Upconverting Nanoparticle‐Loaded Microneedles for Near‐Infrared Responsive Delivery of Gonadotropins to Increase Success of In Vitro Fertilization

Abstract

In vitro fertilization (IVF) requires repeated bolus injections of gonadotropins at precise times, making it a painful and stressful experience for women. Polymeric microneedles (MNs) present a painless alternative to hypodermic needles, but controlling the release of drugs through MNs remains a challenge. Here, we report the development of a hydrogel MN patch integrated with upconverting nanoparticles coated with light‐breakable mesoporous silica (UCNP@LBMS) that releases the IVF gonadotropin, leuprolide, in response to near‐infrared (NIR) light, while retaining the UCNPs within the MN patch. In rats, a 5 min NIR pulse of 2 W cm⁻² results in a significant increase in leuprolide release in both skin and plasma, with MN length playing an important role in drug bioavailability. Beyond IVF, UCNP‐integrated MN patches offer a painless, automated alternative in any application that requires daily, repeated, time‐sensitive hypodermic needle injections, such as diabetes or multiple sclerosis.

This paper describes the development of an upconverting particle‐integrated microneedle patch that can release the hormone leuprolide when the patch is exposed to near‐infrared light. This system represents proof‐of‐concept of a painless and automated method of delivering drugs at specific times when repeated time‐sensitive injections are required, such as with in vitro fertilization.

1. Introduction

One in every 6 couples struggles with infertility,^{[ 1 ]} a number that is expected to grow due to rising maternal age at first birth.^{[ 2 ]} As a result, many are turning to in vitro fertilization (IVF) as an alternate means of reproduction. The IVF process consists of controlled ovarian hyperstimulation (COH), during which women inject themselves daily with gonadotropins for 2 weeks to increase the number of oocytes available, before the oocytes are retrieved, fertilized, and implanted in the patient.^{[ 3 ]} The COH phase, in particular, imposes a great physical and psychological burden on women due to its time‐sensitive nature and use of hypodermic needles for subcutaneous injections.^{[ 2 ]} In fact, the burden of COH is cited as the primary reason for withdrawal from treatment.^{[ 4 ]} In addition, the possible variability in timing or dosage with hypodermic needle self‐administration could severely hinder the success rate of the procedure, further increasing the stress associated with a treatment priced at $15,000 ‐ $20,000 CAD/cycle.^{[ 5 ]}

Painless transdermal microneedles (MNs) deliver drugs in a minimally invasive manner by penetrating the stratum corneum without activating the pain response from nerves located deeper in the dermis.^{[ 6} , ^{7 ]} Of the materials used to manufacture MNs (e.g., glass, silicon), certain polymers are preferred as they elicit no immunogenic response, degrade into non‐toxic products in a physiological environment, and have sufficient mechanical strength to penetrate skin.^{[ 8} , ⁹ , ^{10 ]} Typically, polymeric MNs rapidly dissolve into the skin within minutes, releasing their payload in a bolus.^{[ 11} , ¹² , ^{13 ]} Tailoring the type of polymer used can delay this release over a month^{[ 14 ]} and layering polymers with different degradation times can provide pulsatile or delayed burst release.^{[ 15 ]}

However, few systems can release drugs at a specific time, as required for the time‐sensitive delivery of IVF hormones.^{[ 9} , ^{16 ]} For example, one such system features UV‐cleavable drug conjugates that are crosslinked within polymeric MNs to release the drug upon UV radiation. However, the use of UV light, which is carcinogenic, precludes it from being used for biomedical applications.^{[ 17 ]} An improvement on this system features the inclusion of silica‐coated lanthanum hexaboride (LaB₆@SiO₂) nanoparticles within polycaprolactone (PCL) MNs. When exposed to a near‐infrared (NIR) laser, the light is absorbed by the LaB₆@SiO₂ nanoparticles and converted to heat, causing the PCL microneedles to melt at 50 °C and release the drug into the skin.^{[ 18 ]} While this system allows for precise control over the timing of the drug release with the use of non‐carcinogenic NIR light, the dissolving nature of the MNs necessarily releases the nanoparticles into the skin. Although these nanoparticles were cleared from the skin into the lymphatic system, their long term fate within the lymphatic system remains unclear.^{[ 18} , ^{19 ]} In fact, the main barrier to the clinical translation of nanoparticle‐based drug delivery systems is their unknown interaction with biological targets within the human body, as this interaction changes based on various physicochemical characteristics (e.g., size, charge) of each nanoparticle^{[ 20 ]} as well as tissue types.^{[ 21} , ^{22 ]} The other disadvantage of using dissolving MNs is that the release of a drug bolus can only be triggered at one point in time, a limitation for applications that require several doses over time.

One way to prevent the release of nanoparticles into the body is to retain them within a hydrogel network, which still allows drugs to flow through. This can be accomplished with MNs made of hydrogel‐forming polymers that can swell in the skin, such as cross‐linked poly(methyl vinyl ether/maleic acid) (PMVE/MA).^{[ 13} , ²³ , ^{24 ]} When dry, these hydrogel MNs have sufficient mechanical strength to penetrate skin^{[ 25} , ^{26 ]} and slowly release drugs as they swell upon uptake of interstitial fluid.^{[ 27} , ²⁸ , ^{29 ]}

Thus, to address the many challenges associated with the COH phase of IVF, this study introduces a hydrogel MN patch integrated with lanthanide‐doped upconverting nanoparticles (UCNPs) coated with a light‐breakable mesoporous silica shell (LBMS), henceforth called UCNP@LBMS/MN, for the light‐triggered release of the gonadotropin agonist, leuprolide. UCNPs convert NIR irradiation into high‐energy photons in the UV–vis region via a multiphoton process known as upconversion.^{[ 30} , ^{31 ]} When combined with a photolabile drug carrier, they produce a system wherein drugs can be released on demand using NIR irradiation.^{[ 32} , ³³ , ^{34 ]} Our group recently showed that UCNP@LBMS can release the vitamin D3 precursor, 7‐dehydrocholesterol, on demand, when exposed to NIR; the NIR radiation is upconverted by the UCNPs into the UV light required to break the nitrobenzyl chains within the MS shell, triggering the release of the drug loaded within the mesopores.^{[ 35 ]}

Instead of painful self‐administered hormonal injections, we envision this UCNP@LBMS/MN patch being applied to the skin and releasing a controlled bolus of leuprolide upon NIR excitation. As shown in Figure 1 , the NIR light is absorbed by the UCNPs and upconverted to the UV light used to break apart the UV‐responsive mesoporous silica structure. The hormone loaded within the mesopores is then released into the hydrogel, flowing through the swollen pores into the skin, while the UCNPs are retained within the MNs. Unlike dissolving MNs that release all the drug at one time point, the hydrogel MNs retain their structure and could be used for repeated dosing given a NIR pulse automated to occur repeatedly at a specific point in the day. This would eliminate the stress of remembering to inject oneself at a certain time while decreasing human error that could negatively impact IVF success rates.

Figure 1.

Schematic illustration of UCNP@LBMS/MN for the light‐controlled release of Cy5‐leuprolide.

Our results show that UCNP@LBMS can be embedded into MNs at various concentrations without hindering drug flow and that the resulting system responds to an NIR trigger to release both a model drug, the fluorescent dye cyanine‐5 (Cy5) conjugated to polyethylene glycol (PEG), and Cy5‐leuprolide in both ex vivo and in vivo models. The controlled timed delivery of a clinically relevant agonist, leuprolide, without the release of UCNPs into the skin demonstrates the translational potential of our system for IVF treatment.

2. Results

2.1. Synthesis, Characterization, and Optimization of UCNP@MS/MNs

2.1.1. Synthesis of UCNP@MS/MN and Verification of Successful UCNP@MS Integration

We synthesized LiYbF₄:Tm³⁺@LiYF₄ UCNPs (Figure 2a), UCNP@MS (Figure S12a, Supporting Information), and UCNP@LBMS (Figure 2b) using a previously reported protocol^{[ 35 ]} and confirmed that the upconversion emission spectra and UV–vis absorption spectra of UCNP@MS and UCNP@LBMS matched what was previously reported (Figure S12b,c, Supporting Information).^{[ 35 ]} We also measured the hydrodynamic diameters of UCNP@MS and UCNP@LBMS using Dynamic Light Scattering as 160 ± 1 and 240 ± 20 nm, respectively (Figure S12d, Supporting Information). These values are slightly larger than the diameters measured using TEM (67 ± 2 and 67 ± 4, respectively), as commonly seen when comparing hydrodynamic diameters to the dried TEM sample, due to additional factors captured when measuring particles in solution, such as particle shape and charge, solvent interactions, and size of hydration shell.^{[ 36 ]} Furthermore, as UCNP@LBMS are likely to have fewer SiO⁻groups than UCNP@MS and thus are less negatively charged, they may be more prone to aggregation, which may explain the larger hydrodynamic diameter observed.

Figure 2.

TEM images of (a) UCNP and (b) UCNP@LBMS. c) Bright field and (d) fluorescence confocal microscopy images (λ_ex = 980 nm) of 1 UCNP@MS/MN tip. e) Upconversion emission spectra as read from bright spots seen in (d). f) SEM image of UCNP@MS/MN. g) Average force required to compress UCNP@MS/MNs by 0.4 mm as measured from mechanical compression tests (n = 10). Statistical significance was determined using a one‐way ANOVA. h) Average needle height before (solid bars) and after (striped bars) compression testing (n = 40). 10 tips from the front row were measured per MN patch and 4 MN patches were measured for each UCNP concentration. Statistical significance for each UCNP concentration was determined using an unpaired t‐test. i) Fluorescence confocal microscopy image (λ_ex = 980 nm) of 1 UCNP@MS/MN tip using layered technique.

As the LB linker required multiple organic synthesis steps, we first performed all characterization with UCNP@MS/MNs to ensure that the UCNPs were successfully integrated within the MNs and to determine if there was a limit to the concentration of UCNPs added before compromising the MN properties. After this concentration was determined, we then loaded UCNP@LBMS into the MNs to test system functionality.

To demonstrate that the UCNP@MS were successfully integrated within the MNs, we cast MN hydrogel mixtures with and without UCNP@MS (1 mg MN⁻¹) into 10 × 10 PDMS MN molds with a MN height of 500 µm, and after cross‐linking in an oven for 48 h, removed the MNs from their molds to form MN tips with a square pyramidal shape. Upon removal from the molds, the height of the MNs ranged from 340 ± 26 to 400 ± 39 µm, as shown in Figure 2h (variation between MN patches shown in Figure S13, Supporting Information); a slight reduction in MN height compared to mold height has been previously reported.^{[ 37} , ^{38 ]} Then, we cut the MN tips using a focused ion beam (FIB) to image the tip cross‐sections with high‐resolution SEM. The UCNP@MS/MN cross‐sections revealed the presence of ≈60 nm particles, which were not present in the MNs with no UCNPs (Figure S14, Supporting Information), suggesting successful integration of UCNP@MS within the MNs.

To visualize the hydrated pore structure of the hydrogel, we immersed the UCNP@MS/MNs in water for 5 min, cut the swollen MN tips with a razor blade to reveal cross‐sections of the hydrogel, then lyophilized and imaged the tip cross‐sections via SEM (Figure S15a, Supporting Information). The UCNP@MS/MN hydrogel cross‐sections after swelling similarly showed the presence of UCNP@MS aggregates that were not present on the smooth surface of the control MNs (Figure S15b,c, Supporting Information), suggesting that the UCNPs could be retained in the hydrogel MNs even in its swollen state.

Finally, to confirm that these aggregates were indeed UCNP@MS, we took confocal fluorescence images of the UCNP@MS/MN tips under 980 nm excitation. The square pyramidal shape of the emission (Figure 2d) matches the bright field image of the MN tip (Figure 2c), indicating that the UCNP@MS were filling the whole tip, with brighter spots indicating higher concentrations of UCNP either due to agglomeration or uneven UCNP distribution. The emission spectra taken from the MN tips (Figure 2e) precisely match the spectra expected from UCNP@MS (Figure S12, Supporting Information), confirming that UCNP@MS had been successfully integrated within the MN tips. Furthermore, the emission increases in intensity with increasing UCNP@MS loading concentrations, suggesting a correlation between the amount of UCNP@MS loaded and the amount actually incorporated within the MN (Figure S16, Supporting Information). SEM images of the UCNP@MS/MN tips show that the incorporation of UCNP@MS did not affect tip morphology (Figure 2f; Figure S17, Supporting Information).

2.1.2. Mechanical and Swelling Properties of UCNP@MS/MNs

We carried out compression testing on MN patches with varying theoretical UCNP@MS concentrations (0 to 6.25 mg UCNPs loaded per MN patch) to find the maximum amount of UCNP@MS that could be included in the MN patches without compromising their ability to penetrate skin. To assess MN tip compressive strength, we measured the force required to press the MN patch perpendicularly against a stainless‐steel base to a distance of up to 0.4 mm using the experimental setup shown in Figure S18 (Supporting Information).

As seen in Figure 2g, the force required to compress the MNs decreased from 0.25 ± 0.07 to 0.09 ± 0.11 N needle⁻¹ with increasing UCNP@MS concentration. Furthermore, we found a significant difference in needle height before and after compression, starting at UCNP@MS loadings of 3.75 mg MN⁻¹ (Figure 2h), with the maximum reduction in needle height being 57 ± 5 µm at a theoretical UCNP@MS concentration of 6.25 mg MN⁻¹.

We hypothesized that the presence of UCNPs induces regions of local stress within the hydrogel, which weaken its mechanical strength^{[ 39 ]} and thus, there is a limit to the number of UCNPs that can be incorporated within the MNs without significantly compromising its mechanical properties. As previous studies have shown that a force of 0.083 N needle⁻¹ is sufficient to insert 50% of the MN length into excised porcine skin,^{[ 17} , ^{40 ]} we concluded that UCNP@MS/MNs with the highest inclusion concentration could still penetrate porcine skin as they could withstand 0.09 N needle⁻¹ of compressive force. This was later confirmed by observing the puncture marks left in porcine skin by the MNs after insertion (Figure 3d).

Figure 3.

a) Cy5‐PEG release curve from UCNP@MS+Cy5PEG/MNs (6.25 mg MN⁻¹) and Cy5PEG/MNs (0 mg MN⁻¹) in BSA‐PBS. b) Schematic of experimental setup for ex vivo analysis of Cy5‐PEG release into porcine skin. c) Calibration curve that relates the mass of Cy5‐PEG in porcine skin to the change in fluorescence reading of the portable reader. d) Fluorescent image of porcine skin after 30 min of UCNP@MS+Cy5PEG/MN (6.25 mg MN⁻¹) insertion with Cy5 filter (λ_ex = 630 nm, λ_em = 675 nm). e) Cy5‐PEG release curve from UCNP@MS+Cy5PEG/MNs (6.25 mg MN⁻¹) and Cy5PEG/MNs (0 mg MN⁻¹) in porcine skin. Triplicates from 3 different MNs were taken for each time point.

Though increasing UCNP@MS inclusion impacted the mechanical properties of the MN patch, we did not find a significant difference in swelling properties as a function of UCNP@MS inclusion after immersion in water for 5 min (Figure S19a–c, Supporting Information). The % mass changes found in this study correspond to what Laszlo et al. reported as swelling ratios for hydrogel MNs with the same polymer composition, assuming linear swelling over time.^{[ 41 ]} Thus, we hypothesized that drug flow would also not be affected by UCNP@MS inclusion.

To see if the UCNP@MS would be released from the UCNP@MS/MNs after swelling, we measured the upconversion emission spectra of the water in which the MNs had been allowed to swell for 5 min to check for the presence of UCNP@MS. We also cut the MN tips of one of the patches to simulate the case of a broken tip before immersion in water and subsequent swelling. As shown in Figure S19d (Supporting Information), no UCNP@MS could be detected in the swelling liquid, as evidenced by the lack of peaks in the emission spectra, in both experiments. These results indicate that the UCNP@MS were not merely physically entrapped within the hydrogel, but were likely interacting through hydrogen bonding forces with the uncrosslinked COOH groups of the Gantrez polymer. Hydrogen bonding between hydrogen bond donating groups found on drug molecules and the uncrosslinked COOH groups of the Gantrez polymer has been previously reported for other Gantrez MNs.^{[ 42 ]}

2.1.3. Optimization of UCNP@MS/MN Synthesis Protocol

The one drawback of the highest UCNP@MS loading was the increased likelihood of tip breakage after insertion, as seen by the reduction in MN height (Figure 2h) and observed MN breakage (Figure S20, Supporting Information) after compression. In order to ensure no UCNP@MS would be left in the skin in case of needle breakage, we decided to optimize our UCNP@MS/MN synthesis protocol so that no UCNPs were located within 100 µm of the tip, a distance well above the maximum reduction in needle height measured (Figure 2h). We achieved this by centrifuging a layer of non‐UCNP loaded hydrogel into the tips and letting it dry overnight to form a plain tip with no UCNPs before proceeding with centrifuging the UCNP‐containing hydrogel as per normal.

Confocal imaging under 980 nm excitation showed that with the modified protocol, the UCNPs were located at ≈100 µm from the tip of the needle (Figure 2i). Furthermore, we measured the emission under 980 nm excitation from a hole created by the MN tip in a piece of rat skin under 980 nm excitation after insertion with the layered UCNP@MS/MNs and did not detect any measurable emission from UCNPs in the skin (Figure S21, Supporting Information). In contrast, small peaks attributable to UCNP emission could be observed from the skin penetrated with the non‐layered UCNP@MS/MNs (Figure S21, Supporting Information).

Given the above findings, we decided to use a theoretical UCNP loading concentration of 6.25 mg MN⁻¹ (corresponding to an actual concentration of 0.87 ± 0.01 mg MN⁻¹ (Figure S11c, Supporting Information)) combined with the layered method to synthesize all subsequent versions of UCNP@MS/MN and UCNP@LBMS/MN.

2.2. Drug Flow Through UCNP@MS+Cy5‐PEG/MNs

2.2.1. In Vitro Evaluation of Cy5‐PEG Flow Through UCNP@MS+Cy5‐PEG/MNs

To see if the UCNP@MS inclusion would impact drug flow through the MNs into interstitial fluid, we synthesized UCNP@MS+Cy5PEG/MN, in which the model drug, Cy5 conjugated to PEG of molecular weight 1000 g mol⁻¹, and UCNP@MS were loaded separately in the MN (see Table S1, Supporting Information for abbreviation descriptions) to simulate the condition where all of the drug had been released from the UCNPs. Cy5‐PEG was used as a model drug for leuprolide for all preliminary optimization experiments due to its easy, well‐known, and reproducible synthesis and fluorescent properties,^{[ 43} , ⁴⁴ , ⁴⁵ , ^{46 ]} while PEG of a similar molecular weight to leuprolide (1209 g mol⁻¹) was chosen to best mimic the flow properties of leuprolide. We then measured Cy5‐PEG flow in both in vitro and ex vivo models. No significant difference in flow rate into BSA‐PBS was detected between MNs with and without UCNP@MS (Figure 3a). This was expected as the MN swelling properties remained unchanged with UCNP@MS inclusion (Figure S19, Supporting Information); it also suggests that any released drug would not be re‐adsorbed onto the surface of the UCNP@MS but rather flow through the hydrogel MNs with a similar rate to MNs without UCNPs.

2.2.2. Ex Vivo Evaluation of Cy5‐PEG Flow Through UCNP@MS+Cy5‐PEG/MNs

To see if these findings would extend to an ex vivo porcine skin model, we first measured a calibration curve to relate Cy5 fluorescence in porcine skin and the amount of Cy5‐PEG. To do this, we used dissolving PVA‐PVP MNs loaded with known amounts of Cy5‐PEG. The PVA‐PVP MNs were allowed to dissolve in the skin, and Cy5 fluorescence in skin was measured with a portable fluorescence reader before and after MN dissolution (Figure 3b) to correlate the change in fluorescence with the amount of Cy5‐PEG released in skin (Figure 3c).

This calibration curve was then used to evaluate whether the highest inclusion of UCNP@MS/MN interfered with Cy5‐PEG flow into the porcine skin. Successful MN penetration into the skin was verified as seen by a clear array pattern of puncture holes in the skin (Figure 3d). Quantification of Cy5‐PEG flow over time revealed that the highest loading concentration of UCNP@MS in the MN patches had no significant impact on drug flow through the MNs in the ex vivo model (Figure 3e), confirming what was found in the in vitro model (Figure 3a).

Different from what was observed in BSA‐PBS (Figure 3a), the release of Cy5‐PEG was the highest at ≈30 min (Figure 3e), and decreased at subsequent timepoints. While in the in vitro model, the continual refreshing of medium allowed up to 80% of loaded Cy5‐PEG to be drawn out of the swollen MN tips, the ex vivo skin only allowed for release as long as there was a concentration gradient. With no mechanism driving continual drug release, Cy5‐PEG released in the skin most likely was reabsorbed into the MNs after the 30 min time point due to the super‐swelling nature of the hydrogel MNs. In fact, super‐swelling hydrogel MNs are more widely used to withdraw fluid, as shown by Laszlo et al.^{[ 41 ]} As a result, only 35 ± 11% of the total Cy5‐PEG used to manufacture the MN patch was able to flow into the skin. We estimate this will also be the maximum amount we will measure in the skin once the NIR trigger is used in the functional light‐responsive UCNP@LBMS/MNs. This maximum amount is comparable to the 27 ± 9% released by the hydrogel MNs over 24 h observed in a study by Courtenay et al.^{[ 28 ]}

2.3. Light‐Triggered Drug Release from UCNP@LBMS‐Cy5leu/MNs

2.3.1. NIR‐Triggered Transdermal Delivery of Cy5‐Leuprolide from UCNP@LBMS‐Cy5leu/MN into Ex Vivo Porcine Skin Model

After characterizing the effect of UCNP@MS inclusion in the hydrogel MN on drug flow, we then proceeded to evaluate the system's ability to release the drug in response to NIR excitation by first loading the model drug, Cy5‐PEG, into UCNP@LBMS, before MN synthesis to form UCNP@LBMS‐Cy5‐PEG/MN (see Table S1, Supporting Information for description of abbreviations).

As shown in Figure 4a, no significant differences in Cy5‐PEG release were seen at short timepoints (10 and 15 min NIR exposure), but starting at 60 min, the UCNP@LBMS‐Cy5‐PEG/MN exposed to NIR irradiation released significantly more Cy5‐PEG than the MN left in the dark. In our previous work, significant differences in drug release due to NIR exposure of UCNP@LBMS were only observed after 30 min irradiation, indicating that at least 30 min are needed to break the LBMS shell and allow the drug to diffuse out of the LBMS pores.^{[ 35 ]} After release from the UCNPs, the drug also has to diffuse through the hydrogel MN pores into the skin, which further explains the time delay observed here. However, the fact that we observed increasing Cy5‐PEG concentration in the porcine skin over time, as opposed to a decrease as observed in Figure 3e, indicated that the continuous Cy5‐PEG release from the NIR‐irradiated UCNP@LBMS‐Cy5‐PEG could maintain the concentration gradient between the MN and porcine skin, preventing reuptake by the MN. Although such long NIR exposure times are not viable in an in vivo setting,^{[ 47} , ^{48 ]} they were necessary in an ex vivo model to demonstrate the full release kinetics of UCNP@LBMS/MNs under NIR exposure, as an ex vivo model lacks the fresh blood flow required to draw out any released drug from the MNs into the skin. In fact, this methodology of choosing longer exposure to stimuli in an ex vivo model versus an in vivo model to show the full drug kinetics of the MN has been reported frequently in past studies.^{[ 49} , ^{50 ]}

Figure 4.

a) Cy5‐PEG release from UCNP@LBMS‐Cy5‐PEG/MN and (b) Cy5‐leuprolide release from UCNP@LBMS‐Cy5leu/MN in ex vivo porcine skin model with and without continuous NIR irradiation (λ_ex = 976 nm, 2 W cm⁻²), shown as a function of time. Each datapoint represents triplicate readings taken with the portable reader from 3 MN patches. Statistical significance was determined using unpaired t‐tests for each time point. Fluorescent microscopy images (Cy5 filter) of ex vivo porcine skin after insertion and removal of UCNP@LBMS‐Cy5leu/MN (c) without NIR irradiation and (d) after 120 min of NIR irradiation. Scale bar = 500 µm.

Encouraged by these results, we repeated this experiment with UCNP@LBMS‐Cy5leu/MNs and saw a similar increase in Cy5‐leuprolide release in skin with NIR exposure as compared to the control condition (Figure 4b), consistent with what was also observed when exposing UCNP@LBMS‐Cy5leu to NIR in an in vitro model (Figure S22, Supporting Information). A maximum of ≈32 and 35% release for Cy5‐PEG and Cy5‐leuprolide respectively, was observed after 2 h of irradiation. This release is similar to the maximum 35% drug release observed in Figure 3e when Cy5‐PEG was freely dispersed within the hydrogel, indicating that most of the drug loaded in the UCNP@LBMS was released after 2 h of continuous NIR exposure.

Fluorescence images of the porcine skin after removal of the NIR‐exposed UCNP@LBMS‐Cy5leu/MN show an array of fluorescent spots after 120 min of excitation corresponding to the drug released from each MN tip (Figure 4d), while no fluorescence was observed in the porcine skin left in the dark (Figure 4c). This further confirmed that Cy5‐leuprolide was released from the MN tips in response to NIR excitation. Variable fluorescence intensity at each MN tip spot may be due to the different depths of MN tip insertion into the skin, the non‐homogeneous dispersion of UCNPs within the MN tip (Figure 2h), or the inconsistency of MN application via manual pressure.^{[ 14 ]}

2.3.2. In Vivo NIR‐Triggered Release of Cy5‐Leuprolide in Rat Skin and Plasma

To demonstrate the NIR‐triggered release of Cy5‐leuprolide in vivo, we applied UCNP@LBMS‐Cy5leu/MNs to rat dorsal skin (Figure 5a) and measured Cy5‐leuprolide release in both skin and plasma over a 48 h period after initially exposing the MNs to a 5 min NIR pulse (980 nm, 2 W cm⁻²) (Figure 5b). Instead of continuous NIR irradiation as in the ex vivo experiment, we were limited to a 5 min pulse as in other in vivo studies,^{[ 51 ]} so as not to burn the rat skin. We also tested MN patches with a needle height of 800 µm (UCNP@LBMS/Cy5leu/MN‐800) in addition to the 500 µm patches (UCNP@LBMS/Cy5leu/MN‐500) that we used in the in vitro and ex vivo experiments to see if an increase in MN height would improve drug release in vivo. These 800 µm MNs had a needle height of 624 ± 86 µm after removal from their molds (Figure S23a, Supporting Information) and an overall lower mechanical strength compared to the 500 µm MNs (Figure S24, Supporting Information), a finding consistent with other studies showing that increasing MN length decreases mechanical strength.^{[ 52 ]} Similar to the 500 µm MNs (Figure 2g), increasing UCNP concentration decreased the MN mechanical strength (Figure S24, Supporting Information). MNs with no UCNP@LBMS (MN‐500) and with free Cy5‐leuprolide loaded within the MN (Cy5leu/MN‐500) were used as negative and positive controls, respectively to evaluate how the Cy5‐leuprolide release from our NIR‐triggered system compared to the minimum and maximum values of possible drug release. Finally, plasma was extracted from one control rat to evaluate the background fluorescence of rat plasma without any treatment.

Figure 5.

a) Schematic showing the process of in vivo testing of UCNP@LBMS‐Cy5leu/MN. b) Digital picture showing NIR exposure of the MN patch applied on the rat dorsal skin. c) Cy5‐leuprolide release in rat skin from MN‐500, Cy5leu‐MN‐500, UCNP@LBMS‐Cy5leu/MN‐500 (with and without 5 min NIR pulse), and UCNP@LBMS‐Cy5leu/MN‐800 (with and without 5 min NIR pulse). Cy5‐leuprolide fluorescence was measured by a portable fluorescent reader after MN removal and converted to Cy5‐leuprolide mass using the calibration curve in Figure 3c. Triplicate readings were taken for each rat, and 3 rats were used per condition. d) Fluorescent microscopy image (Cy5 filter) of histological cross‐sections of rat skin after removal of NIR‐exposed UCNP@LBMS‐Cy5leu/MN‐800. e) Cy5‐leuprolide release in rat plasma after insertion with UCNP@LBMS‐Cy5leu/MN‐800 (with and without 5 min NIR pulse) and Cy5leu/MN‐500, as measured by Cy5 fluorescence emission (λ_ex = 630 nm, λ_em = 675 nm). Triplicate readings for each rat were taken per time point, and 3 rats were used per condition. Orange shaded box represents data points that fall under the background fluorescence of baseline rat plasma. f) Semi‐quantitative image analysis of Cy5‐leuprolide accumulation in rat organs after 48 h measured from IVIS images of ex vivo organs. Statistical significance between conditions for all graphs was determined using an unpaired t‐test.

After UCNP@LBMS‐Cy5leu/MN‐500 patches were exposed to the initial 5 min pulse, we quantified the amount of Cy5‐leuprolide released in skin 48 h after the pulse using the calibration curve developed earlier (Figure 3c), and noted a significant increase of 0.83 ± 0.7 µg in the NIR‐exposed rats compared to the non‐exposed rats (Figure 5c). Similar to the 500 µm MNs, the 5 min NIR pulse also increased the amount of Cy5‐leuprolide found in the skin for the UCNP@LBMS‐Cy5leu/MN‐800, although the increase was not significant (Figure 5c). Furthermore, increasing the needle height to 800 µm significantly increased the ability of Cy5‐leuprolide to flow into the skin with or without NIR irradiation. In fact, when the 800 µm MN patches were exposed to the 5 min NIR pulse, they released almost half (43%) of the total amount that flowed into the skin after 48 h when the same amount of Cy5‐leuprolide (348 ± 5 µg) was loaded directly into the MN patches (Cy5leu/MN‐500). These results were confirmed qualitatively by visualizing Cy5 fluorescence using an in vivo imaging system (IVIS) with a Cy5 filter (λ_ex = 655 nm, λ_em = 716 nm) (Figure S25, Supporting Information). This increased Cy5‐leuprolide flow into the skin with the 800 µm MNs is likely due to an increased depth of penetration; as shown in Figure S26 (Supporting Information), penetration with the 800 µm MNs left puncture marks ≈125 µm in depth, while no such marks were seen after penetration with the 500 µm MNs. Fluorescence microscopy imaging of the excised cross‐sections of skin after penetration with the 800 µm MNs shows further evidence of both MN penetration and fluorescence from the Cy5‐leuprolide release from the MNs (Figure 5d).

Following removal from the skin after the 48 h timepoint, the MNs were bent, and some of them lost their tips (compare Figure S23a,b, Supporting Information). Despite this, the lateral dimensions (width and distance between needles) did not significantly change (Figure S23c, Supporting Information). This is quite different from the significant swelling observed when the MNs were fully submerged in water for 5 min (see Figure S19, Supporting Information). This difference may be ascribed to the dissolved solutes present in interstitial fluid, which would reduce the osmotic driving force for liquid absorption in the MNs, and to the viscoelastic properties of skin, which would exert mechanical resistance, thus reducing hydrogel swelling.^{[ 42} , ^{53 ]}

While the ex vivo results demonstrate how the system responds to continuous irradiation, the in vivo results suggest that even a short 5 min pulse of NIR light can initiate breakdown of the LBMS shell and subsequent release of drug into skin. Furthermore, the higher amount of drug released in the skin in vivo (43%) compared to the ≈35% maximum released ex vivo illustrates how the continual renewal of plasma due to blood flow is important for maintaining the concentration gradient needed to continually pull any released drug from the MN into the skin.

Histological analysis of the skin post NIR‐treatment showed no significant differences between the irradiated and non‐irradiated groups in skin architecture or inflammatory cells, indicating that the 5 min pulse did not induce tissue damage (Figure S26, Supporting Information). Administration of the MN patches and subsequent NIR irradiation was well tolerated by the rats, with little erythema, edema or other signs of irritation (Figure S27, Supporting Information).

As shown in Figure 5e, pharmacokinetic data reveal a significant increase in plasma fluorescence due to NIR exposure of the 800 µm MNs at all timepoints (p ≤ 0.01 when comparing UCNP@LBMS‐Cy5leu/MN‐800 with and without NIR). On the other hand, the pharmacokinetic curves measured on the plasma of the rats that were treated with UCNP@LBMS‐Cy5leu/MN‐500 appeared almost identical with and without NIR exposure, with both curves being below the baseline fluorescence of rat plasma (Figure S28, Supporting Information).

For the rats in the UCNP@LBMS‐Cy5leu/MN‐800 group exposed to NIR, plasma concentrations reached a peak of 1.6 ± 0.1 ng mL⁻¹ at 30 min after NIR irradiation, then slowly decreased to 1.2 ± 0.1 ng mL⁻¹ at 48 h (Figure 5e). The plasma drug concentration at 48 h was 80% lower than the concentration measured for the Cy5leu/MN‐500 group (1.5 ng mL⁻¹ after 48 h), which represents the plasma concentration at 48 h if all the loaded Cy5‐leuprolide had been released from the UCNP@LBMS. The exact numbers should be interpreted with caution, however, as the amounts of Cy5‐leuprolide detected in plasma were very low, and could be affected by the baseline variation of background fluorescence in plasma^{[ 54 ]} over time, as seen by the observed fluorescence in the rat applied with an MN with no Cy5‐leuprolide (MN‐500).

As leuprolide accumulates in the reticuloendothelial system of the mice at concentrations ≈35 times greater than in blood,^{[ 55 ]} we wanted to see if the difference in Cy5‐leuprolide release due to the NIR trigger would be more apparent when looking at Cy5‐leuprolide accumulation in the rat organs. To do this, we used IVIS microscopy to image the organs (heart, lung, liver, spleen, kidney) of each rat and compared the fluorescence of accumulated Cy5‐leuprolide after 48 h. As seen in Figure S29 (Supporting Information), we observed an increase in fluorescence in the organs of the rats inserted with the UCNP@LBMS‐Cy5leu/MN‐800 exposed to NIR light for 5 min compared to the group not exposed to NIR. Semi‐quantitative analysis of the organ fluorescence confirmed that this increase due to the NIR pulse was indeed significant in the 800 µm group (Figure 5f), but not in the rats treated with 500 µm MNs, indicating that a difference of 300 µm in MN height significantly impacts the amount of drug released into both skin and plasma.

Finally, to verify that no UCNPs were released from the MNs, we performed microwave plasma atomic emission spectroscopy (MP‐AES) analysis on the organs after acid digestion to quantify the amount of accumulated Yb³⁺. As shown in Figure S30 (Supporting Information), the amount of Yb³⁺ found in the digested organs, representing the mass of UCNP@LBMS present, was negligible in both the UCNP@LBMS‐Cy5leu‐500/MN and UCNP@LBMS‐Cy5leu/MN‐800 groups exposed to NIR (≤ 20 µg Yb³⁺/g of tissue). Previous mice studies have shown that accumulated amounts of 48 – 260 µg of UCNPs in organs had no long‐term toxic effect of up to 90 days; our system releases significantly less than this amount.^{[ 56 ]}

3. Discussion

This work presents a proof‐of‐concept that UCNP‐integrated hydrogel MNs can be used for NIR‐triggered delivery of leuprolide. However, one of the biggest barriers to overcome before eventual clinical translation is the limited dose of the drug available in the bloodstream. Clinical trials on the use of leuprolide in IVF show that a 1.88 mg subcutaneous dose via syringe injection, resulting in a peak plasma concentration of 19.1 ± 5.4 ng mL⁻¹ is required for successful oocyte growth and retrieval,^{[ 57 ]} although other studies have reported peak concentration values of 4.6 to 10.2 ng mL⁻¹.^{[ 58 ]} After optimization of the MNs to include the maximum number of UCNP@LBMS while still being able to penetrate skin, we were able to load 174 ± 3 µg of Cy5‐leuprolide into a single MN patch, which could deliver a theoretical amount of 348 ± 5 µg when 2 patches are administered. However, from our in vivo plasma results, only a small fraction of this amount was released upon NIR irradiation, with an even smaller amount able to diffuse across the epidermis/dermis to enter the bloodstream. This was the reason for a plasma concentration level (≤ 2 ng mL⁻¹) below the human therapeutic level reported in clinical trials.

In order to increase the amount of drug available in the blood, we have to consider three events where the amount of drug released could potentially be limited: 1) release from UCNP@LBMS after NIR light irradiation; 2) flow from hydrogel MNs into the skin; and 3) diffusion from the skin into the bloodstream. For each of these three events, we propose the following strategies to increase drug release:

• 1)Increase the amount of drug released from UCNP@LBMS after NIR light irradiation: while drug release could be increased by irradiating for longer, NIR pulses longer than 5 min cause tissue overheating.^{[ 47} , ⁵⁹ , ^{60 ]} An alternate irradiation schedule involving multiple pulses, with periods of rest in between to allow the tissue to cool down, could be considered, as has been previously shown for MN delivery of lidocaine.^{[ 51 ]} To increase the amount of drug released without increasing irradiation time, one could also decrease the thickness of the UV‐breakable shell, as shown in our previous work.^{[ 35 ]} From a more fundamental angle, research is being done to improve the overall efficiency of upconversion, as current quantum yields remain low (< 1%)^{[ 61} , ⁶² , ^{63 ]} and present a significant barrier to using small amounts of NIR irradiation to release sufficient drug amounts.
• 2)Increase the amount of drug that flows through the hydrogel MNs into the skin: we specifically chose hydrogel MNs as opposed to dissolving MNs so that they can retain the UCNPs, both to prevent UCNP accumulation in organs with any associated side effects^{[ 64 ]} but also to allow for repeated dosing. This would be helpful, for example, for gonadotropins, which need to be delivered daily over the course of 2 weeks. However, a maximum of ≈30% of loaded Cy5‐leuprolide was able to flow from the MNs into the skin in our ex vivo model (Figure 3e), and in vivo, only ≈3% of Cy5‐leuprolide was detected in the skin when the drug was loaded free inside the hydrogel MN (Figure 5d). The challenge of low dosage is common in the field of MNs,^{[ 65} , ^{66 ]} given the small size of the patch, and researchers have proposed increasing the patch area, number of patches or MN density to achieve clinically relevant dosing levels.^{[ 15 ]} To this end, the manufacturing and administration of larger MN patches (e.g., 6 × 20 cm²)^{[ 67 ]} has already been demonstrated successfully with human volunteers.^{[ 26 ]} As well, modifying the MN gel by changing crosslinking density^{[ 68 ]} or polymer composition^{[ 69 ]} may allow for more drug to flow through the hydrogel pores.
• 3)Enhance the ability of the released drug to diffuse from the skin into the capillaries: of the Cy5‐leuprolide released, not all of it was able to diffuse from the epidermis into the capillaries, as shown by the fluorescence still found in the rat skin after 48 h (Figure 5d). In a study by Ito et al. that measured leuprolide in skin and plasma after release from dissolving MNs, only 31 – 33% of the loaded drug was found in the plasma, while the rest was found in the skin tissue homogenate.^{[ 70 ]} Previous studies have shown that the hydrophobicity^{[ 71 ]} and molecular weight^{[ 72 ]} of a drug impact its diffusion through the layers of skin, thus influencing bioavailability in the bloodstream versus skin after delivery from MNs. In our work, we showed that increasing MN height from 500 to 800 µm increases drug release into both skin (Figure 5d) and plasma (Figure 5e), and accumulation of drug in organs (Figure 5f). Thus, future work could look at increasing MN height without inducing pain (up to 1000 – 1200 µm)^{[ 73 ]} or chemically modifying the drug to enhance its bioavailability after release from MNs. Though previous MN studies have used needle lengths ranging from 500 to 1000 µm,^{[ 74} , ⁷⁵ , ^{76 ]} none to our knowledge have directly compared how different needle lengths impact drug flow into the skin^{[ 77 ]} and few compare the amount remaining in the skin versus released in the blood as a factor determining MN efficacy, most likely due to the difficulty of quantifying drug remaining in skin. Our findings thus illustrate the need for a better understanding of how MN properties, such as needle length, impact not just penetration of skin but the flow of drugs from the epidermis into the bloodstream.

4. Conclusion

In this work, we designed a light‐responsive UCNP@LBMS/MN composite system that releases Cy5‐leuprolide upon NIR irradiation while retaining UCNPs within the MN. Previous light‐responsive MN systems either used UV light^{[ 17 ]} or released the nanoparticles into the skin along with the drug,^{[ 18} , ^{78 ]} both of which reduce its biocompatibility and clinical translatability. We ensured that the maximum UCNP inclusion within the MNs did not adversely affect its ability to penetrate the skin or allow drug flow and observed marked release into an ex vivo model of both Cy5‐PEG and Cy5‐leuprolide under continuous NIR irradiation. A difference in Cy5‐leuprolide release was still observed even with only a short 5 min pulse of NIR light in an in vivo model; this difference was enhanced by increasing MN needle length from 500 to 800 µm. While the leuprolide amount released from our system after 5 min of NIR light did not reach clinically relevant plasma levels, it represents the first device for the painless, controlled, time‐sensitive delivery of hormones for IVF treatment.

These findings open up the possibility for UCNP‐integrated hydrogel MNs to become vehicles for drug delivery from other types of functional nanoparticles without needing to release them into the skin. This would avoid the need for assessment of nanoparticle toxicity in the human body and thus improve the commercial and regulatory outlook of such devices,^{[ 20 ]} speeding up their clinical translation. Integrating stimuli‐triggered nanoparticles would also offer the option of repeated, automated light‐controlled dosing, which could overcome problems like patient compliance and human error, while the promise of painless delivery could drastically change the experience of patients who require repeated administration of drugs via needles for conditions beyond IVF, including, for example, diabetes, rheumatoid arthritis, or multiple sclerosis.

5. Experimental Section

Materials

Y₂O₃ (REacton, 99.999%), Yb₂O₃ (REacton, 99.998%), Tm₂O₃ (REacton, 99.999%), trifluoroacetic acid (99%), 1‐octadecene (ODE, 90%), and oleic acid (OA, 90%) were purchased from Alfa Aesar (USA). Lithium trifluoroacetate (LiCO₂CF₃, 98%), oleylamine (OM, 70%), cetyltrimethylammonium bromide (CTAB, 99+%), tetraethyl orthosilicate (TEOS, >99%), ethanol (EtOH), chloroform (CLF, >99%), sodium hydroxide (NaOH, >98%), hydrochloric acid (HCl, 37%), acetone (>99.5%), hexane (>95%), sodium hydride (NaH, 90%), 5‐hydroxy‐2‐nitrobenzyl alcohol (97%), N,N‐dimethylformamide (DMF, anhydrous, 99.8%), triethoxysilane (95%), platinum (0)‐1,3‐divinyl‐1,1,3,3‐tetramethyldisiloxane complex solution (Karstedt's catalyst; Pt ≈2 wt % in xylene), allyl bromide (97%), 2‐tert‐butyl‐1,1,3,3‐tetramethylguanidine (Barton's base, >97.0%), sodium sulfate (anhydrous, Na₂SO₄, >99%), poly(vinyl alcohol) (10 kDa), polyethylene glycol (10 kDa), bovine serum albumin (BSA, >98%), malondialdehyde bis(phenylimine) monohydrochloride, diisopropylethylamine (DIPEA), nitromethane (CH₃NO₂), iodomethane (MeI), 2,3,3‐trimethylindolenine, disuccinimidyl carbonate (DSC), potassium iodide (KI) and sodium hydrogen carbonate (Na₂CO₃, >99.7%) were obtained from Sigma‐Aldrich (Saint Louis, MO). Acetic anhydride (Ac₂O), acetonitrile (ACN), and diethyl ether (Et₂O) were purchased from Fisher Scientific (Waltham, MA). Dimethyl sulfoxide (DMSO, anhydrous) and 6‐bromohexanoic acid were purchased from Acros Organics (Morris Plains, NJ). Poly(N‐vinylpyrrolidone) (3.5 – 7 kDa) (PVP K‐12) was kindly provided by BASF (Ludwigshafen, Germany). Methoxy‐poly(ethylene glycol) amine (MeO‐PEG‐NH₂) (1 kDa) was purchased from JenKem Technologies (Plano, TX). Sephadex G‐15 was purchased from GE Life Sciences (Mississauga, ON). Gantrez S‐97, a copolymer of methyl vinyl ether and maleic acid (PVME/MA), was kindly provided by Ashland (Kidderminster, UK). Square pyramidal female MN molds made of room temperature vulcanizing silicone, 10 x 10 array, 250 µm x 250 µm x 600 µm (W x L x H) with peak‐to‐peak spacing of 500 µm were purchased from Micropoint Technologies Pte. Ltd. (Singapore). Leuprorelin acetate (leuprolide, >90.0%, HPLC) was purchased from TCI America. Compressed argon gas was obtained from Linde Gas at >99.9% purity. Porcine skin was generously provided by Prof. Alexandre Thibodeau from the Microbiology and Pathology department (Faculty of Veterinary Medicine, University of Montreal) and kept frozen at −20 ^°C until use.

Synthesis of Cy5—1,2,3,3‐Tetramethyl‐3H‐Indol‐1‐ium Iodide (S1)

S1 was synthesized using a previously described procedure.^{[ 43 ]} Briefly, 2,3,3‐trimethylindolenine (2.52 mL, 15.7 mmol, 1 eq) was diluted in 9 mL CH₃NO₂. MeI (1.95 mL, 31.4 mmol, 2 eq) was added dropwise, and the reaction mixture was stirred at room temperature for 20 h under argon. The crude product was precipitated using 60 mL of Et₂O, and the resulting purple solid was collected by vacuum filtration and washed 3× with 15 mL of Et₂O. A pale purple solid was obtained (4.17 g, 13.8 mmol, 88% yield). ¹H NMR (Varian, 400 MHz, DMSO‐d6): δ = 7.91 – 7.88 (m, 1H), 7.82 – 7.80 (m, 1H), 7.63 – 7.57 (m, 2H), 3.96 (s, 3H), 2.75 (s, 3H), 1.51 (s, 6H) ppm (shown in Figure S1, Supporting Information along with synthesis scheme, HPLC chromatogram and MS spectrum).

Chromatographic separation was performed on an Agilent 1290 Infinity LC system (Agilent Technologies, Santa Clara, CA, USA) equipped with a reverse‐phase C18 column (100 mm × 2.1 mm, 1.7 µm particle size). The mobile phase consisted of solvent A (95% H₂O, 5% ACN, 0.1% formic acid) and solvent B (5% H₂O, 95% ACN, 0.1% formic acid). The flow rate was set to 1.5 mL mi⁻¹n, and the following gradient program was used: 0–0.5 min, 5% B; 0.5–1.5 min, linear increase to 95% B; 1.5–3 min, hold at 95% B; 3–4.5 min, re‐equilibration at 5% B. The column temperature was maintained at 24 °C, and the injection volume was 10 µL.

Mass spectrometric detection was carried out using an Agilent 6120 quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA, USA) equipped with an atmospheric‐pressure chemical ionization (APCI) source operating in positive ion mode. The optimized APCI parameters were as follows: capillary voltage, 3000 V; nebulizer gas pressure, 60 psi; drying gas flow rate, 5 L min⁻¹; and drying gas temperature, 300 °C. Data were acquired in full scan mode over a mass range of 100–1000 m z⁻¹, with a scan time of 1.06 s.

1‐(5‐Carboxylpentyl)‐2,3,3‐Trimethyl‐3H‐Indol‐1‐ium Iodide (S2)

S2 was synthesized by adapting a previously described procedure.^{[ 37 ]} Briefly, 2,3,3‐trimethylindolenine (1.60 mL, 10 mmol, 1 eq) and 6‐bromohexanoic acid (1.95 g, 10 mmol, 1 eq) were dissolved in 6 mL ACN. KI (1.66 g, 10 mmol, 1 eq) was added and the reaction mixture was refluxed at 85 °C for 16 h under argon. The mixture was filtered, concentrated under reduced pressure, and the crude product was precipitated by adding 40 mL of 1:1 EtOAc/DCM and cooling to −20 °C. The resulting purple solid was collected by vacuum filtration and washed 3× with 5 mL of 1:1 EtOAc/DCM. A light purple solid was obtained (2.38 g, 5.9 mmol, 59% yield). ¹H NMR (Varian, 400 MHz, DMSO‐d6): δ = 12.00 (br. s, 1H), 7.99 – 7.94 (m, 1H), 7.85 – 7.81 (m, 1H), 7.63 – 7.58 (m, 2H), 4.44 (t, J = 7.8 Hz, 2H), 2.84 (s, 3H), 2.21 (t, J = 7.2 Hz, 2H), 1.83 (m, 2H), 1.58 – 1.50 (m, 8H), 1.45 – 1.39 (m, 2H) ppm (shown in Figure S2, Supporting Information along with synthesis scheme, HPLC chromatogram and MS spectrum).

Cyanine‐5 (Cy5)

Cy5 was synthesized by adapting a previously described procedure.^{[ 36 ]} Briefly, S2 (734 mg, 1.8 mmol, 1 eq) and malondialdehyde bis(phenylimine) monohydrochloride (521 mg, 2.0 mmol, 1.1 eq) were dissolved in 6 mL of Ac₂O, and the mixture was refluxed at 120 °C for 30 min under argon, then cooled to room temperature. S1 (660 mg, 2.2 mmol, 1.2 eq) was suspended in 6 mL of dry pyridine and added to the reaction mixture. The reaction was stirred for 18 h at room temperature, with protection from light. The solvents were removed under reduced pressure at 60 °C. The resulting blue solid was dissolved in 20 mL DCM and washed 3× with 40 mL dH₂O, 1× with 40 mL of 1 M HCl_(aq), and 1× with 40 mL of sat. NaCl. The organic layer was dried over Na₂SO₄ and concentrated under reduced pressure. The crude mixture was purified by column chromatography, eluting with a gradient of 0 – 10% MeOH in DCM. Cy5 was obtained as a metallic red‐blue solid (486 mg, 0.93 mmol, 51% yield). ¹H NMR (Varian, 400 MHz, DMSO‐d6): δ = 12.11 (br. s, 1H), 8.31 (t, J = 13.1 Hz, 2H), 7.60 (d, J = 7.4 Hz, 2H), 7.41 – 7.35 (m, 4H), 7.26 – 7.19 (m, 2H), 6.55 (t, J = 12.3 Hz, 1H), 6.27 (dd, J = 17.0 Hz, 13.9 Hz, 2H), 4.07 (t, J = 7.2 Hz, 2H), 3.58 (s, 3H), 2.18 (t, J = 7.2 Hz, 2H), 1.76 – 1.61 (m, 14H), 1.58 – 1.49 (m, 2H), 1.40 – 1.32 (m, 2H) ppm (shown in Figure S3, Supporting Information along with synthesis scheme, HPLC chromatogram and MS spectrum).

Synthesis of Cyanine‐5 N‐Hydroxysuccinimide Ester (Cy5‐NHS)

Cy5 was activated as Cy5‐NHS by adapting a previously described procedure.^{[ 36 ]} Briefly, Cy5 (486 mg, 0.93 mmol, 1 eq) was dissolved in 10 mL of dry DCM. DIPEA (326 µL, 1.87 mmol, 2 eq) and DSC (536 mg, 2.06 mmol, 2.2 eq) were added, and the reaction was stirred for 24 h at room temperature, with protection from light. The mixture was diluted with an additional 10 mL DCM and washed 4× with 40 mL dH₂O, 1× with 40 mL of 1 M HCl_(aq), and 1× with 40 mL of sat. NaCl. The organic layer was dried over Na₂SO₄ and further dried under reduced pressure to give Cy5‐NHS as a metallic red‐blue solid (544 mg, 0.88 mmol, 95% yield). The purity of the product was confirmed by HPLC‐MS analysis. ¹H NMR (Varian, 400 MHz, DMSO‐d6): δ = 8.32 (t, J = 13.1 Hz, 2H), 7.60 (d, J = 7.4 Hz, 2H), 7.41 – 7.35 (m, 4H), 7.26 – 7.19 (m, 2H), 6.54 (t, J = 12.3 Hz, 1H), 6.26 (t, J = 14.1 Hz, 2H), 4.07 (t, J = 7.2 Hz, 2H), 3.58 (s, 3H), 2.80 (s, 4H), 2.69 – 2.65 (m, 2H), 1.76 – 1.61 (m, 16H), 1.50 – 1.45 (m, 2H) ppm (shown in Figure S4, Supporting Information along with synthesis scheme, HPLC chromatogram and MS spectrum).

Synthesis of Cy5‐PEG‐1k

To more easily quantify drug release via fluorescence, the dye Cyanine‐5 (Cy5) was conjugated to the gonadotropin agonist, leuprolide. For preliminary proof‐of‐concept experiments, Cy5‐PEG was used as a model drug, with PEG‐1k being used to approximate the size of leuprolide (1209 g mol⁻¹). Cy5 was chosen due to its relatively simple synthesis and high biocompatibility, with one of its structural analogs approved by the FDA for in vivo use.^{[ 79 ]}

Cy5‐PEG was synthesized through amine coupling of an activated N‐hydroxysuccinimide (NHS) Cy5 ester to a 1 kDa methoxy‐polyethylene glycol‐amine (MeO‐PEG‐NH₂) polymer as shown in a previous protocol^{[ 45 ]} (see Figure S5, Supporting Information for reaction scheme). Briefly, Cy5‐NHS (18.5 mg, 0.045 mmol, 1.5 eq) and MeO‐PEG‐NH₂ (1 kDa) (30 mg, 0.03 mmol, 1 eq) were dissolved in anhydrous DMSO (1 mL), and the mixture was stirred in the dark for 24 h. Following this, the mixture was diluted in milli‐Q H₂O (20 mL) and lyophilized. The resulting residue was dissolved in milli‐Q H₂O (0.5 mL) and purified on G‐15 Sephadex. The fractions were combined and lyophilized, yielding Cy5‐PEG‐1k as a dark blue solid. The purity of the product was confirmed by ¹H NMR, ¹³C NMR, and HPLC‐MS analysis (Figure S6, Supporting Information).

Synthesis of Cy5‐Leuprolide

Cy5‐leuprolide was synthesized by conjugating the arginine side chain in leuprolide to the NHS ester of Cy5 by selectively acylating the arginine residue present in leuprolide, using a strong base to deprotonate the arginine to favor acylation, before subsequent conjugation to Cy5‐NHS as per a previous arginine modification protocol^{[ 80 ]} (see Figure S7, Supporting Information for reaction scheme). Cy5‐NHS (9.5 mg, 0.015 mmol, 1.5 eq) and leuprolide (12.4 mg, 0.01 mmol, 1 eq) were dissolved in 1 mL DMF. 20.2 mL of Barton's base was added to the solution, which was then stirred in the dark for 1 h at 40 °C. The mixture was then diluted in 7.5 mL 1 M HCl and 15 mL milli‐Q H₂O before lyophilization. The resulting residue was dissolved in a 3:1 mixture of milli‐Q H₂O and MeOH, then purified on G‐15 Sephadex. The fractions were combined and lyophilized, yielding Cy5‐leuprolide as a blue solid. The purity of the product was confirmed by ¹H NMR, ¹³C NMR, and HPLC‐MS analysis (Figure S8, Supporting Information).

Synthesis of Core/Shell LiYbF₄:Tm³⁺/ LiYF₄ (UCNP)

UCNP synthesis was performed using a protocol previously developed by T. Cheng et al.^{[ 81 ]} The following steps are described in more detail below: a) preparation of rare earth (RE) trifluoroacetate precursors; b) LiYbF₄:Tm³⁺ first nuclei (FN) synthesis; c) stabilization of FN; and finally d) shelling of FN with a passive LiYF₄ shell to form UCNPs.

Precursor Preparation

RE trifluoroacetate precursors were prepared by mixing RE₂O₃ (RE = Yb, Y, Tm) with 10 mL trifluoroacetic acid/distilled water mixture (1:1 v/v) in a 50 mL three‐neck round‐bottom flask. To synthesize the precursors for 2.5 mmol LiYbF₄:Tm³⁺ (0.5%), first nuclei (FN), 1.1875 mmol (467.97 mg) Yb₂O_3, and 0.00625 mmol (2.4 mg) Tm₂O₃ were used. In a separate three‐neck flask, 2.5 mmol (564.5 mg) Y₂O₃ was used to synthesize the precursors for 2.5 mmol of the LiYF₄ passive shell. For each flask, the slurry was refluxed under vigorous stirring at 80 °C until a clear solution formed. The temperature was then lowered to 60 °C to evaporate the solvent until dry. The obtained dried solids in the flask with Yb and Tm were used as precursors for the FN synthesis, while the Y shelling precursors were stored until the shelling of the stabilized FN.

LiYbF₄:Tm³⁺ First Nuclei (FN) Synthesis

Two solutions were prepared for FN synthesis. Solution A, a mixture of 7 mL OA, 7 mL OM, and 14 mL ODE, was prepared in a 100 mL three‐neck round‐bottom flask. Solution A was stirred and degassed under vacuum at 110 °C for 30 min first. Then its temperature was raised to 330 °C under an Ar atmosphere. Meanwhile, Solution B was prepared. Solution B was a mixture of 2.5 mmol LiCO₂CF₃, dried Yb and Tm precursors synthesized above, 3 mL OA, and 6 mL ODE prepared in a 50 mL three‐neck round‐bottom flask. Solution B was then degassed under vacuum at 110 °C for 30 min. After the dried precursors were dissolved under vacuum in an OA/ODE mixture, 3 mL OM was added into solution B. Once Solution A had reached the stable desired temperature of 330 °C, Solution B was injected into Solution A using a pump‐syringe system (Harvard Apparatus, Pump 11 series) with an injection rate of 1.5 mL min⁻¹. After 1 h stirring at 330 °C, the reaction mixture was cooled down to room temperature under an Ar atmosphere. The majority of the synthesized product was stored in a Falcon centrifuge tube (50 mL) under Ar while a small portion (0.1 mL) of the product was washed for structural characterization by first precipitating with acetone, centrifugation at 8000 RCF for 15 min (OHAUS FC5718 Frontier 5000 Series), then washing with a mixture of hexane/acetone (1:4 v/v) twice before finally redispersing in hexane for TEM preparation.

LiYbF₄: Tm³⁺ FN Stabilization

To stabilize the sub‐10 nm FN cores that were synthesized in the presence of a high concentration of OM, it was necessary to add a stronger coordinating ligand, like OA, for slightly larger but thermodynamically stable UCNPs. As the FN had solidified after storage at room temperature, the Falcon tube containing the FN was placed in a 40 °C water bath to return it to a liquid form. Then, 1.1 mmol FN obtained in the step above was mixed with OA and ODE (1:1 v/v) for a total final volume of 40 mL. The solution was stirred and degassed under vacuum at 110 °C for 30 min. Then, the temperature of the solution was raised and maintained at 315 °C for 1 h under an Ar atmosphere. After cooling to room temperature, the stabilized nuclei were stored in a Falcon centrifuge tube (50 mL) under Ar for further shelling.

LiYF₄ Shelling for Core/Shell LiYbF₄:Tm³⁺/LiYF₄ (UCNP)

To maximize the upconversion emission and reduce the energy loss to surrounding ligands, the stabilized cores were shelled with a LiYF₄ passive shell. Again, two solutions were prepared. Solution A was prepared in a 100 mL three‐neck round‐bottom flask. The stabilized FN from step 1.5.2 was heated in a 40 °C water bath until liquefied, and 0.18 mmol was added to Solution A with equal volumes of OA and ODE to a final volume of 20 mL. Solution B consisted of 2.5 mmol of the shelling precursors (Y₂O₃) as already prepared in Step 1.5.1. 309 mg LiCO₂CF₃ and 10 mL each of OA and ODE for a final volume of 20 mL. Both solutions were stirred and degassed under vacuum at 110 °C for 30 min. After the degassing procedure, Solution A was placed under an Ar atmosphere and the temperature was raised to 315 °C. Solution B was injected into Solution A at a 1 mL min⁻¹ injection rate under vigorous magnetic stirring. After the injection, the mixture was kept at 315 °C and allowed to react for 1 h with continuous vigorous stirring. Subsequent steps included cooling down to room temperature and washing as described for the FN synthesis. Shelled LiYbF₄:Tm³⁺/LiYF₄, which UCNP was henceforth called, was stored in 20 mL hexane in a glass vial (60 – 65 mg mL⁻¹).

Synthesis of the Light‐Breakable Linker (LB)

The synthesis of the light‐breakable linker (LB) was adapted from a previously established synthesis by Picchetti et al.^{[ 82 ]} In a first step, the diallyl derivative of 5‐hydroxy‐2‐nitrobenzyl alcohol (Compound 1) was prepared by the addition of NaH (311 mg, 12.98 mmol) to a stirring solution of 5‐hydroxy‐2‐nitrobenzyl alcohol (1.00 g, 5.91 mmol) in dry DMF (10 mL) at 0 °C. After complete NaH addition, the reaction mixture was allowed to stir for a further 10 min at 0 °C. Subsequently, allyl bromide (1.17 mL, 13.52 mmol) was added dropwise to the stirring reaction mixture at 0 °C. The reaction was stirred for 1 h at 0 °C and for a further 1 h at room temperature before quenching with 5 mL of water. The aqueous layer was extracted with ethyl acetate (3×, 20 mL), and the organic layers were washed with a saturated Na₂CO₃ solution before drying over anhydrous Na₂SO₄. The solvent was then evaporated under reduced pressure, and the crude oil purified by column chromatography (silica gel, EtOAc/cyclohexane = 1:2), yielding Compound 1 as a yellowish oil (yield = 68%).

Light‐breakable bis‐alkoxysilane, triethoxy(3‐(4‐nitro‐3‐((3‐(triethoxysilyl)propoxy)methyl)phenoxy)propyl)silane (LB), was obtained via dihydrosilylation of Compound 1 using triethoxysilane and the hydrosilylation catalyst (Figure S9, Supporting Information). Briefly, to the product obtained above (1.00 g, 4.01 mmol) in dry toluene (9 mL), triethoxysilane (2.02 mL, 10.95 mmol) and subsequently platinum(0)‐1,3‐divinyl‐1,1,3,3,‐tetramethyldisiloxane complex solution (Karstedt's catalyst, Pt ≈2 wt% in xylene, 125 µL) was added dropwise. The reaction was stirred at 50 °C for 12 h. An additional Karstedt's catalyst solution (100 µL) was added, and the mixture was stirred for a further 12 h at 50 °C. The organic solvent was evaporated under reduced pressure, and the crude product was purified with column chromatography (silica gel, EtOAc/cyclohexane = 1:7), yielding LB as a yellowish oil (yield = 60%). ¹H NMR (400 MHz, CDCl₃): δ 8.14 (d, 1H, J = 9.1 Hz), 7.31 (m, 1H), 6.83 (m, 1H), 4.90 (s, 2H), 4.04 (t, 2H, J = 6.7 Hz), 3.86–3.81 (m, 12H), 3.58 (t, 2H, J = 6.8 Hz), 1.97–1.90 (m, 2H), 1.84–1.77 (m, 2H), 1.23 (t, 18H, J = 7.0 Hz), 0.79–0.75 (m, 2H), 0.73–0.69 (m, 2H) ppm. The ¹H NMR and ¹³C NMR spectra have been previously reported in Tam et al.^{[ 35 ]}

Synthesis of Light‐Breakable Mesoporous Silica‐Coated Upconverting Nanoparticles (UCNP@LBMS) and Control UCNP@MS

UCNP@LBMS were synthesized using a protocol from a previous work.^{[ 35 ]} 7 mg of UCNP in hexane as described above was first precipitated with EtOH and redispersed in 167 µL of chloroform. In a separate 20 mL glass vial, 16.7 mg CTAB was dissolved in 3.33 mL deionized water and heated at 70 °C in a pre‐heated oil bath until clear. Then, the dispersed UCNP in chloroform was added to the CTAB solution at room temperature, vortexed until milky in appearance, sonicated for 10 min, and stirred at 400 RPM for 20 min. The mixture was then transferred to the oil bath at 70 °C under magnetic stirring at 400 RPM for 15 to 20 min until the mixture formed a transparent suspension.

6.67 mL water, 1 mL EtOH, 50 µL 2 M NaOH, and the transparent UCNP suspension (3.5 mL) from above were added to a three‐neck 100 mL round‐bottom flask. The three‐neck flask was then heated under reflux at 70 °C, with stirring at 1000 RPM for 5 min. Then, an 80:20 mol% (i.e., 1.55 µL TEOS/mg LB linker) mixture of TEOS/LB linker was prepared and pipetted up and down until homogeneous. In this case, this corresponded to 32 µL TEOS and 20.7 mg LB linker in total.

UCNPs were also coated with an MS coating with no LB linker (UCNP@MS) for incorporation into MNs for initial characterization experiments (e.g., mechanical, swelling properties). UCNP@MS was prepared in the same way but without the addition of the LB linker (66.7 µL total amount of TEOS with no LB linker). The TEOS solution, with or without LB linker, was then pipetted to the suspension dropwise, spaced 10 s apart. The suspension was left to stir under reflux for 10 min before being transferred into an ice bath to stop the silica network formation.

The UCNP@LBMS suspension was precipitated with EtOH, centrifuged at 8000 RCF for 20 min (OHAUS FC5718 Frontier 5000 Series), then washed once with EtOH/acetone (1:1 v/v) and redispersed in 10 mL EtOH. To remove the CTAB from the mesopores, 10 µL concentrated HCl was added to the UCNP@LBMS suspension and left to reflux overnight under magnetic stirring (400 RPM) at 90 °C. After reflux, the suspension was centrifuged to remove the extracted CTAB and redispersed in 10 mL EtOH. To measure concentration, an empty Eppendorf was weighed, and 1 mL of the UCNP@LBMS suspension was dried under vacuum overnight at room temperature. After drying, the Eppendorf with dried UCNP@LBMS was weighed again, and the difference in mass was then divided by the initial volume (i.e., 1 mL) to obtain the final concentration.

Cy5‐PEG/Cy5‐Leuprolide Loading into UCNP@LBMS

To load UCNP@LBMS with either Cy5‐PEG or Cy5‐leuprolide, 25 mg UCNP@LBMS, as prepared in section 1.7, were fully dispersed in 500 µL of 0.5 mg mL⁻¹ Cy5‐PEG or Cy5‐leuprolide (0.25 mg) under sonication for 30 min. The Cy5‐PEG solutions were made in distilled H₂O, while the Cy5‐leuprolide solutions were in 80:20 % v/v H₂O/DMSO. Then, 750 µL 4 M CaCl₂ was added to cap the pores of the mesoporous silica to prevent premature drug release as previously demonstrated by Kang et al.^{[ 83 ]} After an additional 30 min of sonication, the loaded UCNP@LBMS (UCNP@LBMS‐Cy5‐PEG or UCNP@LBMS‐Cy5leu) were centrifuged, washed once with H₂O by redispersion and centrifugation to remove any unloaded drug and finally redispersed in 0.5 mL H₂O.

To measure the in vitro release of Cy5‐leuprolide from UCNP@LBMS, 12.5 mg UCNP@LBMS were dispersed in 1 mL 5% w/w BSA‐PBS and exposed to continuous NIR irradiation (980 nm, 2 W cm⁻²). At various timepoints, 250 µL of the solution was removed and centrifuged to remove UCNP@LBMS, while the absorbance spectrum of the supernatant liquid was measured using a Spark multimode fluorescence microplate reader (Tecan Group, Ltd., Mannedorf, Switzerland). Then, the particles were redispersed in the same liquid and added to the original solution to continue NIR exposure. The control experiment involved measuring the absorbance spectrum of the supernatant at the same time points of particles stirred in the dark.

Synthesis of UCNP@MS Embedded Hydrogel Microneedles (UCNP@MS/MNs)—Initial Manufacturing Procedure

To test how different concentrations of UCNP@MS affected the mechanical properties, swelling behavior, and pore size of the MNs, UCNP@MS/MNs were synthesized with varying amounts of UCNP@MS. UCNP@MS/MN patches were produced using aqueous polymeric solutions containing 20% w/w Gantrez S‐97, 7.5% w/w PEG, 10 kDa, 3% w/w Na₂CO₃ with varying concentrations of UCNP@MS according to the procedure developed by Donnelly et al.^{[ 25 ]} First, 0.44 g PEG and 0.22 g Na₂CO₃ were dissolved in 4.5 mL milliQ‐H₂O by vortexing 1–2 min until the solution became clear. Then, 0.5 mL of an aqueous suspension containing 5, 10, 15, 20, or 25 mg of UCNP@MS was added to the PEG/Na₂CO₃ solution for a final volume of 5 mL. 1.44 g Gantrez S‐97 was added and immediately vortexed to dissolve before centrifugation for 10 min at 4000 RPM (Sorvall ST 16R, Thermo Fisher Scientific, Waltham, MA). The hydrogel solution was then cast into 4 PDMS 10 × 10 female MN molds with square‐pyramidal tips and centrifuged at 4000 RPM for 5 min in 24‐well cell culture plates (Sarstedt AG & Co., Numbrecht, Germany). This centrifugation was repeated 3 more times with a 180^° rotation in between each centrifugation step. Then, the MN arrays were dried in their molds for 24 h at room temperature, before being placed in an 80 ^°C oven for 48 h and demolded. Verification of successful UCNP@MS integration within the MNs, as well as preliminary characterization (Section 2.1.1 and 2.1.2) was performed with UCNP@MS/MN manufactured according to this procedure.

Layered Manufacturing Procedure

Given that needle tip breakage of up to 100 µm was observed after mechanical testing, the protocol was modified above to generate a 100 µm layer with no UCNP@MS in the MN tips, thus ensuring that there was no leakage of UCNP@MS during drug release due to tip breakage. All subsequent UCNP@MS/MNs were synthesized with this layered method (results discussed starting from Section 2.2).

To do this, 5 mL of the aqueous polymeric solution containing 20% w/w Gantrez S‐97, 7.5% w/w PEG, 10 kDa, 3% w/w Na₂CO₃ was centrifuged at 4000 RPM for 5 min into 4 PDMS molds and left to dry overnight at room temperature to form the tips. The next day, an additional 5 mL of the Gantrez polymeric solution with the specified amount of UCNP@MS was added, then centrifuged into the molds as per the initial protocol, followed by drying and demolding process as described in section 1.9.1 above (Figure S10, Supporting Information).

Characterization of UCNP@MS/MNs—Optical Microscopy of MN Arrays

MN arrays were visualized using bright‐field microscopy (Axio Zoom V16, Zeiss, Toronto, ON) at either 0° or 90° to the stage, and dimensions were measured using the Zen 2 software (Zeiss, Toronto, ON) Magnification levels of ≈12.5× were used. Images were also taken using a Cy5 filter to visualize Cy5 fluorescence (λ_ex = 630 nm, λ_em = 675 nm).

Mechanical Testing

The mechanical properties of the MN arrays loaded with varying UCNP concentrations were characterized with a texture analyzer (TA.XT‐plus, Stable Micro Systems, Surrey, UK). Compression testing was performed by fixing the base of the MN array using double‐sided tape on a flat‐surfaced 6 mm cylindrical probe with the MN tips facing the stainless‐steel base of the analyzer. Then, the analyzer was set to move the probe toward the base at a speed of 0.5 mm s⁻¹ and compress the MN tips to a 0.4 mm distance with a 0.049 N trigger load. The peak force used to achieve this compression was then recorded. Optical images were taken before and after mechanical testing, and ImageJ was used to record the average heights of the first row of needles.

Swelling Behavior

The swelling behavior of the MN arrays was characterized by measuring the MN mass before and after swelling in water. Prior to immersion in water at 21 °C, the dry MN array weight was recorded. MN arrays were then placed in 50 mL Falcon tubes, and 5 mL of milli‐Q H₂O was added. After 5 min of swelling, the MNs were removed from the tube, and the swollen weight was recorded. The difference between dry and swollen weight was used to determine MN array swelling capabilities using the following equation:

$$\mathrm{Swel}\mathrm{ling}\%=\frac{\mathrm{Mass}\mathrm{of}\mathrm{sw}\mathrm{olle}n\mathrm{MN}}{\mathrm{Mass}\mathrm{of}\mathrm{dr}y\mathrm{MN}}\times 100\%$$ (1)

Scanning Electron Microscopy (SEM) of Pore Size

The swollen hydrogel MNs were lyophilized, mounted on an aluminum SEM stub with carbon tape, and then coated with ≈20 nm of carbon using a Leica EM ACE600 carbon sputter coater. The MNs were then imaged with a FlexSem1000 II (Hitachi, Canada) under an accelerating voltage of 10 kV and a spot size of 40 nm. The width and height of 100 pores from each sample were taken using ImageJ.

High‐Resolution SEM of MN Cross‐Sections

To visualize the NPs within the MN cross‐sections, swollen and lyophilized hydrogel MNs were coated with ≈20 nm of carbon using a Leica EM ACE600 carbon sputter coater, then imaged with a FEI Quanta 450 (Thermo Fisher, USA) under an accelerating voltage of 10 kV and a spot size of 2.5 nm. Focused ion beam scanning electron microscopy (FIB‐SEM) with a FEI Helios Nanolab 660 DualBeam was also used as a complementary technique to obtain images of a higher resolution. The MN cross sections were first mounted on a 45° SEM stub and coated with 0.5 nm thickness of platinum. Then, the cross‐section was milled with a gallium‐focused ion beam of an accelerating voltage of 30 kV, starting with a beam current at 21 nA. The current was reduced to 9.3 nA, and the final milling was performed at 2.5 nA until the sample surface became smooth (3 – 10 µm). These FIB‐milled regions were then imaged at an accelerating voltage of 2 kV and beam current of 0.4 nA.

Visualization of UCNP@MS Emission in UCNP@MS/MN

UCNP@MS emission from the UCNP@MS/MNs was performed using a hyperspectral confocal microscope (CIMA, Photon Etc., Montreal, Canada) equipped with a femtosecond tunable laser (MaiTai, SpectraPhysics, California, USA). To obtain images, the wavelength of the laser source was tuned to 980 nm to excite the particles, and luminescence signals were collected through a 300 mm triple grating imaging spectrometer (Acton SpectraPro SP‐2300, Princeton Instruments, USA) in a wavelength range of 200 – 850 nm with an exposure time of 0.01 s.

Synthesis of UCNP@MS+Cy5‐PEG/MNs

To characterize Cy5‐PEG flow through the UCNP@MS/MNs, Cy5‐PEG was co‐loaded into the UCNP@MS/MNs to simulate the condition of the MNs after exposure to NIR irradiation, in which the UCNP@LBMS had broken to release free drug within the MNs.

The UCNP@MS+Cy5‐PEG/MNs were synthesized by modifying section 1.9 in the following ways. After the tips were dried, an additional 5 mL of the Gantrez polymeric solution with 25 mg UCNPs and 1.25 mg Cy5‐PEG was centrifuged into the molds, and the rest of the procedure was followed as outlined in section 1.9.

In Vitro Evaluation of Cy5‐PEG Release from UCNP@MS+Cy5‐PEG/MNs

Cy5‐PEG flow through the UCNP@MS+Cy5‐PEG/MNs was evaluated by immersing the MNs in PBS with 5 % w/w bovine serum albumin (BSA) to simulate the interstitial fluid environment in the skin.^{[ 84 ]} The UCNP@MS+Cy5‐PEG/MNs synthesized in section 1.11 were immersed in 50 mL falcon tubes filled with 5 mL of BSA‐PBS. At each time point, 450 µL of BSA‐PBS was removed and replaced with fresh liquid to model the introduction of fresh fluid in vivo. The Cy5 fluorescence intensity (λ_ex = 630 nm, λ_em = 675 nm, gain 125) in the BSA‐PBS aliquots was measured using a Spark multimode fluorescence microplate reader (Tecan Group, Ltd., Mannedorf, Switzerland).

Preparation of PVA‐PVP Dissolving MNs for Development of Ex Vivo Calibration Curve

To quantify the amount of drug released from the MNs into an ex vivo porcine skin model, a calibration curve of Cy5 fluorescence in porcine skin was developed using PVA‐PVP dissolving MNs containing known amounts of Cy5‐PEG. After these known amounts of Cy5‐PEG were released into the ex vivo skin model, a calibration curve was created that correlated intensities read from a custom‐built portable fluorescence reader^{[ 45 ]} with these known amounts (laser diode λ_ex = 638 nm, band‐pass filter λ_em = 668 ‐ 702 nm). This calibration curve was used in all subsequent ex vivo and in vivo experiments to quantify drug release in the skin via Cy5 fluorescence.

Dissolving MNs were produced using a procedure previously described by Babity et al.^{[ 85 ]} Briefly, PVP K‐12 (0.8 g) and PVA (0.8 g) were added to 2 mL of distilled H₂O and mixed thoroughly. This mixture was heated in an oven at 80 ^°C for 30 ‐ 40 min, then centrifuged at 4000 RPM for 5 min to collect the polymer solution at the bottom of the tube. Cy5‐PEG was added to the matrix solution to make varying concentrations of Cy5‐PEG solution (0.05, 0.1, 0.25, 0.35 mg mL⁻¹). Using a 1 mL syringe, roughly 100 µL of the Cy5‐PEG matrix was cast into the PDMS molds, and these molds were secured with tape in 6‐well cell culture plates (Sarstedt AG & Co., Numbrecht, Germany) then centrifuged at 4000 RPM for 5 min. The addition of Cy5‐PEG matrix to the molds and centrifugation was repeated 3 more times with a 180^° rotation in between each centrifugation step. Any excess Cy5‐PEG matrix was removed from the molds with a spatula, after which they were placed in a vacuum chamber at 150 mbar for 30 min. Finally, a backing layer of the polymeric solution without Cy5‐PEG was added to the top of the molds, which were left to dry for 48 h at 25 ^°C before demolding.

The actual amount of loaded Cy5‐PEG in the dissolving MNs of varying Cy5‐PEG concentrations was verified by re‐dissolving each MN patch in 5 mL distilled H₂O (n = 3) and quantifying Cy5 fluorescence (λ_ex = 630 nm, λ_em = 675 nm) using a Spark multimode fluorescence microplate reader (Tecan Group, Ltd., Mannedorf, Switzerland). The Cy5‐PEG concentration thus measured was then correlated to the fluorescence measured in porcine skin as follows.

Prior to use, the skin was thawed in PBS and shaved. Baseline values of porcine skin fluorescence were first obtained (n = 3) with the portable fluorescent reader (laser diode λ_ex = 638 nm, band‐pass filter λ_em = 668 ‐ 702 nm). Then, one dissolving MN patch of each Cy5‐PEG concentration was applied to the porcine skin with 10 s of manual pressure and allowed to dissolve for 30 min. After removal of the MN from the skin, fluorescence readings were taken again (n = 3) with the portable reader, and the difference in fluorescence intensities before and after MN insertion was plotted against the known Cy5‐PEG concentrations to create an ex vivo calibration curve of Cy5 in porcine skin.

Ex Vivo Evaluation of NIR‐Triggered Cy5‐PEG/Cy5‐Leuprolide Release into Porcine Skin

To determine the amount of drug released from the UCNP@LBMS‐Cy5‐PEG/MN and UCNP@LBMS‐Cy5leu/MNs after NIR irradiation, 25 mg of UCNP@LBMS loaded with either Cy5‐PEG or Cy5‐leuprolide, as prepared in section 1.8, were then loaded into the MNs via the layered method as outlined in section 1.9.2.

Then, the MNs were applied to porcine skin using 10 s of manual pressure and either kept in the dark for the control condition or exposed to 2 W cm⁻² of NIR irradiation (976 nm 10 W multi‐function detachable diode laser, BWT Beijing, China) for different amounts of time (15, 30, 60, 90, 120 min). The differences in fluorescence intensities of the porcine skin as measured by the portable fluorescence reader before and after MN insertion in both control and test conditions were used to calculate the amount of drug released via the calibration curve obtained in section 1.13.

$$\begin{array}{ccc}& & \frac{\%ofleuprolidereleasedafterxmin}{MNpatch}\hfill \\ & & =\frac{Massofleuprolideinskinafterxmin}{\frac{Massofleuprolide}{UCNP@LBMS}\times \frac{ActualUCNP@LBMSmass}{MNpatch}}\hfill \end{array}$$ (2)

As the microneedle tips swelled after insertion into porcine skin, they could not be re‐inserted after removal, requiring a new MN patch for each time point. Thus, 3 MN patches were used to generate the average amount and standard deviation of drug released into the skin at each time point. Optical microscopy images of the MNs were taken before and after skin insertion, and ImageJ was used to calculate the average height difference after skin insertion.

Microwave Plasma Atomic Emission Spectroscopy (MP‐AES) Analysis of NP Concentration in Organ Tissue

To examine organ tissue after MN insertion for the presence of UCNP@LBMS, the organ was first digested in 10 mL 6 M HCl at 100 ^°C overnight. Then, the liquid was filtered through a 0.2 µm syringe filter, and 1 mL of the filtered solution was added to 9 mL 2% HNO₃ before Yb analysis with MP‐AES (Agilent 4210 MP‐AES, Agilent, USA). A calibration curve of Yb³⁺ emission at 328 nm was obtained using standard concentrations of 0, 1, 2.5, and 5 ppm (Figure S11a, Supporting Information).

In a control experiment to confirm that MP‐AES could reliably detect the presence of UCNPs using Yb³⁺ detection at 328 nm, a known amount of UCNPs was digested using the same procedure, and the Yb³⁺ concentration was measured. The Yb³⁺ concentration in the digested UCNPs was close to the theoretical 25% used for the UCNP synthesis and increased linearly with increasing UCNP concentrations, suggesting complete decomposition of UCNPs after this strong acid treatment (Figure S11b, Supporting Information). Four UCNP@MS/MN patches with a theoretical UCNP@MS loading of 6.25 mg MN⁻¹ were similarly digested to analyze the actual UCNP concentration within each patch, which was calculated to be 0.87 ± 0.01 mg MN⁻¹ (Figure S10c, Supporting Information). The approximate sevenfold reduction of UCNP from theoretical to actual loading was due to the loss of polymeric solution during the centrifugation process, as only a small amount was centrifuged into the molds while the rest was discarded.

In Vivo Evaluation of NIR‐Triggered Cy5‐Leuprolide Drug Release in Rats

The animal experiment protocol was approved by the Animal Care Committee of the Faculty of Pharmacy at the University of Montreal (protocol: 22–070). All procedures were conducted in accordance with the relevant guidelines and regulations. Sprague Dawley rats (females, 151–200 g, Charles River Laboratories, Saint Constant, QC., Canada) were used to evaluate the Cy5‐leuprolide release in rat skin and plasma after NIR exposure. An acclimation period of 1 week was allowed before the beginning of the study. The day before the MN application, the dorsal hair of each rat was shaved using an electric razor under isoflurane‐induced anesthesia (3 – 4% for induction, 2 – 3% for maintenance), and depilatory cream was applied for 1 min.

On the day of the experiment, the fluorescence of the dorsal area before MN application was measured three times with the portable fluorescence reader to obtain a baseline fluorescence for each rat. Then, two patches of UCNP@LBMS‐Cy5leu/MNs, each containing a theoretical amount of 6.25 mg UCNP@LBMS loaded with 1.25 mg Cy5‐leuprolide, were inserted into the rat skin using 10 s of manual pressure. The procedure was done under isoflurane‐induced anesthesia. After insertion, the patches were secured using medical tape to ensure that the MNs remained adhered to the skin of the rat. For the control group, plain MNs (no Cy5‐leuprolide or UCNP@LBMS) were used, while for the Cy5‐leu/MN group, 1 mg of Cy5‐leuprolide was loaded per MN patch to simulate the condition where all the loaded Cy5‐leuprolide was released from the UCNP@LBMS. 3 rats were used per condition.

The rats with the UCNP@LBMS‐Cy5leu/MNs were further divided into 2 groups – one group was exposed to 5 min of NIR irradiation (2 W cm⁻²) on each MN patch, while the other was not exposed to any NIR light. Then, at each time point (30 min, 1 h, 2 h, 4 h, 24 h, 48 h), 150 µL of blood was taken from the saphenous vein using a microvette‐EDTA (Fisher Scientific) and immediately centrifuged at 3200 RPM for 5 min at 4 °C. The plasma fraction was removed and placed in a microcentrifuge tube. 10 µL of plasma at each time point was used for analysis of Cy5 fluorescence using the fluorescence microplate reader to determine Cy5‐leuprolide concentration in plasma. The NIR/no NIR experiment was also repeated with UCNP@LBMS‐Cy5leu/MNs with a needle height 800 µm. Finally, the plasma from 3 rats with no treatment was measured to evaluate background fluorescence.

After 48 h, the rats were sacrificed, the MN patches were removed from the skin, and the organs (heart, lung, liver, spleen, kidney) were imaged via a LabeoTech OiS300 In Vivo Imaging System (IVIS, Labeo Technologies Inc.) with excitation and emission filters of 655 and 716 nm respectively before digestion and analysis for the presence of UCNPs (section 2.15).

Images of the rat organs (heart, lung, liver, spleen, kidney) taken by the IVIS microscope were saved in. TIFF format and then analyzed using ImageJ. To ensure consistent display of all images, the images were converted to 12‐bit images with a grayscale set to a linear scaling from min‐max to 0–10000, a range that was then propagated to all images. Furthermore, a threshold was empirically selected by analyzing the images one by one, resulting in the range of 1000–3800, which most clearly identified the increase in fluorescence in the organs from all rat experiments, while minimizing the background variations between images. Once all images had the same threshold and grayscale, to measure the mean fluorescence intensity was proceeded after selecting the image area representing the fluorescence from all the ex vivo organs extracted from one rat.

To evaluate any damage to the skin caused by the treatment, the section of the skin penetrated by the MN was excised and prepared for histological analysis by fixation in 10 % formalin before sectioning into 5 µm sections (microtome RM2255, Leica) and staining with hematoxylin and eosin on an automated slide stainer (DRS‐601, Somagen) using conventional protocols.

Statistical Analysis

All data were presented as the mean and standard deviation, and all statistical analysis was performed in GraphPad Prism 8 (GraphPad Software Inc.). For each experiment, the type of statistical test used for the comparison is included in the figure caption. A p‐value less than 0.05 is denoted with one star (*), a p‐value less than 0.01 with 2 stars (**), and a p‐value less than 0.001 with 3 stars (***).

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Tam V., Trana R., Nieto‐Arguello A., et al. “Upconverting Nanoparticle‐Loaded Microneedles for Near‐Infrared Responsive Delivery of Gonadotropins to Increase Success of In Vitro Fertilization.” Small 22, no. 1 (2026): e13138. 10.1002/smll.202513138

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.