Rational design of an “all-in-one” phototheranostic

We report here porphodilactol derivatives and their corresponding metal complexes. These systems show promise as “all-in-one” phototheranostics and are predicated on a design strategy that involves controlling the relationship between intersystem crossing (ISC) and photothermal conversion efficiency following photoexcitation. The requisite balance was achieved by tuning the aromaticity of these porphyrinoid derivatives and forming complexes with one of two lanthanide cations, namely Gd3+ and Lu3+. The net result led to a metalloporphodilactol system, Gd-trans-2, with seemingly optimal ISC efficiency, photothermal conversion efficiency and fluorescence properties, as well as good chemical stability. Encapsulation of Gd-trans-2 within mesoporous silica nanoparticles (MSN) allowed its evaluation for tumour diagnosis and therapy. It was found to be effective as an “all-in-one” phototheranostic that allowed for NIR fluorescence/photoacoustic dual-modal imaging while providing an excellent combined PTT/PDT therapeutic efficacy in vitro and in vivo in 4T1-tumour-bearing mice.


Introduction
Theranostics (or theragnostics) are emerging as an attractive alternative to the classic "one medicine ts all" approach to disease management. 1,2 The combination of both diagnostic and therapeutic components into one single system provides a strategy that can image diseased tissue, monitor drug delivery, and evaluate therapeutic efficacy. This provides the ability to tailor treatments to an individual patient (personalised medicine). 3 Owing to the non-invasive, high precision and controllable nature of light, optical-based phototheranostics have garnered increasing attention of late. [4][5][6] In these systems, the near-infrared (NIR) optical window (650-1700 nm) has been frequently targeted in an effort to achieve desirable deep tissue light penetration and high resolution imaging with good sensitivity. 7,8 Previous systems have adopted an "assembly" approach that combines various imaging/therapeutic components into one platform. [9][10][11] Although an attractive strategy, the inherent complexity and the potential for unknown toxicities could serve as impediments to clinical translation. [12][13][14] Single molecule-based phototheranostics with various imaging and therapeutic properties could prove easier to prepare and use. However, such systems are under-explored. [15][16][17][18][19][20] In this study we use a combination of structural tuning and control over excited state energetics to create a porphyrinoid derivative, Gd-trans-2, that shows promise as an "all-in-one" phototheranostic. As detailed below, Gd-trans-2 permits NIR uorescence/PA imaging for tumour diagnostics and provides for PTT/PDT efficacy in vitro and in vivo.
Controlling the excited state features was considered key to the successful design of Gd-trans-2. In general, when a uorophore is promoted to its excited state, there are a number of ways for the energy to dissipate. This includes the emission of light, intersystem crossing (ISC) and non-radiative relaxation. In terms of phototheranostics, these dissipation pathways can result in favourable uorescence (FI), phosphorescence (PI) and photoacoustic imaging (PAI) properties, as well as the ability to promote desirable light-based outcomes, such as photodynamic therapy (PDT) and photothermal therapy (PTT) (Scheme 1a). 21 Each pathway is connected to one another and can be modulated through the population of an appropriate excited state. Therefore, the ability to control the dissipation of excited state energy within a given putative phototheranostic might allow it to be tailored to a specic application and ultimately the benet of individual patients. However, to the best of our knowledge, there are few clear design criteria that allow for productive correlations between molecular structure and the corresponding dissipation of excited state energy. A goal of the present study was to achieve such control as embodied in the preparation of a potentially useful phototheranostic.
A predicative requirement for an effective phototheranostic is high molar absorptivity in the NIR region. Next, an appropriate population of the singlet and triplet excited states is required. The associated balance is typically determined by the extent of ISC, a key process that obeys Fermi's golden rule. 21,22 The efficiency of ISC is dependent upon the spin-orbit coupling (SOC) matrix element (hT i |H SO |S 1 i) and the energy gap between the lowest singlet excited state and the corresponding triplet excited state (DE S-T ). 23 To be effective, phototheranostics generally incorporate useful photodynamic properties. In this context, it is important to consider so-called Type I or Type II pathways for the production of reactive oxygen species (ROS). Type II relies on the production of singlet oxygen. In contrast, Type I activation produces radicals directly and could thus allow for the creation of effective PDT systems that operate in hypoxic environments (e.g., the interior of tumours). 24,25 Bacteriochlorin represent a class of easy to derivatise and naturally occurring tetrapyrrole cofactors with excellent NIR absorptivity 26,27 that have shown potential as PDT agents. 28 However, few have been reported as theranostic agents. 29 Over the years, we have been interested in modulating the aromaticity of these tetrapyrrole cofactors and evaluating their photophysical properties. [30][31][32][33] In this work, we show that it is possible to balance the energy dissipation (BED) between excited states in appropriately functionalised lanthanide(III) bacteriochlorin derivatives as a rational approach to creating "all-in-one" phototheranostics (Scheme 1b). This effort culminated in the preparation of the gadolinium(III) porphodilactol Gd-trans-2, a species that could be incorporated into mesoporous silica nanoparticles (MSN) to generate constructs that permits the uorescence/PA monitoring of PTT/PDT-based treatments of 4T1-tumour-bearing mice.

Results and discussion
Design and synthesis of photosensitisers Porpholactones represent a class of synthetic porphyrinoids, in which a b-pyrrolic double bond is replaced by a lactone moiety. 34 The resulting oxazolone unit affords a porphyrinoidbased platform from which a library of tetrapyrrole-based biomimetics may be generated. 35,36 This ease of derivatisation, coupled with their excellent NIR absorptivity, lead us to consider that porpholactones could be used as the starting point for developing an "all-in-one" phototheranostic.
Theoretical calculations were rst performed on a series of porpholactone and porphyrin derivatives (F 20 TPP, trans-1, F 20 -TPPLac and trans-2; Fig. 1a) using B3LYP/6-311G(d) to estimate the highest occupied molecular orbital (HOMO, H) and lowest unoccupied molecular orbital (LUMO, L) distributions and the DE S-T values (see details in the ESI †). In addition, the aromatic character of these porphyrinoids was estimated by calculating the corresponding isotropic nucleus independent chemical shi (NICS(1)) values. As shown in Fig. 1a, the order of aromaticity is F 20 TPP > trans-1 > F 20 TPPLac > trans-2 as inferred from the NICS(1) calculations. The pyrrolic N-H proton shis determined by 1 H NMR spectroscopy proved consistent with this trend (Fig. S2 †).
On the basis of our calculations and consistent with previous observations, 37 the carbonyl units were found to be in conjugation with the macrocycle. As a result, reduction of the lactone unit of, e.g., trans-1 to its corresponding b-hydroxyl derivative, trans-2, leads to a strong decrease in the aromaticity. 38 This reduction in aromaticity serves to narrow the H-L gap and red-shi the lowest energy absorption. On this basis trans-2 was considered attractive as a potential phototheranostic. Unfortunately, however, the reductive conversion of trans-1 to trans-2 leads to an increase in the calculated DE S-T gap from 0.84 to 1.50 eV. Such an increase was not expected to favour efficient ISC as needed for effective PDT.
To improve the PDT properties of trans-2, formation of various metal complexes was considered. The inclusion of metal ions into chromophores has previously been shown to enlarge the SOC matrix element and enhance the efficiency of ISC. 39 Heavy metals, such as Au, Ir, and Pt, have large SOC constants (z > 3000 cm À1 ). This leads to uorescence-based energy dissipation pathways being blocked. The net result is effective PDT behaviour coming at the cost of the optical features considered desirable for a phototheranostic. 39,40 Therefore, lanthanide ions, Gd 3+ and Lu 3+ are oen used to create porphyrinoid complexes since they have z values 39 that are expected to preclude uorescence quenching while enhancing the ISC efficiency. 41 Based on these considerations, the Gd 3+ and Lu 3+ complexes of trans-2 were prepared. While regioisomerically pure to the limits of our analysis, it is to be noted that these complexes are a mixture of stereoisomers. They were studied as such.

Photophysical properties
The absorption and uorescence emission spectra of the compounds included in the present study were recorded in dichloromethane (DCM) (Fig. 2). Corresponding photophysical data are included in Table 1. It was found that trans-2 displayed bacteriochlorin-type absorption features, including split Soret bands at 300-400 nm, a weak Q x band absorption at 500-550 nm, and an intense Q y band at 650-750 nm. The corresponding Gd 3+ , Lu 3+ and Zn 2+ complexes displayed red-shied (ca. 30 nm) Q y (0,0) bands (l max ¼ 756 nm). Gd-cis-2 displayed a blue-shied (by 23 nm) Q y (0, 0) absorption band (l max ¼ 738 nm) and a lower molar absorption coefficient 3 ¼ 9.8 Â 10 4 M À1 cm À1 relative to the trans-isomer (l max ¼ 756 nm, 3 ¼ 1.5 Â 10 5 M À1 cm À1 ). This was believed to be the result of a regioisomeric effect seen previously in the case of cis/trans-1. 30 The photophysical properties of Gdtrans-2 and Gd-trans-3 proved very similar (Fig. 2).
Each compound associated with the present study, namely trans-2, and Zn-, Gd-, and Lu-trans-2, Gd-cis-2, and Gd-trans-3, was found to uoresce in the $700-900 nm NIR region with decay lifetimes of 1.1-2.5 ns in DCM ( Fig. 2 and Table 1). No  phosphorescence was detected either in degassed solution or at 77 K. The uorescence quantum yield of the metal-free ligand trans-2 was 5.9%, which was the highest within the set. Upon photoexcitation (l ex ¼ 395 nm) trans-2 emits with maxima at 733 and 820 nm and is characterised by a uorescence lifetime of 2.5 ns (Fig. S3 †). In contrast, the corresponding metal complexes were found to display uorescence emission features that were red-shied by 30 nm relative to trans-2 and quantum yields in the range of 2.0 to 4.2%. The increase in quantum yield values were seen to correlate with a decrease in the metal SOC constants (z), an increase in the DE S-T gap and introduction of a b-hydroxyl group. As shown in Table 1, the larger SOC constants for the Gd 3+ and Lu 3+ ions (z s ¼ 1653 and 1151 cm À1 , respectively) compared to Zn 2+ (z ¼ 390 cm À1 ) 17 resulted in porphyrinoids with lower uorescence quantum yields (F FL 2.0 and 2.9%) relative to the Zn 2+ complex (F FL 4.2%). Traditionally, Lu 3+ and Gd 3+ porphyrin-or porphodilactone-based complexes emit phosphorescence rather than uorescence. 48 In contrast, the present trans-2 species were found to be uorescent. This contrasting emission behaviour serves to underscore the importance of the DE S-T gap (a function of porphyrinoid ligand aromaticity) and the SOC constants of each metal ion. Notably, Gd-trans-3 displayed an increased lifetime (1.4 ns) and an enhanced uorescence emission quantum yield (i.e., F FL -1.6fold greater) relative to Gd-trans-2. This nding is rationalised in terms of high-energy O-H vibrations in Gd-trans-2 that serve to quench slightly the excited singlet state.

Evaluation of ROS generation and the photothermal effect
To determine the applicability of the present metal complexes as PDT photosensitisers (PSs), we rst examined their ability to generate 1 O 2 upon light irradiation (760 nm, CHCl 3 ). The 1 O 2 quantum yields (F D s) were determined through the emission of 1 O 2 at $1270 nm ( 1 D g / 3 S g transition of 1 O 2 ). Tetraphenylporphyrin (H 2 TPP, F D ¼ 0.55 in CHCl 3 (ref. 46)) was used a reference. 47 As shown in Table 1, Fig. 3a and S4, † each metal complex was shown to have an approximately 20-40% higher F D s than the free ligand (trans-2). This nding is consistent with the design expectation, namely that metal complexation facilitates population of the triplet excited state, which in turn is able to transfer energy to O 2 via a Type II pathway to produce 1 O 2 . In previous research, we found that Gd-based porphodilactones have large HOMO-LUMO energy gaps, which results in excellent 1 O 2 production (up to 95% quantum yield). 48 However, these prior systems lack the capacity to generate other ROS (e.g., through a Type I pathway).
As noted above, Type I ROS production is highly desirable, particularly under hypoxic conditions. Therefore, we turned our attention towards testing whether other ROS, such as the superoxide anion, O 2 c À and the hydroxyl radical, HOc, could be   This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 8204-8213 | 8207 generated from trans-2 and its complexes. This was done by using hydroethidine (DHE) as a uorescent probe for O 2 c À (Fig. 3b), 49 and 5-tert-butoxycarbonyl-5-methyl-1-pyrroline-Noxide (BMPO) as a spin trap for HOc (Fig. 3c) with monitoring by means of electron paramagnetic resonance (EPR) spectroscopy. 50 As these detection protocols required the use of aqueous media, we encapsulated each compound into mesoporous silica nanoparticles (MSN) (Scheme S2 and Fig. S4-S6 †). 51 This MSN strategy was shown to provide formulations that were welldispersed in aqueous media, characterised by excellent biocompatibility, and photophysical properties that were basically unchanged relative to the free systems (Fig. S7 †). The introduction of the lactol unit was found to promote the production of Type I ROS. The O 2 c À quantum yields of MSN-Luand Gd-trans-2 were calculated to be 23% and 21%, respectively (methylene blue was used a reference) (Fig. 3b). 52 A remarkable 6-to 7-fold increase in O 2 c À generation was observed compared to MSN-Gd-trans-3 or free base trans-2 (3.6% and 3.1%). Similar results were obtained for HOc generation (Fig. 3c). Next, we used the 1 O 2 scavenger, sodium azide (NaN 3 ) to exclude any potential conversion of 1 O 2 to other ROS, during these experiments. These studies revealed that the production of HOc is unaffected by and independent of 1 O 2 formation (Fig. S8 †). This was taken as further evidence of ROS being produced via a Type I pathway.
It is important to appreciate that in addition to producing ROS via a Type I process, both MSN-Gd-trans-2 and Gd-trans-3 are able to promote the production of 1 O 2 via a Type II pathway (F D s $ 30%, Fig. 3a). Previous studies that have sought to tune the ratio between Type I and Type II pathways have resulted in signicant changes to the electronic structures as well as photophysical properties of the photosensitiser. [53][54][55] This has made optimisation of various putative phototheranostics difficult. Gratifyingly, the lanthanide complexes of this study appear to fall in an appropriate "sweet spot".
Similar photophysical properties were seen for MSN-Lutrans-2 and MSN-Gd-trans-2. Preliminary experiments revealed both Zn-trans-2 and Lu-trans-2 suffered from poor stability in solution (dichloromethane), whereas good stability was observed for the corresponding Gd 3+ construct (Gd-trans-2) (Fig. S9 †). Additionally, while not a focus of the present study, the paramagnetic Gd 3+ centre present in MSN-Gd-trans-2 might make this system of interest as a possible MRI contrast agent. Therefore, MSN-Gd-trans-2 was chosen for further in-depth study with an emphasis being placed on understanding the role, if any, of the b-hydroxyl subunit.
Our working hypothesis is that the presence of the free bhydroxyl unit accelerates the electron transfer processes critical to Type I reactions between ground-state oxygen and the Gdtrans-2 excited state(s). To test this hypothesis, nanosecond transient absorption spectroscopic studies of both Gd-trans-2 and its O-methylated derivative, Gd-trans-3, were carried out. Each compound was excited using a 355 nm laser in deaerated toluene. As shown in the difference spectra, the triplet state features include maxima at 370, 470, and 570 nm and groundstate bleaching of the Soret-and Q-band absorptions at 340, 400, 700 and 760 nm (Fig. 2e). The triplet state of Gd-trans-2 was shown to decay via a double-exponential, with lifetimes of 0.37 ms (68%) and 2.3 ms (32%), respectively. Similar analyses were then carried out in the presence of air. Under these latter aerobic conditions, Gd-trans-2 displayed a faster decay characterised by a double-exponential function (lifetimes ¼ 0.35 ms (67%) and 0.75 ms (33%), respectively, Fig. 2f).
While not a proof, such a nding is consistent with an equilibrium between two distinct excited states where one (ca. 33%) is sensitive to oxygen (2.3 ms / 0.75 ms) and the other not. From these experiments, the oxygen quenching rate constant (k q ) and quenching efficiency (F q ) were estimated to be 0.69 Â 10 9 M À1 s À1 and 69%, respectively. 45 Similar studies were carried out with Gd-trans-3 yielding k q and F q values of 0.11 Â 10 9 M À1 s À1 and 31%, respectively (Fig. S10 †). The difference in the decay kinetics for Gd-trans-2 and Gd-trans-3 provides support for the notion that the b-hydroxyl group plays a role in mediating the oxygen-induced deactivation of the triplet excited states. Previously, it has been shown that enzymes and their rationally designed synthetic models oen incorporate hydroxyl groups into their structures to accelerate electron transfer processes via proton coupled electron transfer (PCET) mechanisms. 56 On this basis, we suggest the role of the b-hydroxyl unit is to promote Type I ROS generation via PCET between the triplet excited state(s) of the metalloporphodilactol Gd-trans-2 and molecular oxygen.
The paramagnetic properties of Gd 3+ cations have been extensively exploited for the development of T 1 -contrast MRI agents. 57 As a result, we determined the longitudinal relaxivity (r 1 ) value for MSN-Gd-trans-2. As shown in Fig. 4a, r 1 was measured to be 2.95 mM À1 s À1 ; moreover, the contrast in phantom MR images of MSN-Gd-trans-2 as an aqueous dispersion increased with increasing concentration. In spite of these intriguing ndings, a decision was made to focus the present study on exploring the phototheranostic potential of MSN-Gd-trans-2. Recently, molecular systems with excited states that undergo effective nonradiative (NR) decay have been explored as photoacoustic imaging (PAI) and photothermal therapy (PTT) agents. 58,59 Both MSN-Gd-trans-2 and MSN-Gd-trans-3 were found to be easily dispersed in water. Photo-irradiation (760 nm, 6 minutes, 100 mW cm À2 ) of the resulting solutions was found to increase the temperature by $12 C with a photothermal conversion efficiency (h) of ca. 30% (Fig. 4b). 60 No signicant difference between the Gd-trans-2 and Gd-trans-3 MSN formulations was observed, reecting energy dissipation pathways that are not affected by the b-hydroxyl unit. In comparison a smaller increase in the solution temperature was seen for the free ligand MSN-trans-2 (7.0 C, h ¼ 23%). The better performance seen for the Gd 3+ complexes is believed to reect a heavy atom effect, which enhances nonradiative conversion thereby improving the photothermal performance. 61 Lastly, as seen in Fig. 4c, IR thermal images were produced aer laser irradiation (760 nm, 5 min, 100 mW cm À2 ), a result underscoring the potential of metalloporphodilactols, such as Gd-trans-2 and Gd-trans-3, for IR photothermal imaging as well as PTT applications. Overall, these results provide support for the contention that Gd-trans-2 and Gd-trans-3 could serve as "all-in-one" phototheranostics capable of supporting NIR uorescence/PA imaging and PTT/ PDT-based therapies.

In vitro experiments
To evaluate the potential of Gd-trans-2 and Gd-trans-3 to function as phototheranostics in biological settings, the "dark" toxicity and phototherapeutic efficacy of the corresponding MSN formulations were evaluated using a breast cancer 4T1 cell line with a Cell Counting Kit-8 (CCK-8). In the absence of light, a negligible cytotoxic effect was observed (Fig. S11 †). In contrast, under conditions of photo-irradiation (760 nm, 7.5 mW cm À2 , 30 min), a dose-dependent therapeutic effect was seen with MSN-Gd-trans-2 outperforming both the free ligand MSN-trans-2 and the O-methylated derivative MSN-Gd-trans-3 as reected in IC 50 values of 4.1 AE 0.2, 13.4 AE 0.4 and 8.1 AE 0.4 mM, respectively (Fig. 5b). Similar studies were then carried out under lower oxygen tension. A decrease in photocytotoxicity was observed as expected for a process wherein reaction with oxygen contributes to the overall phototoxicity (Fig. S12 †).
The cellular localisation of a therapeutic oen determines its efficacy and selectivity. 24 Therefore the localisation of MSN-Gdtrans-2 and several control systems within living cells was probed using confocal uorescence microscopy. As shown in Fig. 5a, an intense uorescence signal from MSN-Gd-trans-2 was observed. Co-localisation experiments using commercial LysoTracker® Green (Pearson's coefficient ¼ 0.74) provided support for the accumulation of MSN-Gd-trans-2 in lysosomes. Both MSN-trans-2, and MSN-Gd-trans-3 were shown to have This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 8204-8213 | 8209 similar Pearson's coefficients > 0.70 with LysoTracker® Green (Fig. S13 †).
Subsequently, the light-mediated generation of various ROS, namely total ROS, 1 O 2 and O 2 c À , was determined in vitro by confocal uorescence microscopy using 2 0 ,7 0 -dichlorodihydro-uorescein diacetate (H2DCFDA), a commercially available Singlet Oxygen Sensor Green (SOSG) assay, and DHE as uorescent markers for these ROS, respectively. Upon photoirradiation (760 nm, 7.5 mW cm À2 , 30 min), 4T1 cells treated with MSN-Gd-trans-2 (Fig. 5c, column C) produced obvious green (lane I), green (lane II) and orange (lane III) uorescence responses, respectively. This was taken as evidence for the production of each corresponding ROS. This inference was further conrmed by various control experiments where a minimal increase in the uorescent intensity was observed (cf., e.g., column A: photo-irradiation in the absence of Gd-trans-2).
A sharp contrast was seen between MSN-Gd-trans-2 and MSN-Gd-trans-1 in terms of their respective Type I and II ROS production. For instance, only a strong uorescence intensity was observed for 1 O 2 production mediated by MSN-Gd-trans-1 (Fig. 5c, Column E). In contrast, MSN-Gd-trans-3 and MSN-trans-2 displayed a much smaller increase in the uorescence ascribed to the production of O 2 c À but produced a 1 O 2 signal comparable to that of MSN-Gd-trans-2 (Columns D and B). For each compound, we estimated the generation of total ROS, 1 O 2 and O 2 c À from the ratio of uorescence intensity of the putative photosensitiser vs. the corresponding blank (Column A). As shown in Fig. 5d (quantied using the images in Fig. 5c), similar 1 O 2 generation was observed for all compounds. However, overall, a greater level of total intracellular ROS generation was seen for MSN-Gd-trans-2 relative to the other agents considered in this study. This nding thus mirrors what was seen in the predicative aqueous solution studies discussed above (Fig. 3,  supra).
To conrm that ROS generation is essential for the phototherapeutic efficacy seen in the case of MSN-Gd-trans-2, we used sodium azide (NaN 3 ), Mn(II)-tetrakis-(4-N-methylpyridiniumyl) porphyrin (MnTPyP), dimethyl sulfoxide (DMSO) or sodium pyruvate (NaP) as ROS scavengers for 1 O 2 , O 2 c À , HOc and H 2 O 2 species, respectively. These ROS scavengers were found to inhibit only partially the photocytotoxicity (Fig. 5e). This was taken as evidence that along with both Type I and II PDT, PTT effects contribute to the overall phototherapeutic efficacy seen for MSN-Gd-trans-2.
Further and more quantitative analyses of the therapeutic effect of MSN-Gd-trans-2 were made via ow cytometry using a commercially available Annexin V Apoptosis Detection Kit. A relatively high percentage of apoptotic cells, ca. 40-50%, was seen aer photo-irradiation (4 mM, 760 nm, 7.5 mW cm À2 , 30 min), whereas fewer than 1% of apoptotic cells were observed under "dark" conditions or in the absence of MSN-Gd-trans-2 (Fig. S14 †). The activity of the apoptosis related biomarkers caspase-3 and poly ADP-ribose polymerase (PARP) were determined by western blot (Fig. S15 †). The ratio of cleaved caspase-3 and PARP to b-actin increased by 1.8 and 4.7-fold in comparison to that of the dark control group (p < 0.01). These results are interpreted in terms of the induction of an apoptosis pathway during MSN-Gd-trans-2-mediated phototherapy.

In vivo experiments
We next turned our attention to evaluating the potential of MSN-Gd-trans-2 to serve as a phototheranostic in vivo. Recently, increasing attention has been devoted to multimodal imaging. 62,63 Therefore, MSN-Gd-trans-2 was rst evaluated for its ability to be used as multimodal imaging agent in vivo. This would generate the information required for tumour diagnosis and identication of the optimal time point post-injection to carry out light-based therapy. NIR uorescence imaging was performed by recording the uorescence intensity (780 nm) at various time points aer MSN-Gd-trans-2 was administered via intravenous injection (1 mg kg À1 , tail vein) in 4T1-tumourbearing mice (Fig. 6a). A remarkable increase in the uorescence intensity was observed at the tumour. A maximum value was seen ca. 24 h post-injection, as would be expected for an agent that accumulates gradually at the tumour site. To support this observation, the biodistribution of MSN-Gd-trans-2 was analysed 24 h post-injection by means of ex vivo NIR uorescence imaging. As can be seen from an inspection of Fig. 6b, the heart, lung, liver, spleen and kidney produced weak uorescence signals, whereas a high uorescence intensity was seen for the tumour. Similar results NIR imaging results were obtained in the case of MSN-Gd-trans-3 and MSN-trans-2. This observation was supported by ICP-MS analyses; cf. Fig. S16-S18. † We next examined the capability of MSN-Gd-trans-2 to produce an in vivo PA signal in 4T1-tumour-bearing mice. As shown in Fig. 6c and S19, † the PA signal was shown to be enhanced in a statistically signicant fashion 24 h postinjection compared to pre-injection. In contrast, the control group (PBS only), showed no apparent change in the PA signal (Fig. S20 †). These ndings provide a complement to the uorescence data discussed above and support the conclusion that MSN-Gd-trans-2 localises well in tumours. Considered in concert with the other imaging studies, these results also serve to underscore the fact that MSN-Gd-trans-2 could have a role to play in multimodal tumour imaging.
The in vivo phototherapeutic efficacy of MSN-Gd-trans-2 was then explored using a 4T1-tumour bearing mouse model. As a rst step in this evaluation, IR thermal images at the tumour site were recorded following treatment with MSN-Gd-trans-2, MSN-trans-2, and MSN-Gd-trans-3, respectively, under conditions of photo-irradiation for 0, 1, 3, 5 min (1 mg kg À1 PS, 760 nm, 100 mW cm À2 ; Fig. 7a). A rapid temperature rise (22-32 C) at the tumour site was seen for MSN-Gd-trans-2, MSN-trans-2, and MSN-Gd-trans-3 (Fig. S21 †). In contrast, a negligible temperature (DT < 1 C) rise was observed in the case of the PBS control group and only modest photothermal conversion efficiency was seen for the free ligand MSN-trans-2 (DT ¼ 4 C). Again, these ndings serve to highlight the importance of metal complexation for achieving a robust PTT effect.
The anti-tumour PDT/PTT efficacy of each PS was then evaluated in 4T1 tumour-bearing mice. A total of 42 mice were used. They were divided into 6 groups: (1) PBS and dark, (2) MSN-Gd-trans-2 and dark, (3) PBS + light, (4) MSN-trans-2 + light, (5) MSN-Gd-trans-2 + light, (6) MSN-Gd-trans-3 + light. Based on the localisation studies presented in Fig. 6, the animals in Group 3-6 were subjected to photo-irradiation at the tumour site 24 h post-injection (1 mg kg À1 PS, 760 nm laser, 100 mW cm À2 , 5 min). The size of the tumours and body weights of the mice in each group were monitored twice weekly. As shown in Fig. 7b, rapid tumour growth was observed for the PBStreated-group, as well as the groups treated with one of the test photosensitisers but without subjecting to photoirradiation. In contrast, under conditions of photo-irradiation MSN-Gd-trans-2 was shown effective for the inhibition of tumour growth. In the case of MSN-trans-2, mice treated with light were found to have the tumour growth partially inhibited, particularly for the rst 1-2 weeks; however, rapid tumour regrowth was subsequently observed. Suppression of tumour growth was seen for the group treated with MSN-Gd-trans-3 and subject to photo-irradiation, although the benet was less than that seen in the case of MSN-Gd-trans-2 (Fig. 7c). These results thus recapitulate the relative efficacy for these two species seen in vitro (Fig. 5, supra). Since MSN-Gd-trans-2 and MSN-Gd-trans-3 are characterized by similar F D values and give rise to similar PTT effects, we conclude that a Type I PDT effect plays an important role in mediating the overall phototherapeutic efficacy.
No evidence of body weight loss or other abnormalities was seen in the case of all groups, leading us to conclude that the drug-loaded MSN formulations may be subject to minimal side effects (Fig. S22 †). In addition, the uorescence signal gradually diminished over time, which may reect the eventual clearance of MSN-Gd-trans-2 or its metabolism to less-emissive species (Fig. 7d and S22 †). Therefore, in order to understand further the therapeutic effect and evaluate any potential toxicity of each PSs, the tumour tissues and major organs of each mouse group (heart, liver, spleen, lung and kidney) were isolated and subject to hematoxylin and eosin (H&E) staining (Fig. S23 †). Notably, evidence of necrosis and apoptosis was observed in tumours treated with a combination of PS and laser irradiation, while those in the PBS or dark groups showed little change. Slices of organs revealed large and spindle shape nuclei in the case of all groups, leading us to conclude that treatment with MSN-Gdtrans-2 engenders no obvious toxicity in the major organs.
As a test of whether the diagnostic components of phototheranostic MSN-Gd-trans-2 might allow the therapeutic outcome of PTT/PDT-based treatments to be monitored, in vivo uorescence images were obtained on days 1, 7, 14, 21, and 28 following a single injection on day 0 of MSN-Gd-trans-2. Similar monitoring was carried out in the case of a control injection with PBS. As shown in Fig. 7d, over the course of the phototherapy process, the NIR uorescence signal gradually decreased (l ex ¼ 740 nm, l em ¼ 780 nm). On this basis we propose that MSN-Gd-trans-2 acts as a stand-alone phototheranostic that shows promise of both treating and monitoring anti-tumour effects in vivo. It is appreciated, however, that optimisation of the treatment protocol, such as exploring the use of multiple injections spread out of multiple days, may be l ex ¼ 760 nm, 100 mW cm À2 , 5 min) (b) tumour growth profiles for different mice groups (**p < 0.01). (c) Standard Kaplan-Meier curves for different mice groups following treatment. Blue and red arrow represents the injection (0) and irradiation (1 st ) day, respectively; "+" and "À" denote with and without photo-irradiation (l ex ¼ 760 nm, 100 mW cm À2 , 5 min), respectively. (d) In vivo NIR fluorescence images recorded on days 1, 7, 14, 21, and 28 day post-injection. Column A and B reflect treatment with PBS (Group 3) and MSN-Gd-trans-2 (Group 7), respectively. Animals were injected intravenously once on day 0, tail vein; l ex ¼ 740 nm, l em ¼ 780 nm.
This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 8204-8213 | 8211 necessary to maximize the diagnostic and therapeutic utility of MSN-Gd-trans-2.

Conclusions
Porphodilactol derivatives and their corresponding metal complexes have been explored as "all-in-one" phototheranostics. To create an effective system, namely MSN-Gdtrans-2, we felt it was important to balance the inherent photophysical properties with the ISC efficiency. Theoretical calculations, combined with photophysical measurements, revealed that a decrease in aromaticity serves to reduce the HOMO-LUMO energy gap and cause a red-shi in the absorption maximum. Unfortunately, this decrease in aromaticity leads to an increase in DE S-T gap, which reduces the overall ISC efficiency. However, this latter limitation may be overcome by means of appropriate metal complexation. In the case of the porphodilactols of the present study, this could be achieved by formation of a Lu 3+ or Gd 3+ complex. The latter system displayed slightly more favourable photophysical features and proved more stable and was thus studied in depth. It was found that free b-hydroxyl units were essential for promoting Type I ROS production, a desired feature for the development of effective PDT-based agents capable of operating at lower oxygen tensions. A biological evaluation of the metalloporphodilactol Gd-trans-2 as a phototheranostic was carried out aer encapsulation in mesoporous silica nanoparticles. The resulting construct displayed excellent uorescence and photoacoustic imaging capabilities. It also showed good PTT/PDT efficiency both in vitro and in vivo in 4T1-tumour-bearing mice. We thus propose that MSN-Gd-trans-2 warrants further study as an "allin-one" phototheranostic. Indeed, efforts are underway to netune the administration regimes so as to exploit fully the therapeutic and diagnostic potential of this and other metalloporphodilactols.

Ethical statement
All the animal experiments were performed in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the Peking University First Hospital (Beijing, China).

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