Designing novel peptide vehicles for cell-impermeable organic fluorescent probes
We first synthesized nine peptide vehicles; among them, PV-1, PV-2, PV-3, PV-4, and PV-5 were novel peptide vehicles designed by us. All of them are disulfide-bonded dimers of short peptides but contain different peptide sequences or fluorophores (Fig. 1a). TP10, GALA, Penetratin (Pene), and dfTAT were selected from previous reports because they all achieved efficient cytosolic delivery of proteins or nucleic acids by simple coincubation (fig. S1 and notes S1-2)17-20. However, whether these existing peptide vehicles can be used to deliver cell-impermeable organic fluorescent probes into live cells is still unknown because the sizes and structures of organic fluorescent probes are much different from those of protein or nucleic acids and because their delivery capability greatly depends on the chemical structures of their cargoes17-19, 21, 22. We also synthesized Tubulin-FITC, an organic fluorescent probe containing a docetaxel scaffold that selectively binds to microtubules23 and a cell-impermeable dye, FITC (fluorescein isothiocyanate) (table S1).
PV-1 efficiently delivers Tubulin-FITC into live cells.
a The chemical structures of the peptide vehicles. r, d-Arginine; -s-s-, a disulfide bond; RhB, rhodamine B; Rh6G, a rhodamine derivative (fig. S1). b A schematic of the intracellular delivery of PV-1. c Confocal microscopy images of live U-2 OS cells after a 1-h coincubation with Tubulin-FITC (5 μM) and the indicated peptide vehicles at the indicated optimal concentrations for probe delivery. Scale bar: 50 μm. d Percentages of live cells labeled by Tubulin-FITC after a 1-h coincubation with the indicated peptide vehicles (mean + s.d., n = 1000 cells from triplicate experiments). e The mean fluorescence intensity of Tubulin-FITC inside one cell after a 1-h coincubation with Tubulin-FITC (5 μM) and the indicated peptide vehicles (mean + s.d., n = 1, 000 cells from triplicate experiments), a.u.: arbitrary units. f U-2 OS relative cell viability after a 1-h incubation with the indicated peptide vehicles. The error bars represent the standard deviations of triplicate experiments. The concentrations of the peptide vehicles used in (d-f) were the same as those used in (c). g Confocal microscopy images of various live cells after a 1-h coincubation with Tubulin-FITC (5 μM) and PV-1 (4 μM). Scale bar: 10 μm
Live cells were coincubated with Tubulin-FITC and the abovementioned peptide vehicles (Fig. 1b) at different concentrations for 1 h and then imaged by spinning disk confocal microscopy. Our results showed that it was difficult for Tubulin-FITC to enter live cells on its own (Fig. 1c). After coincubation of cells with Tubulin-FITC and GALA, Pene, PV-3, or PV-5, almost no fluorescence from Tubulin-FITC was observed inside cells (Fig. 1c and fig. S2). After coincubation of cells with Tubulin-FITC and TP10 or dfTAT, a weak fluorescence signal from Tubulin-FITC was detected, but not enough to produce clear images (Fig. 1c and fig. S2). Increasing concentrations of these peptide vehicles had little effect on their delivery efficiency (fig. S2), and obvious cytotoxicity was observed when higher concentrations were used (data not shown). By contrast, following coincubation of cells with PV-1 and Tubulin-FITC, approximately 82% of cells were labeled by Tubulin-FITC, and a strong fluorescence intensity of Tubulin-FITC inside cells and specific labeling of microtubules were clearly observed, suggesting an excellent delivery efficacy of PV-1 for Tubulin-FITC (Fig. 1c, d, e). Different peptide sequences, such as TP10, GALA, Pene, PV-1, and PV-4, resulted in significant differences in deliver efficacy (Fig. 1c, d, e), suggesting that the chemical structure of the peptide clearly affects the delivery efficiency, which is consistent with previous reports21. Compared with dfTAT, PV-1 contains a different fluorophore, rhodamine B (RhB), which is covalently conjugated to the peptide segment via the ortho-carboxyl group, instead of tetramethylrhodamine (TMR) via the meta-carboxyl group in dfTAT, resulting in a 6-fold increase in the labeling percentage and a threefold increase in the mean fluorescence intensity (Fig. 1d, e). For PV-1 and PV-2, which contain the same peptide sequence, RhB and Rh6G, respectively are conjugated to the peptide segment via the ortho-carboxyl group, which can promote the formation of the lactone isomer24-26. The delivery efficacy of PV-1 was much better than that of PV-2 (Fig. 1d, e). These results suggest that the chemical structure of the fluorophore in the vehicle clearly affects the delivery efficiency, whereas in the previous report17, 27, TMR was only used as a marker for fluorescence imaging. PV-1 and PV-4 have the same peptide sequences as PV-3 and PV-5, respectively, but contain additional RhB. The delivery efficacy of PV-1 and PV-4 were much better than that of PV-3, PV-5, or a mixture of RhB and PV-3, suggesting that the covalent conjugation of RhB plays an essential role in the delivery function of these peptide vehicles (Fig. 1c-e, and figs. S2-3). We also incubated RhB or TMR with Tubulin-FITC. In both cases, no obvious fluorescence of Tubulin-FITC was found inside cells, indicating that the peptide segment is important for the delivery function (fig. S3). It is worth noting that after conjugation with the peptide segment via the ortho-carboxyl group, the fluorescence intensity of RhB in PV-1 decreased dramatically, and almost no clear fluorescence signal of RhB was detected inside the cells (fig. S3, EX: 561 nm) due to the formation of a nonfluorescent "closed" lactone isomer24-26, thus minimizing the fluorescence interference from the vehicle to the benefit of multicolor imaging applications. By contrast, dfTAT, when excited by a 561 nm laser, emitted a strong fluorescence signal assigned to TMR after it entered live cells (fig. S3).
Then, we measured cell viability to test the cytotoxicity of these peptide vehicles. Among these peptide vehicles, TP10, Pene, and PV-2 had relatively high cytotoxicity (Fig. 1f). The viability of cells treated with PV-1 at the optimal concentration for probe delivery exceeded 85%, highlighting its low cytotoxicity (Fig. 1f). Therefore, in terms of the delivery efficacy, cytotoxicity and fluorescence background, PV-1 was the optimal candidate among all of the peptide vehicles tested for delivering cell-impermeable organic probes. In addition, PV-1 maintained its excellent delivery capability in various cell lines, including Cos-7, ARPE-19, HeLa, and U-2 OS (Fig. 1g).
To explore the cellular uptake mechanism involved in the delivery of PV-1, we first decreased the incubation temperature to 4 ℃, a condition that typically inhibits endocytosis21. After a 1-h coincubation with PV-1 and Tubulin-FITC at 4 ℃, very weak fluorescence was detected inside live cells, suggesting that endocytosis plays an essential role in the delivery of PV-1 (fig. S4). We then used three inhibitors (filipin, chlorpromazine (CPZ), and 5-(N-ethyl-N-isopropyl) amiloride (EIPA)) of endocytosis to block individual endocytic pathways21, 28. The results showed that filipin (an inhibitor of caveolae-mediated endocytosis) and CPZ (an inhibitor of clathrin-mediated endocytosis) had little effect on the delivery of PV-1, and EIPA (an inhibitor of macropinocytosis) significantly inhibited the delivery of PV-1, suggesting that micropinocytosis plays an essential role in the efficient delivery of PV-1 (fig. S4). Interestingly, after coincubation with PV-1, EGFP (enhanced green fluorescent protein) was also delivered into live cells, implying that PV-1-mediated transport is not selective for proteins and small-molecule compounds (fig. S5). The presence of FBS (fetal bovine serum) in the cell incubation medium significantly reduced the delivery efficiency of PV-1 (fig. S6), probably because positively charged PV-1 binds to negatively charged FBS17, 29.
PV-1 delivers organic probes containing different fluorophores and recognition units
Next, we investigated whether PV-1 could deliver cell-impermeable organic probes other than Tubulin-FITC containing different organic fluorophores or recognition units into live cells to achieve specific labeling by simple coincubation. A series of cell-impermeable fluorescent probes (15 probes in total) were synthesized and tested: six probes for microtubules (i.e., Tubulin-Atto 488, Tubulin-Atto 514, Tubulin-Atto 565, Tubulin-Cy3B, Tubulin-Alexa 488, and Tubulin-Alexa 647), four probes for nuclei (i.e., Hoechst-Alexa 488, Hoechst-Alexa 647, Hoechst-Cy3B, and Hoechst-Atto 514), and five probes for lysosomes (i.e., Morph-Alexa 488, Morph-Alexa 647, Morph-Cy3B, Morph-Atto 514, and Morph-Atto 488) (table S1 and notes S3-7). Our results showed that the cellular uptake of all of these probes on their own was poor (fig. S7). However, after co-incubation with PV-1, each of these probes was efficiently delivered into live cells and specifically labeled the respective organelles (Fig. 2a-c, figs. S8-10). There were some off-target punctate signals detected, as shown in Fig. 2a. Our colocalization studies with Tubulin-Alexa 647 (fig. S11) revealed that these punctate vesicles colocalized with LysoTracker Green, a probe that labels lysosomes, but not with Rab5b-GFP, a marker protein of early endosomes30, implying that only a relatively small amount of the probes was trapped inside lysosomes, while most reached the cytosol and labeled the target organelles. The PV-1 delivery method is limited in labeling punctate structures except lysosomes unless off-target punctate signals can be excluded by colocalization with lysosomal trackers. In sum, PV-1 can deliver cell-impermeable organic probes containing different fluorophores and recognition units into live cells, implying a wide range of promising applications of PV-1 in live-cell studies.
PV-1 efficiently delivers diverse cell-impermeable organic fluorescent probes into live cells.
a Confocal microscopy images of live U-2 OS cells after a 1-h coincubation with PV-1 (4 μM) and the indicated probes for microtubules with different colors (5 μM). Scale bars: 10 μm. b Confocal microscopy images of live U-2 OS cells after a 1-h coincubation with PV-1 (4 μM) and Hoechst-Alexa 488 or Hoechst-Alexa 647 (3 μM). Scale bars: 20 μm. c Confocal microscopy images of live U-2 OS cells after a 1-h coincubation with PV-1 (4 μM) and Morph-Atto 488 or Morph-Alexa 647 (5 μM). Scale bars: 5 μm. The labeled organelles are indicated in the bottom left of the first images. d Confocal microscopy images of live U-2 OS cells expressing SNAP-Sec61β (top row, four first images), CLIP-Sec61β (top row, last image), SNAP-Actin, SNAP-Cox8a, SNAP-H2B, CLIP-Lifeact or CLIP-Actin after a 1-h coincubation with PV-1 (4 μM) and the indicated SNAP or CLIP probes (5 μM). Scale bars: 5 μm
PV-1 delivers commercial cell-impermeable SNAP/CLIP probes
SNAP-tag/CLIP-tag labeling technologies provide simple, robust, and versatile approaches for specific labeling of fusion proteins with organic fluorescent probes; therefore, they are powerful tools for imaging proteins and can be used in a wide range of experimental applications31-34. A series of SNAP probes (i.e., SNAP-TMR, SNAP-SiR647, SNAP-Atto 488, SNAP-Alexa 488, SNAP-Dy 549, and SNAP-Alexa 647) and CLIP probes (i.e., CLIP-TMR, CLIP-Atto 488, and CLIP-Dy 547) bearing fluorophores with excellent optical properties have been commercialized (table S1); however, most of them (except SNAP-TMR, SNAP-SiR647, and CLIP-TMR) are cell-impermeable and thus are primarily restricted to labeling proteins on the surface of live cells. SNAP-TMR, SNAP-SiR647, and CLIP-TMR are cell-permeable but they have similar excitation wavelengths, which limits their applications in multicolor imaging.
To address this issue, we investigated whether PV-1 could be used to deliver these commercial cell-impermeable SNAP/CLIP probes into live cells. U-2 OS cells expressing SNAP-Sec61β decorating the endoplasmic reticulum (ER) were incubated with SNAP probes in the absence or presence of PV-1 for 1 h. Remarkably, after coincubation with PV-1, all of the SNAP probes were efficiently translocated into live cells, and specific labeling of the ER was clearly observed (Fig. 2d). By contrast, no obvious fluorescence was detected inside live cells for any of the SNAP and CLIP probes in the absence of PV-1 (fig. S7). Using PV-1 and the SNAP or CLIP probes, we also successfully labeled F-actin, mitochondria, and nuclei in live cells expressing SNAP-Actin, SNAP-Lifeact, CLIP-Actin, CLIP-Lifeact, SNAP-Cox8a, or SNAP-H2B (Fig. 2d and fig. S12). To investigate the labeling specificity, we performed colocalization studies using the anti-actin antibodies Alexa 568, Hoechst 33342, BFP-KDEL8, 35, and MitoTracker Deep Red as the standard markers for actin, nuclei, the ER, and mitochondria, respectively. The results showed that these SNAP and CLIP probes can label the desired targets with high specificity (fig. S13). Considering the universality of SNAP-tag/CLIP-tag technologies for protein labeling, efficient delivery of these commercial cell-impermeable SNAP/CLIP probes by PV-1 into live cells will greatly benefit the imaging of intracellular proteins in live cells for a wide range of applications.
PV-1 simultaneously delivers up to three cell-impermeable organic probes
Multicolor fluorescence imaging has greatly advanced our understanding of the interactions among biomolecules in live cells. To determine whether PV-1 could be used to deliver more than one organic probe into live cells simultaneously for multicolor imaging applications, we incubated PV-1 with Tubulin-FITC and Hoechst-Alexa 647 at the same time. Notably, both Tubulin-FITC and Hoechst-Alexa 647 were efficiently delivered into live cells, and microtubules and nuclei were simultaneously visualized in the same live cells (Fig. 3a). Under normal circumstances, more than 90% of live cells are in interphase; therefore, most of the labeled cells are interphase cells (Fig. 3a), but cells in other stages of the cell cycle (e.g., prophase, metaphase, anaphase, and telophase) can also be labeled and observed (fig. S14). We also successfully labeled microtubules and nuclei, the ER and nuclei, or microtubules and the ER simultaneously by delivering two corresponding probes with PV-1 into live cells (Fig. 3b and fig. S15) Most notably, we simultaneously stained microtubules, nuclei, and the ER in the same live cells by concurrent coincubation of Tubulin-Atto 565, Hoechst-Alexa 647, and SNAP-Surface Atto 488 with PV-1, and we obtained triple-color live-cell SIM superresolution images of these three organelles (Fig. 3c). These results demonstrate that PV-1 can be used to simultaneously deliver two or more organic probes into live cells for multicolor imaging applications.
PV-1 simultaneously delivers up to three cell-impermeable organic fluorescent probes into live cells.
a Confocal microscopy images of live U-2 OS cells after a 1-h coincubation with PV-1 (4 μM), Tubulin-FITC (5 μM), and Hoechst-Alexa 647 (3 μM). Scale bar: 50 μm. b Confocal microscopy images of live U-2 OS cells wild-type (top) or expressing CLIP-Sec61β (bottom left) or SNAP-Sec61β (bottom right) after a 1-h coincubation with PV-1 (4 μM) and the two indicated probes (Hoechst-Alexa 647: 3 μM; the others: 5 μM). Scale bars: 10 μm. c A triple-color SIM image of live U-2 OS cells expressing SNAP-Sec61β after a 1-h coincubation with PV-1 (4 μM), Tubulin-Atto 565 (5 μM), SNAP-Atto 488 (5 μM), and Hoechst-Alexa 647 (3 μM). Scale bar: 2 μm. The labeled organelles are indicated in the panels
Live-cell SIM imaging reveals the dynamic interactions between organelles
PV-1, together with these organic fluorescent probes (table S1), provided more live-cell compatible fluorescent probes for SIM. By using PV-1 and these cell-impermeable fluorescent probes, we obtained SIM images of microtubules, the ER, and nuclei in live cells (Fig. 4a-d and fig. S16). Notably, microtubule and the ER dynamics were continuously recorded in live cells with SIM without obvious photobleaching for up to ~4 or 6 min (Fig. 4b, d, and movie S1-4). Moreover, we were able to record the dynamic interactions between lysosomes and microtubules (Fig. 3e-g and movie S5-7) and between lysosomes and the ER (Fig. 4h-k and movie S8-11) in live cells over a time course of ~3 min with dual-color SIM. Our imaging captured the trafficking of a lysosome along the microtubules (Fig. 4f and movie S6) and a lysosome trapped within the microtubule network (Fig. 4g and movie S7). Most interestingly, these consecutive SIM superresolution images allowed us to identify three types of interactions between lysosomes and the ER: (1) lysosomes were tightly surrounded and trapped within the ER network (Fig. 4i and movie S9); (2) ER tubules captured a lysosome by gradually narrowing the tubule ring (Fig. 4j and movie S10); and (3) a lysosome was bound to the end of an ER tubule and moved together with the tubule (Fig. 4k and movie S11).
Live-cell SIM imaging reveals the dynamics of organelles.
a SIM images of live U-2 OS cells after a 1-h coincubation with PV-1 (4 μM) and Tubulin-Atto 488, Tubulin-Atto 565, or Tubulin-Alexa 647 (5 μM). Additional frames are shown in movie S1. b Enlarged time-lapse images of the boxed region in (a). Representative images of consecutive SIM frames are displayed, and additional frames are shown in movie S2. c The cross-sectional profile of microtubules in (b) with full-width at half-maximum (FWHM) of 108 nm. d An SIM image of live U-2 OS cells expressing SNAP-Sec61β after a 1-h coincubation with PV-1 (4 μM) and SNAP-Alexa 488 (5 μM), and enlarged time-lapse images of the boxed region. Representative images of consecutive SIM frames are displayed, and additional frames are shown in movies S3 and S4. e A dual-color SIM image of live U-2 OS cells after a 1-h coincubation with PV-1 (4 μM), Tubulin-Atto 488 (5 μM), and Morph-Alexa 647 (3 μM). Additional frames are shown in movie S5. f, g Enlarged time-lapse images of the boxed regions in (e). Representative images of consecutive SIM frames are displayed, and additional frames are shown in movies S6 and S7. h A dual-color SIM image of live U-2 OS cells expressing SNAP-Sec61β after a 1-h coincubation with PV-1 (4 μM), SNAP-Alexa 488 (5 μM), and Morph-Alexa 647 (3 μM). Additional frames are shown in movie S8. i Enlarged time-lapse images of the boxed region in (h) reveal dynamic interactions between lysosomes and the ER. Representative images of consecutive SIM frames are displayed, and additional frames are shown in movie S9. j, k Representative time-lapse SIM images reveal different types of dynamic interactions between lysosomes and the ER in live U-2 OS cells expressing SNAP-Sec61β after a 1-h coincubation with PV-1 (4 μM), SNAP-Alexa 488 (5 μM), and Morph-Alexa 647 (3 μM). Additional frames are shown in movies S10 and S11. Scale bars: a, e, and h, 2 μm; b, d, f, g, i, j and k, 1 μm