In Vivo Activation of Duocarmycin–Antibody Conjugates by Near-Infrared Light

Abstract

Near-IR photocaging groups based on the heptamethine cyanine scaffold present the opportunity to visualize and then treat diseased tissue with potent bioactive molecules. Here we describe fundamental chemical studies that enable biological validation of this approach. Guided by rational design, including computational analysis, we characterize the impact of structural alterations on the cyanine uncaging reaction. A modest change to the ethylenediamine linker (N,N′-dimethyl to N,N′-diethyl) leads to a bathochromic shift in the absorbance maxima, while decreasing background hydrolysis. Building on these structure–function relationship studies, we prepare antibody conjugates that uncage a derivative of duocarmycin, a potent cytotoxic natural product. The optimal conjugate, CyEt-Pan-Duo, undergoes small molecule release with 780 nm light, exhibits activity in the picomolar range, and demonstrates excellent light-to-dark selectivity. Mouse xenograft studies illustrate that the construct can be imaged in vivo prior to uncaging with an external laser source. Significant reduction in tumor burden is observed following a single dose of conjugate and near-IR light. These studies define key chemical principles that enable the identification of cyanine-based photocages with enhanced properties for in vivo drug delivery.

Short abstract

A method to precisely deliver a potent antitumor agent using near-IR light is reported. We have defined key features of the cyanine-based photocaging group, enabling the identification of a construct suitable for in vivo imaging and treatment.

Introduction

Molecular tools that respond to near-IR light enable the diagnosis and treatment of various disease states. There has been particular emphasis on addressing certain types of solid tumors. Targeted fluorescent markers enable emerging image-guided surgical interventions, providing real-time definition of tumor margins.^1,2 In the context of light-based treatment, photodynamic therapy (PDT) methods that rely on the local generation of toxic reactive oxygen species (ROS) have been investigated extensively.^3,4 While useful, there may be substantial benefits to methods that use near-IR light to site-specifically release bioactive molecules. Imparting potent pharmacological agents with high spatial control could mitigate systemic toxicity, while delivering otherwise unattainable local drug concentrations.⁵ However, the development of such methods presents a significant chemical challenge.⁶ Uncaging reactions initiated by easily attainable single-photon flux of near-IR light remain rare relative to their counterparts that rely on UV or blue light.^7,8 While recent progress presents tangible opportunities in this area, significant effort is still required.⁹⁻¹⁹ Molecularly well-defined approaches that are well tolerated and stable following systemic administration are needed. Moreover, it would be highly advantageous to be able to evaluate target accumulation prior to release of a potent payload molecule (i.e., theranostic applications).²⁰

To enable the union of fluorescence imaging with targeted small molecule release, we have sought to convert the heptamethine cyanine scaffold into a photocaging group. Benefiting from useful near-IR fluorescent properties and excellent biological compatibility, heptamethine cyanines constitute the chemical component of extensive preclinical and clinical imaging efforts.²¹⁻²⁴ We have demonstrated that C4′-dialkylamine-substituted cyanines undergo small molecule release upon exposure to light in the 690 nm range.²⁵ As shown in Figure 1A, the mechanism of uncaging comprises photochemical and thermal reaction components. The photochemical process, which was previously associated with cyanine photodegradation, entails regioselective photooxidative cleavage of the cyanine polyene via dioxetane intermediates formed from self-sensitized ¹O₂.^26,27 The thermal phase entails C4′-N hydrolysis and then intramolecular cyclization to release phenol payloads. In an initial communication, we applied this method to a near-IR light-activated antibody–drug conjugate (ADC) strategy (Cy-Pan-CA4, Figure 1B).²⁸ However, several aspects of our approach required refinement prior to pursuing in vivo efficacy studies (see below).

Figure 1.

(A) Mechanism of the cyanine uncaging reaction and (B) evolution of the near-IR light-activated ADC strategy.

Here we define key structure–function relationships of the cyanine caging group. These fundamental chemical studies enable the identification of a cyanine–antibody conjugate capable of efficient in vivo drug delivery. We have found that altering the linker domain and cyanine heterocycles provides meaningful improvements in stability, while inducing a significant bathochromic (red) shift in the absorbance maxima (λ_max). Building on these observations, we prepare antibody conjugates that release a derivative of the DNA-alkylating natural product duocarmycin. Enabled by studies of Boger, the duocarmycin class of natural products are exceptionally potent small molecule cytotoxins finding application as ADC payloads.^29,30 The optimal cyanine conjugate, CyEt-Pan-Duo (Figure 1B), displays light-dependent cellular activity in the picomolar range and can be readily activated with 780 nm light. Studies in mouse models show that the conjugate is well tolerated, can be readily visualized using fluorescence imaging, and displays significant antitumor efficacy following external near-IR irradiation. These studies provide chemical insights that enable the identification of cyanine-based photocages capable of modulating biological outcomes in live animal settings.

Results and Discussion

Optimization of Scaffold

A retrospective analysis of the first-generation ADC scaffold, Cy-Pan-CA4²⁸ (Figure 1B), identified several issues to be addressed prior to additional in vivo study. These include enhancing the potency of the payload molecule, enabling higher degree of labeling (DOL), and improving the therapeutic index (defined here as the difference between irradiated and unirradiated IC₅₀). To address the latter, we envisioned decreasing background hydrolysis reactions encountered over the long timeframes required for in vivo applications. Finally, given that heptamethine cyanines frequently exhibit λ_max approaching 800 nm, we speculated that constructs activated by light in this range might be identified.

We predicted that modifications to the ethylenediamine linker domain might reduce the rate of background hydrolysis, while also extending the λ_max further into the near-IR range. In regard to the former, previous reports suggest, perhaps unsurprisingly, that the steric environment of the carbamate functional group can dramatically impact cleavage kinetics.³¹ For the more complex issue of cyanine λ_max, we first took note of an observation that the conversion of a C4′-primary amine to a C4′-secondary amine is accompanied by a bathochromic shift.³² To provide additional insight, quantum mechanical calculations comparing C4′-dimethylamine and diethylamine-substituted heptamethine cyanines were carried out (Figure S1). Calculations at the ORMAS-PT2-PCM (water solvent)³³⁻³⁵ level predict a 31 nm bathochromic shift between the C4′-dimethylamino (λ_computed = 699 nm) and C4′-diethylamino (λ_computed = 730 nm) congeners. Examination of the three-dimensional structure suggests that this effect arises from decreased C4′-N lone pair donation into the cyanine polyene in the more sterically demanding diethyl analogue (see Supporting Information for detailed discussion).

To test these design concepts, we generated cyanine-caged compounds that release the fluorogenic molecule 4-methylumbelliferone (Umb) upon irradiation.^25,36 Using these compounds, we evaluate three key parameters of the uncaging process through the following experimental approaches (Figure 2). First, in method A, we measure the loss of the near-IR cyanine absorbance as a function of 690 nm irradiation. These values are referred to below as photooxidation efficiency and interrogate the photochemical reactivity of the cyanine scaffold. Of note, the values in Figure 2 are presented without correction for differences in absorbance to facilitate comparison under operational conditions. Relative rate constants corrected for absorbance are presented in Table S1. In method B, we measure Umb release by first carrying out an exhaustive photooxidation at 0 °C in a 96-well plate. The plate is transferred to 37 °C, and Umb release, which is typically negligible initially, is then measured over time. This approach serves to separate the cyanine photooxidation process from the hydrolysis/cyclization thermal reaction cascade, enabling quantitative comparison of the latter. Finally, method C, the analogous experiment without irradiation, examines background hydrolysis. The data for the three experiments are reported both as k_rel relative to compound 1, which contains the linker and cyanine scaffold used in our previous studies,²⁵ and as half-lives (t_1/2) for methods A and B and time to 10% Umb release (t_10%) for the slower process captured in method C. The synthesis of compounds 1–8 entails a 2-step or 3-step sequence from the corresponding C4′-chloro precursor and is described in the Supporting Information.

Figure 2.

Impact of modifications to the cyanine caging scaffold on photooxidation efficiency (method A), uncaging kinetics (method B), and background hydrolysis (method C). Method A: Absorbance traces at the λ_max of a 1 μM solution of each cyanine cage in pH 7.4 PBS irradiated with a 690 ± 20 nm LED (20 mW/cm²). Method B: Umb release (measured from a standard curve using fluorescence) from cyanine cage (1 μM, pH 7.4 PBS). Exhaustive photooxidation at 0 °C (15–180 J/cm²) was followed by monitoring Umb release at 37 °C (30 min intervals). Method C: Umb release at 37 °C (2.5 h intervals) from cyanine cage (10 μM, pH 7.4 PBS). For methods A and B, the experiments were run to completion and the data was fit to one phase decay parameters. For method C, the slow rate of background Umb release precluded this analysis. The reported values for the time to 10% reaction conversion (t_10%) is either directly observed or extrapolated from the initial slope (see Supporting Information for details). *Less than 20% Umb release yield after 5 h.

We first examined installation of the N,N′-diethylethylenediamine linker to provide compound 2. To our delight, 2 exhibits a 40 nm bathochromic shift (674 nm vs 714 nm) relative to 1. The photooxidation and background hydrolysis rates were both moderately improved (increased in the former to k_rel = 2.8 and decreased in the latter to k_rel = 0.73), while uncaging kinetics are somewhat reduced (k_rel = 0.81). Of note, we briefly pursued changes to the ethylenediamine linker to provide additional steric bulk. Efforts in this area were generally circumvented by synthetic considerations. For example, installing a gem-dimethyl group within the linker proved problematic due to decreased reactivity of the key secondary amine synthetic intermediate.

We then explored modifications to the central carbocyclic ring of the cyanine scaffold. Exchange of the central cyclohexenyl ring for a cyclopentenyl ring to provide 3 more than tripled the photooxidation rate (k_rel = 3.5), but nearly completely suppressed uncaging with regard to both rate (k_rel = 0.013) and yield (<20% after 5 h). Examining relative mass spectral ion counts over time, we found that the expected carbonyl intermediates (Figure 1A) are formed but do not efficiently undergo the key hydrolysis event (Figure S2).²⁵ Efforts to synthesize the cycloheptenyl analogue of 1 were thwarted by hydrolytic instability of key cyanine intermediates.

We have also prepared compounds where one (4) or both heterocycles (5) were modified to a benzothiazole ring. These studies were guided by the notion that such derivatives might undergo photooxidation with improved efficiency, which had been observed by Hahn in fluorophore studies.³⁷ In line with these observations, 4 and 5 underwent photooxidation with k_rel’s of 3.7 and 6.8, respectively. However, to our surprise, these compounds were both dramatically more susceptible to background hydrolysis (k_rel’s of 4.7 and 8.5, respectively). Moreover, compound 5 released Umb with significantly reduced efficiency. Motivated principally by the requirements of in vivo drug delivery (i.e., to reduce background release reactions), these modifications were not pursued further.

A critical challenge in the generation of these cyanine-containing antibody conjugates is to reduce aggregation upon biomolecule conjugation.³⁸ To address this problem for fluorescence imaging applications, cyanines are often modified with aryl sulfonate functional groups.³⁹ We initially found that the congener resulting from indolenine ring sulfonation, 6, exhibited significantly reduced photooxidation efficiency and increased background hydrolysis (k_rel’s of 0.43 and 1.3, respectively), albeit with remarkably enhanced release kinetics (k_rel = 4.2). Consequently, we installed a sulfonated benz[e]indole ring, a modification used in fluorescence applications to provide a bathochromic shift in λ_max.^40,41 We also installed an alkyne handle to facilitate ultimate bioconjugation.⁴² The resulting compounds 7 and 8, which incorporate N,N′-dimethyl and N,N′-diethyl linkers, respectively, exhibit 16 and 18 nm bathochromic shifts relative to 1 and 2. Promisingly, release kinetics are similar to those of 1, while background hydrolysis is significantly decreased in both cases and most dramatically with 8 (k_rel = 0.28). While the photooxidation efficiency is slightly reduced relative to 1, we noted that, particularly with 8, the 690 nm LED was not optimally matched to its λ_max. Using a 740 nm LED under otherwise identical conditions, we observe photooxidation efficiency that is moderately improved relative to compound 1 (k_rel = 1.2, t_1/2 = 13 min, Figure S3). Consequently, scaffolds 7 and 8 were selected for further evaluation as small molecule delivery agents. Overall, this first foray into structure–function relationships of cyanine photocages reveals that rational design can significantly alter, and, more importantly, augment, key aspects of the photooxidation and release processes.

Synthesis of Cyanine-Caged Duocarmycin Conjugates

Using insights gained in these scaffold optimization studies, we then pursued antibody conjugates with a derivative of the natural product duocarmycin. Two major considerations informed the choice of payload. First, these natural products and their congeners exhibit potency in the picomolar range against a variety of cell lines, even those that exhibit significant drug resistance to more conventional cytotoxic therapies.^29,43 Although not typically suitable for use as untargeted agents due to toxicity concerns, duocarmycin–antibody conjugates created using conventional linker technologies show promising efficacy, including against tumor types exhibiting only modest levels of antigen expression.⁴⁴ Second, modification through the phenol of the seco-form of the alkylation subunit provides a convenient handle for conjugation that also traps the compound in an initially inactive form. While not previously used for the generation of a photoactivatable linker domain, this property has been exploited in several contexts.⁴⁴⁻⁴⁸

The synthesis of key cyanines 18 and 19 is shown in Scheme 1. In brief, the C4′-chloro cyanine 11 was assembled through the typical method from 9 and 10. S_RN1 reaction of 11 provides C4′-N,N′-dimethyl 12 and N,N′-diethyl 13 in 76% and 60% yield, respectively.⁴⁹ The union of cyanines 12 and 13 to the commercial duocarmycin DM payload^50,51 was accomplished via acylation of the secondary amine with the 4-nitrophenyl mixed carbonate 14. Copper-catalyzed [3 + 2] cycloaddition with commercial azide 15 provides 16 and 17 in suitable yield over two steps after reversed phase purification. Final conversion to the NHS ester with TSTU provided the fully functionalized cyanines 18 and 19, which are suitable for antibody conjugation. The optical properties of cyanines 16 and 17 are shown in Table 1. In addition to a 45 nm bathochromic shift relative to 16, N,N′-diethyl cyanine 17 is almost 2-fold brighter due to a higher molar absorptivity.

Scheme 1. Synthesis of Cyanines 18 and 19.

Table 1. Optical Properties of 16 and 17.

	λabs (nm)a	λem (nm)a	ε (M–1 cm–1)a	ΦF (MeOH)	ΦF (PBS)	ε × ΦF (PBS)	rel brightness (PBS)
16 (R = Me)	705	803	49,000	0.10	0.04	1960	1.0
17 (R = Et)	750	807	70,000	0.11	0.05	3500	1.8

^aMeasured at 5 μM in 50 mM PBS pH = 7.4.

We prepared and evaluated the corresponding cyanine–antibody conjugates. Panitumumab, a clinically used monoclonal anti-EGFR antibody, was chosen as the antibody component.²⁸ Cyanines 18 and 19 were conjugated to panitumumab using conventional conditions (pH 8.5 PBS buffer) with 4.5 equiv of the small molecule and purified using preparative size-exclusion chromatography (SEC) to provide CyMe-Pan-Duo and CyEt-Pan-Duo (DOL 4.0–4.3). The purity of the conjugates was confirmed by SDS–PAGE.

In Vitro and In Vivo Drug Delivery

We characterized the in vitro efficacy of CyMe-Pan-Duo and CyEt-Pan-Duo using EGFR+ (MDA-MB-468) and EGFR– (MCF-7) cells. Binding specificity was confirmed using fluorescence microscopy. Both conjugates preferentially labeled and internalized into the receptor-positive cells, with no fluorescence signal observed in receptor-negative cells (Figure S4). Using these same cell lines, we next examined cell viability in a light- and antigen-dependent manner. Cells were incubated with each antibody conjugate for 24 h, the media were replaced, and then the cells were exposed to 20 J/cm² of 690 nm light from a LED source. After 72 h, irradiation of both conjugates inhibited viability of MDA-MB-468 cells with similar potency to free duocarmycin DM (Table 2, see Figure S5 for full dose response curves). This activity was substantially diminished in the absence of irradiation for MDA-MB-468, and in the presence and absence of irradiation for MCF-7 cells. We also initially examined the role of O₂ in the drug delivery process, given the known propensity of solid tumors to local hypoxia and the role of O₂ in the uncaging reaction.⁵² Carrying out the identical set of experiments in an incubator equilibrated to 1% O₂ provided nearly identical IC₅₀ values to those in Table 2 (Table S2). Overall, these in vitro studies indicate an approximately 2-fold improvement in the therapeutic index for CyEt-Pan-Duo relative to CyMe-Pan-Duo, consistent with the moderately improved hydrolytic stability observed in our kinetics studies described above. Moreover, the therapeutic index of CyEt-Pan-Duo is nearly 8× improved relative to our previous conjugate, Cy-Pan-CA4 (Figure 1), and exhibits an over 400× improvement in potency.²⁸

Table 2. In Vitro Assessment of CyMe-Pan-Duo and CyEt-Pan-Duo^a.

compd	cell line and irradiationb		IC50 (nM)c	fold Δd
CyMe-Pan-Duo	MDA-MB-468	+hν	0.039 ± 0.0011	1.0
		–hν	12 ± 1.4	310
	MCF-7	+hν	52 ± 1.0	1300
		–hν	90 ± 1.9	2300
CyEt-Pan-Duo	MDA-MB-468	+hν	0.026 ± 0.00090	1.0
		–hν	15 ± 1.2	580
	MCF-7	+hν	50 ± 1.0	1900
		–hν	>100	>3900
Duo DM	MDA-MB-468	±hν	0.012 ± 0.00050
	MCF-7	±hν	0.019 ± 0.0017
Pan	MDA-MB-468	±hν	>100
	MCF-7	±hν	>100

^aNear-IR light-dependent growth inhibition of MDA-MB-468 (EGFR+) and MCF-7 (EGFR−) cells treated with each antibody conjugate, free duocarmycin DM (Duo DM), or free panitumumab (Pan). Cells were either irradiated with 20 J/cm² of 690 nm light or kept dark.

^b20 J/cm² (15 mW/cm², 22 min).

^cAverage IC₅₀ value ± standard deviation (n = 4).

^dRatio value/(MDA-MB-468 + hν).

Noting the increased molar absorption coefficient and 45 nm bathochromic shift of the N,N′-diethyl variant, we evaluated the impact of irradiation light dose and wavelength on cell viability (Figure 3A). MDA-MB-468 cells were incubated with 100 pM of each antibody conjugate for 24 h, the media were replaced, and the cells were exposed to varied doses of 690 or 780 nm light from LED sources (Figure 3B,C). At least 20 J/cm² of both 690 and 780 nm light was required for CyMe-Pan-Duo to inhibit cell viability by >50%. By contrast, only 10 J/cm² of 690 nm light and 5 J/cm² of 780 nm light were needed to achieve similar viability effects with CyEt-Pan-Duo. The observation that CyEt-Pan-Duo can be activated with 780 nm light should ultimately prove advantageous for in vivo applications. Given the improved therapeutic index and near-IR light sensitivity of CyEt-Pan-Duo, this compound was chosen for in vivo study.

Figure 3.

Effect of light dose and wavelength on the in vitro efficacy of CyMe-Pan-Duo and CyEt-Pan-Duo. (A) The relationship between absorbance spectra and the irradiation wavelength. (B) MDA-MB-468 (EGFR+) cell viability under control conditions (vehicle, free duocarmycin DM (Duo DM), and unirradiated ADC). (C, D) Cell viability as a function of irradiation light dose and wavelength in the presence of 100 pM of each ADC (solid line, CyMe-Pan-Duo; dashed line, CyEt-Pan-Duo). Error bars represent standard deviation (n = 4).

Using an EGFR+ MDA-MB-468 xenograft tumor model, we examined the effect of DOL on the biodistribution of CyEt-Pan-Duo. Antibody conjugates with DOL 1, 2, and 4 were prepared as described above and were administered via tail vein injection (100 μg). In vivo fluorescence imaging (800 nm) was used to gauge tumor localization and stability (Figure S6). Time and labeling density-dependent accumulation of the conjugate occurred at the tumor, the liver, and the bladder. Tumor-to-background ratio was greatest for the DOL 4 conjugate and reached a maximum 3–4 days post-conjugate dosing. No loss of fluorescence signal was observed up to 7 days after dosing, indicating significant stability of CyEt-Pan-Duo. The ability to deplete the fluorescence signal using external irradiation was next examined. CyEt-Pan-Duo (DOL 4, 100 μg) was administered by tail vein injection and allowed to accumulate in the tumor for 4 days, after which the tumor region was selectively exposed to 690 nm light from a commercial 800 mW/cm² PDT laser system (Figure S7). Following application of 80 J/cm² (1.7 min irradiation), the 800 nm fluorescence intensity of the tumor was nearly completely ablated (∼80%). This light dose was used for further study and is similar to, or even less than, those required for existing in vivo PDT applications.⁵³

We next evaluated the impact of CyEt-Pan-Duo (DOL 4) on tumor burden and survival. MDA-MB-468-luc tumor-bearing mice were randomized into 5 treatment groups of 9 animals each. These included vehicle (group 1), 100 μg conjugate without irradiation (group 2), and 10, 30, or 100 μg of conjugate with a single irradiation (690 nm, 80 J/cm² for groups 3–5). Antibody conjugate was administered 10 days post cell injection and light applied 4 days post-conjugate dosing (Figure 4A). Cyanine uptake and then cyanine signal depletion, as a marker for drug release, were confirmed using fluorescence imaging (Figure 4B). We observe that the conjugate reaccumulates in the tumor following the initial irradiation, suggesting that further irradiation might provide additional payload delivery. Effects on tumor proliferation following the single light dose were assessed through luminescence imaging of luciferase activity (short-term) and caliper measurement of tumor size (long-term). Immediate decreases in luciferase activity were observed for groups 4 (30 μg + light) and 5 (100 μg + light) relative to group 1 (vehicle), with statistical significance at day 7 post-irradiation (Figures 4C,D and S8). Statistically significant decreases in tumor volume were also observed for these groups relative to group 1 at 17 days post-irradiation (Figure 4E). Importantly, these effects on tumor growth correlate with increased survival at the conclusion of the study for group 4 and 5 mice compared to group 1 (Figure 4F). While group 2 (100 μg without irradiation) did not exhibit statistically significant differences in luciferase activity relative to group 1, improvements in survival and reductions in tumor volume were observed at the conclusion of the study. This may reflect slow light-independent cleavage of the payload molecule from the conjugate, as well as some photolytic release from animals that were not extensively shielded from ambient light. All the treated mice exhibited no statistically significant differences in body weight relative to untreated animals (Figure S9). We note that additional dosing options beyond the simple strategy used here—a single application of conjugate and light—are feasible and will be the subject of future investigation.

Figure 4.

In vivo efficacy of CyEt-Pan-Duo (DOL 4) in MDA-MB-468-luc tumor-bearing mice. (A) Conjugate dosing (i.v.), irradiation (690 nm, 80 J/cm², 800 mW/cm²), and imaging regimen. (B) Fluorescence images at 800 nm. (C) Bioluminescence images of luciferase activity. (D) Luciferase activity as a function of time post-irradiation, relative to initial. (E) Tumor volume as a function of time post-irradiation. (F) Survival as a function of time post-irradiation. For D–F, vehicle (black), 100 μg of CyEt-Pan-Duo – hν (red), and 10 μg (green), 30 μg (purple), and 100 μg (orange) of CyEt-Pan-Duo + hν. n = 9 mice per condition. *p < 0.05, **p < 0.01, Dunnett’s test with ANOVA (D, E) with error bars representing standard error of the mean (SEM) or log-rank test (F).

Conclusion

Combining fluorescence imaging with the release of a therapeutic agent is an enticing prospect. The clinical translation of fluorescence-guided surgical methods, often using heptamethine cyanines as the light-harvesting component, illustrates the potential for new optical tools to progress in a clinical context.^2,54 Novel optical drug delivery methods may find use in isolation, or provide momentum for the notion of integrating localized pharmacological treatment into a surgical context to address challenging tumor margins. The latter option is particularly appealing given the broad use of surgical debulking as first line treatment for many solid tumors.⁵⁵ However, it is noteworthy that there are several significant tumor classes that are readily irradiated with existing PDT infrastructure.^4,53

The studies above provide key chemical underpinning, as well as biological demonstration, for efforts to recast the cyanine scaffold as a targeted drug delivery modality. Rational modifications to the scaffold provide a bathochromic shift in the λ_max, while also improving hydrolytic stability. These chemical observations translate to a cellular setting through efficient drug delivery with 780 nm light and enhanced therapeutic index. The near-IR fluorescence properties of the cyanine scaffold, and the irradiation-induced loss of that signal, provide markers for in vivo target accumulation and drug release, respectively. Significant antitumor efficacy is seen with only a single dose of light and conjugate.

The approach reported here, which combines well-tolerated monoclonal antibodies with organic small molecule entities, has certain promising features for further use. This combination may offer benefits relative to existing optical drug delivery methods that rely on metallic nanomaterials.^18,19 Translation of these methods has been complicated by the requirement of high light doses, issues relating to clearance and toxicity, and modest target accumulation.⁵⁶⁻⁵⁸ By contrast, small molecule/antibody conjugates find significant translational application, including as optical diagnostic and treatment modalities.^54,59−61 It is notable that the standard criteria for the antibody component of an ADC—high tumor selectivity and excellent cellular internalization⁶²—are not obligate requirements for this method. The broad availability of optimized monoclonal antibodies, along with other targeting methods, means that a variety of tumor-associated antigens might be pursued.

Additional systematic study on the link between cyanine structure and consequent uncaging/biological function will increase their utility for various biomedical applications. For example, issues related to tumor uptake and clearance will require further assessment, and can likely be improved through modifications to the cyanine scaffold and bioconjugation method.⁶³⁻⁶⁶ Finally, we note that the currently configured uncaging method is likely best suited for use with a targeting ligand. Targeting serves both to promote tumor accumulation and to spatially restrict intermediates in the uncaging process. Without using a localization strategy, which could be desirable for certain applications, the diffusion of partially uncaged intermediates will likely hamper spatially controlled delivery of the small molecule payload. While further optimization studies may provide improved release kinetics, it may be that relying on hydrolysis and cyclization steps imposes an upper limit on the rate of uncaging. Efforts to create and deploy optimal cyanine constructs for in vivo drug delivery, as well as to identify alternate cleavage chemistries, are ongoing.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscentsci.7b00026.

• Experimental methods, supplementary figures and tables, synthetic procedures, NMR spectra, and Cartesian coordinates (PDF)

Author Contributions

R.R.N., A.P.G., H.K., and M.J.S. conceived the key concepts. R.R.N., A.P.G., and T.Y. conducted the synthesis and characterization experiments. A.P.G. conducted the in vitro cellular studies and analysis. T.N. and H.K. performed in vivo studies and analysis. J.I. conducted the computational studies. R.R.N., A.P.G., and M.J.S. wrote the paper.

The authors declare no competing financial interest.