Plain PLGA NPs produced by the methodology followed in this work are much widely characterized in the literature^11,17–19. Specifically, on THC-PLGA NPs, a thorough characterization was previously reported by our team^11,12, while the attachment of Tf to PLGA NPs through the carbodiimide strategy is also a commonly stablished approach^20–23. Thus, the physicochemical characterization of the NPs produced in this work was focused on confirming the successful production of plain and Tf-modified formulations, to subsequently analyze the influence of the Tf modification on the later performance of the NPs.

Regarding previous works, the results of size and ZP of the plain PLGA NPs and THC-PLGA NPs here presented correlate with them^11,17–19. The similar values of size and ZP of plain blank and plain THC-loaded PLGA formulations (PLGA NPs and THC-PLGA NPs) indicate a negligible drug adsorption of Δ⁹-THC onto the surface of NPs, being the majority of drug molecules incorporated into the NPs matrix¹¹. In addition, the high EE values obtained are supported by the lipophilic character of the drug, which prevents its partitioning into the aqueous phase during the formulation process^11,24.

Tf interaction with its receptor has been a widely established and exploited approach for an effective cellular specific uptake for several drugs aiming to target cancer cells and the blood brain barrier (BBB) mainly^25–28, due to the differentiated high expression of the receptor in the target locations. In this work, Tf was coupled to plain PLGA NPs and THC-PLGA NPs via a carbodiimide strategy, a mild conjugation approach. High yields of Tf-coupling to the surface of NPs were obtained (Table 1), which are in agreement with the data reported in the scientific literature obtained with the same carbodiimide strategy²². The decrease in ZP values of NPs after Tf-coupling (Table 1) was expected since this chemical modification implies the participation of carboxylic acid groups on the surface of NPs that account for the negative surface charge prior to Tf-coupling, and thus this ZP modification further confirmed the attachment of Tf ligands to the NPs^22,29,30.

The similarity between the EE values obtained with plain THC-PLGA NPs and Tf-THC-PLGA NPs indicate that no leakage of encapsulated drug is to be expected from the Tf-coupling procedure, most probably due to the same reduced probability of the partitioning of the lipophilic drug to the aqueous media responsible for the high EE and DL values^11,31,32.

Concerns often raise in scientific discussion when it comes to label NPs by conjugating a fluorophore to their constitutive materials or by encapsulating it; a conjugated tracking molecule would provide information about where the carriers or their degradation products are located, but not about the delivered drug. On the other hand, encapsulated fluorophores would provide an approximation of the location and distribution expected from the encapsulated drug, but not about the fate of the carrier itself. In this work, a double labeling was proposed in order to differentiate between the tracking of the NP itself and its cargo; a fluorescent dye was conjugated to the backbone of the polymer to track the NPs, while additional dyes were encapsulated to illustrate the expected behavior of encapsulated drugs. Given the lipophilic character of Δ⁹-THC, a lipophilic dye NR was expected to provide a closer resemblance to the delivery process of these formulations. Nonetheless, both a lipophilic and a hydrophilic fluorophores were alternatively used to also allow for a more accurate comparison with the available literature, where all these labeling strategies are used indistinctively.

The same carbodiimide approach followed for Tf-conjugation was applied to the attachment of FITC to the polymer backbone prior to the production of NPs³³, leading in this case to a fluorescently-labeled polymer. Although a higher mass ratio of coupled molecule with respect to the polymer was obtained with FITC (12.90 μg/mg of NPs for FITC vs. 5.1–7.8 μg/mg of NPs for Tf, Table 1), it should be noted that all the Tf molecules were conjugated to the surface of NPs, while FITC molecules where conjugated to the polymer before NPs production and thus a high percentage of molecules are expected to be finally located in the core of NPs. Nonetheless, considering the initial amount of FITC employed in the reaction, the final yield of product proved to be significantly less efficient than in the case of Tf (1.29% coupled FITC vs. 59–71% coupled Tf). However, the final mass ratio of conjugated FITC proved to be sufficient for NP tracking while not inducing differences in particle size after NPs production (Table 1). The ZP of FITC-NPs presented a decrease with respect to PLGA NPs (− 30 mV of PLGA NPs vs. − 18 mV of FITC-PLGA NPs, Table 1), which again is consistent with the implication of negatively charged PLGA carboxylic acid groups in the conjugation reaction. The ZP decrease due to the coupling of FITC to PLGA is consistent with the literature³⁴, although comparisons should be made carefully when the conjugation is addressed with different coupling reactions.

For the double fluorescent labeling of NPs, NR was encapsulated in FITC-PLGA NPs as a lipophilic model molecule, and Rh as a hydrophilic model molecule^11,18,35. The particle size and ZP values obtained with both NR- and Rh- loaded FITC-PLGA NPs were similar to blank FITC-PLGA NPs (p < 0.05) (Table 1), indicating that no molecules of fluorophores were expected to be located adsorbed on the surface of NPs but in the inner core of the carriers. The subsequent Tf-coupling to the surface of double-labeled NPs led to a decrease in ZP values with respect to FITC-PLGA NPs, either blank, NR- or Rh- loaded, which was once again expected and attributed to the neutralization of FITC-free carboxylic acid groups on particle surface after the attachment of the protein by conjugation. The lower values of CE% of Tf obtained with Tf-FITC-PLGA NPs with respect to Tf-PLGA NPs (Table 1) were attributed to the lower number of PLGA carboxylic acid groups available for Tf-conjugation, since they are also the substrate of the FITC-conjugating reaction that takes place in a step previous to NP production and Tf-coupling. In the specific case of Tf-Rh-FITC-PLGA NPs, which led to the lowest CE% value (51%), the decrease was attributed to residual molecules of Rh that could remain adsorbed on the surface of the NPs due to the partitioning to the aqueous phase during NPs production, effect that would be minimize in the case of NR due to its lipophilic character.

Finally, SEM imaging (Fig. 1) revealed a monodispersed population of spherical nanostructures in agreement with the DLS characterization results (Table 1) and our previous works on THC-PLGA NPs^11,12, with no observable difference induced by the Tf coupling procedure or the presence of this moiety on the nanoparticle surface.

The biphasic release profile exhibited by both THC-PLGA-NPs and Tf-THC-PLGA (Fig. 2) was in agreement with previous results¹¹, where the initial fast release phase was attributed to the loss of drug molecules located near the NPs surface and the following slower phase attributed to polymer degradation and/or drug diffusion through the polymeric matrix. The lack of significant differences between plain and Tf-modified NPs showed that the Tf-coupling procedure did not affect the release profile of the formulation, and hence probably neither the initial drug distribution in the polymer matrix. Again, this could be due to the lipophilic character of the drug and subsequently minimized probability of partitioning to the reaction aqueous medium^11,31, which is also consistent with the preserved values of EE and DL.

Caco-2 cells are heterogeneous human epithelial colorectal adenocarcinoma cells employed as colorectal cancer (CRC) cell model expressing CB₁ subtype cannabinoid receptors (CBRs)^36–39 as well as cell model expressing Tf receptors^40–42, and therefore they were selected as cell line for the studies. Cell viability assays confirmed the absence of toxicity of blank formulations, widely reported in the literature; PLGA is a biocompatible and biodegradable polymer and thus no toxicity was expected from PLGA-based nanoparticles^10,17,18,43. On the other hand, free Δ⁹-THC, presented a statistically significant increase (p < 0.05) in cell viability at the first time point of incubation assayed (2 h) (Fig. 3) at 100 and 150 μM, which is consistent with previous results reported by our group^11,12. It should be considered that, although cannabinoids are known to inhibit cell growth and induce apoptosis in tumour cells⁴⁴, their effects on cell viability and proliferation are dependent on the specific cell culture assayed and the doses administrated^36,45,46. Specifically, the dose employed is known to be crucial; low cannabinoid concentrations enhance cell proliferation in vitro through the stimulation of several pathways⁴. In in vivo conditions, this pro-tumorigenic effect is not observed, but the activation of these cellular pathways due to low cannabinoid concentration leads to the generation of resistance to THC antitumor action^4,47. In this work, the encapsulation of Δ⁹-THC into PLGA NPs avoided this cell proliferation effect from free Δ⁹-THC in vitro at the first time point assayed (2 h, Fig. 3), while decreasing cell viability at the highest concentration tested (150 μM) at 24 h and 48 h compared to the free drug (24 h and 48 h, Fig. 3). These observations are in agreement with a previous work from our group¹¹, where longer incubation times enabled a higher cytotoxic effect from plain THC-PLGA NPs compared to free Δ⁹-THC. These results are attributed to the controlled release of the drug from the nanoparticles, and hence lower but sustained delivered dose of cannabinoid to the cells¹¹. Therefore, testing whether this effect is translated to in vivo would represent an interesting strategy against the development of resistance to Δ⁹-THC antitumor action.

Furthermore, Tf-THC-PLGA NPs provided a noticeable different result. This formulation did not only avoid the cell proliferation effect of free Δ⁹-THC at the first 2 h of incubation, but also provided decreased viability at 24 h (from 75 to 150 μ equivalent concentration), achieving the lowest cell viability values obtained in the study at 48 h (16% at the maximum dose analysed). These are considerably much shorter time periods than needed to achieve a similar effect with either Δ⁹-THC or plain THC-PLGA NPs (up to 3.5-fold)^11,12.

For a better understanding of the mechanisms by which the Tf surface modification improved the cytotoxicity profile of the nanocarriers to such extent, assays with double-fluorescently-labeled NPs, either plain or Tf-modified, were carried out. When comparing the data reported in the literature, careful consideration must be paid to the details of experimental procedures; the variations on the methods of fluorescent labeling (encapsulation, chemical conjugation to the carrier constitutive materials, conjugation to encapsulated drugs, conjugation to targeting molecules, etc.), targeting molecule incorporation to the system (chemical coupling reaction, surface adsorption, dye encapsulation), the cell line assayed, incubation times and even the type of cell culture support (smooth or permeable, with or without additional chambers for drug diffusion), to mention a few, may induce differences in the output to a greater or lesser extent^48,49. In this work, the double fluorescent labeling strategy presented, along with the use of either lipophilic or hydrophilic encapsulated dyes, and performing uptake studies with a wide time assay period, were meant to provide a wider perspective and comparison with scientific literature records.

Size, surface charge and hydrophobicity of NPs are widely acknowledged to influence the endocytic pathway in cells⁵⁰. These parameters were considered similar for this matter between the formulations assayed in this piece of work (Table 1), and thus they were not regarded as influencing variables in the discussion of uptake studies.

Overall, the uptake of plain PLGA NPs has been identified in the literature as a combination of fluid phase pinocytosis and clathrin-mediated endocytosis. The first one would follow a linear progression while the latter is saturable, resulting in biphasic kinetics of cellular uptake⁵¹. The nanoparticles would be found later in lysosomes, from which they would escape to the cytosol as a result of the protonation of carboxylic groups in the acidic environment, exposing a hydrophobic surface that would interact with the membrane of the vesicles^52,53. In addition, biphasic kinetics of PLGA NPs exocytosis have also been proposed, which would involve both endosomal escape to cytosol and exocytosis and also a recycling endosomal pathway, the latter avoiding passage through the cytosol⁵¹. On the other hand, Tf and Tf-coupled molecules interact with the Tf receptor and are subsequently internalized via clathrin-mediated endocytosis to follow the transferring recycling pathway, according to the literature^54,55. Overall, two recycling pathways are differentiated; a faster one involving early endosomes and recycling endosomes, and a slower one involving lysosomes and the scape of the material to the cytosol followed by exocytosis⁵⁴. As a general rule, recycled material follow both pathways with a preference related to its molecular weight and diffusion rate, where small molecules would follow the faster route and large molecules the slower one⁵¹. Hence, Tf-modified nanoparticles would be expected to follow the slow recycling pathway, also involving lysosomal scape to the cytosol as their non-modified counterparts, with the difference of the involvement of the Tf-receptor mediated interaction. Finally, it is worth considering that nanoparticles decorated with a receptor targeting ligand are generally expected to be uptaken at a higher rate⁵⁵ than their non-modified counterparts.

The results of uptake studies from flow cytometry assays are depicted in Fig. 4. The uptake of FITC-PLGA NPs assumed from the fluorescent signal of FITC (Fig. 4a) revealed a time-dependent uptake of NPs with a rapid exponential phase in the first 2 h that then turned into a plateau, in agreement with the above discussed literature. The correspondent NR and Rh signals from the same formulations showed a similar profile (Fig. 4b), in accordance with published works. For instance, a linear time-dependent uptake profile was reported by Reix et al.⁵⁶ with FITC-insulin loaded PLGA NPs in Caco-2 cultures in a 6 h study. Other sources reported different outcomes nonetheless, which could be ascribed to several experimental variables. Prior studies in our group presented a near 0% uptake values of NR-loaded plain PLGA NPs in Caco-2 cells in the first 6 h of incubation, that were increased at 24 h¹¹. This could be probably due to differences in the methodology followed; in the present study, the cells were treated and fixed right at the end of the incubation period. However, in our previous studies¹¹, the NPs suspension was replaced by culture media at the end of each incubation period, in order to treat and analyze the living cells at the end of the study. Exocytosis of nanoparticles from culture cells has been reported in several studies^51,57, especially when the nanoparticles are removed from the incubation media and hence the concentration gradient is reversed^56,58. Therefore, and regardless of the experimental specifics of each case, the maintenance of cells in NP-free media after incubation with the formulations could be overall considered a factor inducing decreased value of total uptake, due to nanoparticle exocytosis.

Surprisingly, Tf-FITC-PLGA NPs, on the other hand, presented a delayed internalization with respect to plain FITC-PLGA NPs (Fig. 4a), with intensity values around 10% after 2 h and maximum values at the final incubation period (48 h), therefore without a conclusive observation of uptake saturation within the time assayed. The corresponding signal from encapsulated dyes (Fig. 4b) showed a similar trend, differentiated from their non-Tf-modified counterparts, but with higher variations between values obtained with either NR or Rh. This observation could be due to some extent of dye release from the particles and its cell internalization dissociated from the carrier. In that case, higher internalization would be expected from NR compared to Rh, due to its lipophilic character, which would explain the slightly higher values of uptake from Tf-NR-FITC NPs vs. Tf-Rh-FITC NPs (Fig. 4b). Overall, in the case of Tf-modified formulations, the tracking of chemically linked dye (FITC, Fig. 4a) vs. encapsulated dyes (NR and Rh, Fig. 4b) led to a similar conclusion, i.e. delayed internalization of Tf-modified formulations vs. plain ones, albeit with differences in the obtained values in the case of loaded dyes. With regard to published literature, even though particle surface decoration with receptor ligands is generally reported to increase uptake rate⁵⁵, some works agreed with decreased uptake due to Tf-surface modification⁵⁹. Specifically, Zhang et al.⁵⁹ reported coumarin-loaded, Tf-modified polyestyrene nanoparticles showing slower but more efficient internalization and faster transfer of the particles from the lysosome to the cytosol in HepG2 cells. In addition, Sahoo et al.⁶⁰ reported a delayed but higher sustained uptake of coumarin and Taxol loaded, Tf-conjugated PLGA NPs vs. plain NPs in MCF-7 cells.

Overall, the clear difference in cell internalization rate of Tf-modified PLGA NPs vs. plain PLGA NPs is especially worth of attention, since it could play an important role in the improved cytotoxicity revealed by the MTT assays. It should be noted that CB₁ receptors, whose activation is responsible for the antitumor effect of Δ⁹-THC^4,61, are located at the surface of Caco-2 cells. Thus, the enhanced effect of Tf-THC-PLGA NPs could be based on a mechanism more complex than an increased intracellular accumulation of NPs, opposed to the underlying principle sustaining other strategies^62,63. Initially, a higher residence time of the particles near the cell surface, and hence a sustained release of cannabinoid closer to its receptor, could be considered a reason for the higher efficacy of the formulation. On one hand, a slower internalization process via Tf receptor-mediated endocytosis vs. non-specific endocytosis could be responsible for this behavior. On the other hand, faster recycling and exocytosis of the particles through a Tf-related pathway could also be considered as potential cause for decreased particle cell accumulation, and hence increased presence in the external medium. Both hypotheses would need further studies to be conveniently assayed.

In an attempt to clarify the mechanisms involved in this interaction, further studies with endocytosis inhibitors were carried out. When hypertonic sucrose media was employed to inhibit clathrin-mediated endocytosis^64,65, the uptake rate of Tf-FITC-PLGA NPs increased (Fig. 4c) at the first time points compared to standard conditions (Fig. 4a), while no significant differences were observed for plain FITC-PLGA NPs. Once again, tracking of NR and Rh signals (Fig. 4d) led to a similar general profile as with FITC tracking, albeit with differences in the obtained values at 1 h and 2 h. Given that the Tf internalization pathway through its Tf receptor involves the formation of clathrin-coated pits^50,62,63, the results obtained supports that the delayed internalization observed for Tf-modified particles in standard conditions, potentially responsible for their enhanced antitumor effect, is due to a Tf-related cellular interaction. On the other hand, the increased uptake rate in the presence of sucrose points out that the nanoparticles are also internalized by alternative routes to clathrin-mediated endocytosis, which would allow for a faster cell uptake. Several endocytosis pathways have been suggested in the literature⁵², especially fluid phase pinocytosis⁵¹, as previously mentioned. While these pathways take place simultaneously in standard conditions, the density of the targeting ligand on nanoparticle surface would determine the predominant pathway⁶⁶. It should be noted that, even when clathrin-mediated internalization is inhibited, including Tf-receptor mediated internalization, the uptake profile of Tf-modified nanoparticles still remained relatively slower than plain nanoparticles (Fig. 4c,d). Therefore, the presence of Tf moieties on particle surface may still exert an influence in non-clathrin-mediated internalization, which could be due to non-specific interactions related to its hydrophilic character, as opposed to plain FITC-PLGA NPs surface.

Following, the inhibition of caveolae-mediated endocytosis was performed by including genistein in the culture media. In this case, plain FITC-PLGA NPs presented a similar profile with respect to standard conditions (Fig. 4e vs. a), while the uptake rate of Tf-FITC-PLGA NPs was increased (Fig. 4e), resembling the results obtained with sucrose hypertonic media (Fig. 4c). Similarly, the values obtained through tracking of encapsulated dyes led to a similar profile to FITC tracking (Fig. 4f vs. e), with some differences between NR and Rh signals. Regarding plain nanoparticles, the results are supported by the previous work from Panyam et al.⁶⁷ with HASMCs cells, where the the inhibition of actin polymerization by cytochalasin D did not affect the uptake of coumarin-6-loaded PLGA NPs, confirming no involvement of caveolae-mediated endocytosis. Regarding Tf-FITC-PLGA NPs, the observed increased uptake was surprising, since Tf-receptor internalization is known to be mediated by clathrin⁵⁴, and even plain PLGA NPs are not expected to be internalized via caveolae-mediated mechanisms^51,53. It should be noted that, although genistein does inhibit SV40 induced vesicle formation from caveolae, and thus it is widely used as inhibitor for caveolae-mediated endocytosis, this inhibition is not strictly specific for caveolae-mediated processes, since it has been reported to inhibit some clathrin-associated routes of receptors that imply tyrosine phosphorylation or F-actin recruitment⁴⁹. Even though no studies are available in the literature to shed light on this specific matter for Tf-modified PLGA NPs interacting with Caco-2 cells, the closest comparison can be found, to the best of our knowledge, in the work of Chang et al.⁶⁸. The authors assayed PLGA NPs coated with adsorbed FITC-Tf on their surface in a co-culture of brain endothelial cells and astrocytes as a blood brain barrier (BBB) cell model. The results showed an inhibition of Tf-adsorbed NPs uptake by filipin, an inhibitor of cholesterol-dependent mechanisms which involve caveolae-mediated uptake as well as other clathrin-mediated mechanisms such as micropinocytosis⁴⁹.

Finally, the near 0% values of fluorescence obtained with the inhibition of energy-dependent processes (Fig. 4g,h) confirmed that the internalization of all the formulations assayed took place by energy-dependent mechanisms irrespective of the Tf-functionalization^69,70.

Qualitative assessment of the interaction of NPs with the cells by fluorescent microscopy showed a similar distribution of both FITC and NR or Rh signals in the cell culture (Fig. 5a), suggesting that the encapsulated dyes were internalized in the cells still loaded in the nanoparticles, with minimal absorption of released dye. It should be considered nonetheless that microscopy imaging provides only qualitative information, and hence some degree of free dye absorption, as flow cytometry results suggested, should not be discarded. Still, in the light of these results, tracking of encapsulated dyes was overall confirmed to be a valid strategy to track the nanoparticle structure inside the cells in this case. In addition, the higher intensity of fluorescence obtained in cells incubated with plain formulations vs. Tf-modified formulations (Fig. 5a,b) was in agreement with the delayed internalization observed by flow cytometry in the case of Tf-modified NPs. On the other hand, the nanoparticles were located both overlapping and not-overlapping with Lysotracker-dyed regions both with plain and Tf-modified formulations (Fig. 5a,c), indicating partial distribution of the nanocarriers inside and outside lysosomes, in agreement with previous studies on PLGA NPs⁵¹. It should be noted that Lysotracker dyed acidic organelles, and thus it could not be assayed whether nanoparticles outside of lysosomes were located inside early or recycling endosomes, or in the cytosol. However, even though fluorescent microscopy imaging did not clarify further the preferred internalization and trafficking pathway of the nanoparticles, it did confirm the involvement of lysosome-associated routes, which is in agreement with the results obtained from flow cytometry analysis.

To sum up, in light of the flow cytometry analysis results and fluorescent microscopy imaging, the internalization mechanism potentially involved in the improved cytotoxic effect of Δ⁹-THC-loaded Tf-PLGA NPs seemed dependent on cholesterol-associated and clathrin-mediated mechanisms, which apply for endocytosis of receptors that are dependent on tyrosine phosphorylation, including the Tf receptor⁵⁴. In addition, both fluorescent labeling strategies, either chemical linkage or dye encapsulation, allowed to withdraw overall similar conclusions, although the latter was subject to differences in the values obtained in each case. Hence, chemical linkage of the fluorophore seems the best suited approach to track the nanoparticle itself, while dye encapsulation seems to provide a better approximation to the fate of the loaded drug.

It should be noted that the presented cell studies were carried out in FBS-free medium, in order to minimize variability from growing cells. The presence of proteins is known to strongly influence the interaction of nanocarriers with cells due to the formation of the protein corona, which is considered nowadays a main factor for discordance between in vitro and in vivo results^71,72. However, in vitro-formed corona has been reported to be radically different from in vivo conditions, thus being the use of FBS strongly discouraged in these studies⁷². Specifically, Tf-modified polyestyrene nanoparticles were recently reported to lose their targeting ability with the formation of an in vitro corona, but retain it in vivo with high affinity to the target cells⁵⁹. Hence, a similar behavior would be plausible in the case of Δ⁹-THC-loaded Tf-PLGA NPs. Nonetheless, additional studies should be performed to further asses the interaction mechanism of the proposed formulation with the target cells, as well as testing it in vivo.