Stimuli-responsive nucleic acid-functionalized metal–organic framework nanoparticles using pH- and metal-ion-dependent DNAzymes as locks

Drug-loaded DNA-capped metal–organic framework nanoparticles are unlocked by pH or Mg2+ ions/ATP triggers, resulting in the release of the loads.


Introduction
Metal-organic frameworks (MOFs) represent a broad class of organic/inorganic porous materials that have attracted substantial research efforts in recent years. 1 Different applications of the porous MOFs as functional carriers have been demonstrated, and these include their use as carriers for catalysts, 2 matrices for the storage of gases, 3 frameworks for carrying drugs and controlled delivery systems, 4 materials for sensing, 5 and functional materials for improving fuel cells performance. 6 The base sequence of nucleic acids encodes substantial structural and functional information into the biopolymer. 7 This includes the signal-triggered reconguration of nucleic acids such as the pH-induced reconguration of cytosine-rich nucleic acids into i-motif structures, 8 or the assembly of guanosine-rich sequences into G-quadruplexes, in the presence of ions, e.g. K + , Pb 2+ , or Sr 2+ ions. 9,10 Similarly, sequence-specic nucleic acids reveal catalytic properties, and metal-iondependent catalytic nucleic acids (M n+ ¼ Mg 2+ , Pb 2+ , Hg 2+ , Cd 2+ and more) have been developed. 11 These unique functions of nucleic acids were applied for the development of stimuliresponsive drug carriers. Mesoporous SiO 2 nanoparticles, 12 hydrogels, 13 microcapsules, 14 or liposomes 15 were loaded with drugs (or drug models) and locked by stimuli-responsive nucleic acid capping units. In the presence of appropriate triggers, such as pH, 16 enzymes, 17 aptamer/ligand complexes, 18 or metal-ion-dependent DNAzymes, 19 the nano/micro carriers were unlocked, resulting in the release of the loads.
The high porosity of MOFs, the ability to modify the MOFs using surface functionalities, and particularly, the feasibility of synthesizing nano-sized metal-organic framework nanoparticles (NMOFs), turn these nanomaterials into ideal drugcarrying matrices. Although several studies reported on the synthesis of stimuli-responsive MOFs using chemical capping functionalities, 20 the preparation of stimuli-responsive DNA/ MOF hybrid systems is scarce and relatively unexplored. Only recently, we reported on the synthesis of substrate-loaded microcrystalline MOFs capped by pH-responsive or K + ionstabilized G-quadruplex capping units. 21 The pH-or crownether-induced unlocking of these MOFs was demonstrated. Nonetheless, these microcrystalline MOFs cannot permeate into cells, and their use as drug-delivery systems is limited. Accordingly, the development of stimuli-responsive nucleic acid-functionalized nano-sized metal-organic frameworks, NMOFs, is needed.
In the present study we report on the synthesis and loading of nucleic acid-functionalized, stimuli-responsive NMOFs. The nucleic acid capping units are composed of pH-responsive nucleic acids or a metal-ion-dependent DNAzyme unit. The loads trapped in the NMOFs are uorescent dyes (acting as drug analoges) or the doxorubicin anti-cancer drug. In the presence of appropriate triggers, pH or metal ions, the NMOFs are unlocked, leading to the release of the loads. It should be noted that the previously reported nucleic acid capping units, 21 used to lock the microcrystalline MOFs, could not be adapted to cap the loaded NMOFs (leakage of the loads was observed in the absence of an external trigger). This forced us to develop a new strategy to synthesize the stimuli-responsive, gated, DNA/NMOF systems. This involves the covalent modication of the NMOF with nucleic acid tethers using the "click-chemistry" principle. We further demonstrate the incorporation of the NMOFs into cancer cells and study the cytotoxicity of doxorubicin-loaded NMOFs towards cancer cells. We also discuss other applications of the stimuli-responsive NMOFs as sensors, sense-andtreat systems, and functional components for logic-gate operations. It should be noted that the related nucleic acid-based locked mesoporous SiO 2 nanoparticles as drug carriers were previously reported. 12 However, the NMOFs reveal the following advantages over the SiO 2 nanoparticles: (i) the ordered repeat unit cell ions/ligands in NMOFs allow the precise modication of the ligand with the nucleic acid, and (ii) nucleic acid-modied mesoporous SiO 2 nanoparticles tend to aggregate and precipitate, while the nucleic acid-functionalized NMOFs are stable as suspendable mixtures. This turns the NMOFs into superior drug carriers for medical applications.

Results and discussion
The synthesis of the DNA-gated metal-organic framework nanoparticles (NMOFs) is depicted in Fig. 1(A). Aminefunctionalized triphenyl carboxylic acid (1) was reacted with ZrCl 4 to yield the porous NMOFs. 22 The size of the porous nanoparticles was in the range of 250-300 nm. The NMOFs had a bipyramidal structure, as shown in Fig. 1(B). BET measurements indicated a surface area of ca. 1160 m 2 g À1 and pore size corresponding to ca. 1.58 nm. The amine functionalities associated with the NMOFs were transformed to azide functionalities in the presence of t-butyl nitrite and trimethylsilyl azide and modied with a nucleic acid using the click chemistry principle. The amine-modied nucleic acid (2) was reacted with dibenzocyclooctyne-sulfo-N-hydroxysuccinimidyl ester (DBCOsulfo-NHS ester), and the resulting DBCO-modied nucleic acid (3) was conjugated to the azide groups linked to the NMOFs. Fig. 2(A) depicts the loading of the NMOFs with the dye/drug loads and the pH-induced unlocking and release of the loads. The (3)-functionalized NMOFs were loaded with the respective dye/drug and capped by the hybridization with the nucleic acid (4) that is complementary to the surface-linked promoter nucleic acid (3). The hybrid duplex (3)/(4) is designed to include in the strand (4), a cytosine (C)-rich sequence that recongures at acidic pH into an i-motif structure. This hybrid was designed with the vision that an acidic pH exists in cancer cells, and thus, reconguration of (4) into the imotif structure would unlock the NMOFs and release the anticancer drugs. Accordingly, the NMOFs were loaded with the methylene blue dye as a drug model, or with the doxorubicin anti-cancer drug. Fig. S1 † depicts the time-dependent release of methylene blue from the NMOFs upon subjecting the nanoparticles to pH ¼ 7.4 and pH ¼ 5.0. Evidently, the release of the dye from the NMOFs is substantially more efficient at pH ¼ 5.0. Aer ca. 180 minutes, the uorescence intensity of the dye levels off to a saturation value, implying that the dye trapped in the NMOFs was fully released. Using an appropriate calibration curve, we estimate that ca. 72.2 mmol methylene blue were released per gram of NMOFs (for the evaluation of the loading of methylene blue in the NMOFs, see Fig. S2, ESI †). It should be noted that the loading of methylene blue associated with the nucleic acid-functionalized NMOFs is substantially higher (ca. 4-fold) compared to that of the nucleic acid-functionalized mesoporous SiO 2 nanoparticles. 19 The time-dependent release of the dye at pH ¼ 7.4 also reaches a saturation value. This background release is attributed to crystalline domains with incomplete or defective capping. Fig. 2(B) depicts the uorescence spectra of doxorubicin, DOX, released from the NMOFs at pH ¼ 5.0 aer time intervals of 30 and 60 minutes. These uorescence spectra are compared to the uorescence spectra of DOX released from the NMOFs at pH ¼ 7.0. Evidently, the release of DOX from the NMOFs at pH ¼ 7.0 is very low. Fig. 2(C) depicts the time-dependent release of DOX from the NMOFs at pH ¼ 5.0 and pH ¼ 7.4. From the saturated uorescence intensity of DOX at pH ¼ 5.0, and using an appropriate calibration curve, we estimate that ca. 52.8 mmol DOX were released per gram of nanoparticles (for the evaluation of the loading of DOX in the NMOFs, see Fig. S3, ESI †). The enhanced release of the different loads at pH ¼ 5.0 is attributed to the reconguration of (4) into the i-motif structure that leads to the unlocking of the duplex capping units and the release of the dye/drug.
The doxorubicin-loaded NMOFs were further modied to include a cancer cell targeting element. Towards this end we made use of the fact that certain cancer cells include the nucleolin receptor to which the AS1411 aptamer was developed. 23 Accordingly, the (3)-modied dye/doxorubicin-loaded NMOFs were hybridized with the nucleic acid (5), which includes in its sequence the cytosine-rich, pH-responsive domain and the AS1411 aptamer domain, shown in Fig. 3(A). Fig. S4 † depicts the pH-induced release of the methylene blue dye from the NMOFs that are locked with the (3)/(5) capping units. While at pH ¼ 7.4 inefficient release of the dye occurs, at pH ¼ 5.0 effective release of the dye proceeds, as a result of the unlocking of the capping units through the formation of i-motif structures. Fig. 3(B) and (C) show the pH-induced release of DOX from the (3)/(5)-capped NMOFs. Fig. 3(B) shows the uorescence spectra of the released DOX at pH ¼ 5.0, aer time intervals of 30 and 60 minutes, in comparison to the uorescence spectra of the released DOX aer the same time intervals at pH ¼ 7.4. While the release of DOX at pH ¼ 7.4 is negligible, effective release of DOX is observed at pH ¼ 5.0. Fig. 3(C) shows the time-dependent release proles of DOX from the NMOFs at pH ¼ 5.0 and pH ¼ 7.4.
We then examined the cytotoxicity of the (    AS1411 aptamer conjugate reveals a detectable, enhanced cytotoxicity towards the cancer cells. Presumably, the nanosized dimensions of the MOF nanoparticles allow the preferential permeation of the nanoparticles into the cancer cells via the specic binding of the overexpressed nucleolin receptor, illustrated in Fig. 4 (d) and (e), respectively. Interestingly, the DOX-loaded NMOFs modied with the i-motif/AS1411 conjugates show a two-fold higher cytotoxicity, as compared to the DOX-loaded NMOFs capped with the (3)/(4) units, which include only the i-motif-responsive unlock units.
We have further searched for the development of stimuliresponsive DNA-capped, substrate-loaded, NMOFs. Toward this end we applied metal-ion-dependent catalytic nucleic acids as gating units. Metal-ion-dependent catalytic nucleic acids, DNAzymes, have attracted substantial interest in recent years, and different metal ion/nucleic acid conjugates, e.g. M n+ ¼ Mg 2+ , Zn 2+ , Pb 2+ , Hg 2+ , Ni 2+ , and Cu 2+ , have been developed. 11 Different applications of metal-ion-dependent DNAzymes were reported, including their use as amplifying labels for sensing events, 24 catalytic unlocking units for the release of drugs from mesoporous SiO 2 NPs, 25 for the dissolution of hydrogels and the programmed release of enzymes, 26 and their use as catalytic optical transducing outputs of logic-gate operations and computing circuits. 27 Fig. 6(A) depicts the loading of the NMOFs with Rhodamine 6G, their capping with the Mg 2+ -dependent DNAzyme, and the principle of unlocking the NMOFs. The NMOFs were functionalized with the nucleic acid (6), using the "click chemistry" method, loaded with Rhodamine 6G, and capped with the Mg 2+ ion-dependent DNAzyme sequence (7). The nucleic acid (6) is composed of a nucleobase-modied sequence acting as substrate for the DNAzyme. In the presence of Mg 2+ ions the DNAzyme is activated, leading to the cleavage of the substrate, unlocking of the NMOFs, and to the release of the dye. Fig. 6(B) shows the uorescence spectra of the released Rhodamine 6G resulting from the treatment of the NMOFs with variable concentrations of Mg 2+ , for a xed time of 30 minutes. As the concentration of Mg 2+ ions increases, the release of the dye is enhanced, and this is consistent with the higher degree of unlocking of the NMOFs by the DNAzyme. In the presence of Mg 2+ ions, the release of the load reaches a saturation value aer ca. 120 minutes. Using an appropriate calibration curve, we estimate that ca. 62.7 mmol Rhodamine 6G were released per gram of NMOF nanoparticles. The release of the load in the absence of Mg 2+ ions reaches a saturation value aer ca. 120 minutes and no further release is observed aer a longer time interval. The low-level release of the dye in the absence of the Mg 2+ ions is attributed to the release of the load from incomplete (or "defective") locks associated with surface domains on the NMOFs. The unlocking of the NMOFs is selective towards Mg 2+ ions, Fig. S6, † and other metal ions do not unlock the NMOFs. A similar concept was applied to unlock methylene blue-loaded NMOFs capped by the Pb 2+ -iondependent DNAzyme, shown in Fig. 7(A). In this system, the NMOFs were functionalized via the click-chemistry principle, with the ribonucleobase-containing nucleic acid (6), acting as substrate for the Pb 2+ -ion-dependent DNAzyme. The (6)-modi-ed NMOFs were loaded with methylene blue and capped with the Pb 2+ -ion-dependent sequence (8). In the presence of Pb 2+ ions, the cleavage of the (6)/(8) duplex capping units proceeds, leading to the unlocking of the NMOFs, and the release of methylene blue. Fig. 7(B) depicts the uorescence spectra of the methylene blue released, aer a xed time interval of 30 minutes, using variable concentrations of Pb 2+ ions as unlocking release triggers, and the respective calibration curve is shown in Fig. S7. † As the concentration of Pb 2+ -ions increases, the release process is enhanced, which is consistent with the enhanced unlocking of the capping units. Fig. 7(C) shows the time-dependent release of the methylene blue loads upon subjecting the NMOFs to 100 mM Pb 2+ (curve b), in comparison to the release of the loads in the absence of the Pb 2+ ions (curve a). The release prole of methylene blue in the presence of Pb 2+ ions reveals that aer ca. 120 minutes the release process  reaches a saturation value, implying that the unloading of the dye is completed. Using an appropriate calibration curve, we estimate that ca. 61.8 mmol methylene blue were released per gram of nanoparticles within this time interval. As expected, the unlocking of the NMOFs is selective, and other metal ions did not induce the triggered release of the dye from the NMOFs, as shown in Fig. S8. † The metal-ion-dependent selective release of dye loads from the NMOFs provides a versatile means to detect metal ions, and particularly, a method for the multiplexed analysis of ions using mixtures of NMOFs loaded with different uorescent labels capped by different metal-ion-dependent DNAzymes. Furthermore, the use of a mixture of NMOFs loaded with two different uorophores and unlocked by two different metal ions, acting as unlocking triggering units, provides a means for the multiplexed analysis of different ions, where the uorophore released from the respective NMOF provides the readout signal for the sensing event. Thus, the metal-ion-modied NMOFs may act as modules for logic-gate operations. The two different ion-triggers might act as inputs, and the uorescence of the released dyes provides the outputs of the logic-gates. Fig. 8 depicts the application of a mixture of the NMOFs loaded with Rhodamine 6G or methylene blue, and capped with the Mg 2+ -ion-dependent DNAzyme and the Pb 2+ -ion-dependent DNAzyme as capping units, respectively, as a model assembly for multiplexed sensing and logic-gate applications. The Mg 2+ and/or Pb 2+ ions act as inputs for the system. In the absence of the ions, only the background uorescence corresponding to the leakage of the uorophores from the NMOFs is observed (panel I). In the presence of only Mg 2+ ions or Pb 2+ ions, the uorescence of only Rhodamine 6G (l em ¼ 550 nm) or methylene blue (l em ¼ 682 nm) is intensied, shown in panel II and panel III, respectively.
In the presence of the inputs, Mg 2+ and Pb 2+ ions, the two kinds of NMOFs are unlocked, giving rise to the uorescence of the two uorophores, shown in panel IV. Besides demonstrating the ability of the NMOFs to act as functional carriers for multiplexed analysis of ions, the system follows an "AND" logic-gate operation.
We then applied the DOX-loaded capped NMOFs as functional units for the development of a NMOF carrier responsive to a cancer cell biomarker. In a previous study, we reported on ATP-responsive/Mg 2+ -dependent DNAzyme capped SiO 2 nanoparticles for the triggered release of the anti-cancer drug doxorubicin. 25 In this study we made use of the basic elements: (i) ATP is over-expressed in cancer cells due to the enhanced metabolism in these cells, and (ii) for optimal catalytic activities of the Mg 2+ -dependent DNAzyme the precise base-sequence in the loop-region is important. The introduction of a foreign nucleic acid sequence into the loop decreases the activity of the Mg 2+ -dependent DNAzyme, due to the exibility of the loopregion that reveals a lower affinity toward the Mg 2+ ions. Nonetheless, it was demonstrated that a foreign sequence corresponding to an aptamer sequence (e.g. the anti-ATP aptamer sequence) can, upon formation of the aptamer-ligand complex (e.g. the ATP-aptamer complex), act as an auxiliary trigger for the stabilization of the Mg 2+ -ion-dependent DNAzyme loop, and cause re-activation of its high catalytic function. Accordingly, we examined the possibility of designing "smart" ATP-responsive Mg 2+ -ion-dependent DNAzyme-capped DOX-loaded NMOFs for the selective treatment of cancer cells, as shown in Fig. 9(A). In this system, the (6)-functionalized NMOFs (where (6) consists of the ribonucleobase-modied nucleic acid sequence acting as substrate for the Mg 2+ -dependent DNAzyme) were loaded with doxorubicin (DOX) and capped with the sequence (9), which consists of an extended loop composed of the Mg 2+ -dependent DNAzyme subunits, separated by the ATP-aptamer sequence. The resulting (6)/(9)-capped DOX-loaded NMOFs are expected to reveal inefficient Mg 2+ ion induced triggered release of the DOX drug, but effective release of the loads in the presence of ATP and Mg 2+ ions, due to the stabilization of the catalytic loop by the formation of the ATP-aptamer complex. Accordingly, we examined the ATP-triggered release of DOX from the (6)/(9)capped NMOFs and compared the release features of the "smart" NMOFs to the (6)/(7)-capped DOX-loaded NMOFs ( (7) is the regular Mg 2+ -dependent sequence). Fig. S9 † depicts the uorescence spectra of released DOX upon treatment of the (6)/ (7)-capped DOX-loaded NMOFs with different concentrations of Mg 2+ ions for a xed time interval of 60 minutes. As the concentration of Mg 2+ ions increases, the uorescence intensity of the released DOX increases, consistent with the enhanced unlocking of the NMOFs. Fig. S10 † shows the time-dependent release of DOX upon treatment of the (6)/(7)-capped NMOFs with 2 mM Mg 2+ ions, and the release prole is compared to the time-dependent release of DOX from the NMOFs in the absence of added Mg 2+ ions. Evidently, effective release of DOX proceeds upon treatment of the NMOFs with Mg 2+ ions. The release of DOX from the NMOFs in the absence of Mg 2+ ions reaches a low saturation value aer ca. 100 minutes, and this is attributed to incomplete, (or "defective") capping sites on the NMOFs. Fig. 9(B) depicts the uorescence spectra of the released DOX upon the treatment of the (6)/(9)-capped NMOFs with variable concentrations of Mg 2+ ions and a xed concentration of ATP (1 mM) for a xed time interval of 60 minutes. As the concentration of Mg 2+ ions increases, the amount of released DOX is higher. Control experiments reveal that in the absence of ATP the unlocking of the (6)/(9)-capped NMOFs is inefficient in the Mg 2+ -ion concentration range of 1 mM to 50 mM, as shown in Fig. S11. † Furthermore, the unlocking of the (6)/(9)-capped NMOFs is also controlled by the concentration of ATP, and at a xed concentration of Mg 2+ ions (2 mM) the release rate of DOX increases as the concentration of ATP is increased, as shown in Fig. S12. † These results indicate that the unlocking of the (6)/(9)-capped NMOFs is cooperatively triggered by Mg 2+ and ATP. While Mg 2+ ions are essential to induce the release process, the ATP ligand cooperatively enhances the release process. These conclusions are further supported by following the time-dependent release of DOX from the (6)/(9)-capped NMOFs in the presence and absence of added ATP, shown in Fig. 9(C). While the release of DOX in the presence of 2 mM Mg 2+ ions is only slightly enhanced as compared to the natural leakage from the NMOFs with no Mg 2+ ions, the addition of ATP signicantly enhances the DOX release, and a 2-fold higher release prole of DOX is observed (for the comparison of the release proles of the (6)/(7)-and (6)/(9)-capped NMOFs see Fig. S13 †).
We then examined the cytotoxicity of the (6)/(9)-capped NMOFs toward MDA-MB-231 breast cancer cells and MCF-10A epithelial normal breast cells, and the results are shown in Fig. 10. While no cytotoxic effect of the DOX-unloaded (6)/(9)capped NMOFs on the MDA-MB-231 or MCF-10A was observed aer a time interval of ve days, the DOX-loaded (6)/(9)-capped NMOFs had a cytotoxic effect only on the MDA-MB-231 cancer cells. Aer three days, cell death corresponding to 30% of cells was observed, and this increased to a cell death of 40% aer ve days. The non-detectable toxicity towards the normal MCF-10A cells is attributed to inefficient permeation of the NMOFs into the normal cells and to the lack of release of DOX by any permeated NMOFs, due to the low concentration of ATP in the cells. The toxicity towards the cancer cells is attributed to the cooperative functions of the over-expressed ATP in cancer cells in the unlocking of the NMOFs and the release of DOX. It should be noted that the (6)/(7)-capped DOX-loaded NMOFs revealed lower cytotoxicity toward the MDA-MB-231 cancer cells, as compared to that of the (6)/(9)-capped DOX-loaded NMOFs (15% cell death as compared to 30% cell death, respectively). This difference is attributed to the low concentration of intracellular Mg 2+ ions, which results in the unlocking process and release of DOX being inefficient. Presumably, the cooperative ATP-driven rigidication of the loop in the (6)/(9)-capped NMOFs enhances the affinity of intracellular Mg 2+ ions toward the loop domain, thereby improving the DOX release process. Similar results were obtained upon examining the timedependent cytotoxicity of the (6)/(7)-and (6)/(9)-DOX-loaded NMOFs on aggregated 3D spheroid cancer cells (cf. Fig. 11 and accompanying discussion).
The synergistic superior cytotoxic effect of ATP and Mg 2+ ions on the cytotoxicity of (6)/(9)-capped NMOFs toward the MDA-MB-231 cells, as compared to the cytotoxicity of (6)/(7)capped NMOFs unlocked only by Mg 2+ ions, were conrmed by following the time-dependent apoptosis of the cell aggregates using the cell IncuCyte system, 28 as shown in Fig. 11. In this experiment, cell spheroids are treated with the respective NMOFs and the IncuCyte® Red cytotoxicity reagent that probes the apoptosis of cells at the time interval of incubation (see details in Experimental section). Fig. 11(A) depicts the colors of the spheroid cell aggregates treated with different NMOFs for different time intervals (the intensity of the red color reects the degree of cell death). Evidently, the cancer cells spheroids treated with the DOX-loaded, (6)/(9)-capped NMOFs reveal substantially high cell death as compared to the spheroids treated with (6)/(7)-capped NMOFs. Fig. 11(B) shows the timedependent apoptosis of the MDA-MB-231 cells upon treatment with the different NMOFs. While the DOX-vacant NMOFs capped with (6)/(9) do not show cytotoxicity (curve (b)), the DOX loaded (6)/(9)-capped NMOFs show high cytotoxicity towards the cancer cells (curve (d)). The cytotoxicity of the (6)/(9)-capped NMOFs is superior to the cytotoxicity of the (6)/(7)-capped NMOFs (curve (d) vs. curve (c), respectively).

Conclusions
In summary, the present study has introduced nucleic acidfunctionalized metal-organic framework nanoparticles, NMOFs, as stimuli-responsive carriers of drugs or analoges of drugs. We have developed a versatile methodology to covalently link nucleic acids to the NMOFs, integrate the loads in the NMOFs, and use the nucleic acid tethers as anchoring sites for capping of the NMOFs with stimuli-responsive units that allow the unlocking of the NMOFs and the release of the loads. Two different stimuli-responsive capped NMOFs were introduced. One NMOF carrier included a pH-responsive nucleic acid duplex as a lock. At an acidic pH, the duplex DNA capping units were separated through the reconguration of one of the duplex constituents to the i-motif structure, a process that unlocked the capping units and allowed the release of the loads. The  second stimuli-responsive NMOF involved the association of a sequence-specic metal-dependent DNAzyme sequence on ribonucleobase modied tethers linked to the NMOFs. In the presence of Mg 2+ ions or Pb 2+ ions the catalyzed cleavage of the locks led to the release of the loads. Besides the new fundamental mechanisms to unlock the loaded NMOFs, a method to target the stimuli-responsive NMOFs to the cancer cells was developed via the conjugation of the AS1411 aptamer sequence to the stimuli-responsive nucleic acid. The AS1411 aptamer sequence binds to the nucleolin biomarker associated with the cell boundaries of different cancer cells thereby enhancing the permeation of the NMOFs into cancer cells. Furthermore, the design of "smart" Mg 2+ -ion-dependent DNAzyme-capped NMOFs that are unlocked by cooperative triggers (Mg 2+ /ATP) represents a novel approach for a sense-and-treat anti-cancer drug carrier.
The study has highlighted the loading of the different NMOFs with the anti-cancer drug doxorubicin, and the cytotoxicity of the drug-loaded NMOFs toward MDA-MB-231 breast cancer cells and MCF-10A epithelial normal breast cells. Preliminary cell experiments revealed impressive selective cytotoxicity towards the cancer cells. This selective cytotoxicity was attributed to the enhanced permeation of the NMOFs into the cancer cells by the receptor-mediated endocytosis and to the targeted permeation of the NMOFs into the cells. These results suggest that many other drugs, and specically anti-cancer drugs, could be incorporated into the stimuli-responsive NMOFs. Also, many different NMOFs could act as stimuli-responsive carriers. Beyond the use of the nucleic acid-capped NMOFs as selective drug carriers, the application of the NMOFs for multiplexed sensing and logic-gate operation was demonstrated.

Loading of the nucleic acid (3)-functionalized NMOFs
For the (3)-functionalized NMOFs, 5 mg of NMOFs was incubated with methylene blue (0.5 mg mL À1 ) or the anti-cancer drug doxorubicin (1.0 mg mL À1 ) for 24 h in 2 mL of PBS buffer solution (10 mM, pH ¼ 7.4), and this corresponded to the unlocked form of the nucleic acid (3)-functionalized NMOFs. Subsequently, the NMOFs were transferred to a buffer solution and hybridized with nucleic acid 4 or 5, respectively, to yield the locked (3)/(4) or (3)/(5) DNA-functionalized NMOFs loaded with the dye or drug. Aer 12 h, the NMOFs were washed several times to remove the excess and non-specically bound methylene blue or doxorubicin.
pH-induced unlocking of the (3)/(4)-or (3)/(5)-functionalized NMOFs and the release of the encapsulated loads The pH-responsive methylene blue-loaded NMOFs (1 mg mL À1 ) were subjected to buffer solutions, at pH ¼ 7.4 or pH ¼ 5.0. At appropriate time intervals, samples of the mixture are centrifuged to precipitate the NMOFs (10 000 rpm for 10 minutes). The uorescence of the released load in the supernatant solution was measured using a Cary Eclipse Fluorescence Spectrophotometer (Varian Inc.).
Metal ion-induced unlocking of nucleic acid (6)functionalized NMOFs and the release of the encapsulated loads The detailed protocol for the loading the (6)-functionalized NMOFs with Rhodamine 6G, methylene blue or doxorubicin is provided in the ESI. † The duplex DNA-capped, Rhodamine 6G or methylene blue loaded functionalized NMOFs, at a concentration corresponding to 1 mg mL À1 , were subjected to the respective ions to unlock the NMOFs and release the loads. The NMOF solutions were treated with different concentrations of Mg 2+ or Pb 2+ ions, respectively. At time intervals, the respective sample solutions were centrifuged to precipitate the NMOFs (10 000 rpm for 10 minutes), and the uorescence of the released loads in the supernatant solutions was measured.
Parallel operation of the NMOFs using Mg 2+ ions and Pb 2+ ions as inputs The parallel operation of the mixture of (6)/(7)-and (6)/(8)-capped NMOFs by unlocking the capping units with Mg 2+ ions and Pb 2+ ions as inputs is described in the ESI. † The AND gate operation was examined in a composite solution that included the Mg 2+ -responsive Rhodamine 6G-loaded NMOFs and the Pb 2+ -responsive methylene blue-loaded NMOFs. The appropriate input triggers were Mg 2+ and/or Pb 2+ ions: (0, 0), (1, 0), (0, 1), and (1, 1). Digital "1" levels of Mg 2+ and Pb 2+ were 25 Â 10 À3 mM and 100 Â 10 À6 mM, respectively. Aer incubation of the respective samples for 30 minutes, the solutions were centrifuged to precipitate the NMOFs, and the uorescence of the released loads in the supernatant solution was measured.
Mg 2+ and ATP-induced unlocking of the (6)/(9)-functionalized NMOFs and the release of the encapsulated loads The (6)/(9)-capped NMOFs loaded with Rhodamine 6G or DOX (1 mg mL À1 ) were treated with the appropriate concentrations of Mg 2+ ions and ATP. At dened time intervals, samples of the NMOFs were centrifuged (10 000 rpm) to precipitate the NMOFs, and the uorescence of the released loads in the supernatant solutions was measured.
A detailed description of the experiments probing the cytotoxicity of the different DOX-loaded NMOFs to the MDA-MB-231 malignant breast cells and MCF-10A epithelial breast cells is described in the ESI. † Evaluation the cytotoxicity of cancer cell 3D-spheroids The NMOF-induced cytotoxicity of 3D-spheroids of the MDA-MB-231 cancer cells was evaluated using the IncuCyte Zoom system (Essen Bioscience). 28 The MDA-MB-231 cells were seeded in a 96 well ULA plate (Corning 7007) at a density of 2 Â 10 3 , and allowed to culture for an additional 3 days to form the spheroids. The resulting spheroids were then treated with the respective NMOFs and the IncuCyte® Red cytotoxicity reagent (Essen Bioscience Cat #4632) to probe the apoptosis of cells. Cell images were recorded every two hours, and the resulting red color, reecting cell death, was analyzed with the IncuCyte soware. This allowed the evaluation of the time-dependent apoptosis of the cells.