Near-Infrared Organic Structures with Extended Lifetimes to Unlock the Potential of In Vivo Fluorescence Lifetime Imaging

Abstract

In vivo fluorescence lifetime imaging (FLI) is an advanced optical imaging modality with the capability to reveal additional biological information compared to fluorescence intensity (FI) imaging. In addition to enabling tissue discrimination based on microenvironmental variations, lifetime contrast allows the differentiation of spectrally overlapping fluorophores, thereby expanding the high potential of multiplexing imaging. Until now, most of the studies evaluating the benefits of in vivo FLI in terms of enhanced sensitivity and specificity for tumor imaging with fluorescent contrast agents have been conducted using heptamethine cyanine dyes. While these fluorophores are highly successful for conventional in vivo fluorescence imaging, they are characterized by short singlet excited-state lifetimes (generally below 1 ns), making it challenging to achieve satisfying tissue contrast and to implement multiplexing imaging strategies based on their lifetime differences. Here, we report a bioconjugatable, water-soluble NIR-emissive aza-BODIPY dye with a 100% longer lifetime than IRDye 800CW in physiological conditions. We demonstrate its suitability for in vivo FLI when conjugated to cetuximab and reveal how conjugation methods impact the lifetime characteristics of fluorescent antibody conjugates.

Introduction

Fluorescence lifetime imaging (FLI) is an advanced optical imaging technique that quantifies the average time a fluorophore resides in its excited electronic state following photon absorption before relaxing to the ground state. Unlike fluorescence intensity, fluorescence lifetime is an intrinsic photophysical parameter that remains largely independent of fluorophore concentration, excitation power, and instrument configuration, making FLI a robust and quantitative imaging modality. Because the lifetime of a given fluorophore can also be sensitive to its local molecular microenvironment, FLI enables discrimination of tissues based on differences in biochemical composition and metabolic activity, providing access to new information in comparison to conventional intensity-based methods. Moreover, fluorescence-lifetime-based contrast enables the separation of multiple fluorophores sharing similar spectral profiles, thereby facilitating effective multiplexing , and paving the way for new imaging strategies beyond the visualization of a single tracer in vivo.

While the use of FLI in microscopy (commonly referred to as FLIM) is well-established for studying various biological processes at the cellular and subcellular levels, in vivo macroscale FLI leveraging autofluorescent signatures or exogenous near-infrared (NIR) fluorescent agents is increasingly gaining clinical relevance as a tool for diagnostic imaging and real-time surgical guidance. Notably, the group of Kumar has demonstrated in both mouse models and ex vivo human samples that FLI, when combined with the NIR tracers indocyanine green (ICG) and cetuximab-IRDye800CW, can improve the differentiation between tumor signals and those originating from surrounding tissues. − This enhanced contrast results in superior tumor-versus-normal classification accuracy compared with conventional fluorescence imaging. Unfortunately, most organic NIR fluorescent dyes used for in vivo and translational fluorescence imaging applications are heptamethine-derived cyanine dyes, which exhibit short lifetimes, typically around or below 1 ns. These short lifetimes complicate data analysis and substantially limit the potential for multiplexing. ,

We focused our attention on a family of BODIPY derivatives, namely the aza-BODIPYs. These fluorophores have already demonstrated strong potential for in vivo optical imaging in both the NIR-I and NIR-II spectral windows. , We selected the diphenyldithienyl aza-BODIPY (DPDTAB) scaffold, previously reported by the Xiao group, as our starting point. Indeed, this bis-thienyl derivative exhibits interesting photophysical properties within the NIR-I spectral range (λ_abs/λ_em = 710/732 nm, ε = 108 600 M^–1·cm^–1 and Φ_F = 46% in MeCN), along with an unusually long fluorescence lifetime of 3.6 ns. A major limitation of native aza-BODIPYs for bioimaging is their pronounced lipophilicity mitigating their potential in vivo. To address this, we used the well-established boron substitution strategy previously developed in our group to prepare a biocompatible and bioconjugatable water-soluble aza-BODIPY derivative (Wazaby), while preserving its favorable lifetime properties in aqueous media. In this work, we introduced a short PEG (polyethylene glycol) linker bearing either a bioconjugatable group or a macrocyclic polyazapolycarboxylate chelator, DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid), which not only serves as a pharmacomodulator but can also be used in future studies to incorporate a radiometal suitable for additional imaging modality, such as PET (Positron Emission Tomography) or SPECT (Single-Photon Emission Computed Tomography). The resulting FLT-Waza was subsequently conjugated to the anti-EGFR antibody cetuximab to allow targeted imaging of EGFR-expressing tumors. Beyond assessing this innovative tracer for in vivo macroscale FLI, we also aimed to understand how the selected labeling strategy alters the lifetime properties of the resulting conjugates, ultimately providing a novel way to control fluorophore lifetimes in the context of antibody-based imaging.

Results and Discussion

Synthesis of Water-Soluble thienoWaza by Dyes

DPDTAB was first synthesized following a previously reported procedure. The next step involved the use of a Grignard reagent derived from N,N-dimethylpropargylamine allowing substitution of the two fluorine atoms to give the derivative thienoWaza in 85% isolated yield. The practical implementation of the postsynthetic water-solubilizing strategy was initially assessed with the preparation of bis-cationic and zwitterionic aza-BODIPY dyes, thienoWaza-1 (bearing two quaternary ammonium moieties) and thienoWaza-2 (bearing two sulfobetaine moieties) (see Supporting Information for synthetic details and Scheme for their structures). These two water-soluble model compounds exhibit long fluorescence lifetime values (ca. 3.0–3.2 ns) in DMSO as well as in PBS with 5% BSA (ca. 2.8–2.9 ns) (Table S2). These encouraging preliminary results confirm that such functionalization did not adversely affect the lifetime value of the aza-BODIPY core and it is therefore of particular interest for developing water-soluble and bioconjugatable derivatives. To this end, FLT-Waza-01 (for site-specific bioconjugation to cetuximab) and FLT-Waza-02 (for random bioconjugation to cetuximab) were prepared as detailed in Scheme . In order to incorporate both a water-solubilizing group and a terminal reactive handle within the structure of thienoWaza, we selected a short PEG linker containing a Boc-protected amino group and a bromine as a leaving group (N-Boc-PEG₄-Br) to achieve successful quaternization of N,N-dimethylaminopropargyl arms under conventional conditions of S_N2 reactions. The resulting bis-PEGylated product was not purified and was directly subjected to unusual aqueous (aq.) acidic conditions (a mixture of aq. 1.0 M HCl and MeCN, room temperature), fully compatible with thienoWaza scaffold stability, allowing the challenging removal of only one Boc group. The monoamine intermediate was purified by semipreparative RP-HPLC and thereafter underwent N-acylation with DOTA-tris­(tBu) ester NHS ester in the presence of TEA in dry DMF. Further acidic treatment with aq. 4.0 M HCl at 40 °C for 48 h resulted in complete removal of the Boc group and deprotection of the tert-butyl esters, without compromising the integrity of the boron complex within the aza-BODIPY scaffold. Purification by semipreparative RP-HPLC yielded pure FLT-Waza-01 as a TFA salt, with an overall yield of 34% after four steps. Its structure was confirmed by detailed measurements, including ESI-HRMS and NMR spectroscopic analyses (Supporting Information). The availability of a primary amine-terminated PEG linker within the core structure of FLT-Waza-01 enabled its direct use in transglutaminase-mediated bioconjugation, with the aim to achieve site-specific labeling of antibodies. As an additional functionalization, FLT-Waza-01 was also converted into the monosquaramide derivative FLT-Waza-02 by reacting with diethyl squarate in the presence of DIPEA in EtOH. Indeed, this amine-reactive dye is suitable for random labeling on lysine residues. ,

1. Synthesis of FLT-Waza-01 and FLT-Waza-02 [DIPEA = N,N-diisopropylethylamine, O/N = Overnight, Pr-SO₃H = 3-Propylsulfonic acid, RT = Room Temperature, TEA = Triethylamine].

Photophysical Studies of thienoWazaby Dyes

The photophysical studies of the different compounds were conducted in DMSO, in PBS alone, or with additives modeling body fluid (i.e., BSA protein or fetal bovine serum (FBS)). All spectroscopic data are gathered in Table and Table S2, and the corresponding overlay of electronic absorption, excitation, emission spectra, and fluorescence lifetime decay curves are available in (Supporting Information Figure S43-47). Compared to the parent aza-BODIPY dye, DPDTAB, alkynyl substitution at the boron atom and subsequent N-quaternization of the two N,N-dimethylaminopropargyl arms do not significantly impact absorption and emission profiles, with maxima centered at ca. 720/740 nm and 715/730 nm in DMSO and PBS (with additives), respectively (Figures S31–S42). However, the lack of additional hydrophilic arms within the structure of thienoWaza leads to undesired aggregation behavior in aqueous buffer. The hypsochromic shift and the broadening of its NIR absorption band are consistent with the formation of nonemissive H-aggregates exhibiting face-to-face molecular stacking (Figure S32). A quite similar but much less pronounced trend is noted with the bis-cationic derivative thienoWaza-1 and the bis-zwitterionic thienoWaza-2, as supported by the imperfect match between their absorption and excitation spectra (Figures S34 and S37). As expected, the double N-quaternization of thienoWaza with a short amine-terminated PEG linker and the mono-N-acylation with hydrophilic DOTA led to a fully water-soluble NIR emissive derivative (FLT-Waza-01) with highly valuable spectral features under simulated in vivo conditions, close to those determined in organic solvent (DMSO). Its fluorescence brightness (ε × Φ_F) value of 10 600 M^–1 cm^–1 in PBS is quite remarkable and therefore supports the efficacy of the solubilizing strategy considered in this work.

1. Photophysical Properties of Aza-BODIPY Dyes Studied in This Work, Determined at 25 °C .

Aza-BODIPY	Solvent	Abs max (nm)	Em max (nm)	Stokes shift (cm–1)	ε (M–1 cm–1)	ΦF (%)	τ (ns)
DPDTAB	MeCN	710	732	423	108 600	46	3.6
	DMSO	728	746	331	118 300	20 (42)	3.4
thienoWaza	DMSO	718	739	396	78 600	6	1.9
	PBS	728	-	-	31 700	-	-
	PBS + 5% BSA	716	728	230	68 300	5	2.9
FLT-Waza-01	DMSO	724	740	298	108 300	20	2.9
	PBS	715	734	362	105 500	10	1.5
	PBS + 5% BSA	715	733	343	107 600	10	1.8
	PBS + 10% FBS	715	729	268	103 800	11	1.6

^a[BSA = bovine serum albumin, FBS = fetal bovine serum, fwhm = full-width half-maximum, PBS = phosphate-buffered saline (100 mM phosphate +150 mM NaCl, pH 7.5)]

^bStock solutions (1.0 mg/mL) of fluorophores were prepared in DMSO.

^cOnly the 0–0 band of the S₀–S₁ transition of the aza-BODIPY core is reported.

^dDetermined using an aza-BODIPY dye (BF₂ chelate of [5-(4-methoxyphenyl)-3-phenyl-1H-pyrrol-2-yl]-[5-(4-methoxyphenyl)-3-phenyl-pyrrol-2-ylidene]-amine) as a standard (Φ_F = 36% in CHCl₃, Ex at 670 nm). ,

^eNanoLED excitation was at 671 nm.

^fValues determined and reported by Zhang et al.

^gA broad Abs band (fwhm of ca. 125 nm), assigned to nonemissive aggregates, is observed.

^hNonfluorescent.

Finally, in order to assess the suitability of this novel water-soluble aza-BODIPY dye for in vivo FLI, singlet excited-state lifetime measurements through time-correlated single-photon counting (TCSPC) were performed in different solvents (Figure S46). The lifetime of FLT-Waza-01 in DMSO (2.9 ns) was consistent with that of its parent nonconjugatable derivatives; however, as commonly observed with other organic-based fluorophores, the lifetime of FLT-Waza-01 decreased to 1.5 ns in an aqueous buffer solution like PBS and increased back to 1.6–1.8 ns in the presence of serum proteins. This is likely attributed to interactions between the fluorophore and the hydrophobic pocket of BSA proteins, which limit nonradiative decay processes. Nevertheless, these lifetimes remain substantially higher than that of IRDye 800CW (τ = 0.9 ns in PBS + 5% BSA), one of the most popular cyanine-based NIR dyes used in fluorescence molecular imaging.

Preparation and Spectral Characterization of Cetuximab–thienoWaza Conjugates

Next, the ability of bioconjugatable aza-BODIPY dyes FLT-Waza-01 and FLT-Waza-02 to be covalently grafted to the cetuximab antibody (a chimeric human/mouse IgG1 monoclonal antibody targeting the epidermal growth factor receptor (EGFR), a cell surface receptor overexpressed in many types of cancer) was investigated in order to obtain targeted imaging agents for in vivo applications. It is now well established that the way in which a fluorophore is covalently attached to a biological vector, such as an antibody, has a major impact on both in vitro and in vivo behavior of the resulting bioconjugate. Usually, higher fluorescence brightness and tumor accumulation are observed when the construction of the bioconjugate is achieved in a site-specific and controlled manner. , However, to the best of our knowledge, the influence of the bioconjugation method on the fluorescence lifetime performance of immunoconjugates has never been investigated. In order to study the impact of the bioconjugation strategy on the lifetime values, the dyes were bioconjugated either randomly on lysines for FLT-Waza-02, based on squaramide ester chemistry, or using a transglutaminase-mediated site-specific approach for FLT-Waza-01. This latter technique involves, after removal of the glycan moieties of the heavy chains, the use of microbial transglutaminase (MTGase), an enzyme that is able to catalyze the transamidation reaction between the glutaminyl 297 residue and the amine-terminated PEG linker of FLT-Waza-01. Both approaches resulted in the formation of Cetux-FLT-Waza-02 (for random labeling) and Cetux-FLT-Waza-01 (for site-specific bioconjugation) with an average degree of labeling (DOL) of 2, but with full control of conjugation sites for the enzyme-mediated site-specific approach (Figures S48 and S49).

Lifetime measurements on the fluorescent cetuximab conjugates were then in PBS (Figure ). Notably, the site-specific conjugation leads to a higher average lifetime value (2.0 ns) compared to random conjugation (1.6 ns), highlighting how the labeling strategy can affect the final conjugates’ lifetime, as the lifetime values of FLT-Waza-02 are similar to those found for FLT-Waza-01 (Table S2).

1.

(A) Bioconjugation strategies were used for the preparation of immunoconjugates Cetux-FLT-Waza-02 and Cetux-FLT-Waza-01 [O/N = overnight]. Fluorescence lifetime decay curves and fluorescence lifetime values of cetuximab conjugates in PBS at 25 °C. (B) Measurements in PBS (concentration: 1.0 mg/mL) at 25 °C. NanoLED excitation at 671 nm; Em monochromator set to 740 nm with a bandpass of 14 nm for Cetux-FLT-Waza-02 (χ² = 1.1, biexponential fitting) and a bandpass of 6 nm for Cetux-FLT-Waza-01 (χ² = 1.3, biexponential fitting).

The affinity of the two bioconjugates toward EGFR was also assessed by surface plasmon resonance (SPR). All conjugates showed similar binding affinity (K _d (Cetux-FLT-Waza-01) = 9.90 nM and K_d (Cetux-FLT-Waza-02) = 8.29 nM), demonstrating that neither the grafted dye nor the conjugation strategy had a negative effect on the targeting capabilities of cetuximab (Figure S21 and Table S3, Supporting Information).

In Vivo Fluorescence Lifetime Imaging of Cetuximab–thienoWaza Conjugates

To evaluate the impact of both labeling strategies on the tumor uptake of the tracers and their in vivo fluorescence lifetime, preclinical imaging experiments were conducted in mice bearing EGFR-positive subcutaneous tumors of hypopharyngeal squamous cell carcinoma (FaDu cells) (Figure ), using a 128 × 128-pixel VUB tauCAM, a novel macroscale imaging system capable of both FI and FLI in the NIR-I region. The randomly labeled Cetux-FLT-Waza-02 tracer showed discernible tumor accumulation only after 24 h postinjection (p.i.), reaching its maximum signal intensity and tumor-to-background ratio (TBR) (1.83 ± 0.22) at 48 h p.i. In contrast, Cetux-FLT-Waza-01 showed clear tumor accumulation as early as 6 h p.i. with TBR exceeding 2.1 (Figure A,B). Notably, the tumor signal for the site-specific bioconjugate was 1.5- to 2.5-fold higher than that of the randomly conjugated tracer. At all time points, the fluorescence lifetime values measured within the tumor were significantly longer than the lifetime of autofluorescent signals measured preinjection (1.05 ± 0.13, p < 0.05), with the tumor fluorescence lifetime for Cetux-FLT-Waza-01 being, on average 300 ps, longer than for Cetux-FLT-Waza-02 (1.93 ± 0.06 and 1.63 ± 0.07 ns, respectively, p < 0.05) (Figures C, Figure S54). Furthermore, for both tracers, the tumor fluorescence lifetimes were longer than those measured in background tissue at the level of the shoulder (1.69 ± 0.04 vs 1.42 ± 0.02 for Cetux-FLT-Waza-02 and 1.95 ± 0.09 vs 1.55 ± 0.06 for Cetux-FLT-Waza-01 at 48 h p.i., p < 0.05) or at other sites with nonspecific intensity signals such as the lower legs and tail base (Figure A,D,E). These differences are quite high compared to values reported for cyanine-based probes. , While shorter lifetime values were also observed in the surrounding tumor margins (for Cetux-FLT-Waza-02 at all time points, for Cetux-FLT-Waza-01 only after 24 h), these differences were not statistically significant. Nevertheless, segregation between the different regions is feasible based on the histogram profiles depicting the full distribution of the fluorescence lifetime values (Figure F and G).

2.

(A) Representative in vivo reflectance, fluorescence intensity, and FLI images of mice bearing an EGFR+ subcutaneous tumor (right lateral view), acquired at 6, 24, and 48 h after intravenous administration of Cetux-FLT-Waza-02 or Cetux-FLT- Waza-01. On the reflectance images, ROIs drawn around the tumor (T), the tumor margin (TM), and the shoulder (S) (background normal tissue) are indicated. Sites with nonspecific fluorescent signals on the legs and tail base are indicated with arrows. (B) Fluorescence intensity of the tumor and tumor-to-background ratio over time for Cetux-FLT-Waza-02 and Cetux-FLT-Waza-01. (C) Fluorescence lifetime of the tumor over time for Cetux-FLT-Waza-02 and Cetux-FLT-Waza-01. (D, E) Fluorescence lifetime of the tumor, tumor margin, and shoulder over time for Cetux-FLT-Waza-02 (D) and Cetux-FLT-Waza-01 (E). Data are presented as the mean ± standard deviation (n = 3 per group). * and ^# p-values <0.05 are considered statistically significant. (F, G) Representative histograms of the fluorescence lifetime measured at the tumor, tumor margin, and shoulder of a mouse, 48 h after administration of Cetux-FLT-Waza-02 (F) and Cetux-FLT-Waza-01 (G).

Conclusions

In conclusion, we developed a bioconjugatable, water-soluble NIR-I dye, FLT-Waza-01, which exhibits remarkably longer lifetimes (>1.6 ns) compared to previously reported heptamethine cyanine dyes such as ICG or IRDye800CW (below 1 ns). The chosen aza-BODIPY scaffold was structurally optimized to achieve its facile conjugation to antibodies, as well illustrated with cetuximab, with two distinct labeling strategies: a site-specific enzymatic method using MTGase or a random derivatization of lysine residues using squaramide chemistry. The lifetime properties of the resulting immunoconjugates highlight, for the first time, the importance of the conjugation method for FLI with aza-BODIPYs and illustrate a new way to regulate the lifetime in the context of antibody conjugates. Indeed, Cetux-FLT-Waza-02 displayed a lower fluorescence lifetime in comparison to its site-specific analog, Cetux-FLT-Waza-01. This difference was further confirmed during in vivo studies, which highlighted the significant benefit of the site-specific conjugation strategy. Nevertheless, using both conjugates, we were able to easily and specifically distinguish tumor from background regions or tissue autofluorescence (characterized by an average lifetime value around 1 ns before injection). The development of water-soluble, NIR-emissive molecules with extended lifetimes is also a major and essential step toward the implementation of in vivo multiplexing studies using dyes that are photoactive within the same spectral range but characterized by different fluorescence lifetime values. , These findings establish a new approach for tuning fluorescence lifetime in antibody-based probes and highlight the promise of long-lifetime NIR fluorophores for multiplexing in vivo imaging.

Experimental Section

General Information

Unless otherwise noted, all commercially available reagents and solvents were used without further purification. Reactions were carried out in analytical-grade solvents from Carlo Erba or Fisher Chemical under a normal atmosphere unless specified otherwise. TLC was carried out on Merck Millipore or Supelco DC Kieselgel 60 F-254 aluminum sheets (Merck Group). The spots were directly visualized or illuminated with a UV lamp (λ = 254/365 nm). All purifications by column chromatography were performed on silica gel (60–200 μm) from VWR. HPLC-grade DCM, MeCN, and THF were dried over alumina cartridges immediately prior to use (water content: 50, 65, and 75 ppm, respectively, determined by Karl Fischer titration) using a solvent purification system, PureSolv PS-MD-5 model, from Innovative Technology. DMF was purchased from Fisher Chemical (>99%, lab reagent grade), dried by storage over activated 3 Å molecular sieves, and kept under an Ar atmosphere. HPLC-gradient grade MeCN was obtained from Fisher Chemical. Aza-BODIPY dye, used as a standard for the measurement of fluorescence quantum yields (BF2 chelate of [5-(4-methoxyphenyl)-3-phenyl-1H-pyrrol-2-yl]-[5-(4-methoxyphenyl)-3-phenyl-pyrrol-2-ylidene]-amine) [490035–88–6], was prepared according to literature procedures. Indocyanine green (ICG or IR-125) was provided by Acros Organics (laser grade, #412541000). All aqueous buffers used in this work and aqueous mobile phases for RP-HPLC analyses/purifications were prepared using water purified either with a PURELAB Ultra system or a Chorus PURELAB system (ELGA, VEOLIA, purified to 18.2 MΩ·cm). Purity was determined from HPLC analysis and was >95% for all the injected compounds. FBS was purchased from PAN-Biotech (P30–3306). A commercial solution of cetuximab (anti-EGFR antibody, Erbitux) was purchased from Merck. Cetuximab was purified with a HiTrap Desalting column (GE Healthcare) and concentrated to 10 mg/mL in PBS. For a detailed description of RP-HPLC systems mentioned throughout the experimental section, see Supporting Information.

Chemical Synthesis

Diphenyldithienyl aza-BODIPY dye (DPDTAB) [1323977–61–2] was synthesized according to a literature procedure.

thienoWaza: In a Schlenk tube fitted with a stirring bar, N,N-dimethylpropargylamine (160 μL, 1.48 mmol, 2 equiv) was dissolved in dry THF (17 mL), and the solution was bubbled with Ar gas. A commercial solution of EtMgBr (1.81 mL, 0.9 M in THF, 1.63 mmol, 2.2 equiv) was then added, and the resulting reaction mixture was heated to reflux for 45 min. In parallel, a second Schlenk tube was filled with DPDTAB (379 mg, 0.74 mmol, 1 equiv) and dissolved in dry THF (17 mL). The mixture from the first Schlenk tube was transferred through a cannula to the second Schlenk tube. The final mixture was then refluxed for 45 min, and the reaction was checked for completion by TLC (DCM/MeOH 9:1, v/v). Thereafter, the reaction mixture was cooled to 0–4 °C and quenched by adding EtOH (3 mL). The solvents were evaporated under reduced pressure, and the resulting residue was purified by column chromatography over silica gel (eluent: a step gradient of MeOH in DCM from 0% to 10%) to give the desired product as a glittery dark red-green powder (346 mg, 0.54 mmol, yield 74%). ¹ H NMR (500 MHz, 298 K, CDCl₃): δ = 8.83 (d, ³ J = 3.8 Hz, 2H), 8.05 (d, ³ J = 7.1 Hz, 4H), 7.61 (d, 2H, ³ J = 5.2 Hz), 7.47–7.41 (m, 6H), 7.27 (overlapping with residual signal of nondeuterated CHCl₃, t, 2H), 7.22 (s, 2H), 3.20 (s, 4H), 2.15 (s, 12H). ¹¹ B NMR (160 MHz, 298 K, CDCl₃): δ = −11.98 (bs). ¹³ C NMR (126 MHz, 298 K, CDCl₃): δ = 149.8, 143.5, 141.7, 135.2, 134.5, 132.4, 131.1, 129.3, 129.2, 129.1, 128.7, 119.7, 95.5, 53.5, 48.7, 43.4. HPLC (system A): t _R = 4.1 min (purity 97% at 260 nm and 97% at 707 nm). HRMS (ESI+) (Da): m/z calcd 318.6247 for [M + 2H]²⁺; found 318.6246.

thienoWaza-1: Compound thienoWaza (30 mg, 0.047 mmol, 1 equiv) was dissolved in DCM (2 mL), and a large excess of MeI (1 mL, 16 mmol, 340 equiv) was added. The resulting reaction mixture was stirred at RT for 30 min. Then, MeI was coevaporated with DCM under reduced pressure. Drying under vacuum provided thienoWaza-1 as an iodide salt in a quantitative yield (43 mg, 0.047 mmol). ¹ H NMR (500 MHz, 298 K, CD₃OD): δ = 8.67 (d, ³ J = 3.8 Hz, 2H), 8.12 (d, ³ J = 7.1 Hz, 4H), 8.00 (d,³ J = 5.2 Hz, 2H), 7.57 (s, 2H), 7.52–7.44 (m, 6H), 7.45 (t, ³ J = 4.6 Hz, 2H), 4.17 (s, 4H), 2.94 (s, 18H). ¹¹ B NMR (160 MHz, 298 K, CD₃OD): δ = −12.14 (bs). ¹³ C NMR (126 MHz, 298 K, CD₃OD): δ = 151.5, 144.2, 143.5, 136.1, 135.0, 134.6, 133.1, 130.7, 130.6, 129.8, 130.4, 121.0, 58.2, 53.2. HPLC (system A): t _R = 3.8 min (purity 99% at 220 nm and 100% at 704 nm). HRMS (ESI+) (Da): m/z calculated 332.6403 for [M]²⁺, found 332.6403.

thienoWaza-2: Compound thienoWaza (30 mg, 0.047 mmol, 1 equiv) was dissolved in dry MeCN (2 mL). 1,3-Propanesultone (12 mg, 0.1 mmol, 2.1 equiv) and anhydrous K₂CO₃ (25 mg, 0.19 mmol, 4 equiv) were sequentially added. The resulting reaction mixture was stirred at 50 °C for 4 h. The reaction was checked for completion by RP-HPLC-MS (system A). Thereafter, the mixture was filtered to remove inorganic salts, and directly purified by semipreparative RP-HPLC (system B, t _R = 33.0–35.0 min). The product-containing fractions were lyophilized to give the TFA salt of thienoWaza-2 as a dark green amorphous powder (18 mg, yield 40%, based on TFA mass = 20%, determined by ion chromatography). ¹ H NMR (500 MHz, 298 K, CD₃OD): δ = 8.71 (d, ³ J = 3.8 Hz, 2H), 8.11 (d, ³ J = 7.2 Hz, 4H), 8.01 (d, ³ J = 5.1 Hz, 2H), 7.56 (s, 2H), 7.52–7.46 (m, 8H), 4.10 (s, 4H), 3.43 (m, 4H), 2.91 (s, 12H), 2.78 (t, ³ J = 6.2 Hz, 4H), 2.11 (m, 4H). ¹¹B NMR (160 MHz, 298 K, CD₃OD): δ = −12.18 (bs). ¹³ C NMR (126 MHz, 298 K, CD₃OD): δ = 151.3, 144.2, 143.3, 136.5, 135.1, 134.8, 133.2, 131.0, 130.6, 129.7, 130.4, 120.9, 63.9, 56.2, 51.0, 48.8, 20.0. HPLC (system A): t _R = 3.8 min (purity 100% at 225 nm and 100% at 707 nm). HRMS (ESI+) (Da): m/z calculated 462.6104 for [M + 2Na]²⁺, found 462.6099 (M = zwitterionic form (C₄₄H₄₄BN₅O₆S₄)).

FLT-Waza-01: Steps 1 (N-alkylation) and 2 (Boc removal) : Compound thienoWaza (100 mg, 0.15 mmol, 1 equiv) was dissolved in dry MeCN (10 mL), followed by the addition of N-Boc-PEG₄-bromide (purchased from BroadPharm company, 190 mg, 0.47 mmol, 3.1 equiv) and NaHCO₃ (66 mg, 0.78 mmol, 5.2 equiv). The reaction mixture was heated under reflux overnight. The reaction was monitored by RP-HPLC-MS (system A), showing the complete consumption of starting thienoWaza and formation of the bis-PEGylated product. Then, the crude mixture was filtered to remove inorganic salts and concentrated under reduced pressure. The resulting residue was dissolved in MeCN (3 mL), and aq. 1.0 M HCl was added (1 mL, 6.7 equiv). The mixture was left stirring at RT overnight. RP-HPLC-MS analysis showed the formation of the monodeprotected derivative as the major product, with the fully deprotected product as the side product. The mono N-Boc derivative was purified by semipreparative RP-HPLC (system B, t _R = 33.0–35.0 min). The product-containing fractions were lyophilized to give the TFA salt of the monoamine intermediate, which was used directly in the next step (80 mg, yield 44%). Steps 3 (N-acylation with DOTA-tris­(tBu) ester NHS ester) and 4 (Boc removal) : TFA salt of the monoamine (40 mg, 0.041 mmol, 1 equiv) was dissolved in dry DMF (2 mL), followed by the addition of DOTA-tris­(tBu) ester NHS ester (provided by CheMatech company, 42 mg, 0.051 mmol, 1.2 equiv) and TEA (24 μL, 0.17 mmol, 5 equiv). The resulting reaction mixture was stirred at 40 °C for 1 h to reach completion (confirmed by RP-HPLC-MS analyses, system A). Thereafter, volatiles were evaporated under reduced pressure. The crude residue was redissolved in MeCN (2 mL), and aq. 4.0 M HCl (1 mL, 117 equiv) was added. The reaction mixture was stirred at 40 °C for 48 h. RP-HPLC-MS analysis showed the only formation of FLT-Waza-01. The latter was purified by semipreparative RP-HPLC (system B, t _R = 25.0–27.0 min). The product-containing fractions were lyophilized to give the TFA salt of FLT-Waza-01 as a dark blue spongy solid (32 mg, yield 34% over four steps, based on TFA mass = 38%, determined by ion chromatography). ¹ H NMR (500 MHz, 298 K, CD₃OD): δ = 8.66 (d, ³ J = 3.9 Hz, 2H), 8.14 (d, ³ J = 7.1 Hz, 4H), 8.02 (d, ³ J = 5.2 Hz, 2H), 7.62 (s, 2H), 7.54–7.48 (m, 6H), 7.43 (t, ³ J = 4.7 Hz, 2H), 4.22 (s, 2H), 4.21 (s, 2H), 3.81 (broad peak, 5H), 3.73 (broad peak, 5H), 3.59 (m, 3H), 3.53–3.33 (m, 40H), 3.25 (2 broad peaks, 8H), 3.04 (m, 3H), 2.95 (bs, 12H). ¹¹ B NMR (160 MHz, 298 K, CD₃OD): δ = −12.10 (bs). ¹³ C NMR (126 MHz, 298 K, CD₃OD): δ = 161.5, 161.2, 150.9, 143.7, 142.9, 135.7, 134.6, 134.4, 132.6, 130.4, 130.0, 129.9, 129.4, 120.5, 118.3, 116.0 (trifluoroacetate), 113.7, 70.9, 70.9, 70.9, 70.8, 70.8, 70.7, 70.7, 70.6, 70.6, 70.6, 69.8, 67.3, 65.1, 65.0, 63.5, 57.0, 51.5, 51.4, 40.0, 39.8. HPLC (system A): t _R = 3.2 min (purity 100% at 220 nm and 100% at 709 nm). HRMS (ESI+) (Da): m/z calculated. 487.5770 for [M²⁺ + H]³⁺, found 487.5570.

FLT-Waza-02: Compound FLT-Waza-01 (15.4 mg, 10 μmol, 1 equiv) was dissolved in absolute EtOH (2 mL). Diethyl squarate (17.85 μL, 91 μmol, 9.1 equiv) and DIPEA (17.85 μL, 105 μmol, 10.5 equiv) were sequentially added. The reaction mixture was stirred at 35 °C for 1 h. RP-HPLC-MS analysis (system A) showed the complete consumption of the starting amine and the only formation of the desired product. Volatiles were then evaporated under reduced pressure. The resulting residue was purified by semipreparative RP-HPLC (system B, t _R = 17.0–19.0 min). The fractions containing the product were lyophilized to give the TFA salt of FLT-Waza-02 as a dark blue sponge (10 mg, yield 62%, calculated without accounting for TFA counterions). The product was solely characterized by HRMS and used directly in bioconjugation experiments. HPLC (system A): t _R = 3.5 min (purity 100% at 250 nm and 100% at 708 nm). HRMS (ESI+) (Da): m/z calcd. 528.9156 for [M²⁺ + H]³⁺, found 528.9155.

Photophysical Measurements

Stock solutions of aza-BODIPY fluorophores . 1.0 mg/mL solutions of fluorophores were prepared in DMSO (UV-spectroscopy grade, Honeywell Riedel-de-Haën) and can be stored at 4 °C for long periods of time without noticing degradation. Solution defrosting was achieved immediately prior to use through a short incubation in a 40 °C water bath. Absorbance and fluorescence measurements . UV–vis absorption spectra were obtained on a Varian Cary 50 Scan (single beam) spectrophotometer (software: Cary WinUV) using rectangular quartz cells (Hellma, 100-QS, 45 × 12.5 × 12.5 mm, path length: 10 mm, chamber volume: 3.5 mL) at 25 °C (using a Lauda Ecoline Recirculating Chiller RE 106 combined with a temperature controller Lauda E100, connected to the spectrophotometer cell holder). Fluorescence Ex/Em spectra were recorded with a HORIBA Jobin Yvon Fluorolog FL3–22 spectrofluorimeter (software FluorEssence) at 25 °C (using a Lauda Ecoline Recirculating Chiller RE 106 combined with a temperature controller Lauda E100, connected to the spectrofluorometer cell holder) with a standard fluorometer cell (Labbox, LB Q, light path: 10 mm, width: 10 mm, chamber volume: 3.5 mL). The absorption spectra of aza-BODIPY derivatives thienoWaza, thienoWaza-1, thienoWaza-2, and FLT-Waza-01 were recorded in the corresponding “physiological” aq. buffer (PBS, PBS + 5% BSA, or PBS + 10% FBS) and DMSO within the concentration range of 1–15 μM (total volume = 3.0 mL, three distinct dilutions for the accurate determination of molar extinction coefficients). Ex/Em spectra were recorded after emission/excitation at the suitable wavelength (set of parameters for Fluorolog: shutter: Auto Open, unless otherwise cited, Ex/Em slits = 5 nm, integration time = 0.1 s, 1 nm step, HV­(S1) = 950 V). All fluorescence spectra were corrected up to 850 nm. Relative fluorescence quantum yields of aza-BODIPY fluorophores thienoWaza, thienoWaza-1, thienoWaza-2, and FLT-Waza-01 were measured in the corresponding buffer/solvent at 25 °C by a relative method using the following standard: aza-BODIPY dye reported by Gorman et al. [BF₂ chelate of [5-(4-methoxyphenyl)-3-phenyl-1H-pyrrol-2-yl]-[5-(4-methoxyphenyl)-3-phenylpyrrol-2-ylidene]­amine, Φ_F = 36% in CHCl₃]. Ex at 670 nm (slit 5 nm), Em 680–850 nm (slit 5 nm) and a dilution by a factor 3 between absorption and fluorescence measurements. The following equation was used to determine the relative fluorescence quantum yield:

$${\mathrm{\Phi }}_F(x)=(A_S{/A}_X)(F_X{/F}_S){(n_X{/n}_S)}^2{\mathrm{\Phi }}_F(s)$$

where A is the absorbance (in the range of 0.01–0.1 A.U.), F is the area under the emission curve, n is the refractive index of the solvents (at 25 °C) used in measurements, and the subscripts s and x represent standard and unknown, respectively. The following refractive indices were used: 1.446 for CHCl₃, 1.478 for DMSO, and 1.337 for PBS, PBS + 5% BSA, and PBS + 10% FBS. Fluorescence lifetime measurements. Decay curves of aza-BODIPY fluorophores thienoWaza, thienoWaza-1, thienoWaza-2, FLT-Waza-01, and FLT-Waza-02 in “physiological” aq. buffer and DMSO (concentration: 1.0 μM) were recorded on the Fluorolog FL3–22 spectrofluorometer equipped with a TBX-04D (300–850 nm) picosecond photon detection module, a DeltaHub high-throughput TCSPC controller, and a NanoLED-C2 controller connected to a NanoLED N-15 670L (peak wavelength at 671 nm, repetition rate = 1 MHz, pulse duration < 200 ps, sync delay = 10 ns). NanoLED-785L (peak wavelength at 777 nm, pulse duration < 200 ps) was used for fluorescence lifetime measurement of ICG in DMSO. Data acquisition was achieved with DataStation 2.7 software. Data processing was achieved with DAS6 (fluorescence decay analysis) software. The following parameters related to data acquisition and optics were used: measurement range = 100 ns, RT preset = 0 ms, peak preset = 10 000 counts. Em monochromator (set to 725 nm, 730 nm, 735 nm, or 740 nm)/bandpass (within the range of 4–14 nm) values were tuned according to the compound studied and the solvent used. Reference measurement/prompt was achieved using LUDOX AM-30 colloidal solution (30 wt % suspension in H₂O, #420875–1L, Sigma-Aldrich). A mono- or biexponential decay with χ² < 1.3 and uniformly distributed standard deviation was chosen as the best fit. In the case of biexponential fitting, the intensity-weighted average fluorescence lifetime is given in order to compare systems regardless of their decay behavior. Please note : given the small volume of solutions available, decay curves of fluorescent antibody conjugates were measured using a Hellma fluorescence cuvette, Ultra Micro (high-performance quartz glass, Suprasil quartz, spectral range 200–2500 nm, path length 3 × 3 mm, chamber volume 45 μL).

Bioconjugation

Deglycosylation of Antibodies

Antibody (2 mg/mL) in PBS (pH 7.4) was incubated with 500 U/mg (one unit is defined as the amount of enzyme required to remove >95% of the carbohydrate from 10 μg of denatured RNase B in 1 h at 37 °C in a total reaction volume of 10 μL) of protein N-glycosidase F (PNGase F from NEB, New England Biolabs) overnight at 37 °C. The enzyme was then removed, and the deglycosylated antibody was concentrated by centrifugation (Amicon Ultra-2 Centrifugal Filter Unit MWCO 50 kDa from Merck, three cycles of centrifugation at 4000 rpm for 15 min). The reaction was checked by RP-HPLC-MS. Recovery yield: 90%.

Site-Specific Conjugation (Cetux-FLT-Waza-01)

Deglycosylated antibody (4 mg/mL, 1.2 mg) in PBS was incubated with 20 equiv of FLT-Waza-01 (30 mM stock solution in DMSO) and with 3 U/mg of antibody of microbial transglutaminase (MTGase from Zedira). The resulting solution was stirred overnight (16 h) using an Eppendorf ThermoMixer C (900 rpm) at 37 °C. The reaction was checked by RP-HPLC-MS (system C). Excess of fluorophore and MTGase were removed by FPLC purification (vide infra). DOL was calculated based on ESI-MS results (DOL = 2.0 ± 0.1, bioconjugation yield = 60%).

Random Conjugation (Cetux-FLT-Waza-02)

Bioconjugation was optimized to get a DOL of around 2 for comparison purposes with the site-specific method. On this account, 7 equiv of FLT-Waza-02 (15 mM stock solution in DMSO) were added to a solution of native antibody (2 mg/mL, 1.7 mg) in 0.2 M aq. bicarbonate buffer (pH 9.2). The reaction mixture was then stirred using an Eppendorf ThermoMixer C (900 rpm) for 6 h at 37 °C. The free fluorophore was removed by FPLC. DOL was calculated based on MALDI-TOF MS results (DOL = 2.1 ± 0.2, bioconjugation yield = 60%).

Fast Protein Liquid Chromatography (FPLC) Purification

FPLC purification of the conjugates was performed on an ÄKTA 25 M system (GE Healthcare Life Sciences) with a Hitrap Mabselect column (MabSelect resin, Protein A, cross-linked agarose, column I.D. seven mm, bed dimensions 7 × 25 mm, bed volume 1 mL). After deposition of the product, the conjugate was washed with 5 CV of PBS, pH 7.4 (1 mL/min), followed by 5 CV of PBS + 0.1%Tween 20 (1 mL/min) and 5 CV of PBS, pH 7.4 (1 mL/min). The conjugate was then eluted with 25 mM aq. acetic acid (1 mL/min) and collected. Purification was monitored at 280 and 700 nm. Thereafter, the solution was transferred to an ultracentrifugal filter device (Amicon Ultra 2 mL, Ultracel cutoff 50 kDa from Merck Millipore) and centrifuged at 4000 rpm for 3 × 15 min in order to condition the mixture in PBS (pH 7.4). Purification monitoring was performed using UNICORN 7.2 interface software (GE Healthcare). The recovered yield and mass amounts of antibody used, with the corresponding DOL, were determined thanks to ESI-MS results (for site-specific conjugation) or MALDI-TOF MS results (for random conjugation) after purification.

SPR Experiments

The binding affinity of cetuximab, deglycosylated cetuximab, Cetux-FLT-Waza-02 and Cetux-FLT-Waza-01 toward EGFR, as well as the association (k _a) and dissociation rate constants (k _d ), were determined using surface plasmon resonance (SPR) on a Biacore T200 system (Cytiva). First, the CM5 sensor chip (Biacore, Series S Sensor Chip CM5; Cytiva) was activated with 1-ethyl-3-(3-(dimethylamino)­propyl) carbodiimide/N-hydroxysuccinimide (EDC/NHS; Amine Coupling Kit; Cytiva). Subsequently, recombinant human EGFR protein (Sino Biologicals; cat. no. 100001-H08H) (10 μg/mL) was immobilized on the chip using 10 mM sodium acetate at pH 4.0 and 4.5 for cetuximab, with a flow rate of 30 μL/min. The SPR measurements were performed at 25 °C in HEPES-buffered saline (HBS, pH 7.4) as the running buffer. The antibody constructs were then injected consecutively in 3-fold serial dilutions ranging from 0.3 nM to 250 nM. The association step was allowed for 100 s, the dissociation step for 600 s, and the runs were repeated three times. Regeneration was performed for 30 s at 30 μL/min using 0.1 M glycine at pH 2.0, followed by a stabilization period of 180 s. The kinetic rate constants were determined by mathematical fitting using the 1:1 binding with drift and RI2 model proposed by the Biacore Evaluation Software (Cytiva), and the k _d/k _a ratio was used to determine the equilibrium dissociation constant (K_D ).

In Vivo Experiments

All in vivo experiments were performed according to the EU Directive 2010/63/EU and Belgian legislation and were approved by the Ethical Commission for Animal Experimentation of the Vrije Universiteit Brussel, Belgium (project nr. 23–272–15). Female Crl:NU-Foxn1nu mice (Charles River, France) were housed in individually ventilated cages (n = 3 mice/cage) and received a standard diet and water ad libitum (SAFE 105, Safe Diet, Augy, France). Mice were inoculated with FaDu cells (2.5 × 10⁶ cells in 150 μL PBS; purchased from ATCC) above the right hind leg and imaged when the FaDu tumors had reached a size of approximately 500 mm³.

Fluorescence Intensity and Fluorescence Lifetime Imaging

FLI was conducted with the VUB tauCAM, a macroscopic fluorescence lifetime imaging system with a 128 × 128-pixel custom fast time-gated CMOS image sensor. The tauCAM was synchronized to a pulsed supercontinuum laser (Rock 500–6, Leukos, Limoges, France) with a custom optical filter setup providing 240 mW of illumination power through an excitation filter centered around 670 nm (ET670/50m, Chroma Technology Corp., Vermont, US). On the day of the experiment, the instrument response function of the tauCAM was determined by measuring reflected excitation light with a neutral density filter (Thorlabs, Newton, NJ, USA) and an exposure time of 5 ms. Thereafter, the neutral density filter was replaced by an emission filter centered around 740 nm (ET740/40m, Chroma Technology Corp., Vermont, US) to block excitation light and capture the fluorescent emission of the bioconjugates. Mice were imaged under isoflurane anesthesia (5% for induction, 2% for maintenance, and 1.5 L/min oxygen flow rate) prior to the injection of the tracer to evaluate autofluorescence. Thereafter, mice were intravenously injected via the lateral tail vein with either Cetux-FLT-Waza-01 (n = 3) or Cetux-FLT-Waza-02 (n = 3) (100 μg in 200 μL). FLI was conducted at 6, 24, and 48 h postinjection (p.i.), with an acquisition time of between 250 and 1000 ms in the left lateral decubitus. 500 ms images were used for further quantification, while post-HDR modus was used for image display. Before analyzing the images in ImageJ, curve fitting of the fluorescent decay was performed using custom-built software, with the IRF implemented in the curve fitting model, and the fluorescence lifetimes per pixel were extracted. To ensure the validity and reliability of the reconstructed images, all data sets were subjected to a series of quality control (QC) tests.

For each pixel, the goodness-of-fit between the measured decay and the fitted model was evaluated using both the reduced chi-square statistic and the reduced Poisson deviance. The reduced chi-square was calculated as ${\chi }_{red}^2=\frac{1}{\nu }\sum _{i=1}^N\frac{(O_i-{\mu }_i{)}^2}{{\mu }_i+\sigma \text{\_}rea{d}^2+\alpha {\mu }_i^2}$ and the reduced Poisson deviance as $D_{scaled}^{red}=\frac{1}{\nu }2\sum _{i=1}^N[O_i\ln \left(\frac{O_i}{{\mu }_i}\right)-(O_i-{\mu }_i)]$ . Here, O_i represents the measured photon counts, μ_i the model-predicted photon counts, ${\sigma }_{read}^2$ the camera read-noise variance, and $\alpha {\mu }_i^2$ an intensity-dependent overdispersion term. The parameter ν denotes the number of degrees of freedom, and N the number of time bins included in the goodness-of-fit evaluation. Pixels were considered reliable when both the reduced chi-square and reduced Poisson deviance were below a threshold value of 2.5. All images satisfied these QC criteria (Figure S52). Thereafter, the fluorescence intensity and FLI images were further processed in ImageJ. FI values are represented in arbitrary units, whereas fluorescence lifetime values are represented in the picosecond-nanosecond range. ROIs were drawn for the tumors, surrounding tumor margins, and at the level of the shoulders based on reflectance images, and average fluorescence lifetime values for the ROIs were calculated. In addition, histograms depicting the normalized pixel counts within each ROI as a function of the measured fluorescence lifetime values were generated to visualize the fluorescence lifetime distribution in the drawn ROIs.

Statistical Analysis

Data were represented as the mean ± standard deviation. Statistical analyses were performed using Prism software v.10.3.1. Two-way ANOVA tests with Geisser–Greenhouse correction and correction for multiple comparisons were used to compare the fluorescence lifetime between groups and over time. Statistical differences (p-value below 0.05) were indicated with * or ^#.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c04430.

• NMR, mass spectra, and RP-HPLC chromatograms, spectroscopic/photophysical characterizations of aza-BODIPY fluorophores, bioanalytical characterization of fluorescent antibody conjugates, and supplementary in vivo images: Tables S1–S3; Figures S1–S54 (PDF)

• Molecular formula strings (CSV)

○.

E.C. and H.A.S. contributed equally to this work. S.H., A.R., C.G., and F.D. conceived the project. S.H., A.R., C.G, F.D., C.P. and H.I. supervised the research. E.C. and H.A. synthesized the fluorescent organic dyes and performed photophysical characterization with assistance of A.R. E.C. performed the bioconjugation studies. S.J., L.M., M.C.M.S., and T.D. performed the in vivo experiments and image processing with assistance of T.V.D., T.L., S.S., M.K., and H.I. were involved in the development of the imaging system used in this study. E.C., H.A., S.J., L.M., T.D., S.H., A.R., C.G., and F.D. analyzed the results and wrote the manuscript with input from all the authors. All authors have approved the final version of the manuscript.

The authors declare no competing financial interest.