A hydrogen sulphide-responsive and depleting nanoplatform for cancer photodynamic therapy

Hydrogen sulfide (H2S) as an important biological gasotransmitter plays a pivotal role in many physiological and pathological processes. The sensitive and quantitative detection of H2S level is therefore crucial for precise diagnosis and prognosis evaluation of various diseases but remains a huge challenge due to the lack of accurate and reliable analytical methods in vivo. In this work, we report a smart, H2S-responsive and depleting nanoplatform (ZNNPs) for quantitative and real-time imaging of endogenous H2S for early diagnosis and treatment of H2S-associated diseases. We show that ZNNPs exhibit unexpected NIR conversion (F1070 → F720) and ratiometric photoacoustic (PA680/PA900) signal responsiveness towards H2S, allowing for sensitive and quantitative visualization of H2S in acute hepatotoxicity, cerebral hemorrhage model as well as colorectal tumors in living mice. ZNNPs@FA simultaneously scavenges the mitochondrial H2S in tumors leading to significant ATP reduction and severe mitochondrial damage, together with the activated photodynamic effect, resulting in efficient suppression of colorectal tumor growth in mice. We believe that this platform may provide a powerful tool for studying the vital impacts of H2S in related diseases.

8. In Figure S20, there are obvious interference signal from NIR-II imaging and NIR imaging. It seems the specific diagnosis of HCT116 tumor is very poor. And the authors should give the detailed explanation on the interference signal in the activated nanoprobe.
Reviewer #3: Remarks to the Author: In the article "Quantitatively Visualizing and Depleting Endogenous H2S Combined with Activated Photodynamic Effect for Synergistical Enhanced Colorectal Cancer Therapy", the authors report a smart, H2S-responsive and depleting nanoplatform (ZNNPs) to quantitatively and real-time imaging endogenous H2S for early diagnosis and treatment of H2S-associated diseases (acute hepatotoxicity, cerebral hemorrhage, and colorectal tumor). This ZNNPs nanoplatform displayed unexpected NIR conversion (F1070→F720) and ratiometric photoacoustic (PA680/PA900) signal responsiveness towards H2S and simultaneously depleted the mitochondrial H2S level in cancer cells leading to remarkable reduction of glycolysis and severe mitochondrial damage, together with the activated photodynamic effect. Sufficient experiments and data were supplied to make this work nice. The nanoplatform is new and the application is important. Thus, I recommend its publication after a minor revision.  2. Although both ZNNPs and ZNNPs@FA showed good H 2 S responsiveness in vivo, their stability in serum-containing solution is lacking, which should be studied.
Reply: Thanks for the reviewer's valuable comments and suggestions. To investigate the stability of ZNNPs and ZNNPs@FA in serum-containing solution, both probes were separately dispersed into 10% serum-containing PBS buffer and then stocked at 37 o C for one week. The hydrodynamic sizes of ZNNPs and ZNNPs@FA were real-time monitored by DLS analysis. As shown in Figure  S5c and S5d, no big change in terms of particle size for both probes was observed, suggesting that both ZNNPs and ZNNPs@FA have great stability in serum-containing aqueous solution. Figure S5. The stability study of ZNNPs (c) and ZNNPs@FA (d) in PBS containing 10% serum at 37℃ over one week. 3. The data show that ZNNPs have good responsiveness towards H 2 S. How about the detection limit of ZNNPs toward H 2 S. What is the expression level of H 2 S in liver or colorectal tumor normally?
Reply: Thanks for your great comments. To determine the detection limit of ZNNPs toward H 2 S, probe ZNNPs (14 μg/mL) was incubated with various concentrations of NaHS ranging from 0 to 500 μM for 10 min followed by detection of the fluorescence (FL 720 ) and ratiometric photoacoustic (PAS 680 /PAS 900 ) signal. Figure S11c and S11d clearly shows that both FL 720 and PAS 680 /PAS 900 were enhanced linearly with the increase of NaHS concentrations. According to the formula LOD = 3σ/s (where LOD was the limit of detection, σ was the standard deviation of blank probe sample measurement, and s was the slope of the calibration pot, values are mean ± SD, n =  3), the lowest detection limits of H 2 S in fluorescence and ratiometric PAS 680 /PAS 900 were eventually calculated to be ~4.7 μM and ~0.68 μM, respectively, which strongly demonstrates that the sensitivity of ratiometric PA is significantly superior to fluorescence in detection of H 2 S. Figure S11. Determination of the detection limit of ZNNPs toward H 2 S in the fluorescence intensity at 720 nm and ratiometric PAS 680 /PAS 900 , respectively.
To know the expression level of H 2 S in liver or colorectal tumor normally, we also searched and read a lot of literature, and found the concentration of H 2 S in mouse liver is around 20 µM, and is ranging from 0.3 to 3.4 mmol L -1 in mouse colorectal tumor. The related references are listed below: 4. Toward in vivo applications, the blood circulation half-life of ZNNPs and ZNNPs@FA should be also studied, and the cytotoxicity of ZNNPs and ZNNPs@FA in mice should carried out, for example, blood hemolysis and hemanalysis.
Reply: Thanks for the reviewer's great comments. It is true that the blood circulation behavior of nanoprobes in body is one of important factors which should be investigated especially for in vivo biological applications. To this end, two groups of healthy mice were intravenously injected with ZNNPs and ZNNPs@FA, respectively. The blood samples were collected from the eyes at different time points for NIR fluorescence measurements. The results shown in Figure S32 indicate that both probes exhibited a short blood circulation time. After calculation, the main  Error bar indicates SD (n = 2). ZNNPs (10 mg/kg) and ZNNPs@FA (10 mg/kg) were i.v. injected into the mice. (100 μL of blood were pretreated by adding 1 μL of 50 mM NaHS solution and incubating at 37℃ for 2 h).
To further evaluate the cytotoxicity of ZNNPs and ZNNPs@FA in mice, we performed the hemolysis analysis for both nanoprobes in mouse blood. The results shown in Figure S15 illustrated that no significant red cell damage was observed for both ZNNPs and ZNNPs@FA even at the concentration of 160 μg/mL, implying that they don't have apparent hemolysis. Meanwhile, we also carried out the blood routine analysis for both ZNNPs and ZNNPs@FA. The blood routine biochemical results shown in Figure S35 and S36 clearly indicate that no significant difference for all parameters (ALP, AST, ALT, UREA, CREA, RBC, WBC, PLT, MCV, MCH, MCHC, HGB, RDW, HCT, MPV, LYM, PCT, and PDW) associated with acute toxicity compared to the control blood was observed, again suggesting our probes have excellent biocompatibility. 5. According to the results in Figure 1d, the FL1070 is gradually decreased with the increasing concentration of H 2 S used, why does the FL1070 almost kept constant for NIR imaging of H 2 S in livers (Fig 3b)? Please make a clarification on it.
Reply: Thanks for your great question! Our in vitro results in Figure 1d are indeed shown that the FL 1070 intensity of the probe gradually decreased with the increasing concentrations of H 2 S. However, when the probes were intravenously injected into the mice via the tail vain, the probes accumulated in the mice liver gradually with the blood flow over time. In fact, the amount of the probes enriched in liver is dynamically changing due to the simultaneous occurring of clearance by the body. Hence, it is difficult to see a clear gradual decrease of FL 1070 in the liver of living mice overall . 6. Some formatting and typo errors present in the manuscript. Please Go through the whole Reply: We highly appreciated you pointed out our errors! We have carefully gone through the whole manuscript and made some revision accordingly.

Response:
We appreciate the reviewer's comments! As the reviewer said, it is true that many nanoprobes have recently been developed to visualize and deplete the endogenous H 2 S for H 2 Sactivated NIR imaging and photothermal/photodynamic therapy of tumors both in vitro and in vivo. However, these previously reported studies mainly focused on the imaging of H 2 S or activated tumor therapy. Differing from all of the previous studies, our novelty and innovation in this work can be summarized in two aspects: First, the key novelty of our study is that we not only developed a new H 2 S-responsive NIR/ratiometric photoacoustic imaging probe for visualization of endogenous H 2 S in vivo, more importantly, taking advantage of the ratiomatric PA feature, we for the first time established a quantitative equation between ratiometric PA signal and the level of endogenous H 2 S in living system. In light of this relationship, we successfully realized the noninvasive and quantitative visualization of endogenous H 2 S in mouse liver, brain as well as colorectal tumor. Furthermore, we could achieve the quantitative evaluation of the drug hepatotoxicity in real-time manner by using our probes. Therefore, our approach may offer a powerful tool for studying the vital impacts of H 2 S in related diseases; Second, we for the first time found our probe could also effectively deplete the endogenous H 2 S in cellular mitochondria, which can cause significant ATP reduction and severe mitochondrial damage, combined with the H 2 S-activated photodynamic effect, eventually leading to efficient suppression of colorectal tumor in living mice. Overall, our study would provide a new insight for achieving precise diagnosis and treatment of colorectal cancer.
Some detailed comments are listed as follows: 1. Subcutaneous tumor is not sufficient for present work, and the in situ tumor model should be used for the imaging and therapy.
Reply: Thanks for your suggestion! To be honest, our current work is mainly focusing on the proof of concept case study for developing a smart sensing platform to quantitatively assess the pathological progression of H 2 S-associated diseases. I agree with the reviewer's opinion, the experiments just based on subcutaneous tumors are not sufficient to fully validate the effectiveness and universality of our approach. Due to the complexity and time consumption of in situ tumor model, in the early study we mainly focus on validating the feasibility of our probes for quantitative detection of endogenous H 2 S in subcutaneous tumors. To further explore the utility of our probes in vivo, we are currently building in situ mouse tumor models to deeply evaluate their potential for quantitative detection of endogenous H 2 S in living mice. The results of the extensive studies will be reported later.
2. Biological characterization is too rough. For example, statistical analysis, cell toxicity, pathological section, blood biochemical analysis, pharmacokinetics and survival curve are missed. Reply: Thanks for the reviewer's valuable comments! To make some biological characterization fully comprehensive, we performed further statistical analysis for some data in the revised manuscript. Additionally, to deeply evaluate the biocompatibility of probe ZNNPs and ZNNPs@FA, we also carried out their blood routine analysis. The blood routine biochemical results shown in Figure S35 and S36 indicate that no significant difference for all parameters (ALP, AST, ALT, UREA, CREA, RBC, WBC, PLT, MCV, MCH, MCHC, HGB, RDW, HCT, MPV, LYM, PCT, and PDW) associated with acute toxicity compared to the control blood was observed, suggesting our probes have excellent biocompatibility.
We next conducted a preliminary study of the biodistribution and pharmacokinetic behavior of both ZNNPs and ZNNPs@FA. The probes were i.v. injected into the heathy mice via the tail vain, and then the major organs including heart, liver, spleen, lung, and kidney were dissected and taken out for fluorescence imaging at 720 and 1070 nm. As shown in Figure S33, fluorescence signals of various organs indicated that the nanoprobes mainly accumulated in the liver and lung at 3 h post-injection. With the time elapses, both probes were slowly excreted from liver and lung into kidney, indicating the probes are gradually excluded from the body through the renal metabolism. The quantitative analysis of these fluorescence images showed almost all probes had been excreted from the body at 48 h post-injection. Furthermore, the organs at various time points were digested and lysed, and the supernatant were subjected for fluorescence measurement at 720 nm. The quantitative results in Figure S34 clearly indicate that approximately 16.5% of nanoprobes presented in every gram of liver of mice at 3 h post-injection, whereas less than 5.1% of them was till retained at 48 h post-injection, strongly implying that our probes can be rapidly metabolized by the body, and they have excellent biological safety. The survival rates of mice receiving different treatments were also analyzed. The results shown in Figure S30 indicate that ZNNPs@FA+660 nm group exhibits greatly improved survival rate without single death over 20 days in comparison with control groups.   ZNNPs (10 mg/kg) (a) and ZNNPs@FA (10 mg/kg) (b). Error bar indicates SD (n = 2). Every 100 μL of organ homogenate was pretreated by adding 1 μL of 50 mM NaHS solution and incubating at 37℃ for 2 h.
3. As the author mentioned, the ZNNPs can be used for brain tissue imaging. How did they broken the blood-brain barrier? The authors should give detailed evidence and explanation. Reply: Thanks for your great questions! As we all known, one of the huge challenges faced for the diagnosis and treatment of brain disease is how the drugs can go through the BBB and accumulate at the focal region effectively. Based on the results in Figure 4i and 4j, we could clearly find that our probe ZNNPs could also light-up the injured region of mouse brain due to the high upregulation of the H 2 S level. To further confirm this conclusion, we further repeated these experiments. Both healthy and brain-injured mice were i.v. injected with ZNNPs (10 mg/kg) through tail vain. After 4 h post-injection, the mouse brain was imaged by both fluorescence and PA imaging systems at several time internals. As shown in Figure S21a, the PA 680 signals enhanced gradually with the increasing time, while only low PA 900 signals were recorded over time. In huge contrast, for the healthy mice almost no PA signals at either 680 nm or 900 nm were detected all the time. Moreover, we also conducted ex vivo fluorescence imaging of the mouse brain resected from the mice. The results in Figure S21b and S21c shows that the injured tissues in the resected brain can be obviously identified by naked eyes and fluorescence both at 720 nm and 1070 nm. However, no fluorescence was detected from the control mice at all, indicating very little probe has entered the mouse brain. Together, these results strongly demonstrate that probe ZNNPs can go through the BBB and entered the injured brain for visualization of the damaged tissues. In light of these solid evidences, we firmly believe that the probes can solely enter the injured mouse brain but difficult for healthy mice. We speculated that the reason for our nanoprobes easily entering the injured mouse brain is because the BBB of brain-injured mice was disrupted, but the healthy mice retain intact and compact BBB, which has been well demonstrated in numerous previously reported papers. To further verify our hypothesis, the brain tissues both from brain-injured mice and healthy mice were resected and sliced for HE and Nissl staining. Figure S21d and S21e shows that much more nerve cells presented in the injured brain tissues as compared to the normal mice. Besides, obvious breakage and loosen could be clearly observed for brain-injured mice from the HE staining results. Collectively, these results strongly support that our probe is more preferable for detection and tracing of brain damage in living mice. fluorescence signals were determined if the cells were pre-incubated with ZnCl 2 (40 μg/mL, 10 min), indicative of lower concentration of H 2 S than the control cells. Moreover, if the cells were pre-incubated with extraneous NaHS (100 µM, 1h) or L-Cys (24 μg/mL, 1h) followed by ZNNPs treatment, the fluorescence signals were dramatically suppressed in comparison with the control cells, which is highly consistent with the observed F 1070 decreasing for the assays containing probes and NaHS solution. Values represent mean ± SD (n = 4, *P < 0.05, **P < 0.01, ***P < 0.001).
5. The accumulation of nanoparticles in tumors was limited, which will affect the therapeutic effect.
Reply: I appreciate the reviewer's question. I fully agree with the reviewer. It is well known that all of nanoparticles actually accumulated within tumors mainly based on the EPR effect, and a large number of studies have already demonstrated that less than 10% of total nanoparticles are ultimately enriched in the tumors, which is indeed not sufficient for achieving efficient treatment of tumors. In fact, this problem is one of the huge challenges faced in nanomedicine currently. Thus, researchers are trying to improve the accumulation of nanoparticles in tumors by modifying the surface of nanoparticles with tumor-targeted groups. In our work, the tumor-targeted group--folic acid was decorated onto the surface of ZNNPs to improve the accumulation of nanoprobes in colorectal tumors. Nevertheless, it still far away from the achievement of highly effective tumor treatment. Therefore, our research is mainly for proof of the concept. Hopefully our findings will provide new insight for efficient treatment of colorectal tumors in clinics. 6. What is the detection limitation of the explored nanoprobe for H 2 S? Reply: Thanks for your great comments! To determine the detection limit of ZNNPs for H 2 S, 14 μg/mL of ZNNPs was incubated with various concentrations of NaHS ranging from 0 to 500 μM for 10 min followed by detection of the fluorescence (FL 720 ) and ratiometric photoacoustic (PAS 680 /PAS 900 ) signal. Figure S11c and S11d clearly shows that both FL 720 and PAS 680 /PAS 900 were enhanced linearly with the increase of NaHS concentrations. According to the formula LOD = 3σ/s (where LOD was the limit of detection, σ was the standard deviation of blank probe sample measurement, and s was the slope of the calibration pot, values are mean ± SD (n = 3)), the lowest detection limits of H 2 S in fluorescence and ratiometric PAS 680 /PAS 900 were eventually calculated to be ~4.7 μM and ~0.68 μM, respectively, which again demonstrates that the sensitivity of ratiometric PA is higher than NIR fluorescence imaging in H 2 S detection.
7. Authors also missed the QY of nanoprobes. Reply: Thanks for the reviewer's valuable suggestion! We further measured the fluorescence quantum yields of nanoprobe ZNNPs in PBS buffer (pH 7.4) with and without treatment of NaHS (500 μM). NIR II dye IR-26 (Ф = 0.05%) and Cy5.5 (Φ = 23%) was chosen as the references. As shown in Figure S11a, after calculation, the fluorescence quantum yields (QY) of ZNNPs with and without NaHS are 17.2% and 1.1%, respectively, at the excitation wavelength of 660 nm. Besides, the QY of ZNNPs with and without NaHS are 0.006% and 0.1%, respectively, at the excitation wavelength of 808 nm. with and without incubation of NaHS (500 μM). All the measurements were carried out at room temperature.
8. In Figure S20, there are obvious interference signal from NIR-II imaging and NIR imaging. It seems the specific diagnosis of HCT116 tumor is very poor. And the authors should give the detailed explanation on the interference signal in the activated nanoprobe. Reply: Thanks for the reviewer's question! I agree with you. It is true that there are obvious non-specific fluorescence signals in both NIR and NIR-II imaging of tumor-bearing mice. We speculate that there are two major reasons: first, as mentioned above, nanoparticles mainly accumulated within tumors based on the well-known EPR effect, only very little amount (less than 10%) of nanoparticles was ultimately enriched within the tumor, a lot of nanoparticles were simultaneously uptake in liver and other organs, which inevitably produces non-specific fluorescence signals. Second, many studies have demonstrated that high concentration of endogenous H 2 S (~20 µM) presents in mouse liver and colorectal tumor (0.3 to 3.4 mM). By contrast, the concentration of H 2 S in tumors is much higher than the one in the liver with up to 10-fold. Besides, our purpose in current work is not only intending to develop a promising probe for diagnosis of colorectal tumor, but also realize noninvasive and quantitative visualization of endogenous H 2 S in vivo by taking advantage of the ratiometric PA imaging feature of our probes. Hence, these interference signals in other organs will have little effect on the quantitative visualization of H 2 S in tumors. 3)In Fig. 3 and Fig. 7, please mark the scale bar in fluorescence and PA images.