of TC-Amp is a promising step toward the use of nonhydrolyzable Ent mimics as the delivery vector for SACs that hijack Ent uptake machinery.

Beyond the demonstration of TC-mediated OM transport of drug cargo, our platform for Ent- and now TC-based conjugates positioned us to address the question of how synthetic mimics employed in SACs can compare to the native siderophore after which they are modeled. As the conjugation handle of our siderophore scaffold is located on the catechol group, rather than the backbone, we could directly compare the TC-based conjugate to the analogous conjugate based on the native siderophore Ent. Early on, we considered that employing a nonnative backbone within a triscatecholamide siderophore could detract from its ability to compete with native siderophores for sequestration of Fe and binding to receptors relative to a derivative of native Ent.^{47, 50–51} Our results from ⁵⁷Fe uptake and Ent competition assays indicate the preference of the STm OM receptors for the native substrate, Ent, over TC. Nevertheless, TC was effective in targeting the same transport machinery for efficient delivery of the drug cargo, as TC-Amp and Ent-Amp showed comparable antibacterial activity and time-kill kinetics, which are markedly better than unmodified Amp. Overall, our comparative analysis of TC-Amp and Ent-Amp provide compelling evidence that a conjugate based on a nonnative but highly mimetic scaffold can achieve activity on par with the analogous conjugate based on the native siderophore.

Finally, we pursued expanded antibacterial activity studies that support the therapeutic potential and preclinical testing for conjugates like Ent-Amp and TC-Amp. We demonstrated that both SACs achieve enhanced antibacterial activity against several Gram-negative pathogens of significant clinical concern. These pathogens included strains with intrinsic Amp resistance, which required co-administration of a β-lactamase inhibitor in the case of Kp but not Ab and Pa, highlighting the potential of siderophore conjugation to combat select resistance mechanisms. We additionally established the potency of Ent-Amp and TC-Amp as commensurate, under the conditions tested, with the FDA-approved and clinically implemented Cfdc. The ability of these proof-of-concept SACs to achieve activity comparable to the optimized scaffold of Cfdc highlights the promise of our platform for Ent- and TC-based conjugates, of which the modular nature can facilitate introduction of more advanced warheads or optimized siderophore moieties to further improve activity and selectivity.

The need for narrow-spectrum antibacterials and new antibiotic delivery approaches grows in parallel with the rise of antibiotic resistance. Exploiting the active transport of essential nutrient acquisition systems such as the siderophore uptake machinery is a promising approach to combat OM impermeability and reduce perturbation of commensal microbes. Enthusiasm toward the field of SACs in recent years has raised the bar for characterizing the activity, biological interactions, mode of action, and downstream effects of these types of conjugates, deepening our collective understanding of their cellular fate and aiding in the ongoing design and improvement of new conjugates. The TC-mediated delivery of Amp presented here provides a proof-of-concept for the use of TC as a delivery vector through the OM of Gram-negative pathogens via Ent uptake machinery. Furthermore, our investigation of the selectivity of TC-Amp, results from the comparative studies of TC-Amp and Ent-Amp, and the demonstrated potency of Ent-Amp and TC-Amp against Amp-resistant strains aid in generalizing principles of siderophore conjugation and reveal an expanded scope of use for these SACs. Moving forward, additional complexities can be considered regarding variations in the siderophore moiety of SACs. For example, alternative siderophore backbone structures can have differential influence on sensing-based regulation of receptor expression,^{57, 100–101} may impact the cellular compartment in which the siderophore ultimately accumulates (periplasm or cytoplasm),^{38, 102} and can affect whether the Fe bound by the siderophore is co-delivered (e.g., released by hydrolysis) with the drug cargo.³⁴ For strains that biosynthesis and transport diverse siderophores beyond Ent and salmochelins (e.g., pyoverdine by Pa, aerobactin by Ec), probing the role that these natural products play in the susceptibility to Ent-inspired SACs will be informative. Looking ahead, we seek to evaluate whether TC provides delivery of antibacterial cargos with cytoplasmic targets across the IM via FepCDG, as well as the extent to which the hydrolytic stability of TC may be a key advantage over Ent-based conjugates. Future endeavors to deliver alternative cargos and tune the selectivity of Ent- and TC-based conjugates will also broaden the therapeutic potential of SACs.

Experimental Methods

Complete experimental methods are provided as Supporting Information.