Rendering hydrophobic nanoclusters water-soluble and biocompatible

Hydrophobic and hydrophilic nanoclusters embody complementary superiorities. The means to amalgamate these superiorities, i.e., the atomic precision of hydrophobic clusters and the water dissolvability of hydrophilic clusters, remains challenging. This work presents a versatile strategy to render hydrophobic nanoclusters water-soluble—the micellization of nanoclusters in the presence of solvent-conjoined Na+ cations—which overcomes the above major challenge. Specifically, although [Ag29(SSR)12(PPh3)4]3− nanoclusters are absolutely hydrophobic, they show good dissolvability in aqueous solution in the presence of solvent-conjoined Na+ cations (Na1(NMP)5 or Na3(DMF)12). Such cations act as both counterions of these nanoclusters and surface cosolvent of cluster-based micelles in the aqueous phase. A combination of DLS (dynamic light scattering) and aberration-corrected HAADF-STEM (high angle annular dark field detector scanning transmission electron microscopy) measurements unambiguously shows that the phase-transfer of hydrophobic Ag29 into water is triggered by the micellization of nanoclusters. Owing to the excellent water solubility and stability of [Ag29(SSR)12(PPh3)4]3−[Na1(NMP)5]3+ in H2O, its performance in cell staining has been evaluated. Furthermore, the general applicability of the micellization strategy has been verified. Overall, this work presents a convenient and efficient approach for the preparation of cluster-based, biocompatible nanomaterials.

(ii) On the other hand, although several hydrophilic nanoclusters are not that precise in terms of structures and compositions (even some of them are poly-dispersed), they are much more suitable for the aforementioned bio-applications owing to their water dissolvability. [43][44][45][46][47][48][49][50][51] In addition, hydrophilic nanoclusters benet from ultra-small sizes, non-toxicity, strong photo-stability, good biocompatibility, and potential anti-cancer activity, and their use in biological applications is thus highly promising. [43][44][45][46][47][48][49][50][51] However, drawbacks of known hydrophilic nanoclusters also exist-their properties are even harder to precisely control than in hydrophobic nanoclusters, and their imprecise structures preclude quantitative tracking in cells or organisms. These drawbacks greatly impede the practical uses of hydrophilic nanoclusters in bio-applications.
Indeed, hydrophobic and hydrophilic nanoclusters embody complementary superiorities. Accordingly, the means to transfer atomically precise, hydrophobic nanoclusters into water, and then exploit them for aqueous-phase applications, is anticipated to overcome the aforementioned drawbacks, and should be an important goal in nanocluster science. Herein, we report a versatile strategy to render hydrophobic nanoclusters water-soluble-the micellization of these nanoclusters in the presence of solvent-conjoined Na + cations. Specically, the [Ag 29 (SSR) 12 (PPh 3 ) 4 ] 3À (SSR ¼ 1,3-benzene dithiol) nanocluster is absolutely hydrophobic; however, in the presence of solventconjoined cations (Na 1 (NMP) 5 or Na 3 (DMF) 12 ) as counterions, Ag 29 @Na compounds ([Ag 29 (SSR) 12 12 ] 3+ , Ag 29 -Na 1 or Ag 29 -Na 3 hereaer because they contain three Na + monomers or one [Na 3 ] 3+ trimer, respectively) show good dissolvability in aqueous solution. A combination of DLS (dynamic light scattering) and aberration-corrected HAADF-STEM (high angle annular dark eld detector scanning transmission electron microscopy) measurements unambiguously shows that the phase-transfer of hydrophobic Ag 29 into water is triggered by the micellization of nanoclusters. Owing to the excellent water dissolvability and stability of Ag 29 -Na 1 , its performance in cell staining is evaluated, and such cluster-based micelles show specic selectivity in staining lysosomes. Furthermore, the general applicability of this micellization method in the nanocluster eld has been veried, based on several other negative-charged nanoclusters, which further demonstrates the signicance of this method in the preparation of cluster-based, biocompatible nanomaterials.
Crystallization of Ag 29 -Na 1 and Ag 29 -Na 3 Single crystals of Ag 29 -Na 1 or Ag 29 -Na 3 were cultivated at room temperature by vapor-diffusing ethyl ether into the NMP solution of Ag 29 -Na 1 or the DMF solution of Ag 29 -Na 3 . Aer 2 weeks, red crystals were collected, and the structure of Ag 29 -Na 1 or Ag 29 -Na 3 was determined.  Micellization of Ag 29 -Na 1 and Ag 29 -Na 3 Specically, 3 mg of Ag 29 -Na 1 or Ag 29 -Na 3 was dissolved in 3 mL H 2 O. In this process, Ag 29 -Na 1 and Ag 29 -Na 3 clusterbased micelles were formed in their aqueous solutions. For measuring the sizes of Ag 29 -Na 1 or Ag 29 -Na 3 micelles in different stages (Fig. 3), 100*n mL of the aforementioned aqueous solution of Ag 29 -based micelles was injected into (3-0.1*n) mL H 2 O each time (n is the number of the stage), and the obtained H 2 O solution was directly used for the DLS analysis.

The general nanocluster micellization method
Specically, 30 mg of each negative-charged nanocluster was mixed with 1 mg of CH 3 COONa and 30 mL DMF, and the obtained mixture was dissolved in 1 mL H 2 O. The precipitate was then removed to produce the cluster-based micelle in H 2 O. Notably, DLS measurements in Fig. 6 were performed in saturated aqueous solutions of these nanoclusters.  24 ] 4À ) were based on the reported methods. [54][55][56][57][58] The solubility of different nanoclusters in aqueous solution The nanocluster@solvent-conjoined cation compound was dissolved in H 2 O, and the solubility of each nanocluster in the aqueous solution was determined. Specically, the solubility of Ag 29 -Na 1 in H 2 O was 6.76 mg mL À1 ; the solubility of Ag 29 -Na 3 in H 2 O was 7.88 mg mL À1 ; the solubility of Au 25 (SC 2 H 4 Ph) 18

Characterization
All UV-vis absorption spectra of nanoclusters were recorded using an Agilent 8453 diode array spectrometer, whose background correction was made using a pure solution blank.
PL spectra were measured on a FL-4500 spectrouorometer with the same optical density of 0.1.
Electrospray ionization mass spectrometry (ESI-MS) measurements were performed using a MicrOTOF-QIII high resolution mass spectrometer. The sample was directly infused into the chamber at 5 mL min À1 . For preparing the ESI samples, nanoclusters were dissolved in NMP/DMF (0.1 mg mL À1 ) and diluted (v/v ¼ 1 : 2) with methanol. 23 Na nuclear magnetic resonance (NMR) spectra were acquired using a Bruker 600 Avance III spectrometer equipped with a Bruker BBO multinuclear probe (BrukerBioSpin, Rheinstetten, Germany). Dynamic light scattering (DLS) was performed with a Malvern Zetasizer Nano ZS instrument. For preparing the DLS samples, the nanocluster@solvent-conjoined cation compounds were dissolved in H 2 O to produce the cluster-based micelles. The DLS result of each nanocluster was repeated 40 times to remove the error.
The Ag 29 -based micelles were imaged with an aberrationcorrected HAADF-STEM (high angle annular dark eld scanning transmission electron microscope) aer the solvent that contained Ag 29 -based micelles was drop-cast onto ultrathin carbon lm TEM grids. The microscope employed was a FEI Themis Z. The electron beam energy was 200 kV. The collection angle of the HAADF detector was adjusted to collect signals scattered between 52 (inner angle) and 200 (outer angle) mrad (camera length of 146 mm). The HAADF-STEM image was obtained with Thermo Scientic Velox soware using 1024 Â 1024 pixels and the dwell time is set to 10 us.

X-ray crystallography
For Ag 29 -Na 1 , the data collection for single crystal X-ray diffraction was carried out on a Stoe Stadivari diffractometer under nitrogen ow, using graphite-monochromatized Cu Ka radiation (l ¼ 1.54186Å). For Ag 29 -Na 3 , the data collection for the single-crystal X-ray diffraction was carried out on a Bruker Smart APEX II CCD diffractometer under nitrogen ow, using graphitemonochromatized Mo Ka radiation (l ¼ 0.71073Å). Data reductions and absorption corrections were performed using the SAINT and SADABS programs, respectively. The electron density was squeezed by Platon. The structure was solved by direct methods and rened with full-matrix least squares on F 2 using the SHELXTL soware package. All non-hydrogen atoms were rened anisotropically, and all the hydrogen atoms were set in geometrically calculated positions and rened isotropically using a riding model.

Cell culture
HepG2 (liver hepatocellular carcinoma) cells were purchased from Shanghai Bioleaf BioBiotech. Co. Ltd. Specically, the cells were incubated in Dulbecco's Modied Eagle's medium (DMEM) containing 10% FBS and 1% antibiotics (penicillin and streptomycin), and the cells were further maintained at 37 C in an atmosphere of 5% CO 2 and 95% air.

Confocal microscopic imaging
Confocal microscopy imaging was performed with a Leica TCS SP8 confocal microscope with an adjustable white laser (470-700 nm) and 63X/100X oil-immersion objective lens. The incubated cells were excited at 470 nm for Ag 29 -Na 1 , and 633 nm for Lysotracker Red. The emission signals were collected at 550 nm for Ag 29 -Na 1 , and 650-700 nm for Lysotracker Deep Red.  absolute back-ground values. The following measurement settings were used: maximum iterations: 40 times; signal-tonoise ratio: 20; quality threshold: 0.05; iteration mode: optimized; brick layout: auto.

Results and discussion
The preparations of both Ag 29 -Na 1 and Ag 29 -Na 3 nanoclusters are shown in the Experimental methods section, and the crystallization of Ag 29 -Na 1 or Ag 29 -Na 3 nanoclusters was conducted by vapor-diffusing ethyl ether into the NMP solution of Ag 29 -Na 1 or the DMF solution of Ag 29 -Na 3 . Structurally, the overall conguration of the Ag 29 (SSR) 12 (PPh 3 ) 4 framework in both Ag 29 -Na 1 and Ag 29 -Na 3 was much like that of the previously reported one; 52,53 however, there were still subtle differences among these Ag 29 frameworks in terms of the corresponding bond lengths (Table S1 †). Specically, except for the bonds between Ag(core shell) and S(motif), all bonds including Ag(core)-Ag(core shell), Ag(core shell)-Ag(core shell), Ag(core shell)-Ag(motif), Ag(motif)-S(motif), and Ag(motif)-P(motif) of Ag 29 -Na 1 and Ag 29 -Na 3 nanoclusters were much longer than those in the bare Ag 29 (-SSR) 12 (PPh 3 ) 4 nanocluster (Ag 29 -Na 0 hereaer). Accordingly, the overall Ag 29 core structures of both Ag 29 -Na 1 and Ag 29 -Na 3 were more expanding than that in Ag 29 -Na 0 .
For the counterions of these [Ag 29 (SSR) 12 (PPh 3 ) 4 ] 3À nanoclusters, the Ag 29 -Na 1 nanocluster comprised three NMPconjoined [Na 1 (NMP) 5 ] + cations per Ag 29 compound (Fig. 1A and B), whereas each Ag 29 compound matched only one DMFconjoined [Na 3 (DMF) 12 ] 3+ cation in the structure of Ag 29 -Na 3 ( Fig. 1C and D). It should be noted that no Na + cation (or other cations) has been observed in the crystal lattice of Ag 29 -Na 0 in both crystals of Ag 29 (SSR) 12 (PPh 3 ) 4 reported by the Bakr and the Pradeep groups, which might result from the high disorder of these Na + counterions. 52,53 Indeed, induced by the xation of the crown ether, Chakraborty et al. captured these Na + cations in a form of Na + @dibenzo-18-crown-6. 59 In this work, for both Ag 29 -Na 1 and Ag 29 -Na 3 nanoclusters, the Na + counterions were xed by oxygen-carrying solvents such as NMP and DMF to generate the solvent-conjoined cations.
Although the Ag 29 -Na 0 nanocluster displayed good solubility in NMP or DMF, it was absolutely hydrophobic. However, owing to the presence of the solvent-conjoined cations such as Na 1 (NMP) 5 and Na 3 (DMF) 12 , Ag 29 -Na 1 and Ag 29 -Na 3 nanoclusters were perfectly soluble in both organic reagents (NMP and DMF) and the aqueous solution ( Fig. 2A). Previous research has investigated the phase transfer of hydrophilic nanoclusters from aqueous to organic phases induced by the addition of counterions (e.g., phase transfer of the Au 22 nanocluster in the presence of tetraoctylammonium cations); 60-62 however, the study of the reverse process (i.e., phase transfer of hydrophobic nanoclusters from organic to aqueous phases) is rather limited in the nanocluster research eld. In this work, based on the Ag 29 (SSR) 12 (PPh 3 ) 4 cluster template, the transfer of hydrophobic nanoclusters from the organic phase to water has been accomplished.
ESI-MS results of both Ag 29 -Na 1 and Ag 29 -Na 3 nanoclusters exhibited ve peaks corresponding to the Ag 29 (SSR) 12 (PPh 3 ) n compounds where n was 0-4 ( Fig. 2B and S1 ‡), demonstrating the dissociation-aggregation pattern of PPh 3 ligands on the Ag 29 nanocluster surface. 63 However, the [Na 1 (NMP) 5 ] + and [Na 3 (DMF) 12 ] 3+ were undetectable in the ESI-MS, which was proposed to result from the weak interactions between Na + cations and NMP/DMP molecules causing such solventconjoined cations to decompose in mass spectroscopy. 23 Na NMR was performed to verify the presence of Na + or Na +conjoined cations in these nanoclusters. As shown in ESI Fig. S2A, † the 23 Na NMP signals of CH 3 COONa were 0.55 and 1.36 ppm in DMF-D7 and NMP-D9, respectively, whereas the signals of Ag 29 -Na 0 (in DMF-D7), Ag 29 -Na 1 (in NMP-D9) and Ag 29 -Na 3 (in DMF-D7) were À3.08, À0.88, and À1.06 ppm, respectively. Such differences also suggested the distinct existing form of each Na + -based cation in the corresponding nanocluster. Besides, the 23 Na NMP signal of CH 3 COONa in D 2 O is located at À0.12 ppm, which was remarkably different from those of Ag 29 -Na 1 or Ag 29 -Na 3 in D 2 O (À3.75 or À3.36 ppm, respectively; Fig. S2B †). In this context, throughout the micellization of Ag 29 clusters, Na + /solvent counterions would not dissociate from the nanoclusters. However, it still remained unknown whether the structures of [Na 1 (NMP) 5 ] + and [Na 3 (DMF) 12 ] 3+ cations retained in the cluster-based micelles because these micelles were hard to analyze at the atomic level.
The optical absorptions and emissions of the Ag 29 nanoclusters in different solutions were compared. As depicted in Fig. 2C, the UV-vis spectrum of the NMP solution of Ag 29 -Na 1 showed an intense absorption at 445 nm and three shoulder bands at 320, 365, and 508 nm, whereas all of these peaks were attenuated when the nanocluster was dissolved in aqueous solution; such a phenomenon was also observed for the Ag 29 -Na 3 nanocluster (Fig. 2D). Despite this attenuation, the optical absorptions of both Ag 29 -Na 1 and Ag 29 -Na 3 in aqueous solution were actually quite similar to those in NMP or DMF. However, remarkable differences existed in terms of the emission wavelength and the photoluminescence (PL) intensity ( Fig. 2C and  D). Specically, the Ag 29 -Na 1 @NMP emitted at 630 nm, whereas the emission peak of Ag 29 -Na 1 @H 2 O is located at 690 nm along with the broadening of the emission wavelength. Besides, the emission of Ag 29 -Na 3 @DMF was centered at 634 nm, and the broadened emission of Ag 29 -Na 3 @H 2 O displayed two peaks at 662 and 708 nm. Of note, compared with the Ag 29 -Na 1 in NMP or Ag 29 -Na 3 in DMF, signicant attenuation on PL intensity was monitored when these Ag 29 nanoclusters were dissolved in H 2 O.
Although all types of Ag 29 nanoclusters (Ag 29 -Na 0 , Ag 29 -Na 1 , and Ag 29 -Na 3 ) presented cubic-like crystals at the macro-level (Fig. 2C,D, insets), 52,53 the emissions of them were entirely different. It has been demonstrated that the crystal of Ag 29 -Na 0 emitted at 670 or 700 nm with different crystalline patterns. 53 However, the emissions of Ag 29 -Na 1 and Ag 29 -Na 3 luminesced at 637 and 694 nm, respectively ( Fig. 2E; and see Fig. S3 † for the emission of Ag 29 -Na 0 ). Besides, the PL intensity of the Ag 29 -Na 3 crystal was slightly stronger than that of the Ag 29 -Na 1 crystal. Such differences in emission wavelength and PL intensity of these Ag 29 crystals resulted from their different crystal lattices (or different inter-cluster interactions).
It has been demonstrated that the hydrophobic Ag 29 (-SSR) 12 (PPh 3 ) 4 nanoclusters can be transferred into water in the presence of solvent-conjoined Na + cations; however, the existence of these Ag 29 nanoclusters in aqueous solution remains mysterious. Herein, the DLS (dynamic light scattering) and aberration-corrected HAADF-STEM techniques have been performed for monitoring their real-time existence.
With the help of DLS, the sizes of these Ag 29 nanoclusters (Ag 29 -Na 1 and Ag 29 -Na 3 ) in aqueous solution were monitored (Fig. 3). The concentration of Ag 29 -Na 1 (or Ag 29 -Na 3 ) in H 2 O of each stage was controlled as (0.1*n/3) mg mL À1 (n is the stage number in Fig. 3, and see the corresponding optical absorptions in Fig. S4 †). For Ag 29 -Na 1 , the measured sizes of these clusters were quite small ($2.69 nm) when their concentration was low (0.03 mg mL À1 ). When the concentration of nanoclusters increased, a remarkable size growth was observed. Finally, the Ag 29 -Na 1 cluster size soared to a plateau of $106 nm (Fig. 3A  and B). A similar variation tendency has been observed for the Ag 29 -Na 3 nanocluster, whose sizes in the aqueous solutions grew from $13.54 nm to $92.76 nm (Fig. 3D and E). Considering that it is not possible for the hydrophobic Ag 29 (SSR) 12 (-PPh 3 ) 4 clusters to form the contact surface with H 2 O molecules, we proposed that the phase-transfer of Ag 29 into water resulted from the micellization of such nanoclusters. The critical micelle concentration (CMC) of Ag 29 -Na 1 and Ag 29 -Na 3 micelles should both occur at stage 6 ( Fig. 3), and thus the CMC values of both Ag 29 micelles were determined as 0.2 mg mL À1 .
As depicted in Fig. 3C and F, the Ag 29 cluster-based micelle was composed of a hydrophobic Ag 29 interior and a hydrophilic Na-NMP (or Na-DMF) surface. That is, owing to the water-soluble Na-NMP (or Na-DMF) surface, the cluster-based micelle displayed good dissolvability and stability in aqueous solution.
The aberration-corrected HAADF-STEM measurements were further performed to verify the generation of the cluster-based micelles. Fig. 4, S5, and S6 † show the selected images of these Ag 29 -carrying micelles. As depicted in Fig. 4A and S5, † Ag 29 -Na 1 cluster entities were discrete in NMP, and were gradually assembled in aqueous solution. With the increased concentration of the dissolved Ag 29 -Na 1 in H 2 O, the sizes of micelles increased gradually. Finally, the sizes of the Ag 29 -Na 1 micelles were stabilized at about 100 nm. Similar size variations have also been observed for the Ag 29 -Na 3 micellization process, where the nal-stage sizes were also determined as about 100 nm (Fig. 4b and S6 †). Such a size growth trend and the nalstage size excellently matched with those derived from the DLS measurement (Fig. 3), further conrming the cluster micellization process. Based on the DLS and STEM results, the aggregation numbers in cluster-based micelles were proposed (Fig. S7 †)-111 360 of Ag 29 -Na 1 in each micelle with a 106 nm diameter and 72 810 of Ag 29 -Na 3 in each micelle with a 92 nm diameter.
Of note, although the structure of the Ag 29 (SSR) 12 (PPh 3 ) 4 molecule is retained during cluster micellization, the structures of [Na 1 (NMP) 5 ] + and [Na 3 (DMF) 12 ] 3+ cations may be altered; however, the water solubility of Ag 29 -Na 1 and Ag 29 -Na 3 activated by the presence of these solvent-conjoined cations indeed renders these hydrophobic clusters biocompatible to some extent, which sheds light on the preparation of atomically precise cluster-based, biocompatible nanomaterials.
Although both Ag 29 -Na 1 and Ag 29 -Na 3 exhibited excellent solubility in water (6.76 mg mL À1 for the Ag 29 -Na 1 and 7.88 mg mL À1 for Ag 29 -Na 3 ), the Ag 29 -Na 3 micelles were prone to coagulation; by comparison, the Ag 29 -Na 1 micelles were quite stable in the aqueous phase (Fig. S8 †). Due to the excellent stability of Ag 29 -Na 1 micelles, their performance in cell staining was evaluated. Specically, 5 mg mL À1 Ag 29 -Na 1 was incubated with live HepG2 cells and imaged directly under a laser confocal microscope. As shown in Fig. 5A, aer incubation for 2 hours, Ag 29 -Na 1 enabled effective uptake in the cytosolic region and displayed a punctate signal. In a parallel experiment (Fig. 5B), the same concentration of Ag 29 -Na 1 incubated with pre-xed cells displayed neglected uptake. These results demonstrated that Ag 29 -Na 1 was not a cell permeable probe, but might be internalized with live cells via an energy-dependent uptake pathway, such as endocytosis. To precisely determine the intracellular compartment where Ag 29 -Na 1 is stained, a colocalization experiment was performed. Live cells were incubated with Ag 29 -Na 1 and labelled with a lysosomal commercial dye, LysoTracker Deep Red. The micrograph in Fig. 5C suggested that the Ag 29 -Na 1 signal highly overlapped with the LysoTracker signal with a Pearson correlation coefficient (R r ) of 0.935, which further conrmed that the cell entry of Ag 29 -Na 1 might follow an endocytosis pathway. These observations also demonstrated that the Ag 29 -Na 1 micelles could stain lysosomes.
Furthermore, the bio-application of Ag 29 -Na 1 in superresolution imaging was evaluated. The chosen single cell was incubated with Ag 29 -Na 1 as described above and imaged under a stimulated emission depletion nanoscope (STED). The threedimensional (3D) micrographs revealed a whole cell lysosome distribution at an unprecedented resolution at both the x-axis and z-axis. This strongly demonstrated the ultra-high photonstability of Ag 29 -Na 1 and reected that such cluster micelles could be utilized for super-resolution-based imaging. Of note, due to the existence of the lipophilic phospholipid bilayer in the cell, the structures of [Na 1 (NMP) 5 ] + and [Na 3 (DMF) 12 ] 3+ cations were hard to retain in the cell staining process. However, the signicant role of these solvent-conjoined cations in corresponding Ag 29 nanoclusters was the phase-transfer effect that rendered these hydrophobic nanoclusters water-soluble, and thus their bio-applications appear to be promising.
In previous studies concerning cation-containing micelles, the role of Na + cations in micellization has been thoroughly researched (e.g., the micellization of alkyl sulfates or the ionic micelles). [64][65][66][67][68] However, the cation-induced micellization has not been reported in the nanocluster eld. Considering that the solvent-conjoined Na + cations can act as general counterions for negatively charged nanoclusters, we perceive a good opportunity to render such hydrophobic nanoclusters water-soluble. Herein, several negative-charged nanoclusters including [Au 25 24 ] 4À nanoclusters were used for evaluating the general applicability of the nanocluster micellization strategy. [54][55][56][57][58] For the preparation, each nanocluster was mixed with CH 3 COONa and minute quantities of DMF, which produced the [cluster] À [Na-solvent] + compounds. As shown in the digital photos in Fig. 6, in the absence of [Na-solvent] + cations, these nanoclusters were well soluble in CH 2 Cl 2 (photo i) but insoluble in H 2 O (photo ii); that is, they were absolutely hydrophobic. By comparison, in the presence of solvent-conjoined Na + cations, all of the obtained compounds showed good dissolvability in aqueous solution (photo iii). Specically, the water solubility of Au 25 (SC 2 H 4 Ph) 18 @Na-DMF, Ag 25 (SPhMe 2 ) 18 @Na-DMF, Pt 1 Ag 24 (SPhMe 2 ) 18 @Na-DMF, Ag 44 (-SPhF 2 ) 30 @Na-DMF, Au 12 Ag 32 (SPhF 2 ) 30 @Na-DMF, Ag 28 Cu 12 (-SPhCl 2 ) 24 @Na-DMF was 5.35, 5.78, 13.42, 27.12, 28.34, and 25.32 mg mL À1 , respectively. As depicted in Fig. 6, for each nanocluster, the optical absorptions in CH 2 Cl 2 and in H 2 O were the same, demonstrating the stability of the nanocluster in the aqueous phase. Furthermore, DLS measurements were performed in the saturated aqueous solutions of these nanoclusters, and all of the size-distribution results demonstrated the generation of cluster-based micelles (Fig. 6, insets). Consequently, the micellization of nanoclusters triggered by the addition of solventconjoined cations is indeed a generally applicable strategy for rendering hydrophobic nanoclusters water-soluble, at least for the negatively charged nanoclusters.

Conclusions
In summary, we presented a versatile strategy to render hydrophobic nanoclusters water-soluble-the micellization of nanoclusters; such a dissolvability variation was triggered by the addition of solvent-conjoined Na + cations. Specically, although several negative-charged nanoclusters (such as Ag 29 (-SSR) 12 (PPh 3 ) 4 , Au 25 (SR) 18 , Ag 25 (SR) 18 , etc.) were absolutely hydrophobic, they showed good dissolvability in aqueous solution in the presence of solvent-conjoined Na + cations. Crystal structures of Ag 29 -Na 1 and Ag 29 -Na 3 demonstrated that such Na + cations were capped by oxygen-carrying solvent molecules, and existed as [Na 1 (NMP) 5 ] 3+ or [Na 3 (DMF) 12 ] 3+ , acting as both counterions of negatively charged nanoclusters and surface cosolvent of cluster-based micelles in the aqueous phase. A combination of DLS and aberration-corrected HAADF-STEM unambiguously identied the generation of micelles of such nanoclusters. Owing to the excellent water solubility and stability of Ag 29 -Na 1 , its performance in cell staining has been evaluated-Ag 29 -Na 1 cluster-based micelles can stain lysosomes in both general imaging and super-resolution-based imaging.
Overall, this work hopefully sheds light on the preparation of atomically precise cluster-based, biocompatible nanomaterials.

Conflicts of interest
There are no conicts to declare.