Morphable ‘Stitched’ Sensors for Simultaneous Spatiotemporal Tracking of Correlated Bioanalytes in Living Cells

Abstract

Correlated changes in molecular levels and distributions are associated with all life processes and importantly regulate key decision-making events in biology. Spatiotemporal dynamics of biomolecules are essential for functions like cell signaling, transport, immunity, and recycling, and are either affected by or cause diseases such as cancers, inflammation, and neurodegeneration. In this backdrop, the ability to catch molecules of life in action using optical imaging is extremely powerful. Simultaneous tracking of biologically correlated analytes in living cells necessitates cell-permeable, multianalyte sensors. Key criteria for achieving simultaneous, correlated tracking of multiple analytes are that sensors for these analytes should enter cells at the same time and concomitantly reach the same location where we want to detect the analytes. Separate sensors, either small-molecule- or macromolecule-based, cannot fulfill these requirements directly. We introduce morphable ‘stitched’ sensors, where fluorescent sensors for single analytes can be strategically joined via native chemical ligation (NCL) on a made-to-order basis. Morphable ‘stitched’ sensors are built from a library of single-analyte sensing units conjugated to short peptide scaffolds. The use of peptide-based scaffolds in combination with NCL allows generation of modular, biocompatible, water-soluble, and importantly cell-permeable multianalyte sensors tailored to address specific biological questions. We report five proof-of-concept multianalyte sensors created from a common single-analyte sensor library using this ‘stitching’ strategy. ‘Stitched’ sensors enable simultaneous imaging and temporal tracking of bioanalytes via time-lapse imaging. In our pilot studies we image different combinations of analytes including protons, hydrogen peroxide, and enzyme activity, in live cells, affording insights into pH- and hydrogen peroxide-dependent enzyme activity. Cellular uptake studies show that the analyte-sensing modules in ‘stitched’ sensors enter cells simultaneously and as rapidly as 5 min, exhibit synchronized uptake dynamics, and are internalized in equal proportions, resulting in a uniform distribution across the cell population. Our novel morphable ‘stitching’ platform therefore offers a universal approach toward live-cell multianalyte imaging.

Introduction

The concerted dynamics of biomolecules and ions form the essence of life. All life processes including cell-signaling, growth, proliferation, transport, shape, and metabolism are intricately managed through regulated fluctuations in concentration and distribution of small molecules, ions, and macromolecules. − Understandably, aberrant fluctuations in the distribution and concentration of biomolecules/analytes are implicated in various pathophysiological conditions. − Hence, the ability to track multiple biological players simultaneously in the playground of life, is key to understanding how life works and what happens when the molecular circuitry of life is disrupted.

Toward the ultimate aim of tracking concerted fluctuations of biomolecules, we set out to develop a multianalyte sensing strategy for imaging more than one analyte simultaneously in living biological systems. How do we achieve simultaneous tracking of multiple bioanalytes in live cells? Our approach was to start with single/monoanalyte sensors which can selectively and sensitively detect and image a single type of bioanalyte. Next, we listed the requirements for translating single-analyte sensors to multianalyte sensing: 1. To track the correlated dynamics of more than one bioanalyte in a specific location within the biological sample, sensors for individual analytes should reach the same location. 2. To obtain information on relative levels of bioanalytes, the concentration of the single-analyte sensors in the targeted biological location should ideally be the same. 3. The single-analyte sensors should be cell-permeable and reach the biological location being investigated at the same time. In summary, we require spatiotemporally controlled delivery of multiple single-analyte sensors into living cells, i.e., cell-permeable ‘stitched’ multianalyte sensors (Figure a).

1.

(a) Highlighting the key advantage of our novel strategy of morphable ‘stitched’ sensors in the context of multianalyte sensing enabling spatiotemporally controlled rapid and direct cellular uptake of single-analyte sensors (right), in comparison to employing separate sensors (left). (b) Illustration of the morphable ‘stitched’ sensor concept where a library of short peptide sequences containing different cargo can be ‘stitched’ in a context-dependent manner to develop multianalyte sensors employing NCL.

However, conjugating single-analyte sensors to obtain multianalyte sensors is synthetically prohibitive. For each biological context, that is, different cell-type/intracellular compartment/combination of bioanalytes, a different synthetic strategy would be required. We asked what if we could develop a universal strategy whereby single-analyte sensors could be ‘stitched’ together to provide cell-permeable multianalyte sensors in a context-dependent manner. For example, we could have a library of single-analyte sensors, like sensors for pH, reactive oxygen species (ROS), enzymes, and metal ions. These sensors can then be ‘stitched’ to afford rapidly cell-permeable, multianalyte sensors on a made-to-order basis (Figure ).

We note that ex vivo/outside cell detection of multiple bioanalytes is possible via the ‘chemical nose’ strategy whereby multiple single analyte sensors do not have to be conjugated. − However, the ‘chemical nose’ concept does not fulfill requirements of in vivo/in-cell multianalyte sensing which requires spatiotemporally controlled rapid cellular uptake of single-analyte sensors. Elegant examples of multianalyte sensing of bioanalytes have been reported by Yamuna Krishnan’s group who have used DNA scaffolds to conjugate multiple sensors (Table S1). − Protein, DNAzyme, and RNA aptamer–dye pair-based methods have also been explored (Table S1). − However, all of these scaffolds are cell-impermeable and rely on complex cell-delivery techniques (Table S1). Apart from macromolecule-based approaches, dual-responsive probes that can detect more than one analyte, have also been reported in literature. − These probes are usually based on single reporters/fluorophores. However, in these sensors, different analytes can attenuate or enhance the signal, and combination of analytes often produce ambiguous responses. As a result, intensity-based readouts become difficult to interpret, making single-dye, multianalyte sensing inherently prone to misinterpretation. Importantly, such sensor designs are customized and not general, making synthesis of modular probes nontrivial. DNA-nanoparticle conjugate, and quantum dot-based single-vector FRET probes have been developed for detecting multiple targets. However, DNA-nanoparticle based systems can be limited by cell impermeability, nanoparticle-based probes may face challenges in biological translation, and quantum dot-based approaches can have issues related to biocompatibility and compositional complexity. , Recently, pattern-generating fluorescence-based, unique identification (ID) probes have emerged, producing unique emission fingerprints to distinguish multiple analytes both in solution and intracellularly. However, unimolecular ID probes with covalently linked sensing units are synthetically challenging, while self-assembled DNA-based ID probes tend to dissociate in cellular environments. To the best of our knowledge, there are no strategies for developing made-to-order, cell-permeable, multianalyte sensors.

We have designed a novel strategy in which a library of single-analyte sensors can be conjugated or ‘stitched’ on an as-needed basis (Figure b). The idea of morphable ‘stitched’ sensors, involves attaching single-analyte sensors to short peptide sequences. Each peptide sequence will have only one reactive amino acid sidechain for attaching the single-analyte sensor, thereby avoiding multiple protection/deprotection steps. The single-analyte peptide blocks can be ‘stitched’ via a biocompatible, mild peptide-ligation method like native chemical-ligation (NCL) on a made-to-order basis (Figure b). This method will have several advantages: (1) the use of short peptide scaffolds which are biocompatible; (2) avoiding multiple protection/deprotection strategies which would be needed if multiple sensors have to be conjugated to the same scaffold; (3) ability to leverage cell-permeability and targetability of peptide sequences; (4) most importantly, morphability of the strategy, i.e., the same library of single-analyte sensors can be used to generate different cell-permeable, multianalyte sensors based on the biological context and the selection of analytes to be detected.

To test the strategy, a library of single-analyte sensing units was developed for sensing H⁺ ions, H₂O₂, and cathepsin B activity which are representative examples of ions, small-analytes, and macromolecules, respectively (Figure ). Different sensor combinations from the same library were ‘stitched’ using NCL to create cell-permeable, multianalyte sensors (Figure ). We have demonstrated five proof-of-concept examples of the strategy for simultaneous tracking of bioanalytes and enzyme activity in living cells. Importantly, the morphable ‘stitched’ sensor strategy is amenable to different sensing schemes (Figures , , and ) and indeed affords simultaneous rapid uptake, localization, and internalization in equal proportions, for the single-analyte sensing modules. In the future, by employing libraries containing single-analyte sensors, therapeutics, and diagnostic agents; biological-context-dependent combination sensing, personalized disease diagnosis, and disease treatment avenues may become accessible via our strategy.

2.

Scheme depicting the designs of the five proof-of-concept multianalyte sensors generated using our novel strategy of morphable ‘stitched’ sensors.

3.

(a) Fluorescence response of individual pH sensor (10 μM) to pH (4 to 8) in HEPES (20 mM) buffer, λ_ex 540 nm. (b) Fluorescence response of MSS1 (10 μM) to pH (4 to 9) in HEPES (20 mM) buffer, λ_ex 540 nm. (c) Fluorescence response of individual H₂O₂ sensor (10 μM) to H₂O₂ (0 μM to 210 μM) in HEPES (20 mM) buffer (pH 7), λ_ex 380 nm. (d) Fluorescence response of MSS1 (10 μM) to H₂O₂ (0 μM to 210 μM) in HEPES (20 mM) buffer (pH 7), λ_ex 380 nm. (e) Scheme depicting the response of MSS1 at different conditions of pH and H₂O₂ upon excitation at 380 and 540 nm, respectively. (f) 3D plot correlating the ratio of emission at 573 nm to that at 514 nm (λ_ex = 380 nm), with change in both pH and H₂O₂ concentration.

6.

(a) Scheme depicting the sensing mechanism of CTSB activity by MSS2 and MSS3. (b) Representative confocal single z plane images of HeLa cells depicting colocalization of MSS2 (pH channel, top panel: λ_ex = 543 nm, λ_em = 575–650 nm; CTSB channel, bottom panel: λ_ex = 405 nm, λ_em = 420–500 nm) with Lysotracker Green (λ_ex = 488 nm, λ_em = 520–560 nm). Cells were incubated with MSS2 (10 μM) for 5 min, washed, incubated with Lysotracker Green (100 nM) for 30 min, washed, and imaged. Scale bar: 10 μm. (c) Representative confocal single z plane images of HeLa cells depicting colocalization of MSS3 (pH channel, top panel: λ_ex = 543 nm, λ_em = 575–650 nm; CTSB channel, bottom panel: λ_ex = 405 nm, λ_em = 420–500 nm) with Lysotracker Green (λ_ex = 488 nm, λ_em = 520–560 nm). Cells were incubated with MSS3 (10 μM) for 5 min, washed, incubated with Lysotracker Green (100 nM) for 30 min, washed, and imaged. Scale bar: 10 μm.

8.

(a) Scheme depicting the sensing mechanism of CTSB activity by MSS4 and MSS5. (b) Representative confocal single z plane images of HeLa cells depicting colocalization of MSS4 (H₂O₂ channel, top panel: λ_ex = 488 nm, λ_em = 520–570 nm; CTSB channel, bottom panel: λ_ex = 405 nm, λ_em = 420–500 nm) with Lysotracker Red (λ_ex = 543 nm, λ_em = 575–650 nm). Cells were incubated with MSS4 (10 μM) for 5 min, washed, incubated with Lysotracker Red (100 nM) for 30 min, washed, and imaged. Scale bar: 10 μm. (c) Representative confocal single z plane images of HeLa cells depicting colocalization of MSS5 (H₂O₂ channel, top panel: λ_ex = 488 nm, λ_em = 520–570 nm; CTSB channel, bottom panel: λ_ex = 405 nm, λ_em = 420–500 nm) with Lysotracker Red (λ_ex = 543 nm, λ_em = 575–650 nm). Cells were incubated with MSS5 (10 μM) for 5 min, washed, incubated with Lysotracker Red (100 nM) for 30 min, washed, and imaged. Scale bar: 10 μm.

Results and Discussion

Blueprint of Morphable ‘Stitched’ Sensors

Simultaneous tracking of more than one small molecule/ion/enzyme activity in living cells requires respective single-analyte sensors to reach the same intracellular location at the same time and in equal proportions. This condition can be fulfilled if multiple single-analyte sensors are integrated into a single cell-permeable molecule (Figure a). However, chemically joining multiple sensor molecules is synthetically challenging and is not general since each multianalyte sensor would require a unique synthetic strategy. Further, tedious optimization would be required to achieve cell-permeability and water solubility, two key requirements for simultaneous live-cell imaging. The morphable ‘stitched’ sensing strategy involves attaching a single-analyte sensor to a short peptide sequence to form peptide-based monomeric sensing units (Figure b). The peptide sequences would be selected such that there would be only one amino acid with a reactive side chain to attach a single-analyte sensor. Hence, the development of single-analyte sensing blocks would ideally require a single step. Peptide blocks, each containing a single cargo, would then be ‘stitched’ together via NCL (Figure b). The strategy of morphable ‘stitched’ sensors not only simplifies synthesis but also allows for flexibility in designing sensors tailored to specific analytes and biological contexts, as different single-analyte sensors can be ‘stitched’ using the generalized strategy in a context-dependent manner (Figure b). A key advantage of using a peptide-based scaffold is rapid cell-permeability and targetability to specific intracellular locations which can be easily achieved by selecting an appropriate peptide scaffold. These features are not easily accessible in other synthetic or biological scaffolds (Table S1). We note that for applications in molecular sensing, long-term intactness of the peptide scaffold in the cellular environment is not strictly necessary. Once the single-analyte sensors are delivered to the requisite biological location, simultaneously and in the correct proportion, intactness of the scaffold is not a stringent condition for the sensing strategy to work. However, if required, the intactness of the multianalyte sensing entity, can be confirmed in live cells by applying Förster resonance energy transfer (FRET)-based sensing.

The primary requirements for NCL are a cysteine residue at the N-terminus of one peptide and a thioester modification at the C-terminus of the other peptide monomeric unit (Figure b and Scheme S1 and S2). Thus, when selecting the peptide sequences, the following conditions should be met: 1. The entire peptide sequence after ‘stitching’ should be cell-permeable. 2. The monomeric peptide blocks should either have an N-terminal cysteine or a thioester-modified C-terminus that are required for NCL. 3. The monomeric peptide blocks should have only one amino acid with a reactive side chain for cargo attachment to avoid side reactions. Peptide sequences can be designed based on previously reported cell permeable peptides while ensuring an N-terminal cysteine in one of the blocks and a single reactive amino acid in each block (Figure ).

While the concept of morphable ‘stitched’ sensors may be applicable to any imaging modality, optical imaging in a fluorescence confocal microscopy setup was selected for the pilot experiments. The reasons for this choice are noninvasiveness and apt spatiotemporal resolution of confocal fluorescence microscopy for live-cell imaging. Importantly, numerous selective and sensitive single-analyte fluorescent sensors have been reported, − enabling a wide range of fluorescent probes from which single-analyte sensors can be selected. This will also widen the scope of this strategy for general applicability in the future. As proof-of-concept examples, we applied the ‘stitched’ sensor approach to investigate the interplay between pH, H₂O₂, and cathepsin B activity in living cells. Since pH and H₂O₂ influence key biological processes and can individually affect cathepsin B activity, − a library of single-analyte sensors for each of these biological molecules/ions was generated. By linking different monoanalyte sensors from this library via a single ligation step, we built five multianalyte sensors that enabled simultaneous imaging of pH, H₂O₂, and cathepsin B in living cells.

Simultaneous Tracking of pH and H₂O₂ Using the Morphable ‘Stitched’ Sensing Strategy

To establish the morphable ‘stitched’ sensing strategy, a multianalyte sensor (MSS1) was developed for simultaneous detection of pH and H₂O₂ inside living cells (Figure ). pH and H₂O₂ play important roles in regulating a multitude of biological processes including cell signaling, metabolism, immune response and autophagy. ,− Moreover, antioxidant enzymes work efficiently at optimum physiological pH and regulate the cellular redox balance. , Therefore, disruption in physiological pH may perturb the cellular redox balance. In diseases like cancers and neurodegenerative disorders like Alzheimer’s disease, disrupted cellular pH has been reported to be associated with elevated oxidative stress. ,, Investigating the inter-relationship between pH and H₂O₂ simultaneously, during different biological processes can therefore enhance understanding of oxidative stress-related cellular physiology and pathology.

For the first morphable ‘stitched’ sensor, MSS1, 1,8-naphthalimide with a boronic ester cap , was selected as a single-analyte sensor for H₂O₂ detection. , 5(6)-carboxytetramethylrhodamine was chosen as the single-analyte sensor for pH detection (Figure ). Many cell-permeable peptide sequences contain lysine (K), alanine (A), and leucine (L). ,− Among the three above-mentioned amino acids, only K has a reactive side chain. Moreover, both selected single-analyte sensors featured carboxylic acid functionalities, which could be attached to the peptide via a lysine side chain. Hence, a sequence containing K, L, and A amino acids was suitable for our strategy. Accordingly, we designed our peptide blocks to contain K, A, and L amino acids (Figure ).

The design of the H₂O₂-sensitive block included a cysteine residue at the N-terminus. The C-terminus of the pH-sensitive block was modified to hydrazine (Scheme S3 and S4). The final complete peptide sequence for the ‘stitched’ sensor (MSS1) was K-A-L-A–C-K-L-A, which was synthesized in two blocks: K-A-L-A (for pH block A) and C-K-L-A (for H₂O₂ block B) (Scheme S3 and S4). The pH-sensitive dye was attached to lysine and the lysine–dye conjugate (2) was attached to peptide 1 to form the pH block A (Scheme S3). The H₂O₂-sensitive dye was also attached to lysine and the lysine–dye conjugate (4) was then attached to peptide 3 to form the H₂O₂ block B (Scheme S4). The hydrazine-modified C-terminus of the pH block A was converted to a thioester in situ and ligated to the H₂O₂ block B, containing an N-terminal cysteine, via NCL in a single-pot, thus completing the synthesis of MSS1 (Scheme S5). MSS1 was then purified using HPLC and characterized using LC/ESI-MS (Figure S1).

MSS1 was completely water-soluble in the tested concentration of up to 2 mM. Therefore, all experiments were carried out in aqueous buffer. The absorption spectra of MSS1 displayed two characteristic peaks at 350 and 553 nm, corresponding to the two dyes under neutral pH conditions. Under acidic pH, the absorbance at 553 nm increased, while the addition of H₂O₂ caused the 350 nm peak to shift to 380 nm (Figure S2). Upon excitation of the individual pH (rhodamine-lysine conjugate) sensor at 540 nm, a fluorescence peak at 573 nm was observed, which increased 2-fold as the pH decreased from 8 to 4 (Figure a). When the sensor was ‘stitched’ to form MSS1 and excited at the same wavelength (540 nm), a similar emission peak appeared, showing a comparable 2-fold increase in intensity as the pH decreased from 9 to 4 (Figure b and Scheme S6). These results demonstrate that ‘stitching’ did not affect the functionality of the pH sensor. When the individual H₂O₂ sensor (naphthalimide-lysine conjugate) was excited at 380 nm, an emission peak at 515 nm was observed, which exhibited a 2-fold increase in intensity with rising H₂O₂ concentration (Figure c). When MSS1 was excited at 380 nm in an aqueous buffer of pH 7, a similar peak at 514 nm was observed (Figure d). The emission peak at 514 nm showed a 2.1-fold enhancement in the presence of increasing concentrations of H₂O₂, attributed to the conversion of the boronic ester group in the H₂O₂-sensitive naphthalimide-based dye to a hydroxyl group (Figures d and S2 and Scheme S6). Hence, ‘stitching’ did not affect the functionality of the H₂O₂ sensor as well. Upon excitation of MSS1 at 380 nm, a shoulder at 573 nm was also observed which became more prominent at acidic pH (Figures d, S2, and S4a). We hypothesized that the peak at 573 nm was due to FRET from naphthalimide to rhodamine (Figures d and S3 and Scheme S6). To get definitive evidence of FRET, fluorescence spectra of the donor naphthalimide-boronic ester dye were recorded independently at pH 7 and pH 4 in the presence of H₂O₂ (210 μM) (Figure S4b). Importantly, the donor emission intensity and spectral shape remained essentially unchanged between these two pH conditions, indicating that the donor itself does not undergo pH-dependent photophysical changes. Using these donor-only spectra as references, the FRET efficiency in MSS1 was quantified. At pH 7, the calculated efficiency was approximately 24%, whereas at pH 4 it increased to 67%, consistent with enhanced energy transfer under acidic conditions (Figure S4c,d).

The pH and H₂O₂ broad end-point responses of MSS1 are summarized in Figure e and Scheme S6. A 3D plot could be constructed (Figure f), correlating the ratio of emission at 573 nm to that at 514 nm (λ_ex = 380 nm) with change in both pH and H₂O₂ concentration to enable simultaneous monitoring of relative changes in pH and H₂O₂.

For live-cell imaging experiments, 405 nm laser (for excitation of the H₂O₂ sensing segment) was used as one of the excitation sources for MSS1, with two emission channels: 500 to 545 nm (green channel) and 575 to 650 nm (FRET channel). Additionally, a 543 nm laser (for excitation of the pH sensing segment) served as the other excitation source, with emission detected from 575 to 650 nm (red channel) (Figure ). The expected response of MSS1 at different levels of H⁺ and H₂O₂ are summarized in Figures e and a. An MTT assay for cellular toxicity in HeLa cells indicated that the sensor was not toxic up to a tested concentration of 20 μM, until the maximum tested incubation time of 60 min (Figure S5). Hence, concentrations lower than 20 μM and incubation times less than 60 min were used for all cell studies. MTT assays were also conducted spearately for cells incubated with 100 μM H₂O₂ and cells incubated with media at pH 4.5 (experimental conditions) without any sensor incubation, confirming that approximately 90% and 100% of cells remained viable, respectively (Figure S6). Because MTT measurements can be unreliable at low pH, propidium iodide (PI) staining was additionally performed to directly assess cell death. The PI data analysis showed a nonsignificant difference in staining intensity between pH 7.4 and pH 4.5 (Figure S6), confirming that the imaging conditions do not compromise cell health.

4.

(a) Representative confocal single z plane images of HeLa cells incubated with MSS1 (10 μM) for 30 min, washed, and imaged (i–iii). H₂O₂ (100 μΜ) was added to the cells and imaged after 30 min (iv–vi). Following that, the external media pH was changed to 4.5, and cells were imaged after 30 min (vii–ix). [First row: λ_ex = 405 nm, λ_em = 500 to 545 nm; second row: λ_ex = 405 nm, λ_em = 575 to 650 nm; third row: λ_ex = 543 nm, λ_em = 575 to 650 nm]. (b) Representative confocal single z plane images of HeLa cells incubated with MSS1 (10 μM) for 30 min, washed, and imaged (i–iii). The external media pH was changed to 4.5, and cells were imaged after 30 min (iv–vi). Following that, H₂O₂ (100 μΜ) was added to the cells and imaged after 30 min (vii–ix). [First row: λ_ex = 405 nm, λ_em = 500 to 545 nm; second row: λ_ex = 405 nm, λ_em = 575 to 650 nm; third row: λ_ex = 543 nm, λ_em = 575 to 650 nm].

5.

(a) Comparing the experimental in-cell response of MSS1 with the expected response. (b) Bar plot showing mean intensities obtained from intensity analysis of four sets of cells at different pH and H₂O₂ levels. Data are presented as SEM, where N = 3, n = 30 in each set.

MSS1 entered cells within 30 min of direct incubation (Figures and S7). The percentage of labeled cells was calculated from fluorescence confocal images of the sensor taken with a wide field of view (Figure S8). In multiple cell plates (n = 45–90 total cells), the percentage of cells labeled by MSS1 was 100%. In resting HeLa cells, low fluorescence was observed in all the channels (green, red and FRET) due to low H₂O₂ levels and neutral cytosolic pH as predicted (Figures a,b­(i–iii), e, and a). Since the fluorescence signal inside the cells was low, we further evaluated the signal-to-noise (S/N) ratio under wash-free conditions after 30 min of incubation. The initial S/N was >4 (Figure S9) and upon H₂O₂ addition, the S/N increased markedly (Figure S9). If significant probe remained outside the cells, the background would have also increased after H₂O₂ treatment, but no such increase was observed. The initial S/N reflects very low resting fluorescence of the probe, however an S/N of ∼4 supports reliable intracellular signal detection. Although MSS1 showed uniform uptake in cells within 30 min of incubation, the fluorescence intensity per cell was low, possibly indicating a low cellular uptake due to the absence of positively charged amino acids.

A few punctate regions with relatively higher emission intensity were observed in the red channel (λ_ex = 543 nm, λ_em= 575–650 nm) (Figure a,b­(iii)). The punctate regions may correspond to acidic compartments within the cells. When cells were exogenously treated with H₂O₂ at pH 7.4, enhanced fluorescence was only observed from the green channel upon 405 nm excitation (Figure a­(iv)). Due to the neutral cytosolic pH, the fluorescence intensity observed from the red channel (λ_ex = 543 nm, λ_em = 575–650 nm) was negligible (Figure a­(vi)). Additionally, there was low fluorescence intensity in the FRET acceptor channel (λ_ex = 405 nm, λ_em = 575–650 nm) (Figure a­(v)).

Following this, the media pH was changed to 4.5. 30 min post media change, due to the lowering of the cytosolic pH, an enhanced emission was observed from the red channel (λ_ex = 543 nm, λ_em = 575–650 nm) (Figure a­(ix)). Using TRapH, a ratiometric pH sensor developed by our group, we directly quantified intracellular pH changes when extracellular media pH was lowered to 4.5 and confirmed that an ∼20 min lag is sufficient for the intracellular pH to stabilize at a lower value (Figure S10). Therefore, the results obtained with MSS1, could be cross-validated with the ratiometric pH sensor TRapH. The fluorescence intensity from the green FRET-donor channel (λ_ex = 405 nm, λ_em = 500 to 545 nm) decreased substantially (Figure a­(vii)) with a concomitant increase in intensity in the FRET acceptor channel (λ_ex = 405 nm, λ_em = 575–650 nm) (Figure a­(viii)). We noted that the in-cell response of MSS1 was exactly in accordance with the predicted response of the sensor (Figures e, , and a).

In order to find out if MSS1 could be used for live and simultaneous spatiotemporal tracking of pH and H₂O₂, time lapse images in living cells were recorded over 60 min. The experimental scheme is depicted in Figure a. We observed gradual increase in fluorescence intensity in the green channel (0–30 min) and low fluorescence intensity in the FRET channel (0–30 min) upon increase in H₂O₂ levels at pH 7.4 (Video S1). At the 30 min time point when the media pH was changed to 4.5, the fluorescence intensity in green channel gradually decreased over time (30–59 min) with concomitant increase in the FRET channel (30–59 min) (Video S1). This result distinctly showed that the ‘stitched’ sensor MSS1 was intact within living cells over a long time period of at least 1h and could spatiotemporally and simultaneously track pH and H₂O₂ levels.

Cell experiments were also performed by reversing the sequence of pH change and H₂O₂ addition. When the external media pH was changed to 4.5 without exogenous H₂O₂ treatment, there was an increase in emission intensity in the red channel (λ_ex = 543 nm, λ_em = 575–650 nm) due to lower cytosolic pH (Figure b­(vi)). Since the H₂O₂ levels were low, negligible fluorescence emission was observed upon 405 nm excitation (Figure b­(iv–v)). Following this, cells were exogenously treated with H₂O₂. The enhanced fluorescence intensity in the red channel (λ_ex = 543 nm, λ_em = 575–650 nm) was maintained without further change (Figure b­(ix)) as there was no further alteration in the cytosolic pH. The fluorescence intensity from the FRET acceptor channel (λ_ex = 405 nm, λ_em = 575–650 nm) increased (Figure b­(viii)) while that of the FRET donor channel remained low due to energy transfer (Figure b­(vii)). Therefore, MSS1 was able to distinguish all four experimental conditions: neutral pH with low H₂O₂, neutral pH with high H₂O₂, low pH with low H₂O₂, and low pH with high H₂O₂ (Figure ).

Given the clear and compelling evidence for simultaneous uptake and colocalization of both sensing units in MSS1, as demonstrated by FRET, the uptake behaviors of representative ‘unstitched’ probes (rhodamine and 1,8-naphthalimide boronic ester) were also examined to investigate the advantage of our ‘stitching’ strategy. The results showed that the pH sensor (rhodamine) entered the cells rapidly whereas the H₂O₂ sensor (1,8-naphthalimide boronic ester) took longer time as compared to the former, displaying distinctly different rates of cellular uptake (Figure S11). It is important to note that the sensors would start sensing analytes while being taken up and differential uptake would lead to completely uncorrelated sensing of the analytes. Additionally, the overlay images at different time-points showed dissimilar spatial distribution of the ‘unstitched’ probes (Figure S11). This observation was supported quantitatively by Pearson’s coefficients of 0.28 at 15 s and 0.38 at 900 s. These findings clearly demonstrate that the ‘unstitched’ single-analyte sensors may enter the cells at different times with distinctly different localization (Figure a), making them unsuitable for the simultaneous detection of multiple analytes within specific intracellular regions. This observation highlights the significance of our ‘stitching’ strategy in simultaneous sensing of biocorrelated analytes.

‘Stitched’ Sensors for Simultaneous Tracking of Cathepsin B Activity and pH

Encouraged by the results with MSS1 for simultaneous tracking of pH and H₂O₂ inside living mammalian cells, the repertoire of morphable ‘stitched’ sensors was expanded. We next sought to monitor the activity of an enzyme along with various bioanalytes simultaneously for understanding of the roles of different analytes on enzyme function. Cathepsin B (CTSB) is a lysosomal cysteine protease which plays a key role in intracellular proteolysis. CTSB enables protein turnover and therefore maintains cellular homeostasis. CTSB has also been reported to participate in antigen processing, thus aiding in immune response. − Dysregulation of CTSB is implicated in cancers, neurological disorders, and inflammatory disorders. − Under physiological conditions, CTSB functions within the lysosome (pH 4.5–5). However, in the case of neurological disorders and inflammatory diseases, CTSB has been reported to translocate from the lysosome to the cytoplasm due to lysosomal membrane permeabilization (LMP). The translocated CTSB then functions at the cytosolic pH (pH 7.4) on substrates different than that in lysosomes, triggering inflammatory pathways, ultimately leading to cell death. , Consequently, it is crucial to monitor CTSB activity across different substrates, as well as the influence of pH on its function. We reckoned that our multianalyte sensing approach can aid in identifying pH-dependent substrate specificity for CTSB and facilitate the development of diagnostic tools for disorders associated with CTSB dysregulation. Additionally, it may assist in designing targeted therapeutics that selectively inhibit the pathophysiological activity of CTSB at pH 7.4, while preserving its normal function within the lysosome (pH 4.5–5).

There are some reports regarding possible pH-dependent substrate-specificity of CTSB, , but direct visualization of CTSB activity on those substrates simultaneously with pH in living systems has not been achieved yet, to the best of our knowledge. In this backdrop, the novel concept of morphable ‘stitched’ sensors presented here was used to develop two multianalyte sensors (MSS2 and MSS3) (Figure ) for simultaneous tracking of intracellular pH and CTSB activity employing two distinct substrates for CTSB.

CTSB exhibits carboxypeptidase activity. Two among the reported peptide substrates of CTSB are phenylalanine (F)-arginine (R)-X and R-R-X, where X represents any amino acid. CTSB cleaves the C-terminal peptide bond (R–X). , F-R-X is cleaved by CTSB across a broad pH range with enhanced cleavage in acidic pH, while R-R-X has been shown to be cleaved by CTSB preferentially at neutral pH. , These two sequences were selected for the CTSB sensitive blocks of MSS2 and MSS3, respectively (Scheme S7). In place of X, 7-amino-4-trifluoromethylcoumarin (ATC) was incorporated as the fluorophore (Scheme S7) inspired by reports of existing CTSB sensors. The constructs would initially be nonemissive due to lack of intramolecular charge transfer within the fluorophore attached to the peptide (Figure a and Scheme S9). Upon CTSB activity, the C-terminal amide bond would be cleaved, releasing the fluorophore, resulting in a turn-on fluorescence response (Figure a and Scheme S9). Cysteine was attached at the N-terminus of both the sequences for NCL (Scheme S7). Therefore, the final sequences of the CTSB-sensitive blocks were C-F-R-ATC (CTSB block C for MSS2) and C-R-R-ATC (CTSB block D for MSS3). The pH block A used for the synthesis of MSS1 was used again in MSS2 and MSS3. This highlights the modularity and morphability of our sensing concept. The pH block A was attached to each CTSB-sensitive block C and D, respectively, via NCL to complete the synthesis of MSS2 and MSS3 (Scheme S8). Both the sensors were purified using HPLC and characterized using LC/ESI-MS (Figure S12 and S13).

MSS2 and MSS3 were completely water-soluble in the tested concentrations of up to 1 mM. Therefore, all experiments were carried out in aqueous buffer. The absorption spectra of MSS2 and MSS3 featured a characteristic peak at 553 nm and a broad peak near 350 nm corresponding to rhodamine and ATC dyes, respectively (Figure S14). Upon excitation at 540 nm, MSS2 and MSS3 displayed a peak with a maximum at 571 nm, which increased 3.3-fold in emission intensity as the pH decreased from 7 to 4 (Figure S15). Upon excitation at 380 nm, both MSS2 and MSS3 exhibited a low-intensity peak at 491 nm corresponding to the quenched ATC unit of the CTSB block (Figure S16). The intensity of the peak at 491 nm did not exhibit any change at lower pH (pH 4) (Figure S17). The expected responses of MSS2 and MSS3 in the presence of analytes of interest are summarized in Scheme S9.

Next, the in-cell responses of MSS2 and MSS3 were studied, using 405 and 543 nm lasers as the excitation sources for the ATC-based dye and the rhodamine-based dye, respectively. An MTT assay for cellular toxicity in HeLa cells indicated that the sensors were not toxic up to a tested concentration of 20 μM, until the maximum tested incubation time of 30 min (Figure S18). Hence, concentrations lower than 20 μM and incubation times less than 30 min were used for all cell studies. Both MSS2 and MSS3 entered cells within 5 min of direct incubation (Figure b,c and ). The faster cellular uptake time observed for MSS2 and MSS3 when compared to MSS1 may be attributed to the presence of arginine residues. The percentage of labeled cells was calculated from fluorescence confocal images of both the sensors taken with a wide field of view (Figure S19). In multiple cell plates (n = 45–90 total cells), the percentages of cells labeled by MSS2 and MSS3 were 100%.

7.

(a) Representative confocal single z plane images of HeLa cells incubated with MSS2 (10 μM) for 5 min, washed, and imaged (first column). Following that, the external media pH was changed to 4.5, and cells were imaged after 30 min (right column). [First row: λ_ex = 543 nm, λ_em = 575 to 650 nm (pH channel); second row: λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel)]. Scale bar: 20 μm. (b) Representative confocal single z plane images of HeLa cells incubated with MSS3 (10 μM) for 5 min, washed, and imaged (first column). Following that, the external media pH was changed to 4.5, and cells were imaged after 30 min (right column). [First row: λ_ex = 543 nm, λ_em = 575 to 650 nm (pH channel); second row: λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel)]. Scale bar: 20 μm. (c) Bar plot representing variation of the in-cell fluorescence of the pH-sensitive unit at neutral and acidic intracellular pH. Data are presented as SEM, where N = 3, n = 30 in each set. (d) Bar plot representing variation of the in-cell fluorescence of the CTSB-sensitive unit at neutral and acidic intracellular pH. Data are presented as SEM, where N = 3, n = 30 in each set. *p < 0.05, **p < 0.01, and ***p < 0.001. (e) Uptake of MSS2 visualized through intensity measurements in both the blue and red channels over time. Inset: ratio of blue channel to red channel intensities during the uptake process. λ_ex = 543 nm, λ_em = 575 to 650 nm (pH channel); λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel). (f) Uptake of MSS3 visualized through intensity measurements in both the blue and red channels over time. Inset: ratio of blue to red channel intensities during the uptake process. λ_ex = 543 nm, λ_em = 575 to 650 nm (pH channel); λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel).

In resting HeLa cells, punctate-like regions were observed in both the channels in case of MSS2 (Figure b). Punctate-like regions were also observed in the red channel for MSS3 (λ_ex = 543 nm; pH channel) (Figure c). The intensities in the red channel (λ_ex = 543 nm; pH channel) in case of both MSS2 and MSS3 in resting HeLa cells were similar to the intensity obtained in the red channel of MSS1 (Figure S20). This result validated that changing the H₂O₂ sensing block of MSS1 to a CTSB sensitive block did not have any effect on the response of the pH block A, highlighting a key advantage of our ‘stitched’ sensors strategy. The fluorescence in the blue channel (λ_ex = 405 nm; CTSB channel) of MSS3 was more spatially spread compared to that in case of MSS2 (Figure b,c). The substrate for CTSB in MSS2 gets cleaved preferentially at acidic pH. Hence fluorescence corresponding to release of ATC leading to punctate regions observed in the blue channel (CTSB channel) of MSS2 were probably due to the CTSB activity in lysosomes. The substrate in case of MSS3 is expected to be cleaved by CTSB at both neutral and acidic pH. Therefore, MSS3 can also possibly track the cytosolic CTSB activity along with activity in lysosomes, resulting in diffused blue fluorescence along with puncta (CTSB channel).

To establish the pH dependent tracking of CTSB activity by MSS2 and MSS3, colocalization studies were performed with Lysotracker Green (λ_ex = 488 nm) (Figure b,c). For both sensors, the high intensity regions in the red channel (pH channel) colocalized well with Lysotracker Green (Pearson’s coefficient = 0.88 (MSS2) and 0.87 (MSS3)) (Figure b,c). This was expected since the pH-sensitive unit exhibits higher fluorescence at acidic pH and lysosomes have a pH of around 4.5–5. In the case of MSS2, we observed excellent colocalization with Lysotracker not only in the red channel but also in the blue channel (CTSB channel) (Pearson’s coefficient = 0.86) (Figure b). This is consistent with the fact that the substrate for MSS2 is preferentially cleaved under acidic pH conditions, , enabling the sensor to track lysosomal CTSB activity, which accounts for the strong colocalization in the blue channel (CTSB channel) with Lysotracker Green. For MSS3, the colocalization in the blue channel (CTSB channel) was significantly lower compared to MSS2 (Pearson’s coefficient = 0.61) (Figure c). This confirms that the substrate of MSS3 is cleaved not only at acidic pH but also at neutral pH, , enabling the sensor to track CTSB activity both in lysosomes and in the cytosol. The reduced colocalization in the blue channel further indicates the presence of CTSB activity outside the acidic lysosomal environment in resting cells. To ensure that there was no bleed-through from the blue (CTSB) to red (pH) channels of MSS2 and MSS3, respectively, cells were incubated with MSS2/MSS3 and imaged at the excitation wavelength 405 nm and red emission channel (575–600 nm) (Figure S21). In both cases, negligible bleed-through fluorescence was observed from the blue channel to the red channel confirming that there was no cross-talk between the responses in live-cell imaging.

When the external media pH was changed to 4.5, there was a significant increase in intensity in the red channel (pH channel) in case of both MSS2 and MSS3, similar to that observed in case of MSS1 (Figures a–c and S22a,b), indicating a decrease in intracellular pH. Concomitantly, a significant rise in fluorescence intensity in the blue channel (CTSB channel) was noted with a decrease in intracellular pH (Figures a,b,d and S22c,d) for both MSS2 and MSS3. In the case of MSS2, the fluorescence in the blue channel became more diffuse, in contrast to the initial punctate-like distribution (Figure S22c). This may be attributed to the lowered cytosolic pH, which allowed MSS2 to track cytosolic CTSB activity more effectively, resulting in a more intense and diffuse fluorescence in the blue channel. Furthermore, the intensity in the blue channel also significantly increased in the cytoplasm in case of MSS3 (Figure S22d). If lowering the intracellular pH did not have any effect on CTSB translocation, then the blue channel (CTSB channel) intensity of MSS3 would have remained unchanged on decreasing cellular pH. The increase in the intensity in the CTSB channel of MSS3 thus indicated that lowering the cytosolic pH led to enhanced translocation of CTSB from the lysosomes to the cytosol. A decrease in cytosolic pH may compromise the integrity of the lysosomal membrane, making it more permeable. This enhanced permeability can in turn facilitate the translocation of CTSB, from the lysosome into the cytosol. − This may be cited as the reason behind the elevated cytosolic levels of CTSB, as evident from the enhanced blue fluorescence in the cytoplasm on lowering intracellular pH in case of MSS3. These findings directly shed light on the pH-dependent substrate specificity of CTSB, further demonstrating that MSS2 and MSS3 could track CTSB activity under different pH conditions.

To further investigate the selectivity of the probes toward monitoring the activity of CTSB, cells were incubated with a CTSB inhibitor followed by the respective probes (Figures S23 and S24). The inhibitor did not affect the response of the pH-sensitive units at physiological pH, yet a few altered responses were observed in some cells at reduced media pH with no significant change in the mean intensity values (Figures S23 and S24). The observed shifts may be attributable to the inhibitor perturbing the physiology of certain cells within the population. However, the CTSB (blue) channel intensities in case of both MSS2 and MSS3 were significantly reduced in inhibitor-treated cells at resting state (Figures S23 and S24). Importantly, a change in media pH did not trigger any increase in fluorescence in the CTSB channel in the presence of the inhibitor (Figures S23 and S24). Therefore, these results confirmed that the probes could selectively track the in-cell activity of CTSB.

Time-lapse images were acquired to track CTSB activity and pH simultaneously in live cells. Upon decrease in external media pH to 4.5, the fluorescence intensity in the red channel (pH) increased with time (0–28 min) in case of both MSS2 and MSS3 with a concurrent overall increase in blue channel (CTSB activity) intensities (0–28 min) in case of MSS2 (punctate-like to diffused distribution) and MSS3 (significant increase in the cytoplasm) (Videos S2 and S3). Lowering cytosolic pH, reduced colocalization between the CTSB channel and Lysotracker (Figures S25 and S26), indicating that CTSB-mediated cleavage also occurred outside lysosomes. Simultaneously, colocalization between the CTSB and pH-sensing channels in the cytosol distinctly increased (Figures S25 and S26). This result confirmed that the cleaved coumarin product and activated pH sensor were present in same regions of lowered cytosolic pH indicating the CTSB-mediated cleavage was occurring outside lysosomes in the cytoplasm when pH was lowered, possibly due to LMP. Hence, using MSS2 and MSS3, we were able to track the pH-dependent activity of CTSB in a spatiotemporal manner, providing insights into the influence of pH on translocation of the enzyme from lysosomes to the cytoplasm.

Further, to check if ‘stitching’ sensors indeed allowed simultaneous uptake of monoanalyte probes as hypothesized in our design of morphable ‘stitched’ sensors, the cellular uptake of the sensors (MSS2 and MSS3) was tracked from 0 to 6 min on both channels (Figure e,f). The fluorescence intensities in both channels increased and saturated simultaneously, indicating synchronized uptake (Figure e,f).

Similar constructs can be developed by merely changing the peptide substrates of CTSB, while retaining the rest of the design. This can be easily and efficiently achieved due to the modularity of our concept. Our approach was effective in elucidating the pH-dependent substrate specificity of CTSB and can contribute to the development of diagnostic tools for disorders linked to CTSB dysregulation. Moreover, it may enable screening of targeted therapeutics that selectively inhibit the pathophysiological activity of CTSB in the cytosol, while maintaining its normal enzymatic function within the lysosome. This would be crucial for treating disorders related to CTSB dysregulation. Taken together, MSS1, MSS2, and MSS3 highlight that the morphable ‘stitched’ sensors strategy can be effectively used to track multiple bioanalytes in living cells in a made-to-order fashion by selecting single-analyte sensing units from the same library.

‘Stitched’ Sensors for Simultaneous Tracking of Cathepsin B Activity and H₂O₂

As the final proof-of-concept to test our morphable ‘stitched’ sensors strategy, we decided to simultaneously monitor H₂O₂ levels and CTSB activity. pH is not the sole factor influencing CTSB activity. ROS such as H₂O₂, participate in the Fenton reaction with lysosomal Fe²⁺ ions, producing hydroxyl radicals that induce lipid peroxidation in the lysosomal membrane. This leads to LMP and the subsequent leakage of CTSB into the cytosol. , To obtain insights into the process of ROS-mediated CTSB translocation, an attractive approach would be to simultaneously monitor both H₂O₂ levels and CTSB activity in lysosomes as well as cytoplasm. Thus, the morphable ‘stitched’ sensors strategy was next applied to develop two multianalyte sensors for this purpose, MSS4 and MSS5 (Figure ).

The CTSB-sensitive blocks used in MSS2 and MSS3 were reused in MSS4 and MSS5, respectively, again highlighting the modularity of the concept. The N-terminals of the CTSB-sensitive blocks contained cysteine (Scheme S7). Therefore, the C-terminal of the H₂O₂- sensitive block should have a hydrazine modification for conversion into a thioester, enabling subsequent NCL (Scheme S10). The peptide sequence (K-A-L-A) of the pH-sensitive block in MSS1, MSS2, and MSS3 were used for developing the H₂O₂-sensitive blocks for MSS4 and MSS5 (Scheme S10). Both ATC and naphthalimide share the same excitation wavelength. Therefore, fluorescein modified with boronic acid , was substituted for naphthalimide as the H₂O₂-sensitive unit (Scheme S10). The fluorescein-based H₂O₂-sensitive dye was attached to lysine and the lysine–dye conjugate (8) was attached to the peptide 1 (K-A-L-A) with hydrazine modified C-terminus to form the new H₂O₂ block E. The H₂O₂ block E was attached to each CTSB block C and D, respectively via NCL to complete the synthesis of MSS4 and MSS5 (Scheme S11). Both the sensors were purified using HPLC and characterized using LC/ESI-MS (Figures S27 and S28).

MSS4 and MSS5 were completely water-soluble in the tested concentrations of up to 1 mM. Hence, all experiments were performed in aqueous buffer. The absorption spectra of MSS4 and MSS5 showed a characteristic peak at 510 nm and a broad peak near 350 nm corresponding to fluorescein–boronic acid and ATC dyes, respectively (Figure S29). Upon excitation at 488 nm for the H₂O₂ sensing unit, MSS4 and MSS5 showed a peak with a maximum at 515 nm, which exhibited a 3-fold increase in emission intensity in case of MSS4 (Figure S30a) and a 2-fold increase in emission intensity in case of MSS5 in the presence of H₂O₂ (140 μM) (Figure S30b). When excited at 380 nm, both MSS4 and MSS5 displayed a low-intensity peak at 491 nm, corresponding to the quenched ATC unit of the CTSB block (Figure S31). The peaks at 491 nm did not show any response in the presence of H₂O₂ indicating that the CTSB blocks were not affected by altered H₂O₂ levels (Figure S32). In MSS4, the CTSB substrate functions at acidic pH. Hence, it was necessary to check that the basal fluorescence of MSS4 remained unaffected at low pH. At low pH (4) upon excitation at both 380 and 488 nm, respectively, there was no change in the fluorescence of MSS4 (Figure S33). The expected response of MSS4 and MSS5 in the presence of analytes is summarized in Scheme S12.

Next, the in-cell responses of MSS4 and MSS5 were investigated using 405 and 488 nm lasers as excitation sources for the ATC-based and fluorescein-based dyes, respectively. An MTT assay for cellular toxicity in HeLa cells indicated that the sensors were not toxic up to a tested concentration of 20 μM, until the maximum tested incubation time of 30 min (Figure S34). Hence, concentrations lower than 20 μM and incubation times less than 30 min were used for all cell studies. Both MSS4 and MSS5 were able to enter live cells within 5 min of direct incubation (Figures and ). The percentage of labeled cells was calculated from fluorescence confocal images of both the sensors taken with a wide field of view (Figure S35). In multiple cell plates (n = 45–90 total cells), the percentages of cells labeled by MSS4 and MSS5 were 100%.

9.

(a) Representative confocal single z plane images of HeLa cells incubated with MSS4 (10 μM) for 5 min, washed, and imaged (first column). Following that, H₂O₂ (100 μM) was added, and cells were imaged after 30 min (right column). [First row: λ_ex = 488 nm, λ_em = 520 to 600 nm (H₂O₂ channel); second row: λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel)]. Scale bar: 20 μm. (b) Representative confocal single z plane images of HeLa cells incubated with MSS5 (10 μM) for 5 min, washed, and imaged (first column). Following that, H₂O₂ (100 μM) was added, and cells were imaged after 30 min (right column). [First row: λ_ex = 488 nm, λ_em = 520 to 600 nm (H₂O₂ channel); second row: λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel)]. Scale bar: 20 μm. (c) Bar plot representing variation of the in-cell fluorescence of the H₂O₂-sensitive unit before and after addition of H₂O₂. Data are presented as SEM, where N = 3, n = 30 in each set. (d) Bar plot representing variation of the in-cell fluorescence of the CTSB-sensitive unit before and after addition of H₂O₂. Data are presented as SEM, where N = 3, n = 30 in each set. *p < 0.05, **p < 0.01, and ***p < 0.001. (e) Uptake of MSS4 visualized through intensity measurements in both the blue and green channels over time. Inset: ratio of blue to green channel intensities during the uptake process. λ_ex = 488 nm, λ_em = 520 to 600 nm (H₂O₂ channel); λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel). (f) Uptake of MSS5 visualized through intensity measurements in both the blue and green channels over time. Inset: ratio of blue to green channel intensities during the uptake process. λ_ex = 488 nm, λ_em = 520 to 600 nm (H₂O₂ channel); λ_ex = 405 nm, λ_em = 420 to 500 nm (CTSB channel).

In resting HeLa cells, a few punctate like regions were observed in both the channels in case of MSS4 (Figure b). In case of MSS5, the green channel (λ_ex = 488 nm; H₂O₂ channel) displayed some punctate regions with high intensity similar to MSS4 (Figure b,c). The blue channel (λ_ex = 405 nm; CTSB channel) for MSS5 showed significantly higher fluorescence intensity (Figures c and S37) compared to MSS4. The blue channel fluorescence in case of MSS5 was also more diffuse and spatially distributed (Figure S36c,d). The in-cell blue channel (λ_ex = 405 nm; CTSB channel) fluorescence intensities of MSS4 and MSS5 at the resting state were similar to that of MSS2 and MSS3, respectively (Figure S37). Hence, the H₂O₂ block did not affect the function of the CTSB sensing blocks. We noted an overlap between the emission spectra of the ATC and fluorescein-based units (Figures S30 and S31). To address this, the sensors were excited at 405 nm and imaged in the green channel, where some fluorescence was detected due to bleed-through from the blue to green channel (Figure S38). Hence, a bleed-through correction was incorporated in the image analysis (Supporting Information Section 6).

The low intensity in the cytosol and punctate high-intensity regions in the green channel (Figure b,c) may reflect low cytosolic H₂O₂ levels and higher H₂O₂ concentrations in the lysosomes, consistent with known lysosomal H₂O₂ accumulation. To verify that high intensity regions in the green channels of both MSS4 and MSS5 and in the blue channel of MSS4 were from lysosomes, colocalization studies were performed with Lysotracker Red (λ_ex = 543 nm) (Figure b,c). In both the cases, the high intensity regions in the green channel (H₂O₂ channel) colocalized well with Lysotracker Red (Pearson’s coefficient = 0.86 (MSS4) or 0.84 (MSS5)) (Figure b,c). This was consistent with earlier reports indicating presence of elevated levels of H₂O₂ in lysosomes. In case of MSS4, colocalization was not only observed in the green channel (H₂O₂ channel) but also in the blue channel (CTSB channel) (Pearson’s coefficient = 0.83) with the Lysotracker channel (Figure b). MSS4 preferentially tracks lysosomal CTSB activity as it contains the peptide substrate that is cleaved at acidic pH, , as evidenced by strong colocalization with Lysotracker Red. In contrast, the CTSB channel of MSS5 showed reduced colocalization with Lysotracker Red (Pearson’s coefficient = 0.59) (Figure c), indicating that the substrate that it contained was cleaved at both acidic and neutral pH, , allowing the probe to track CTSB activity in both lysosomes and cytosol.

Upon adding H₂O₂ to live cells, both MSS4 and MSS5 showed an increase in fluorescence intensity in the green channel (H₂O₂ channel) (Figure a–c), suggesting elevated intracellular H₂O₂ levels. The fluorescence in the green channel also became more diffuse, indicating an increase in cytosolic H₂O₂ (Figure a,b). No rise in fluorescence intensity in the blue channel (CTSB channel) was noted in case of MSS4 (Figures a,d and S36c). This is because even if H₂O₂ had triggered translocation of CTSB to cytosol, MSS4 could not track its activity at neutral cytosolic pH. Therefore, no increase in CTSB channel intensity was observed. However, upon decreasing the cytosolic pH, a significant increase in blue channel intensity was observed (Figures S39 and S36c), as the lowered pH allowed MSS4 to detect and track CTSB activity in the cytosol. The observed increase in intensity in the CTSB channel in this case was greater than that seen with MSS2 when the pH was lowered without the addition of exogenous H₂O₂ (Figure S40). This suggested that excess cellular H₂O₂ may have triggered a more pronounced translocation of CTSB to the cytoplasm compared to low pH conditions.

MSS5 can monitor CTSB activity at neutral cytosolic pH and hence was aptly placed to specifically examine the effect of elevated H₂O₂ on CTSB translocation. Upon addition of H₂O₂ to cells incubated with MSS5, we could observe a significantly enhanced blue fluorescence in the cytosol (CTSB channel) (Figures b,d and S36d). This indicated that elevated H₂O₂ levels indeed triggered the translocation of CTSB to the cytoplasm, which was tracked by MSS5. Moreover, the increase in blue channel fluorescence upon H₂O₂ addition was greater than that observed when intracellular pH was lowered in the case of MSS3 (Figure S41). This further suggested that H₂O₂ may serve as a more potent inducer of LMP, thereby facilitating the translocation of CTSB from lysosomes to the cytoplasm. These findings validate existing literature that implicate the role of ROS in LMP and the subsequent translocation of CTSB. ,

Time-lapse images acquired in blue (CTSB) and green (H₂O₂) channel showed that upon exogeneous H₂O₂ treatment the green channel fluorescence intensity increased with time (0–30 min) for both MSS4 and MSS5 but there was no simultaneous increase in blue channel intensity in case of MSS4 during the same period of time (Videos S4 and S5). On the other hand, in case of MSS5, there was a concurrent increase in blue channel intensity (0–27 min) along with the increase in intensity of the green channel (0–27 min) (Video S5). However, upon changing the media pH to 4.5, in case of MSS4, the blue channel intensity increased significantly (30–59 min) with no concomitant change in green channel intensity (30–59 min) (Videos S4). Thus, MSS4 and MSS5 were able to track the H₂O₂-dependent CTSB activity in a spatiotemporal manner via simultaneous detection, elucidating ROS-mediated translocation of the enzyme.

Given lysosomal colocalization of the probes, the uptake pathway of a representative probe MSS5 was further examined by incubating cells at 4 °C (Figure S45). The reduction in uptake under these conditions indicated an ATP-dependent mechanism, likely endocytosis, although passive diffusion cannot be completely ruled out. Notably, despite endocytic entry, the probes were not confined to endosomes or lysosomes. Upon H₂O₂ or acid exposure, fluorescence increased diffusely throughout the cytoplasm instead of being restricted in punctate vesicles. This demonstrated probe release into the cytosol, confirming their functional accessibility to intracellular analytes.

To further investigate the selectivity of the probes toward monitoring the activity of CTSB, cells were incubated with a CTSB inhibitor followed by the respective probes (Figures S42 and S43). The inhibitor did not affect the response of the H₂O₂-sensitive units to elevated levels of H₂O₂ notably, but a few altered responses were observed in some cells with no significant change in the mean intensity values (Figures S42 and S43). The observed shifts may be attributable to the inhibitor perturbing the physiology of certain cells within the population. However, the CTSB (blue) channel intensities in case of both MSS4 and MSS5 were significantly reduced in inhibitor-treated cells at resting state (Figures S42 and S43). Moreover, addition of H₂O₂ (both in case of MSS4 and MSS5) or change in media pH (in case of MSS4) did not trigger any increase in fluorescence in the CTSB channel in inhibitor-treated cells (Figures S42 and S43). Therefore, these results confirmed that the probes selectively track the in-cell activity of CTSB. Importantly, MSS4 and MSS5 were able to track the H₂O₂-dependent activity of CTSB via molecular imaging for the first time and highlighted the relevance of multianalyte imaging in elucidating biological processes.

Finally, it was necessary to check if ‘stitching’ sensors indeed allowed simultaneous uptake of monoanalyte probes as hypothesized in our design of morphable ‘stitched’ sensors. Hence, the cellular uptake of the sensors (MSS4 and MSS5) was tracked from 0 to 6 min on both channels (Figure e,f). While individual sensors may exhibit different uptake dynamics, upon ‘stitching’, the fluorescence intensities in both channels increased and saturated simultaneously, indicating synchronized uptake (Figure e,f). Moreover, the ratio of intensities remained constant throughout the uptake period (Figure e,f insets), demonstrating that the sensing units in the ‘stitched’ sensors were indeed internalized in equal proportions.

Also, fluorescence intensities from both channels representing emissions from H₂O₂ and CTSB sensors, were collected across multiple cells, and the ratio of these intensities was calculated. The ratio remained consistently similar across the entire cell population (Figure S44). This indicates that ‘stitching’ results in a uniform distribution of the sensors within the cells.

Conclusions and Perspectives

We have developed a novel universal strategy for generating cell-permeable multianalyte sensors for simultaneous tracking of bioanalytes in living systems. Here, we report five proof-of-concept demonstrations of this strategy for simultaneous detection of pH, H₂O₂, and CTSB activity in living mammalian cells, achieved by made-to-order ‘stitching’ of sensing blocks from a library of monoanalyte sensors. The multianalyte sensors developed were all water-soluble and directly cell-permeable. The probes successfully afforded simultaneous imaging and tracking of bioanalytes in living cells and notably provided insights into alterations in the in-cell activity of an enzyme when pH and H₂O₂ levels were altered, highlighting the need for live-cell multianalyte sensing. Since our strategy involves peptide blocks, targetability to specific intracellular locations was also accessible, as demonstrated by sensing the activity of an enzyme by varying peptide substrates that function at different pH. In the future, depending on the biological context, cell permeable peptides with organelle-targeting capabilities can also be utilized for targeting the ‘stitched’ sensors to specific compartments inside the cell.

Why are ‘stitched’ sensors advantageous compared to employing individual sensors separately? To check the unique advantage of our ‘stitching’ strategy, we studied the uptake behavior of representative ‘unstitched’ probes which clearly illustrated that neither the uptake nor the localization of the individual sensors was synchronized. One of the key benefits of the ‘stitching’ concept, is that ‘stitched’ sensors ensure the same subcellular localization for all sensing units, as demonstrated by FRET analysis for MSS1 and colocalization with lysotrackers for MSS2, MSS3, MSS4, and MSS5. This guarantees that multiple analytes are measured precisely within the same cellular compartments. Cellular uptake studies showed that ‘stitched’ sensors entered cells simultaneously and exhibited synchronized uptake dynamics, unlike individual sensors which may be internalized at different rates. Moreover, the fluorescence intensity ratios between the sensing units remained consistent across the entire cell population, indicating uniform distribution and enabling reliable comparisons of sensor responses. Importantly, 'stitched’ sensors enabled simultaneous temporal tracking of multiple bioanalytes via time-lapse imaging in living cells. Together, these key features highlight that ‘stitched’ sensors provide enhanced accuracy, and spatial consistency, for simultaneous multianalyte imaging, offering a powerful approach to study complex biological processes in live cells.

The concept of morphable ‘stitched’ sensors, a first to the best of our knowledge, holds significant potential for advancing our understanding of cellular homeostasis by allowing real-time monitoring of the dynamic distribution and interplay of multiple analytes within cells. Decoding complex relationships among different bioanalytes will provide valuable insights into how cells maintain balance under normal and stressed conditions. Finally, in the future, our novel approach of ‘stitching’ functional molecular entities on peptide blocks will be applicable to live-cell chemical fingerprinting, combination therapeutics, and multianalyte-based diagnostics.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01271.

• Experimental details of synthesis, LC-ESI-MS, fluorescence experiments, and cell studies (PDF)

• Video S1 (MP4)

• Video S2 (MP4)

• Video S3 (MP4)

• Video S4 (MP4)

• Video S5 (MP4)

A.D. and S.K. designed and conceptualized the project and wrote the paper. S.K. performed the experiments and analyzed the data. M.B. participated in the conceptualization, design, and synthesis of MSS4 and MSS5 along with experiments and data analysis. S.K.D. participated in experiments and data analysis. All of the authors checked the results and the manuscript and approved the final version of the manuscript.

The authors declare no competing financial interest.