DRUG DELIVERY 2021, VOL. 28, NO. 1, 2383–2391 https://doi.org/10.1080/10717544.2021.1998249 RESEARCH ARTICLE Ion-dependent slow protein release from in vivo disintegrating micro-granules a,b,c   Patricia Alamo , Eloi Parladec,d, Hector Lopez-Lagunac,d,e, Eric Volta-Duranc,d,e, a,b,c,e c,d,e Ugutz Unzueta , Esther Vazquez , Ramon Manguesa,b,c and Antonio Villaverded,e a Biomedical Research Institute Sant Pau (IIB Sant Pau), Barcelona, Spain; bJosep Carreras Leukaemia Research Institute (IJC Campus Sant Pau), Barcelona, Spain; cCIBER de Bioingenierıa, Biomateriales y Nanomedicina (CIBER-BBN), Madrid, Spain; dInstitut de Biotecnologia i de Biomedicina, Universitat Autonoma de Barcelona, Bellaterra, Spain; eDepartament de Genetica i de Microbiologia, Universitat Autonoma de Barcelona, Bellaterra, Spain ABSTRACT ARTICLE HISTORY Through the controlled addition of divalent cations, polyhistidine-tagged proteins can be clustered in Received 3 October 2021 form of chemically pure and mechanically stable micron-scale particles. Under physiological conditions, Revised 14 October 2021 these materials act as self-disintegrating protein depots for the progressive release of the forming Accepted 18 October 2021 polypeptide, with potential applications in protein drug delivery, diagnosis, or theragnosis. Here we KEYWORDS have explored the in vivo disintegration pattern of a set of such depots, upon subcutaneous adminis- Protein materials; tration in mice. These microparticles were fabricated with cationic forms of either Zn, Ca, Mg, or Mn, microparticles; protein which abound in the mammalian body. By using a CXCR4-targeted fluorescent protein as a reporter depots; self-disintegrating building block we categorized those cations regarding their ability to persist in the administration site materials; tumor targeting and to sustain a slow release of functional protein. Ca2þ and specially Zn2þ have been observed as particularly good promoters of time-prolonged protein leakage. The released polypeptides result is available for selective molecular interactions, such as specific fluorescent labeling of tumor tissues, in which the protein reaches nearly steady levels. Introduction non-drug-containing material that increases the complexity of the system and the fabrication process, often imposing Many pharmacological treatments for chronic diseases, such chemical constraints and toxicological concerns (Sharma as cancer would benefit from a time-prolonged drug supply et al., 2012; Palombo et al., 2014; Aragao-Santiago et al., aiming at reaching constant or nearly constant levels at the 2016; Shen et al., 2017). Recently, self-contained, self-disinte- site of action (Natarajan et al., 2014; Koshy et al., 2018; Cross grating protein material in form of microparticles have been et al., 2019). This is in contrast with the common therapeutic developed (Chen et al., 2020; Sanchez et al., 2020), suited for protocols that are based on repetitive drug administrations, a slow protein release in vivo (Sanchez et al., 2020; Serna frequently given with a few-day time intervals, resulting in pez-Laguna et al., 2021). These artificial struc- et al., 2020; Lo oscillating drug concentrations and irregular therapeutic tures mimic the molecular organization and functionality of impact (Wen et al., 2015; Zou et al., 2020). Approaching the secretory granules from the mammalian endocrine sys- steady drug concentrations in blood is expected to minimize tem, which contain and release peptidic hormones to the side effects in off-target tissues and to support a potent bloodstream (Maji et al., 2009; Mankar et al., 2011; Jacob therapeutic action (Rosen & Abribat, 2005; Wen et al., 2015; et al., 2016; Jacob et al., 2019). Such artificial material is Li et al., 2019; Zou et al., 2020), while non-repetitive adminis- based on pure preparations of a single polypeptide species trations would reduce the exploitation of sanitary resources with a fused histidine-rich peptide, and that is clustered as (Pareek et al., 2019). Diverse strategies for sustained drug granules around the microscale, by the external addition of delivery are under exploration, mainly based on porous Zn2þ. The divalent cation generates cross-molecular interac- materials, hydrogels, matrices, or other types of macroscopic, tions between histidine-rich domains of adjacent polypeptide microscale, or nanoscale containers that hold the drug for its chains, which remain attached by such interactions in form progressive leakage (Gilmore et al., 2016; Li & Mooney, 2016; of a mechanically stable protein network (Lo pez-Laguna Ali & Ahmed, 2018; Ghalei et al., 2018; Koshy et al., 2018; Wu et al., 2020). Upon in vitro incubation under physiological et al., 2018; Meng et al., 2019). Such methodologies involve a conditions or by subcutaneous administration in vivo, the CONTACT Ramon Mangues rmangues@santpau.cat Biomedical Research Institute Sant Pau (IIB Sant Pau), Sant Antoni Ma Claret 167, Barcelona, 08025, Spain; Antonio Villaverde antoni.villaverde@uab.cat CIBER de Bioingenierıa, Biomateriales y Nanomedicina (CIBER-BBN), C/Monforte de Lemos 3–5, Madrid, 28029, Spain These authors contributed equally to this work. Supplemental data for this article can be accessed here. ß 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2384 P. ÁLAMO ET AL. protein granules spontaneously disintegrate, probably by Size determination of secretory granules and the progressive chelation of the clustering ions. This fact allows a released soluble protein slow leakage of the protein building blocks, in a functional The hydrodynamic diameter (in nm) of secretory granules form, ready for interactivity or any other biological activity. and soluble protein released in vitro was determined by Divalent non-toxic cations other than Zn2þ are, in principle, dynamic light scattering (DLS) at 25  C (633 nm), run duration pez-Laguna et al., potentially suited for protein clustering (Lo (0.839 s), the number of runs 15, using the forward scatter 2020). By exploring these alternatives, a diversity of related (for secretory granules) and backscatter (for released soluble materials could be generated whose properties, regarding protein) detectors, in a Zetasizer NanoZS (Malvern the kinetics of protein release, functionality and bioavailabil- Instruments Limited) using ZEN2112 3 mm quartz ity, are presumed to be differential. batch cuvettes. In vivo kinetic biodistribution of fluorescent material by subcutaneously implanted granules Materials and methods In vivo experiments were approved by the Animal Ethics Manufacture of secretory granules and release of Committee at Hospital de la Santa Creu i Sant Pau (proced- soluble protein ure 115_9721) and performed according to European Council Pure soluble T22-GFP-H6 [Figure 1(A), fully described else- directives. To generate the CXCR4þ SW1417 CRC cancer where (Rueda et al., 2015)], was aliquoted in 250 mL at 2 mg/ model, four-week-old female mice of the Swiss nude strain, mL in the storage buffer (166 mM NaCO3H þ 333 mM NaCl). in the 18–20 g body weight range (Charles River, L-Abreslle, Protein precipitation as secretory granules was induced by France) and maintained in pathogen-free conditions, were the direct addition of divalent cations to protein solutions. used. We injected subcutaneously in the mouse flank (n¼) 5  106 CXCR4þ SW1417 CXCR4þ human CRC cells, resus- Different cation: protein molar proportions (and therefore pended in 150 lL of media. When tumors reached approxi- working ion concentrations) were used for the different types mately a 120–200 mm3 volume, animals were randomly of granules, depending on the crosslinking ions. The molar allocated to the different groups and administered in the protein amount refers to the molar amount of histidine resi- subcutis of the mouse lumbar region, at the side contralat- dues in the overhanging H6 tag (that is, the raw protein eral to the tumor site, with a pellet of 1 mg of T22-GFP-H6 molar value x 6). For the construction of Zn-based depots granules suspended in a 150 lL PBS buffer. Inclusion bodies we used a molar ratio 50:1 (at 10 mM Zn2þ); for ZnMn-based formed by a recombinant T22-GFP-H6 and purified from bac- depots, the ratios were 30:1 (Zn2þ) and 70:1 (Mn2þ) (at 6 mM teria or PBS buffer were injected as controls. Zn2þ and 14 mM Mn2þ); for ZnMg-based depots, 30:1 Zn2þ Ten minutes, 5, 10, 24, 48, 120, or 600 h after the adminis- and 470:1 Mg2þ (at 6 mM Zn2þ and 94 mM Mg2þ); for MnMg tration, mice were registered in vivo. The mouse was euthan- depots, 150:1 Mn2þ and 350:1 Mg2þ (at 30 mM Mn2þ and ized, and the kidney, liver, tumor, and tissues surrounding 70 mM Mg2þ); for Ca-based depots, the ratio was 350:1 the injection point were resected. Following, we ex vivo reg- (using 70 mM Ca2þ); for CaMn-based depots, ratios were istered the intensity of the fluorescence emitted by the pro- 200:1 Ca2þ and 100:1 Mn2þ (at 40 mM Ca2þ and 20 mM tein released by the protein depots and biodistributed to the Mn2þ); for CaZn-based depots, 270:1 Ca2þ and 30:1 Zn2þ (at tumor and non-tumor organs and also the fluorescence 54 mM Ca2þ and 6 mM Zn2þ) (see Figure 1(B)). Proportion 1:1 remaining the injection site, using the IVISV 200-Spectrum R refers to 0.2 mM of both protein and ion. Samples were then (PerkinElmer, Waltham, MA, USA). The fluorescent signal was incubated at room temperature for 10 min and subsequently digitalized, displayed as pseudocolor overlay, and expressed centrifuged at 10,000 g and room temperature for 10 min to as radiant efficiency. All experimental data points were cor- separate the soluble from the insoluble protein fractions. The rected by subtracting the mean average radiant efficiency of insoluble fraction (namely secretory granules) was collected each respective tissue from mice injected with PBS. and stored at 80  C for further use, and the soluble fraction Preliminary screening was conducted with one mouse per was quantified by Bradford’s assay (Maniatis et al., 1989) to condition while the final ex vivo assay was, at least, per- determine the amount of precipitated protein in mg. formed in duplicate. Bacterial inclusion bodies formed by T22-GFP-H6 were pre- pared by standard bacterial production procedures Statistical analyses (Cespedes et al., 2020). The release of soluble protein from secretory granules All statistical analyses and data representation were per- was triggered in vitro by the direct addition of 250 mL of stor- formed in Graph Pad Prism (v8.0.2). age buffer (166 mM NaCO3H þ333 mM NaCl) into the thawed insoluble fraction (namely secretory granules). Samples were Results and discussion properly mixed using a pipette for several minutes and cen- trifuged at 10,000 g and room temperature for 10 min to col- To refine the design of such disintegrable protein drug lect the soluble fraction (namely released soluble protein). depots with potential for therapeutic applications we have DRUG DELIVERY 2385 Figure 1. Formulation and physicochemical characterization of secretory granules using different combinations of divalent cations. (A) Schematic representation of the manufacturing process. Architectonically stable protein stages are depicted down below each picture. More details about the organization of final and inter- mediate materials can be found elsewhere (L opez-Laguna et al., 2021). (B) Methodological procedure of secretory granules displaying different types of divalent cations and concentrations. Pictures of the resultant pellets are also displayed. (C) Size determination (in nm) by DLS after the preparation of secretory granules (black). Size determination upon sonication (1 min, 10% amplitude, and 0.5 s on/off) to address mechanical stability (red). SG refers to Secretory Granule. Displayed values correspond to peak sizes. (D) Size determination (in nm) by DLS of released soluble protein from secretory granules (black). Size determination upon thermal exposure (37  C for 24 h) to address thermal stability (brown). All measurements were performed in sextuplicate, and data was represented as mean ± SEM (stand- ard error of the mean). 2386 P. ÁLAMO ET AL. explored here the use of different divalent cations for the generated granules, with sizes ranging between 300 and construction of secretory granules of a reporter fluorescent 600 nm, resulted mechanically stable as they resisted sonic- protein (T22-GFP-H6, Figure 1(A)) produced and purified ation with only slight size modifications (Figure 1(C)). Also, from recombinant bacteria. This has been done to compara- they exclusively leak, in vitro, nanoparticles (but not mono- tively evaluate how the forming building blocks leak from mers) of around 13–15 nm that are also structurally stable the depot at the subcutaneous injection site. Several divalent upon in vitro incubation in the physiological buffer for at cations, namely those of Zn, Ca, Mn, and Mg, found at rela- least 24 h at 37  C (Figure 1(D)). This observation suggested tively high concentrations in living beings, were selected to that the leaked material could be also stable in physio- prevent potential toxicity issues linked to the molecular glue. logical fluids. These physiological linkers were applied to generate, in vitro, For a fast initial screening, 1 mg of each sample was sub- secretory granules of the modular protein T22-GFP-H6, cutaneously injected in the CXCR4-expressing SW1417 through a process in which homomeric nanoparticles are human colorectal cancer mouse model, as a single dose, in a intermediates (Lo pez-Laguna et al., 2021, Figure 1(A)). This contralateral remote area regarding the tumor location polypeptide contains an enhanced GFP as a core protein for (Figure 2(A)). The material remaining at the injection point straightforward monitoring of biodistribution, flanked by a C- was monitored through fluorescence as an indication of pro- terminal hexahistidine tail (H6) and an N-terminal cationic tein loss, during 10 days upon administration. As observed, peptide, T22. T22 is a ligand of the tumoral marker CXCR4 an inclusion body version of the protein that also releases (Tamamura et al., 1993, 1998a, 1998b), that in form of a fluorescent protein material (Unzueta et al., 2018) acting as recombinant fusion version promotes a highly selective bind- naturally produced secretory amyloids (Unzueta et al., 2018; ing and penetration into CXCR4þ cells. Also, it’s in vivo accu- Cespedes et al., 2020), was unable to stay at the injection mulation in tumoral tissues is detectable through point for long time periods post-administration (Figure 2(B)). fluorescence (Cespedes et al., 2018; Falgas et al., 2020a, In contrast, and in general terms, all the clustered materials 2020b; Pallares et al., 2020). The combination of T22 and GFP were observed as valid depots from which the forming pro- in a single modular polypeptide is then suitable to estimate tein was progressively released in comparison to the control, not only the protein leakage from the injection site but also soluble non-clustered protein (Figure 2(B)). However, clear its final fate when using animal models of CXCR4þ cancers. dissimilarities in the protein release rate were also observed. Also, T22-GFP-H6, in contrast to other previously tested pro- While with some differences, Zn- and Ca-containing materials teins (Lopez-Laguna et al., 2021), is well-aggregated in vitro tended to lose fluorescence more progressively. In contrast, by Zn 2þ as mechanically stable materials (Lo pez-Laguna materials in which Mn and Mg participated, including the et al., 2021). In physiological buffer, these clusters release MnMg combination, showed a tendency to release the pro- oligomers of the protein (Figure 1(A)) sizing around tein in a faster way. The MnMg combination resulted in fact 13–14 nm (Lo pez-Laguna et al., 2021), a size very similar to in the material disintegrating more rapidly (Figure 2(B)). that reached by the soluble protein version upon a spontan- Although these data resulted only just from a preliminary eous self-assembling process (around 12 nm) (Lopez-Laguna screening, the rather consistent negative impact of Mn2þ et al., 2019). and Mg2þ on the permanence of the material at the injec- In this regard, seven versions of T22-GFP-H6 (Figure 1(B)), tion site made us presume that the materials resulting from clustered as mechanically stable micron-scale protein gran- those ions are more unstable than their counterparts. Then, ules (Figure 1(C)) were generated through alternative cations looking for a time-prolonged release, we took a deeper or cation mixtures for subcutaneous administration, a route exploration of the depot potential of Zn- and Ca-based that allows entry of any leaked protein into the bloodstream materials. Granules formed with the assistance of either Zn, but also a local permanence of the remaining material as a Ca or a CaZn combination of divalent cations was adminis- depot (Unzueta et al., 2018). The elements selected for clus- tered again in the mouse model of human, CXCR4þ SW1417 tering were Zn and Ca, alone or as mixtures with Mg and colorectal cancer, in a larger number of animals. The admin- Mn, namely ZnMn, ZnMg, MnMg, CaMn, and CaZn (Figure istration was done in the contralateral area relative to sub- 1(B)). These cations were selected for their regular presence cutaneous tumors (Figure 2(A)). The loss of fluorescence at and relative abundance in living beings (Knape et al., 2017; the injection point was monitored during relatively long peri- Pilchova et al., 2017; Al Alawi et al., 2018; Li & Yang, 2018; ods of time in whole animals, spanning from the immediate Pazirandeh et al., 2020; Santos et al., 2020), thus skipping administration to 10 days post-injection (Figure 3). Again, the toxicities potentially linked to rarer oligoelements. These ions immediate visualization of fluorescence at the injection point also cover a wide range in the Irving-Williams (Milicevic was indicative of an extended permanence of the material in et al., 2011) series, which predicts differential stability of the the subcutaneous depot, being Zn2þ the clustering ion sup- complexes formed by those cations through coordination porting a slower protein release. However, the whole animal with histidine residues. In addition, we tested growing cat- imaging was not precise enough for a fine quantification, ion: protein molar ratios to further expand the range of pro- and for observing the fate of the released materials. tein outflow (Figure 1(B)). By using these cations and Expectedly, if our starting hypothesis was correct, the protein mixtures we looked for the generation of granules with dis- resulting from the depot disintegration should generate tinct leakage properties and therefore, different therapeutic steady fluorescence levels in the CXCR4þ tumor, because of potential and applicability in living systems. All the the presence of T22 as a targeting agent (Figure 1(A)). DRUG DELIVERY 2387 Figure 2. Preliminary screening of protein leakage from the secretory granules. (A) Representation of the injection site, in a contralateral area regarding the tumor in the SW1417 colorectal cancer model. (B) Temporal loss of protein material at the injection point monitored in vivo through the emitted fluorescence. In the insets, IVIS images were recorded at 5 days (120 h) post-administration. The scale color bar reflects the radiance expressed as (p/s/cm2/sr). Figure 3. Whole animal IVIS imaging in which the administered material confers fluorescence at the injection point. The administered protein was clustered with divalent cations of either Zn, Ca, or a CaZn mixture as indicated in Figure 1. Numbers at left indicate the particular animal from which the final ex vivo reading was recorded (see Supplementary Figure 1). The scale color bar reflects the radiance expressed as (p/s/cm2/sr). For that, at the times indicated in Figure 3, animals were submicron granules generated with the assistance of ions euthanized for the ex vivo determination of GFP fluorescence allowed a sustained protein release, keeping significant in the depots and in the target tumor (Supplementary Figure amounts of the starting material at the injection point 1). By the analysis of these ex vivo data, the release kinetics 10 days upon administration (Figure 4, Supplementary Figure of each type of depot was determined. As observed (Figure 1). Ca-based materials tended to be less supportive of pro- 4(A,B)), in contrast to inclusion body T22-GFP-H6 that tein permanence than Zn-based counterparts. Then, at a few dropped immediately from the injection site, the artificial hours upon administration, an important leakage of the 2388 P. ÁLAMO ET AL. Figure 4. Kinetics of protein depot leakage upon ex vivo examination, no background subtraction. (A) Individual kinetics of protein leakage upon subcutaneous administration of either Zn-, Ca-, or CaZn-based granules. Soluble T22-GFP-H6 (in form of nanoparticles) was injected as a control. (B) A comparative plotting of the above data. Differences with data at time 0; p  .05, p  .001. protein was observed followed by a smoother release phase values in the two alternative systems. Because of such stabil- (Figure 4(A,B)). In the hybrid CaZn-clustered material, Ca ity, we selected the Ca-based model to evaluate, in parallel, appeared as a negative regulator of protein retention, favor- the occurrence of GFP fluorescence in two main off-targeted ing release. This is an interesting observation as it appears organs, namely the kidney and liver. In this regard, the fluor- that this cation could be used as a modulator to adjust the escence in the tumor was three times higher than that leakage rate if designing very precise kinetics is envisaged. observed in these organs, as evaluated visually by plain kin- Noting this long permanence of the aggregated protein etic curves (Figure 5(D)), or numerically through the area at the injection site, it would be interesting to know the fate below the curve (Figure 5(E)). Therefore, T22-GFP-H6 was not of the released fraction and especially, if the granule-based only released from the subcutaneous depots during at least protein delivery might ensure a steady level of the removed 10 days, but the leaked protein also reached steady levels in protein at the target site. Such target tissue is, in this system, a target tissue, which are significantly higher than back- the CXCR4þ tumor toward which the T22 peptide is directed. ground levels in the kidney and in the liver. The ex vivo analysis of the fluorescence in the tumor Altogether, these data indicate a promising pattern of the revealed steady levels of fluorescence along the monitored slow disintegration of subcutaneously administered secretory time (Figure 5(A); Supplementary Figure 1). The intensity of granules, fabricated in vitro by the use of divalent cations as the fluorescence in the tumor was between 200 and 300 protein clustering agents. Importantly, the offered data indi- times lower than those observed at the injection site. An cates that the selection of the involved ion determines the exception was the animals treated with CaZn granules in strength of the protein-protein contacts and therefore, the which the fluorescence at the depot dropped dramatically rate of protein release during in vivo disintegration of the 10 days upon injection. Interestingly, the observed sets of material (Figures 1–3). Importantly, the protein is released fluorescence values were rather similar in the three types of from these depots for at least 10 days (Figures 2–4), and it tested materials (Figure 5(A)). The protein released from Ca- reaches the target organ in a functional form, ensuring steady based granules was particularly steady, as the fluorescence local levels during the whole experimental time (Figure 5). values in the tumor were very constant from a few hours We had previously demonstrated that bacterial inclusion upon injection on, compared to the slightly fluctuating bodies, a type of non-toxic amyloids found in recombinant DRUG DELIVERY 2389 Figure 5. Protein levels at the CXCR4þ target tumor. Comparative plotting of fluorescence levels at both the injection site and in tumor upon ex vivo analysis, for Zn-based (A), Ca-based (B), or CaZn-based protein depots (C). In Ca-based depots, fluorescence levels in the tumor are plotted together with those in the liver or kidney (D), and the global area under those curves in the monitored time interval is represented for a simple comparison (E). bacteria (de Marco et al., 2019), can release the forming pro- mere therapeutic uses, secretory granules might be also tein in a functional form, upon local (Cespedes et al., 2016; used in theragnosis as the time-prolonged release of tumor- Pesarrodona et al., 2019) or remote (Unzueta et al., 2018; targeted protein markers from remote repositories allows a Cespedes et al., 2020) administration to animals, for a thera- precise visualization of tumor foci. Other strategies have peutic effect in breast and colorectal cancer models been already developed that use metal coordination to sta- (Pesarrodona et al., 2019; Cespedes et al., 2020). However, bilize different categories of nanoparticles for drug delivery inclusion bodies are highly heterogeneous materials regard- (He et al., 2019, 2020), taking advantage of the coordination ing composition, and they trap, during the aggregation of capacities of Zn and other metals (Lo pez-Laguna et al., the recombinant protein in the bacterial cytoplasm and the 2020). On the other hand, metal coordination with polyhisti- further purification process, numerous bacterial molecules dine stretches has allowed the controlled oligomerization of including proteins, nucleic acids, and cell wall components peptide and protein materials in vitro, for the construction of (de Marco et al., 2019). This fact might pose regulatory limi- nanostructured immunogens (Manuel-Cabrera et al., 2016), tations to the clinical applicability of this material, whose nanotubes (Yewdall et al., 2018), and nanowires (Zhang natural formation in the cells cannot be controlled externally pez-Laguna et al., 2020). et al., 2012), among many others (Lo and the disintegration process appears to be too fast for a The exploitation and adaptation of metal and non-metal true time-prolonged delivery system (Figure 2). The materials divalent cations to construct self-disintegrating secretory presented here, that are mimetics of those natural inclusion granules opens a spectrum of possibilities in therapy and bodies (Sanchez et al., 2020; Sanchez et al., 2021), undergo a diagnosis that could be hardly reached by alternative sys- chemically controlled fabrication process from a purified pro- tems, offering in addition an important extent of versatility tein that is clustered, in a controlled process, by physio- logical concentrations of divalent cations present in the regarding the used ion and its capability to retain and leak body. Importantly, the results reported in the present study the protein building blocks. indicate that the release kinetics from these materials, upon Altogether, these data and concepts point out synthetic subcutaneous administration, is time-sustained enough to protein granules composed of ion-clustered his-tagged pro- represent a promising candidate for a slow drug delivery sys- teins as a regulatable secretory platform for clinical applica- tem. Importantly, the disintegration process can be regulated tions that being self-disintegrating, does not need, compared in the upstream section of the fabrication by the proper to alternative approaches (Petlin et al., 2017; Ali & Ahmed, selection of the clustering ion. Among those tested (Figure 2018; Saghazadeh et al., 2018; Stewart et al., 2018; Safdar 1(B)), Zn2þ and Ca2þ resulted in especially promising as they et al., 2019), any scaffold material for drug hosting. favor the retention of the polypeptides in the depot and extend their release into the body (Figure 5). The released protein is fluorescent and fully targeted to the tumor in an animal model of human colorectal cancer, accumulating in Disclosure statement such a fluorescent form in tumor tissues (Figure 5, R.M., E.V., and A.V. are co-founders of NANOLIGENT SL, a company devel- Supplementary Figure 1). This fact indicates that apart from oping tumor-targeted protein-based drugs. 2390 P. ÁLAMO ET AL. Funding Ghalei S, Nourmohammadi J, Solouk A, Mirzadeh H. (2018). Enhanced cellular response elicited by addition of amniotic fluid to alginate We are indebted to AGAUR (INVITA, grant 2020PANDE00003), AGAUR hydrogel-electrospun silk fibroin fibers for potential wound dressing (2017SGR-229) and CIBER-BBN (projects NANOPROTHER) granted to A.V. application. Colloids Surf B Biointerfaces 172:82–9. E.V. received support from AEI (PID2019-105416RB-I00/AEI/10.13039/ Gilmore KA, Lampley MW, Boyer C, Harth E. (2016). Matrices for com- 501100011033) and CIBER-BBN (project NANOREMOTE). R.M. received bined delivery of proteins and synthetic molecules. Adv Drug Deliv support from ISCIII-AEI (PI18/00650, co-funding FEDER), CIBER-BBN Rev 98:77–85. (4NanoMets project), and AGAUR (2017 SGR 865 GRC). U.U. was sup- He Z, Hu Y, Gui Z, et al. (2019). Sustained release of exendin-4 from tan- ported by Miguel Servet contract (CP19/00028) from ISCIII co-funded by nic acid/Fe (III) nanoparticles prolongs blood glycemic control in a European Social Fund (ESF investing in your future) and H.L.L. by a pre- mouse model of type II diabetes. J Control Release 301:119–28. doctoral fellowship from AGAUR (2019FI_B00352). E.V.D. was supported He Z, Nie T, Hu Y, et al. (2020). A polyphenol-metal nanoparticle plat- by a predoctoral fellowship from Ministerio de Ciencia, Innovacio n y form for tunable release of liraglutide to improve blood glycemic con- Universidades (FPU18/04615). A.V. received an ICREA ACADEMIA award. trol and reduce cardiovascular complications in a mouse model of U.U received support from ISCIII-AEI (PI20/00400) co-funded by FEDER ( type II diabetes. J Control Release 318:86–97. a way to make Europe) and Miguel Servet contract (CP19/00028) from Jacob R, Anoop A, Maji S. (2019). Protein nanofibrils as storage forms of ISCIII co-funded by European Social Fund (ESF investing in your peptide drugs and hormones. Vol. 1174. Singapore: Springer, 265–90. future).Protein production was partially performed by the ICTS Jacob RS, Das S, Ghosh S, et al. (2016). Amyloid formation of growth ‘NANBIOSIS,’ more specifically by the Protein Production Platform of hormone in presence of zinc: relevance to its storage in secretory CIBER in Bioengineering, Biomaterials & Nanomedicine (CIBER-BBN)/IBB, granules. Sci Rep 6:23370. at the UAB (http://www.nanbiosis.es/portfolio/u1-protein-production- Knape MJ, Ballez M, Burghardt NC, et al. (2017). Divalent metal ions con- platform-ppp/). The in vivo work was performed by the ICTS NANBIOSIS trol activity and inhibition of protein kinases. Metallomics 9:1576–84. of the CIBER-BBN Nanotoxicology Unit (http://www.nanbiosis.es/port- Koshy ST, Zhang DKY, Grolman JM, et al. (2018). Injectable nanocompo- folio/u18-nanotoxicology-unit/). site cryogels for versatile protein drug delivery. Acta Biomater 65: 36–43. Li C, Wang J, Wang Y, et al. (2019). Recent progress in drug delivery. Acta Pharm Sin B 9:1145–62. Li J, Mooney D. (2016). Designing hydrogels for controlled drug delivery. Data availability statement Nat Rev Mater 1:16071. The data that support the findings of this study are openly available in Li L, Yang X. (2018). The essential element manganese, oxidative stress, DDD (UAB) at https://doi.org/10.5565/ddd.uab.cat/249709 or https://ddd. and metabolic diseases: links and interactions. Oxid Med Cell Longev uab.cat/record/249709. 2018:7580707. Lo  pez-Laguna H, Parlade E, Alamo P, et al. (2021). In vitro fabrication of microscale secretory granules. Adv Funct Mater 31:2100914. pez-Laguna H, Sanchez J, Unzueta U, et al. (2020). Divalent cations: a Lo References molecular glue for protein materials. Trends Biochem Sci 45: Al Alawi AM, Majoni SW, Falhammar H. (2018). Magnesium and human 992–1003. health: perspectives and research directions. Int J Endocrinol 2018: pez-Laguna H, Sanchez JM, Carratala JV, et al. (2021). Biofabrication of Lo 9041694. functional protein nanoparticles through simple His-tag engineering. Ali A, Ahmed S. (2018). A review on chitosan and its nanocomposites in ACS Sustain Chem Eng 9:12341–54. drug delivery. Int J Biol Macromol 109:273–86. Lopez-Laguna H, Unzueta U, Conchillo-Sole O, et al. (2019). Assembly of Aragao-Santiago L, Bohr A, Delaval M, et al. (2016). Innovative formula- histidine-rich protein materials controlled through divalent cations. tions for controlled drug delivery to the lungs and the technical and Acta Biomater 83:257–64. toxicological challenges to overcome. Curr Pharm Des 22:1147–60. Maji SK, Perrin MH, Sawaya MR, et al. (2009). Functional amyloids as nat- Cespedes MV, Cano-Garrido O, Alamo P, et al. (2020). Engineering secre- ural storage of peptide hormones in pituitary secretory granules. tory amyloids for remote and highly selective destruction of meta- Science 325:328–32. static foci. Adv. Mater 32:1907348. Maniatis T, Fritsch EF, Sambrook J, Engel J. (1989). Molecular cloning: a Cespedes MV, Fernandez Y, Unzueta U, et al. (2016). Bacterial mimetics laboratory manual. New York, NY: Cold Spring Harbor Laboratory of endocrine secretory granules as immobilized in vivo depots for Press. Mankar S, Anoop A, Sen S, Maji SK. (2011). Nanomaterials: amyloids functional protein drugs. Sci Rep 6:35765. reflect their brighter side. Nano Rev 2:6032. Cespedes MV, Unzueta U, Avino A, et al. (2018). Selective depletion of Manuel-Cabrera CA, Vallejo-Cardona AA, Padilla-Camberos E, et al. metastatic stem cells as therapy for human colorectal cancer. EMBO (2016). Self-assembly of hexahistidine-tagged tobacco etch virus cap- Mol Med 10:e8772. sid protein into microfilaments that induce IgG2-specific response Chen TY, Cheng WJ, Horng JC, Hsu HY. (2020). Artificial peptide-con- against a soluble porcine reproductive and respiratory syndrome virus trolled protein release of Zn(2þ)-triggered, self-assembled histidine- chimeric protein. Virol J 13:196. tagged protein microparticle. Colloids and Surfaces. B, Biointerfaces Meng D, Lei H, Zheng X, et al. (2019). A temperature-sensitive phase- 187:110644. change hydrogel of tamoxifen achieves the long-acting antitumor Cross LM, Carrow JK, Ding X, et al. (2019). Sustained and prolonged activation on breast cancer cells. Onco Targets Ther 12:3919–31. delivery of protein therapeutics from two-dimensional nanosilicates. Milicevic A, Branica G, Raos N. (2011). Irving-Williams order in the frame- ACS Appl Mater Interfaces 11:6741–50. work of connectivity index 3vv enables simultaneous prediction of de Marco A, Ferrer-Miralles N, Garcia-Fruitos E, et al. (2019). Bacterial stability constants of bivalent transition metal complexes. Molecules inclusion bodies are industrially exploitable amyloids. FEMS Microbiol 16:1103–12. Rev 43:53–72. Natarajan JV, Nugraha C, Ng XW, Venkatraman S. (2014). Sustained- Falgas A, Pallares V, Serna N, et al. (2020a). Selective delivery of T22- release from nanocarriers: a review. J Control Release 193:122–38. PE24-H6 to CXCR4þ diffuse large B-cell lymphoma cells leads to wide Pallares V, Unzueta U, Falgas A, et al. (2020). An Auristatin nanoconju- therapeutic index in a disseminated mouse model. Theranostics 10: gate targeting CXCR4þ leukemic cells blocks acute myeloid leukemia 5169–80. dissemination. J Hematol Oncol 13:36. Falgas A, Pallares V, Unzueta U, et al. (2020b). A CXCR4-targeted nano- Palombo M, Deshmukh M, Myers D, et al. (2014). Pharmaceutical and carrier achieves highly selective tumor uptake in diffuse large B-cell toxicological properties of engineered nanomaterials for drug deliv- lymphoma mouse models. Haematologica 105:741–53. ery. Annu Rev Pharmacol Toxicol 54:581–98. DRUG DELIVERY 2391 Pareek SP, Kumawat S, Sharma V, et al. (2019). Review on sustained Sharma A, Madhunapantula SV, Robertson GP. (2012). Toxicological con- release technology. Int J Pharma Biol Sci Archiv 7:29–38. siderations when creating nanoparticle-based drugs and drug delivery Pazirandeh S, Burns D, Griffin U. (2020). Overview of dietary trace ele- systems. Expert Opin Drug Metab Toxicol 8:47–69. ments [last accessed 25 Jul 2020]. Shen J, Wolfram J, Ferrari M, Shen H. (2017). Taking the vehicle out of Pesarrodona M, Jauset T, Dıaz-Riascos ZV, et al. (2019). Targeting antitu- drug delivery. Mater Today 20:95–7. moral proteins to breast cancer by local administration of functional Stewart SA, Domınguez-Robles J, Donnelly RF, Larran ~eta E. (2018). inclusion bodies. Adv Sci 6:1900849. Implantable polymeric drug delivery devices: classification, manufac- Petlin DG, Tverdokhlebov SI, Anissimov YG. (2017). Plasma treatment as ture, materials, and clinical applications. Polymers 10:1379. Tamamura H, Arakaki R, Funakoshi H, et al. (1998a). Effective lowly cyto- an efficient tool for controlled drug release from polymeric materials: toxic analogs of an HIV-cell fusion inhibitor, T22 ([Tyr5,12, Lys7]-poly- a review. J Control Release 266:57–74. phemusin II). Bioorg Med Chem 6:231–8. Pilchova I, Klacanova K, Tatarkova Z, et al. (2017). The involvement of Tamamura H, Imai M, Ishihara T, et al. (1998b). Pharmacophore identifi- Mg2þ in regulation of cellular and mitochondrial functions. Oxid Med cation of a chemokine receptor (CXCR4) antagonist, T22 Cell Longev 2017:6797460. ([Tyr(5,12),Lys7]-polyphemusin II), which specifically blocks T cell-line- Rosen H, Abribat T. (2005). The rise and rise of drug delivery. Nat Rev tropic HIV-1 infection. Bioorg Med Chem 6:1033–41. Drug Discov 4:381–5. Tamamura H, Kuroda M, Masuda M, et al. (1993). A comparative study of Rueda F, Cespedes MV, Conchillo-Sole O, et al. (2015). Bottom-up the solution structures of tachyplesin I and a novel anti-HIV synthetic instructive quality control in the biofabrication of smart protein mate- peptide, T22 ([Tyr5,12, Lys7]-polyphemusin II), determined by nuclear rials. Adv Mater 27:7816–22. magnetic resonance. Biochim Biophys Acta 1163:209–16. Safdar R, Omar AA, Arunagiri A, et al. (2019). Potential of Chitosan and Unzueta U, Cespedes MV, Sala R, et al. (2018). Release of targeted pro- its derivatives for controlled drug release applications – a review. J tein nanoparticles from functional bacterial amyloids: a death star-like Drug Delivery Sci Technol 49:642–59. approach. J Control Release 279:29–39. Saghazadeh S, Rinoldi C, Schot M, et al. (2018). Drug delivery systems Wen H, Jung H, Li X. (2015). Drug delivery approaches in addressing clin- and materials for wound healing applications. Adv Drug Deliv Rev ical pharmacology-related issues: opportunities and challenges. AAPS J 17:1327–40. 127:138–66.  Wu J, Li P, Dong C, et al. (2018). Rationally designed synthetic protein Sanchez J, Lo pez-Laguna H, Alamo P, et al. (2020). Artificial inclusion hydrogels with predictable mechanical properties. Nat Commun 9: bodies for clinical development. Adv Sci 7:1902420. 620. Sanchez JM, Lo pez-Laguna H, Serna N, et al. (2021). Engineering the per- Yewdall NA, Allison TM, Pearce FG, et al. (2018). Self-assembly of toroidal formance of artificial inclusion bodies built of catalytic b-galactosi- proteins explored using native mass spectrometry. Chem Sci 9: dase. ACS Sustain Chem Eng 9:2552–8. 6099–106. Santos HO, Teixeira FJ, Schoenfeld BJ. (2020). Dietary vs. pharmacological Zhang W, Luo Q, Miao L, et al. (2012). Self-assembly of glutathione S- doses of zinc: a clinical review. Clin Nutr 39:1345–53. transferase into nanowires. Nanoscale 4:5847–51. Serna N, Cano-Garrido O, Sanchez JM, et al. (2020). Release of functional Zou H, Banerjee P, Leung SSY, Yan X. (2020). Application of pharmacoki- fibroblast growth factor-2 from artificial inclusion bodies. J Control netic-pharmacodynamic modeling in drug delivery: development and Release 327:61–9. challenges. Front Pharmacol 11:997.