Life Sciences Discovery and Technology Highlights

Wound Healing

Patch Repair of Deep Wounds by Mobilized Fascia

Mammalian scarring occurs due to specialized fibroblasts migrating into wounds and depositing plugs of extracellular matrix. Abnormal scarring gives rise to either nonhealing chronic wounds or aggravating fibrosis that is a hefty public healthcare burden. The authors previously revealed that a distinct fibroblast subpopulation expressing the engrailed 1 (En1) gene gives rise to scars on the back skin. These cells reside not only in the skin, but also in the underlying fascia layer. Fascia is a gelatinous viscoelastic membrane that facilitates frictionless gliding between the skin and rigid structures below. Whereas fascia in mouse back skin separates skin by the panniculus carnosus (PC) muscle, humans have no interleaving muscle but instead a few thick layers incorporating fibroblasts, lymphatics, fat, and neurovascular, vascular, and sensory neurons.

Using various matrix-tracing techniques, live imaging, genetic lineage tracing, and anatomic fate-mapping models, the authors explore the fundamental mechanisms of scar formation. The authors use live animal subjects inflicted with 6 mm diameter wounds, and chimeric skin and fascia grafts. Each subject was labeled with different fluorescent reporters, TdTomato and green fluorescent protein (GFP), before the wounds were assessed 14 days later. Fascia-derived cells were found to occupy 80% of the wound region. This contrasts with uninjured subjects that showed that the fascial cells resided in their original location. Approximately half these fascial cells expressed a fibroblast phenotype alongside monocytes, lymphatics, endothelium, and nerve cells. The authors then used subjects with labeled En1 lineage fibroblasts (that strongly contribute to scar formation, Engrailed Positive Fibroblast [EPF]), identifying them to be largely in the fascia, whereas a smaller fraction was seen in the dermis. Thus, the fascia contained cells with higher scar-forming potential (EPF).

To assess if these fascia EPFs contribute to wound healing, the authors generated superficial excisional wounds. Just 3 days postwounding, EPF aggregates were observed, which suggested that they traverse dermal layers unobstructed by PC muscles. Thereafter, the authors studied whether there was any correlation with the scar size and the depth of injury. In mice modified with tag En1+ cells, superficial and deep injuries were inflicted. Fourteen days postwounding, deep injuries were found to be 1.7-fold the size of superficial injuries, with twice as many fascial EPFs in deep wounds. The number of EPFs directly correlated with wound size (thus scar severity) and were found to recede 10 weeks after injury. This suggests that the fascia EPFs are responsible for and are a major source of wound fibroblasts. Following this, second harmonic generation (SHG) was used to assess the quality of extracellular matrix, indicating that the wound contained fascia-like collagen compared with stretched and woven dermal matrix.

An in vivo incubation chamber enabled live imaging of fascia tissue, ascertaining that fascia was steered ~2 mm in 7 days. By labeling fascia matrix with NHS esters, fascia matrix was observed to extend upward and plug open wounds, covering ~75% of total wound collagen 7 days postwounding. Further imaging revealed that fascia fibers transformed from a parallel sheet into a highly porous plug. To query the effects of fascia EPFs on scar formation, an impermeable polytetrafluoroethylene membrane was used to obstruct cell movement from the fascia, resulting in wounds remaining open, unlike sham controls, which closed within 21 days. Experiments to mechanically obstruct wound closure further validate this.

Immune cells suggest a regular immunity profile of the open wounds. Instead, these were a result of blocked fascia steering caused by its fibroblasts. This further supports the notion that scar tissue is of fascia origin, since dermal fibroblasts and matrix are unable to repair wounds when the fascia is immobilized. This is further confirmed when experiments to release underlying fascia from the wound site yield similar outcomes. Removing the EPFs through genetic ablation yields similar results, demonstrating that fascia-located EPFs are responsible for steering and plugging deep wounds, working like a rapidly expanding sealant to clog wounds independent of cell proliferation. This fascia-plugging phenomenon was compared with human keloid scars that share their dense collagen fiber structure. Three biomarkers—dpp4, FAP, and NOV—were strongly expressed in the fascia, as in keloid scars, whereas the dermis expressed minimal levels of them. Similarly, deep scars in mice and fascia expressed these three biomarkers more than the dermis, further suggesting that cutaneous scars originate from fascia tissue.

These findings suggest a revision to the existing knowledge of wound closure and remodeling. Whereas superficial trauma leads to dermal remodeling activity, deeper injury gives rise to fascia steering that rapidly seals and plugs wounds before undergoing remodeling. Thus, fascia appears to serve as a reserve matrix that gives rise to hypertrophic and keloid scars upon traumatic injury. Greater layers of fascia in particular regions such as the lower chest, back, thigh, and arm also account for the greater incidence of pathological scarring. (Correa-Gallegos, D.; et al. Nature 2019, 576, 287–292.)

Burn-Related Collagen Conformational Changes in Ex Vivo Porcine Skin Using Raman Spectroscopy

Thermal injuries induce irreversible changes to structural skin elements. These changes alter various properties, including mechanical and optical properties. Burns soften skin, allowing rough discrimination between burnt and unburnt tissue. Yet, there is a lack of understanding of the mechanisms that induce these changes; a greater understanding of these would improve burn diagnosis and treatment. Collagen, constituting 75%–90% of dry dermis weight, is the key constituent of mammalian skin. It provides strength to skin, determining its mechanical response. Thus, investigating burn-induced changes in skin requires close study of associated changes in the underlying collagen molecular structure. Previous studies also report that collagen fibers undergo denaturation upon heating as they shrink from their native triple-helix structure to a random (coiled) structure due to protein unfolding and aggregation. To investigate such structural changes, Raman spectroscopy was employed to noninvasively characterize the molecular structure of skin. Due to its noninvasive nature, it has been used to detect skin cancer and skin features (e.g., dermoepidermal junctions, skin moisture content, and burns). The spectral properties of porcine skin derived from Raman scattering provide unique “fingerprints” to identify molecular changes.

Four major regions are observed in between 500 and 2000 cm⁻¹: ~1590–1710, ~1420–1490, ~1220–1350, and ~800–1030 cm⁻¹. Region 1 (~800–1030 cm⁻¹) is dominated by the amide I band as a result of C=O vibrations. These are essential to the formation of collagen protein. Region 2 (~1220–1350 cm⁻¹) is dominated by CH₂ vibrations associated with proteins and lipids. Region 3 (~1420–1490 cm⁻¹), amide III bands, are associated with the vibration of C–N and N–H stretching and bending. Region 4 is dominated by ν(C–C) vibrations related to amino acids (e.g., proline, hydroxyproline, tyrosine, and tryptophan). The following burn conditions were used: (1) 200 °F for 10 s, (2) 200 °F for 30 s, (3) 450 °F for 10 s, and (4) 450 °F for 30 s. Whereas condition 1 can create superficial second-degree burns, condition 4 creates third-degree full-thickness burns. Analytical calculations also reveal that conditions 2 and 3 transfer intermediate amounts of heat resulting in the monotonic increase in energy transfer from conditions 1 to 4. In the amide III region, the 1448 and 1342 cm⁻¹ bands shifted to a higher wavelength with increased and decreased Raman shifts, respectively, with prolonged and elevated temperature and exposure. In addition, bands at 853, 873, 937, 1317, 1448, and 1662 cm⁻¹ exhibit significant changes.

The 1448 cm⁻¹ peak, characteristic of collagen extracellular matrix, shifts toward a higher wavenumber while decreasing intensity. This is characteristic of decreased lipids and/or proteins resulting from the reduction of CH₂ groups during burning. Similarly, proline/hydroxyproline residues decreased, leading to decreases in the intensity of 853, 873, and 937 cm⁻¹ peaks. A reduction in polypeptide chains in tropocollagen suggests that there is a reduction in stabilizing interactions that forms the collagen triple-helical conformation. Raman spectroscopy analysis shows that burning reduces the quantity of proline/hydroxyproline, resulting in less stable collagen. While the 853 and 873 cm⁻¹ bands keep decreasing, the 937 cm⁻¹ band decreases before plateauing with burn severity (conditions 3 and 4). This suggests that collagen has denatured even before reaching the severest degree of burns. Thus, Raman spectra reveal the effects of burns resulting in a more disordered, less stable collagen structure.

Deconvolution of the amide I region of the Raman spectra provides further information regarding the changes sustained by burns. In pristine collagen, the amide I band is dominated by the α-helix (1662 cm⁻¹) taking up ~26.3% of the total area. With increased burn severity, this shifts toward lower-wavelength numbers—indicative of denaturation. Instead, β-turns and sheets (1663cm-1, 1669cm-1, and 1633cm-1) are observed with an increase in peak intensity with increased burn severity. This further demonstrates that burning gives rise to more disordered and less stable collagen structures. These disordered and less stable structures account for the softening responses observed in burnt skin. (Ye, H.; et al. Sci. Rep. 2019, 9, 19138.)

Heterogeneity in Old Fibroblasts Is Linked to Variability in Reprogramming and Wound Healing

The effects of aging and senescence have been reported to affect cell reprogramming and wound healing, which is systematically studied herein. Using cytokine profiling, plasma and conditioned medium were compared between young (3 months) and old (28–29 months) mice. It was observed that pro-inflammatory (interleukin [IL]-6, tumor necrosis factor [TNF], etc.) and anti-inflammatory (IL-4) cytokines, chemokines, and growth factors are significantly increased, similar to the primary fibroblast cultures. Fibroblast cultures obtained from 108 young, middle-aged, and old mice were reprogrammed by triggering the Oct4, KLF4, Sox2, and MYC genes with reprogramming efficiency measured by alkaline phosphatase (AP) and stage-specific embryonic antigen 1 (SSEA1) staining. While the mean reprogramming efficiency did not significantly change with age, a greater variability in reprogramming was observed in older mice.

Through the use of various analytical methods, transcriptomics, epigenomics, and metabolomics revealed differences between fibroblast cultures that were young, old with good reprogramming properties, or old with poor reprogramming properties. The old fibroblasts were found to have a similar transcriptomic signature to that of activated fibroblasts (myofibroblasts), which were normally involved in tissue repair. The transcription factor EBF2, increased in old fibroblasts, was also enhanced in activated fibroblasts, as well as primary fibroblasts from elderly humans. The proportion of activated fibroblasts was linked to the variability in reprogramming. Old fibroblasts were found to be enriched in activated fibroblasts that strongly expressed α-smooth muscle actin (α-SMA), platelet-derived growth factor receptor α (PDGFRα), THY1, and inflammatory cytokines and were absent of senescence markers (e.g., p16Ink4a). Ebf2 knockdown reduced fibroblast activation, whereas Ebf2 overexpression in young fibroblasts induced cytokine expression. Similarly, activated fibroblasts bearing THY1+/PDGFRα+ fibroblasts were also found in greater proportion in old mice.

Further examination also revealed that the proportion of activated (THY1+/PDGFRα+) fibroblasts correlated positively with reprogramming efficiency and negatively with senescence. Interestingly, activated fibroblasts reprogram less efficiently (intrinsic factors) than unactivated fibroblasts. However, activated fibroblasts secrete (extrinsic) factors that enhance reprogramming efficiency. The role of intrinsic versus extrinsic factors was examined by swapping the conditioned medium between good and bad old fibroblasts. This reduced the difference in reprogramming efficiencies by more than 60%, with intrinsic factors expected to account for the remainder of this effect. The authors then investigated the individual contribution of factors secreted from the old fibroblasts. While IL-6 enhanced reprogramming efficiency, TNF-α and IL-1β inhibited it. As a result, blocking antibodies with IL-6 reduced reprogramming ability, but blocking TNF-α enhanced it. Thus, the authors identified how proportions of old and young fibroblasts, and their individual components (e.g., IL-6 and TNF), drive the variability in reprogramming.

This variability in reprogramming was similarly echoed in mouse wound healing rates, with increased variability observed in older mice. Single-cell RNA-seq identified subpopulations of activated fibroblasts with increased cytokine signaling. In addition, immune cells were also identified in wounds of fast-healing old mice, whereas fibroblasts were dominant in slow-healing old mice. Furthermore, three distinct subpopulations of fibroblasts (A–C) were identified that were enriched in different aspects of activation. While subpopulation A was present in both slow- and fast-healing mice, subpopulation B was more abundant in fast-healing old mice with increased cytokine and signaling expression. Subpopulation C, on the other hand, was associated with slow-healing old mice, exhibiting different cytokine (Ccl11) and transcription (Ebf2) profiles. Differences in activated fibroblast cytokine profiles are likely associated with variability in reprogramming and wound healing trajectories. While this variability (in reprogramming and wound healing) can be detrimental, the authors identify that a subpopulation of activated fibroblasts are largely responsible for this. Consequently, manipulation of their secretory profile may then address issues such as chronic inflammation, wound healing, and reprogramming to assist in treating aging pathologies. (Mahmoudi, S.; et al. Nature 2019, 574, 553–558.)

Pirfenidone Attenuates the Profibrotic Contractile Phenotype of Differentiated Human Dermal Myofibroblasts

Pathological scarring characterized by hypertrophic and keloid scarring is characterized by dysregulated wound healing and reduced extracellular matrix (ECM) turnover, increased ECM production, and the presence of abnormally high numbers of contractile and profibrotic myofibroblasts. Driven by profibrotic transforming growth factor (TGF)-β1 in the wound healing milieu, normal fibroblasts become activated, transforming into highly contractile myofibroblasts. Myofibroblasts influence the ECM through α-smooth muscle actin (SMA) stress fibers attaching via large, focal adhesion complexes. They contribute to skin remodeling by generating excess ECM and reduce turnover. In normal wound healing, myofibroblasts undergo loss of phenotype, whereas pathological scarring conditions give rise to persistent myofibroblasts within the restored wound milieu—contributing to scarring and contractures. Myofibroblasts appear to be convertible to a less activated phenotype. Certain conditions, such as substrate stiffness, have been shown to negate their profibrotic behavior, yet little is known about these mechanisms.

An FDA-approved drug for idiopathic pulmonary fibrosis (IPF), pirfenidone (Pf), is known to possess anti-inflammatory and antifibrotic properties. While the mechanism of action is as yet unknown, it inhibits TGF-β1 expression, also reducing collagen and α-SMA expression. The authors use an in vitro culture of TGF-β1-treated fibroblasts to examine the effects of Pf treatment. TGF-β1 treatment generated α-SMA+ fibroblasts. While α-SMA levels continued to rise after 6 days, the addition of Pf caused α-SMA levels to remain stable. Similarly, F-actin stress fibers demonstrated a similar trend to α-SMA. Even in the case of more mature fibroblasts (treated with TGF-β1 for 5 days), Pf similarly reduced the extent of activation (α-SMA, F-actin expression). Pf strongly attenuated the gene encoding a-SMA; similarly, ITGA3 (which encodes α3 integrins) reduced in expression levels. Connective tissue growth factor (CTGF), a target gene of TGF-β1, was also suppressed with Pf. On the other hand, MMP1, a factor responsible for ECM (digestion) levels, is restored. The authors then tested Pf deactivation of fibroblasts on functional assays. Seeding fibroblasts in collagen lattices and stimulating them with TGF-β1 caused gel contraction.

One-hour Pf treatment inhibited the contractile behavior, with effects observed as early as 10 min. Abrogating the contractile behavior of activated fibroblasts gave rise to a more than threefold increase in gel lattice surface area. Through study of the effects of Pf treatment on TGF-β1-activated fibroblasts (a proxy for myofibroblasts), the gene profile suggests normalizing of the fibroblasts and reduced contractile behavior. This suggests that Pf is worth exploring further as a remedy for scarring. (Wells, A. R.; Leung, K. P. Biochem. Biophys. Res. Commun. 2020, 3, 641–651.)

Tissue Engineering and Regenerative Medicine

Sutureless Repair of Corneal Injuries Using Naturally Derived Bioadhesive Hydrogels

Although 1.5 million cases of corneal blindness are reported annually, corneal transplants only help less than 5% of cases. This is due to a shortage of donors and the financial inaccessibility of the transplant procedure. Corneal damage gives rise to impairment that leads to the compromise of eye structural integrity. While cyanoacrylate glue, tissue grafting, and transplantation are the standards of care, they possess shortcomings. Cyanoacrylates have poor biocompatibility and transparency, are difficult to handle, and integrate poorly with corneal tissue. Grafting is limited by donor tissue, surgical skill, and equipment. Transplants may be rejected, and complications could occur from suturing, tissue death, microbial entrapment, and even allogeneic tissue immune reactions. This makes cost-effective, cell-free biomaterials highly desirable for corneal repair. Ideal candidates should possess the following: (1) biocompatibility and biodegradability, (2) mechanical stability and appropriate stiffness, (3) high transparency, (4) high adhesion to the native tissue, (5) capability for cell support and endogenous tissue regeneration, and (6) clinical compliance for ease of application and use. Biomaterials fall into two categories: (1) synthetic adhesives and (2) naturally derived adhesives. Natural candidates have good biocompatibility but poor mechanical and adhesion properties. Synthetic ones, on the other hand, often suffer in biocompatibility. Currently approved adhesives include ReSure (a PEG adhesive), which is FDA approved for sealing corneal incisions following cataract surgery, although it cannot fill stromal defects. It is also less suitable for wet conditions, poorly adheres, and falls off rapidly. OcuSeal does not seal fast enough to fill stromal defects. Ultraviolet (UV)-activated hydrogels have been considered too, but they risk photodamage of the retina/corneas. Many natural biomaterials include collagen, fibrin, gelatin, alginate, and chitosan, but they lack suitable adhesive properties, retention, optical properties, transparency, and stiffness to integrate with native cornea tissue. Given this situation, the authors engineered a gelatin-photoinitiator bioadhesive designed to cross-link after a short period of exposure to visible light (450–550 nm).

Although various UV-triggered bioadhesives have been considered, GelCORE (gel for corneal regeneration), a visible light system, was designed to overcome safety concerns relating to UV cross-linking. Visible light was able to cross-link hydrolyzed collagen through free radical polymerization in the presence of type 2 initiator Eosin Y, triethanolamine (TEA), and N-vinylcaprolactam (VC) as a co-initiator and co-monomer, respectively. Visible light excites the Eosin Y photoinitiator molecules to a triplet state, allowing them to accept hydrogen from the co-initiator TEA. The deprotonated radicals then undergo vinyl-bond cross-linking with VC to accelerate the gelation of polymeric scaffolds. The GelCORE prepolymer can fill a defect site before triggering polymerization to fill the defect site. Nuclear magnetic resonance (NMR) performed on GelCORE showed that the degree of cross-linking increased from 63% to 89% when exposure to visible light was increased from 1 to 4 min. This gave rise to a concomitant increase in compressive modulus by >3-fold. The concentration of the GelCORE prepolymer gave rise to a 66-fold increase in compressive modulus. Crucially, the compressive modulus was similar to that of native cornea (115 kPa), which is favorable for long-term biocompatibility. Further characterization showed that the degree of hydration was similar to that of native cornea tissue (~85%). GelCORE also allowed the tuning of enzymatic degradation rates to facilitate tissue integration and de novo tissue ingrowth.

Thereafter, GelCORE adhesion properties were analyzed using various ASTM standards for biological adhesives and compared with Evicel and CoSEAL—commercially available surgical sealants. Using ASTM F2392-04 to evaluate burst pressure, GelCORE (20%) adhesives were found to possess 10- to 60-fold higher pressures compared with the commercial candidates. GelCORE burst pressure was found to be similar to that of cyanoacrylate glues, although they have poor biocompatibility. ASTM F2255-05 was then used for testing the shear strength. Once more, 20% (w/v) GelCORE exhibited the highest values, which similarly outperformed the commercial candidates. ASTM F2458-05, to test wound closure, also established favorable adhesion strength values. GelCORE adhesives were then investigated on ex vivo corneal tissues explanted from New Zealand rabbits following the creation of corneal defects (3 mm diameter, >50% deep). These exhibited retention times between 15 and 17 days, remaining uncompromised (with the thickness and spread retained). Slit-lamp biomicroscopy also confirmed that the bioadhesive remained transparent without any observable changes in shape or contour. Anterior segment optical coherence tomography (AS-OCT) confirmed no changes in thickness or shape after 28 days. The authors then assessed whether GelCORE bioadhesives permitted cell migration and cytocompatibility. GelCORE was compared with ReSure by culturing human corneal fibroblast cells (keratocytes). While GelCORE exhibited a high cell viability (>90%), ReSure exhibited <65% viability and lowered metabolic activity. In vitro scratch assays were employed to evaluate wound closure due to cell migration. GelCORE even enhanced cell migration, resulting in a 36% higher cell density after 24 h. Thus, in vitro cultures suggest that GelCORE supports proliferation, adhesion and spreading, and metabolic activity.

These were then explored in live rabbits with 50% deep defects generated in live animals. Following defect creation, 20% (w/v) GelCORE precursor was applied to the defect site, followed by 4 min of visible light exposure. This gave rise to firm adhesion of GelCORE to the corneal defect. AS-OCT confirmed that the hydrogel completely filled the defect and adhered to the stromal bed, remaining transparent 7 and 14 days after application. Cobalt blue slit-lamp imaging of fluorescein staining revealed the progressive reduction of the corneal epithelial defect. This suggests that the bioadhesive encouraged epithelium migration. By day 14, the corneal epithelial defect was completely healed. Histology showed strong adhesion of GelCORE to the stromal tissue following application. At day 14, stromal tissue grew with little signs of inflammatory tissue, for example, fibrous collagenous capsule. In untreated corneas, the stromal layer regenerates poorly. Unlike the heterogenous samples seen in untreated samples, GelCORE gave rise to homogeneous reepithelialization similar to native cornea. CD45+ leukocytes also suggested normal inflammatory responses typically observed in regeneration, with similarities between regenerated tissue and native tissue. Crucially, GelCORE facilitated corneal regeneration and faster patient recovery, reducing the need for visual rehabilitation and potentially avoiding corneal transplantation. This makes it highly promising as a corneal repair candidate. (Sani, E. S.; et al. Sci. Adv. 2019, 5, eaav1281.)

Vascularized Cancer on a Chip: The Effect of Perfusion on Growth and Drug Delivery of Tumor Spheroid

Despite the urgency for drug development in cancer therapy, drug development may last 10 years, with ~43% of them failing to attain their primary objective of impeding cancer advancement. A critical limitation is that animal models poorly mimic the effects on the human body. On the other hand, the conventional cell culture of human-derived cells is comparably rudimentary—without accounting for cell–cell interactions, extracellular matrix interactions, blood supply and flow, and so forth. In the laboratory, multicellular aggregates–spheroids have been used as tumor models because they resemble in vivo tumors with regard to shape, high cell density, and biochemical microenvironment. Incorporating a 3D structure gives rise to “organoids” and may even include cell heterogeneity with a mix of noncancerous cells that give rise to realistic tumor microenvironments (TMEs). One feature incorporated in TMEs is a vascular network generated by endothelial cells that influences the presentation of growth factors, nutrients, oxygen, and so on. Despite this, native TMEs have abnormal blood supply due to leaky blood vessels and increased interstitial fluid pressure from the high cell density, which is not readily recapitulated in culture. To recapitulate native TME conditions, the authors developed a spheroid model (tumor and endothelial cells) containing a perfusable vascular network allowing the maintenance of culture for 24 h. The incorporation of perfusion gives rise to accelerated proliferation and reduced sensitivity to drug concentration.

Various combinations of monoculture, co-culture, and triculture comprised human umbilical vein endothelial cells (HUVECs), human lung fibroblasts (hLFs), and breast cancer cells (MCF-7). Importantly, the triculture organoids remodeled to localize hLFs in the center, surrounded by MCF-7 cells with HUVECs dispersed throughout. They co-localized with hLFs, while small numbers of hLFs were dispersed in the MCF-7 layer. The highest ratio of MCF-7/hLF was found to be 3:1 and subsequently introduced into microfluidic devices. These triculture spheroids were integrated into a microvascular network, which allowed the passage of fluorescent microbeads (3.1 µm diameter) through the spheroids from the entry to exit channels (proving their perfusability). These perfusable cultures retained their 3D structure after 4 days in culture while maintaining molecular features (CD31 and E-cadherin). Perfusion-cultured vascularized trispheroids had a significantly larger tumor area (marked out by E-cadherin expression) than those without vasculature (while maintaining a constant spheroid volume). Increased numbers of proliferating (Ki-67+) cells and a decreasing cell death (ss-DNA+) area were similarly observed in perfusion cultures that exploited the vascular network to proliferate and survive.

The authors then used the physiologically relevant vascularized tumor model for realistic drug testing. In static cultures, the volume of both vascularized and nonvascularized tumor models decreased with increasing paclitaxel. Tumor-positive regions, that is, estrogen receptor areas (ER+), similarly decreased. On the other hand, when triculture tumor spheroids were tested with varying concentrations of paclitaxel (0, 5, or 50 ng/mL) and perfused for 24 h, the overall volume and E-cadherin+ areas did not change. This demonstrated how perfused cultures (resembling the in vivo situation) behaved differently from static cultures, which underscores the need for more realistic in vitro disease models. The perfusion feature may have supported increased proliferation that offsets the cell death caused by paclitaxel exposure. Future adjustments to this physiologically relevant disease model may include perfusion rates, nutrients, oxygen levels, and so forth, to further probe drug activity in realistic conditions. (Nashimoto, Y.; et al. Biomaterials 2020, 229, 119547.)

Muscle Tissue Engineering in Fibrous Gelatin: Implications for Meat Analogs

Currently, there is an incomplete strategy to culture anchorage-dependent cells in order to produce cultured meat products. Meat consists of muscle, fat, and connective tissue with various proportions that differ among various tissue sources. To mimic the native behavior of mature muscle, alignment of densely packed muscle fibers is required. Controlling cell phenotype in large-scale cultures is thus an important consideration of cell bioprocessing for laboratory-grown meat production. Fibrous gelatin biomaterials were identified as a suitable candidate that possessed suitable structural, biochemical features, as well as fitting food regulatory standards. While gelatinous biomaterials have been used as microcarrier cell substrates, these differ from the fibrous nature of native ECM architecture. However, production methods like electrospinning or phase separation are low throughput, limiting their suitability for food production. The team utilized a production method, immersion rotary jet spinning (iRJS), that generates fibers at 2–4 orders of magnitude higher rates compared with electrospinning systems. They hypothesized that gelatin fibers produced by iRJS would be able to support muscle tissue engineering in edible scaffolds at scales required for meat production.

Using a laboratory-scale iRJS, gelatin was extruded through reservoir wall perforations into ethanol solution. These generated fibers were guided into a cylindrical collector at ~2 g/min and the collection was rotated to ensure anisotropic fiber alignment. Twenty percent gelatin solutions initially behave like Newtonian fluids. However, the addition of a food-safe cross-linker (microbial transglutaminase [mTG]) gave rise to a viscous-to-elastic transition before equilibrating after ~10 min. Various gelatin concentrations (4%, 10%, and 20%), precipitation bath compositions (100:0 and 30:70 ethanol/water ratios), and storage solution compositions (postproduction) were achieved. Gelatin fibers consisted of a dense exterior and inner porous region. Fourier transform infrared spectrography also confirmed that peptide bonds were preserved for both non-cross-linked and cross-linked gelatin (chemically and enzymatically). These allowed for the control of gelatin fiber diameter between 1 and 10 µm. Such control allows recapitulation of fibers into architecture resembling native muscle architecture.

By varying the ratio of ethanol and water bath compositions, fiber diameters ranged between 2.9 and 8.7 µm. This bore a resemblance to decellularized muscle tissue, suggesting that the iRJS-produced gelatin fibers can serve as a suitable scaffold for meat products. Sections of gelatin scaffolds, 4 cm² and 1.5 mm thick, were cultured in multiwell plates. Bovine aortic smooth muscle cells (BAOSMCs) and rabbit skeletal myoblasts cells (RbSkMCs) were well attached to the gelatin fibers with F-actin (cytoskeletal) and vinculin (adhesion) proteins. With short-length fibers, greater cell aggregation was observed, whereas longer fibers promoted aligned tissue formation similar to engineered muscle. RbSkMCs seeded in spun gelatin fibers modified with mTG exhibited longer viability than aligned gelatin scaffolds. In the aligned scaffolds, fiber morphology and anisotropic tissue alignment lost structural integrity. In longer-term, 21-day cultures, fibrous gelatin scaffolds were cross-linked with EDC-NHS (N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride-N-hydroxysuccinimide). Both BAOSMCs and RbSkMCs attached to the gelatin fibers, forming 3D tissues. Twenty-eight-day cultured BAOSMCs and RbSkMCs on the spun gelatin scaffolds expressed collagen or collagen-like proteins similar to rabbit muscle, bacon, and ground beef. However, the 3D cultured samples lacked the striated muscle fibers seen in rabbit muscle or bacon. This is likely because RbSkMCs are proliferative cells with limited differentiation capacity and thus limited sarcomere assembly. BAOSMCs are not skeletal muscle cells and thus do not generate the contractile architecture seen in whole, unprocessed meat. Instead, the cultured meat products resembled processed products, such as “fish balls” or ground beef.

The authors finally used texture profile analysis (TPA) to evaluate the laboratory meat products compared with animal meat. Whereas fresh rabbit muscle and beef tenderloin decreased in hardness after cooking, the culture tissues were unchanged. This was likely due to the breakdown of the tissue matrix (collagen dissociation). Cultured meat products, on the other hand, were gelatin based or had been broken or homogenized during processing (ground beef). Thus, gelatin scaffolds were sufficiently versatile to support cell aggregation or the formation of aligned tissue. Their biocompatibility, ability to form complex tissue architecture, and food safety demonstrate their suitability for creating laboratory-grown meat products. (MacQueen, L. A.; et al. NPJ Sci. Food. 2019, 3, 20.)

Drug Delivery

Long-Acting Reversible Contraception by Effervescent Microneedle Patch

Contraception helps women to control the frequency of pregnancies. Regardless, 85 million pregnancies were unintended in 2012. Public expenditure in the United States even cost $21 million in 2010. Ideal contraceptives have some, if not all, of the following characteristics: high safety and efficacy (with minimal side effects); simple, self-administration; long-acting (at least 1 month), infrequent dosing to increase user compliance; generation of minimal biohazardous, sharp waste materials; cost-effectiveness; and high acceptability among women of reproductive age. Current nonhormonal contraceptive options include condoms and diaphragms, but these do not provide persisting protection and have limited effectiveness because of poor user acceptance and adherence. Hormonal options include orally ingested pills, injectables, and (subcutaneous) implantables. Pills are not long-acting and users are not always compliant, while injectables and implantables have reduced acceptability due to their invasiveness and the necessity of healthcare professional involvement.

To address these issues, a microneedle patch containing an effervescent backing was developed. The contraceptive hormone levonorgestrel (LNG) was chosen due to its long history, safety, and efficacy. The effervescent mechanism promotes the rapid release of microneedles from its backing, while biodegradable poly(lactic-co-glycolic) acid (PLGA) polymers allow slow, continuous drug release. The microneedle contains micrometer-sized needles that encapsulate LNG. Pressing the patch into the skin painlessly penetrates the outer barrier, releasing the encapsulated drug. In addition, the microneedle patch enables easy self-administration and does not generate biohazardous sharp waste (due to its in situ dissolving), thus giving rise to greater acceptability in human trials and low-cost manufacturing. Contact of the microneedles with interstitial fluid (ISF) in the skin generates carbon dioxide (CO₂) bubbles that cause the needles and backing to separate within a minute. Microneedle patches were fabricated by casting a PLGA solution containing LNG crystals in a silicone mold. The patches were dried to remove residual solvent and showed 40% loading of LNG in PLGA. The effervescent backing in ethanol was then cast containing water-soluble polyvinylpyrrolidone (PVP), citric acid, and sodium bicarbonate. Moisture then dissolves the patch backing, while CO₂ bubbles weaken the interface, allowing the patch to be removed. The patch consists of a 10 × 10 array of 600 µm needles containing ~0.28 mg of LNG per ~0.5 cm² patch. The failure strength of 0.07 N per needle is sufficient to penetrate skin. In saline solution, the patch backing generated large gas bubbles and enabled needle separation from its backing in ~10 s. This was ~5× to 10× faster than for noneffervescent microneedles.

The effervescent patches were then tested on porcine skin ex vivo. Microneedles loaded with Nile red dye were applied firmly and kept for 50 s. Almost all the dye was found to be remaining, with histological sectioning indicating successful delivery. Approximately 96% of microneedles were found to have separated and ~90% of the encapsulated dye was delivered into the skin. Compared with noneffervescent ones, the effervescent ones delivered ~3-fold more quantities of dye. LNG-loaded microneedles exhibited a rate of ~1.4% release over 60 days with a negligible burst-release profile. In rats administered LNG with effervescent patches, LNG in plasma reached a peak concentration of 0.83 ng/mL at ~98 h after application. This design kept LNG concentration levels above therapeutic levels of 0.2 ng/mL for >30 days before dropping to nonsignificant levels after 60 days. The absorption characteristics exhibited approximately first-order release kinetics. In this 60-day study, effervescent microneedles were well tolerated with no signs of irritation following histological analysis of rat skin.

Microneedles were then applied to the skin of American female volunteers with few signs of skin irritation (e.g., erythema). They even declared their preference for the microneedles compared with oral or injection-administered contraceptives. Further surveys among women of reproductive age from emerging economies, such as India and Nigeria, suggested that they were similarly amenable to using the effervescent microneedles. Compared with voltage-driven and other types of microneedles, the effervescent patches are considered a low-cost manufacturing design with simple and reliable usage. These were also well tolerated with very minor and short-lived signs of skin irritation. Acceptability surveys among a clinical trial population also serve to guide the optimization of microneedle size, application location, and so forth. The study also contained several limitations, including a relatively small cohort of animal and human volunteers. Another issue in product design involved engineering the dosage for 1-month or 6-month longer-term usage. Crucially, the acceptability studies also suggest that the target demographic (women of reproductive age) would prefer effervescent microneedles compared with existing hormonal contraceptives. Several important attributes include their ease of use, discreteness, and short duration of action. Overall, these improve access to long-acting contraception for women. (Li, W., et al. Sci. Adv. 2019, 5, eaaw8145.)

Glucose-Responsive Nanoparticles for Rapid and Extended Self-Regulated Insulin Delivery

Diabetes mellitus gives rise to poorly controlled blood glucose levels. This leads to severe complications, including retinopathy, cardiovascular disease, kidney failure, and even cancer. While intensive insulin therapy can mitigate such issues, multiple blood glucose measurements and insulin injections must be performed daily. Unfortunately, intensive therapy is characterized by decreased patient compliance and increased hypoglycemia, which can give rise to brain damage, seizures, loss of consciousness, and death. Even though insulin products can be rapid-, short-, intermediate-, or long-acting, none of these products respond to glucose levels in a dynamic manner according to the patient’s needs. Such self-regulated delivery technology mimics native insulin production generated by a healthy pancreatic organ.

While this concept has been suggested before, concepts such as continuous glucose sensing coupled with insulin infusion have been shown to have inaccuracies in sensing with lags in signal response. Glucose-responsive biomaterials can react to glucose levels to vary insulin release rates. Glucose detection in such systems is often mediated by the covalent attachment of a glucose-sensing moiety (e.g., boronic acid or the glucose oxidase enzyme [GOx]). Current glucose-responsive formulations exhibit poor sensitivity, suffering from long delays in supplying significant quantities of insulin. The rapid-release profile of insulin using current formulations means that multiple doses are required for glycemic control for therapeutically relevant periods of time (i.e., 10 h). A novel formulation involved the combination of rapid-release and prolonged-release elements comprising acetylated dextran nanoparticles (NPs) encapsulating GOx, catalase, and insulin. Polymers with a high acyclic acetal content degrade quicker, whereas those with a high cyclic acetal content exhibit longer half-lives. Having different periods of dextran synthesis results in a range of cyclic modification from 55% (C55) to 83% (C83). In acidic buffer (pH 4.7), NPs synthesized from the least modified dextran polymer (C55NPs) degraded >80% within the first hour, whereas those with greater amounts of modifications were more stable in acidic solution. The minimally modified C55NPs exhibited minimal degradation in pH 7.4 solution incubated for 24 h.

NPs were next synthesized containing insulin, GOx, and catalase. GOx converts glucose to by-products that give rise to an acidic microenvironment in high-glucose environments. Further analysis suggests that C55NPs in acidic solution reduced in size over the first hour by 33% (using dynamic light scattering and cryotransmission electron microscopy) before becoming undetectable. The use of a combination of rapid release (C55NPs; burst release of the most insulin in 2 h) and delayed insulin release (C71NPs; 2 h delayed onset) resulted in a mixture of fast- and long-acting delivery options. This combination affords rapid insulin release in elevated glucose compared with more prolonged release in physiological glucose concentration levels. Circular dichroism spectra revealed good agreement between released insulin and fresh insulin even after trypsin incubation for 24 h. This suggests that the NP formulation protects insulin from proteolysis and retains its α-helical structure. In fact, released insulin retains bioactivity, enabling AKT phosphorylation through the insulin receptor to a similar extent as fresh insulin. NPs did not demonstrate perceivable hemolysis in the presence of erythrocytes. Implanted in immune-compromised mice, the majority of the NPs were retained at their original location. Pro-inflammatory cytokines—interleukin (IL)-1β, IL-6, IL-12, and tumor necrosis factor (TNF)-α—were not elevated compared with saline injections. Histology showed the presence of macrophages during the short term, but these were not present after 2 weeks.

In streptozotocin-treated mice that mimic type I diabetes, NPs maintained blood glucose levels in a normal glycemic range for 16 h before hyperglycemia was observed 2 days postinjection following a glucose tolerance test. In contrast, mice treated with naked insulin had controlled glucose levels before returning to hyperglycemia levels 3 h after the introduction of glucose (as a simulated meal). This suggested that the NP formulation was suitable as a once-daily treatment. Further analysis of the NP formulation suggests that enzymes and hyperglycemia were critical to trigger enhanced insulin release within “diabetic” mice. This combination of fast and slow (but persistent) release of insulin controls not only hyperglycemia but also the hypoglycemia that ensues due to the oversupply of insulin. This innovation paves the way for the development of self-regulated insulin delivery. (Volpatti, L. R.; et al. ACS Nano 2020, 14, 488–497.)