Living Materials Approach for In Situ Bio‐Polymers Production Using Bacillus Paralicheniformis in Microneedles

Abstract

Living biomaterials, which integrate live organisms with traditional macromolecular scaffolds, function as “live manufacturers” capable of sensing their environment, synthesizing, and releasing biomolecules while remaining stable under physiological conditions. While systems that produce small biomolecules continue to advance, in situ production and secretion of high‐molecular‐weight biopolymers remain relatively underexplored. Here, a microneedle (MN) patch system is presented encapsulating Bacillus paralicheniformis (B. paralicheniformis) – a non‐pathogenic, Gram‐positive bacterium known for its production of γ‐polyglutamic acid (γ‐PGA). The MNs are designed to painlessly penetrate the stratum corneum and reach the dermis. Bacteria are uniformly distributed within the patch, and their presence has minimal impact on the microneedles’ morphology and mechanical integrity. Upon application, B. paralicheniformis is released from the MNs and successfully produced γ‐PGA, with molecular weights ranging from 64 to 563 kDa. Growth studies revealed that Luria‐Bertani (LB) medium supports optimal bacterial proliferation, while E medium enhances γ‐PGA biosynthesis. In vivo studies confirmed that B. paralicheniformis colonized mouse skin following MN administration and secreted γ‐PGA without eliciting toxicity or inflammatory responses. Given the increasing therapeutic use of biopolymers and proteins for treating chronic and acute skin conditions, this living bacterial delivery system offers a promising platform for sustainable and symbiotic dermal therapies.

Dissolvable microneedle (MN) device containing Bacillus paralicheniformis. The polymeric matrix encapsulates and protects the bacteria, preserving their viability while enabling in situ production and release of γ‐polyglutamic acid. The bacteria are delivered into the skin via 500 µm‐long microneedles, and remain detectable on the skin 24 h post‐administration in treated mice, in contrast to control animals receiving bacteria‐free MNs.

1. Introduction

Living biomaterials are the integration of live organisms, such as bacteria or cells, with traditional biomaterials to create functional, responsive constructs.^{[ 1 ]} These systems can act as “live manufacturers”, capable of sensing their environment, producing and releasing biomolecules while maintaining their stability under physiological conditions. By harnessing the metabolic capacity of living cells to produce bioactive compounds from simple precursors, such systems function as dynamic, active therapeutic devices. For example, our group previously encapsulated Bacillus subtilis (B. subtilis), a food‐grade bacterium known for its robust protein secretion mechanism, within a thermoresponsive Pluronic F‐127‐based hydrogel that supports bacterial germination and proliferation.^{[ 2 ]} Upon administration, the transparent formula hardens, forming an in situ “bio‐factory” that continuously produces and releases antifungal agents at the desired site. Other engineered living materials based delivery strategies include encapsulated probiotic Staphylococcus epidermidis for treating acne,^{[ 3 ]} B. subtilis in PVA microparticles for treating open wounds,^{[ 4 ]} engineering bacteria encapsulated in a protective magnetic hydrogel for long‐term disease diagnosis in the intestine.^{[ 5 ]} Finally, when properly encapsulated or engineered for stability, several bacterial species, including Salmonella, Escherichia coli, Clostridium, and Listeria have demonstrated the ability to restrict tumor growth and improve prognosis in preclinical murine models.^{[ 6 ]} However, unlike previous studies focused on the secretion of small antimicrobial molecules, the present work demonstrates the in situ dermal production of high‐molecular‐weight multifunctional biopolymer. To the best of our knowledge, this represents the first demonstration of a living microneedles (MN) system designed to biosynthesize and deliver such large therapeutic polymers directly into the skin.^{[ 7 ]}

Here, we present a living material system designed to biosynthesize and deliver high molecular weight γ‐polyglutamic acid (γ‐PGA) directly into the skin (Scheme 1 ). γ‐PGA has recently attracted attention for its applications in dermal therapies, functioning as a hemostatic adhesive ^{[ 8 ]} and as an active agent for improving skin hydration and reducing wrinkles, outperforming even hyaluronic acid in some studies.^{[ 9 ]} The therapeutic benefits of γ‐PGA have been attributed to its exceptional water retention capacity, biocompatibility and its ability to modulate cell behaviors such as adhesion, migration, and proliferation.^{[ 10} , ^{11 ]} For example, it has been demonstrated that the administration of γ‐PGA directly into the dermis can effectively activate resident dendritic cells, thereby modulating immune responses and alleviating the pathological features of atopic dermatitis .^{[ 12 ]} Moreover, intradermal delivery of high‐molecular‐weight γ‐PGA has been shown to result in prolonged retention within the skin for at least six days, highlighting a significant advantage over conventional topical application, which generally exhibits limited skin penetration and reduced residence time.^{[ 13 ]}

Scheme 1.

MNs array administration. The polymers comprising the device dissolve in contact with water and release the bacterial cargo onto the skin. This allows B. paralicheniformis to produce and secrete γ ‐PGA.

γ‐PGA can be synthesized and secreted by Bacillus paralicheniformis (B. paralicheniformis; formerly Bacillus licheniformis, ATCC 9945A), a Gram positive, facultatively anaerobic bacterium.^{[ 14 ]} These L‐glutamic acid dependent bacteria utilize a racemization pathway to convert L‐glutamic acid into D‐glutamic acid. The resulting monomers are then polymerized by γ‐polyglutamic acid synthetase, an enzyme complex that also mediates the extracellular secretion of γ‐PGA.^{[ 10 ]} To facilitate skin‐targeted delivery, MN were selected as the delivery platform. MN can painlessly penetrate the stratum corneum and epidermis, reaching the dermis layer, without damaging blood vessels or sensory nerves.^{[ 15} , ^{16 ]} Therefore, the encapsulation of live B. paralicheniformis in MN array is hypothesized to maintain bacterial viability, and support continuous, localized γ‐PGA release directly onto the skin.^{[ 17 ]} In this study, MN patches with and without B. paralicheniformis were fabricated using the micromolding technique.^{[ 18 ]} The MN were composed of a blend of poly(vinyl alcohol) (PVA) and poly(vinylpyrrolidone) (PVP), two fast‐dissolving, water‐soluble, and biocompatible polymers known for their mechanical robustness and effective skin penetration capabilities.^{[ 19} , ²⁰ , ^{21 ]} After administration, the polymeric carrier is expected to dissolve in situ, thereby releasing the encapsulated bacteria directly into the skin tissue.

2. Results and Discussion

2.1. Fabrication and Imaging of the Living MNs

MNs patches with and without B. paralicheniformis were fabricated using the micromolding technique from varying bacterial concentrations in a 1:2 (v/v) PVA/PVP mixture (Figure 1A).^{[ 18 ]} Patches were dried overnight at room temperature before being removed from the molds. Each patch measured 64 mm² and was composed of 10 × 10 needle array with a 0.5 mm spacing between adjacent needles (Figure 1B). Scanning electron microscopy (SEM) confirmed the formation of well‐structured pyramid‐shaped MNs with smooth surface typical for PVA and PVP MNs.^{[ 22} , ^{23 ]} All needles appeared structurally consistent, appearing pyramidal square base and sharp tips. No discernible structural or morphological differences were observed between empty and the bacterial MNs (Figure 1C). DAPI staining revealed that B. paralicheniformis was uniformly distributed throughout the needle matrix, from base to tip (Figure 1D). In contrast, patches containing unstained bacteria showed no fluorescent signal under identical Spinning Disc Confocal (SDC) imaging conditions (Figure 1E). To assess the encapsulation efficiency of live bacteria in the MN patches, the number of colony‐forming units (CFU) was quantified. MNs were fabricated as described and subsequently dissolved in LB medium, after which the suspension was plated on agar and incubated overnight. The resulting bacterial count was (5.5  ±  0.4) × 10⁶ CFU(n  =  3), indicating approximately a two‐fold reduction compared to the initial bacterial concentration. This reduction in viable bacterial count may be attributed to two key factors: ^{[ 1 ]} Loss of bacteria during the spin‐coating process, likely due to elevated spinning velocities, and ^{[ 2 ]} Bacterial death during the air‐drying step, a process previously associated with the generation of free radicals and induction of oxidative stress, both of which can impair bacterial viability.^{[ 24 ]}

Figure 1.

MNs fabrication and morphology. A) Schematic Illustration of the fabrication process of B. paralicheniformis loaded MN patches using the micromolding technique. B) Dry B. paralicheniformis patch. C) SEM images of single and 6 needles in a patch, with and without bacteria. D) SDC microscopy image showing bacterial scattering using DAPI. E) Non‐stained bacterial patch image captured in SDC.

Overall, bacterial MNs showed well‐recognized sharp needles structures with bacteria evenly dispersed bacteria. Such uniform dispersion of bacteria within the MNs has previously been shown to preserve shelf life for several months and to enable a more predictable release profile of B. subtilis from PVA/PVP microneedles onto the skin.^{[ 25 ]}

2.2. Mechanical Characterizations

Since needles are required to pass through the stratum corneum^{[ 16 ]} the ability of our system to penetrate the skin layers was examined. After manual application of the patches on the dorsal area of a mice, a confined insertion mark limited to each individual needle can be observed with no impact on the surrounding area. (Figure 2A). A microscopic view of the skin surface after MN insertion using Environmental Scanning Electron Microscope (E‐SEM) reveals the expected array structure, with ≈500 µm spacing between neighboring needles (Figure 2B). Disruption of the skin's surface integrity following MN insertion was also observed using Scanning Electron Microscope (SEM) (Figure 2C), and histologically (Figure 2D). The sliced skin tissue demonstrates rupture of the stratum corneum layer, and the penetration of pyramidal needle to the skin, supporting dermal insertion of the device. Next, the compression forces of 12 needles (n = 4) were measured as a function of displacement up to 0.35 mm (Figure 2E). Compression tests were performed on MNs without bacteria and to MNs with bacteria with different Bacteria:Polymers (v/v) ratios All tested MNs exhibited progressive increase in force with the increase in displacement (Figure 2F). MNs without bacteria (0:1) exhibited the highest compression force, averaging 0.47 N/MN displacement at a 0.35 mm displacement. The 1:2, 1:1 and 2:1 Bacteria:Polymers ratios demonstrated similar compression trends with slightly lower force values of 0.37, 0.38, and 0.33 N/MN, respectively, at a displacement of 0.35 mm. MNs with 0:1 ratio exhibited statistically significant higher force values compared to 1:2 (^** p<0.01), 1:1 (^* p<0.05), and 2:1 (^*** p<0.001) ratio groups, yet no significant difference was observed between all bacterial groups. We note that a force of 0.058 N per needle is widely accepted as the minimum required for effective skin penetration in human subjects.^{[ 26 ]} Therefore, all tested MN compositions in this study exceed this threshold and can be considered mechanically suitable for transdermal application.

Figure 2.

Mechanical behavior of the MNs. A) Photograph showing the dorsal skin of live mouse immediately after administration and removal of the B. paralicheniformis MN patch. B) SEM image showing insertion marks on mouse skin. C) Zoomed‐in SEM image of the interrupted skin tissue after bacterial MNs patch administration and removal. D) A single rupture of the mice skin caused by 1:2 PVA/PVP needle penetration, as seen by H&E staining. E) Schematic representation of the compression test using a texture analyzer (Lloyd TA1). F) Force‐ displacement curve per needle for the tested MNs. Error bars represent one SD from the mean of n = 4. Two‐way ANOVA analysis followed by Bonferroni post‐test was used to determine significance between groups at 0.35 mm displacement, as ^* p<0.05, ^** p<0.01, and ^*** p<0.001. G) SEM images of 1:1 bacteria to polymer MNs after 0.1, 0.2, and 0.3 mm displacements, showing progressive deformation.

The minimal impact of bacterial concentrations on the mechanical properties of the MNs can be explained by the negligible mass of the bacteria relative to the polymers.^{[ 27 ]} To further investigate the effect of compression force on needle integrity, a similar experiment was conducted for the 1:1 formulation, with compression halted at various displacements to assess structural changes using SEM (Figure 2G). At a displacement of 0.1 mm, only minimal deformation of the needle tips was observed. Slightly greater deformation occurred at 0.2 mm, while more pronounced bending of the MN tips was evident at 0.3 mm displacement. This progressive deformation under compression can be attributed to the distribution of force across the entire MN array, which reduces the load borne by each individual needle and results in a gradual rather than abrupt structural change.^{[ 28 ]}

2.3. Bacteria Release and Proliferation

The ability of B. paralicheniformis to be released from the patch and subsequently proliferate was next evaluated. As a glutamic‐acid dependent bacterium, B. paralicheniformis requires specific nutrients to support its growth and γ‐PGA production. To determine the optimal nutrient medium for B. paralicheniformis growth, the bacterium was cultured in three different broths: Luria‐broth (LB), E medium and ATCC #21. The influence each of the tested media on bacterial growth kinetics was assessed by monitoring turbidity at 600 nm (OD₆₀₀) over time (Figure 3A). In LB media, B. paralicheniformis exhibited a relatively short lag phase (≈10 h), compared to ≈16 h in both E medium and ATCC #21. Furthermore, LB supported a sharp exponential growth rate(0.41 1/h), reaching a high cell density of ≈1.7 OD₆₀₀ and significantly higher optical density rates after 10.3 h compared to the other groups (^*** p<0.001). In contrast, growth in E medium and ATCC #21 proceeded with more moderate kinetics (0.27 1/h and 0.28 1/h, respectively), achieving lower peak cell densities of ≈1.2 OD₆₀₀ without significant differences between mediums. The effect of the medium on growth kinetics of B. paralicheniformis can be attributed to the nature of the carbon source provided in each formulation.^{[ 29 ]} LB is a nutrient‐rich medium that supplies a variety of essential carbon sources, primarily derived from protein lysates. In contrast, E medium and ATCC #21 rely on more defined carbon sources such as glycerol and L‐glutamic acid. Moreover, LB contains yeast extract which is abundant in amino acids, peptides, nucleotides, carbohydrates, growth factors and trace elements, giving it a significant advantage over chemically defined mediums like E and ATCC #21.^{[ 30 ]} The live MN patches developed in this study were primarily designed for the transdermal delivery of B. paralicheniformis, a γ‐PGA producing bacterium. To determine the optimal bacterial load, patches with varying bacteria to polymers ratios (bacterial suspension/polymeric solution) were fabricated. Each MN patch was incubated in LB medium and bacterial growth was monitored by measuring the optical density at 600 nm. As expected, no detectable growth was observed for the empty MNs (without bacteria) group, confirming the absence of bacterial contamination during the fabrication process (Figure 3B). Patches fabricated with a 1:1 bacteria‐to‐polymer ratio exhibited significantly higher (^* p<0.05) O.D_600nm values compared to those with 1:2 and 2:1 ratios after 6.6 h, indicating superior bacterial proliferation. Previous studies have shown that dried bacterial formulations require an optimal initial cell concentration to achieve efficient regrowth upon hydration.^{[ 31} , ³² , ^{33 ]} Excessive bacterial loading can reduce the effective surface area of media the bacteria are exposed to, and increase osmotic pressure, whereas insufficient loading may fail to initiate robust growth. In our system, the intermediate 1:1 Bacteria‐to‐polymer ratio provided this optimal balance. This composition also demonstrated favorable mechanical properties, making it the optimal balance in terms of both biological activity and structural integrity. Consequently, the 1:1 ratio was selected for all downstream experiments.

Figure 3.

Effect of different parameters on growth and release of B. paralicheniformis. A) Bacterial growth from overnight starter culture in different mediums. Data are means ± SD, n = 4. B) Bacterial release from MNs containing different bacteria/polymers ratios. n = 4. Two‐way ANOVA analysis was used to determine significance between groups over time, followed by Bonferroni post‐test, as ^* p<0.05, ^** p<0.01, and ^*** p<0.001. C) Area covered by released bacteria colonies as captured by cytation microscopy over 72 h. Data are means ± SD, n = 3. Error bars represent one SD from the mean. D) Representative Cytation imaging showing release morphology of B. paralicheniformis from MNs on E+agar and LB+agar.

The release and proliferation of B. paralicheniformis from the MN patch on a semi solid surface were evaluated using the Cytation 5 imaging system. MNs were positioned in the center of a solid growth media, and bacterial colony expansion was monitored over 72 h. Growth performance was compared between LB agar (LA) and E medium agar. A faster growth was observed for the LA, with colonies covering the entire plate surface within 48 h (Figure 3C,D). The distinct bacterial morphologies observed in this study, branch‐like in LA medium and more rounded, gel‐like structures in E medium, can be attributed to the higher levels of γ‐PGA production in the latter, as further examined in the subsequent assay.

2.4. γ‐PGA Production and Characterization

The ability of B. paralicheniformis to synthesize and secrete γ‐PGA, either in their free state or when encapsulated in MNs (free or MNs) was assessed. First, batch fermentations using free bacteria were conducted over 72 h (Figure 4A). The fermentation broth was centrifuged at 4200 rpm for 15 min and the supernatant was extracted with cold ethanol absolute at a 1:4 (v/v) ratio and incubated overnight at 4 °C. After purification and lyophilization, a soft white polymer was obtained (Figure 4B). In this assay, E medium supported significant higher γ‐PGA production (2.6 ± 0.5 mg mL⁻¹) compared to LB medium (0.51 ± 0.14 mg mL⁻¹) (n = 2). Growth on semi solid media corroborated these findings: in solid E medium, a freely flowing white polymer was observed, spreading beyond the original colony margins (Figure 4C), whereas in LA, the produced polymer retained its shape, showing minimal diffusion beyond the inoculation site (Figure 4D). Finally, a defined amount of commercially available γ‐PGA was dissolved in E medium, and the same extraction procedure was performed to determine recovery yield. The result was 90.0  ±  8.8% (n  =  3), thereby verifying the reliability and efficiency of the extraction method.

Figure 4.

Characterization of γ‐PGA. A) Schematic of γ‐PGA extraction from B. paralicheniformis fermentation broth. B) Lyophilized γ‐PGA following ethanol precipitation and purification. C) Growth of B. paralicheniformis on E+agar plate showing diffuse polymer spread. D) Growth on LA plate with minimal polymer dispersion. E) Image of bacterial MN patch after 72 h incubation demonstrating visible polymer production. F) 1H‐NMR spectrum of γ‐PGA, highlighting characteristic peaks. G) FTIR spectra of γ‐PGA, indicating functional group signatures.

Next, the ability of B. paralicheniformis to synthesize and secrete γ‐PGA while encapsulated within a MN was assessed. A single MN patch was placed at the center of a 6 well plate containing either E medium with agar (E+agar) or LA. The plate was incubated at 37 °C with 90% humidity and γ‐PGA production was monitored as area growth (mm²) over 72 h. Bacterial growth was visible on both media within a few hours; however, a more rapid expansion was observed on LA, with the bacterial colony fully covering the well surface within 48 h (Figure 3C,D).

Nevertheless, after 72 h, only the wells containing E+agar were fully covered with distinct slimy layer (Figure 4E; Video S1, Supporting Information). This observation reinforces our earlier findings that while LB medium supports faster B. paralicheniformis growth, E medium is more conductive to γ‐PGA synthesis, even when bacteria are released from MNs. The superior performance of E medium in promoting γ‐PGA production has been attributed to its high concentration of L‐glutamic acid compared to LB.^{[ 34 ]} Another plausible explanation is that in bacteria, the production of secondary metabolites, such as γ‐PGA, often occurs at the expense of biomass accumulation.^{[ 35 ]} The exclusive and pronounced production of γ‐PGA in E medium aligns well with both our batch fermentation results (Figure 3A) and the understanding that E medium supplies the essential components required for efficient γ‐PGA biosynthesis. Overall, our findings, that LB medium is preferred for the growth of B. paralicheniformis, while E medium enhances γ‐PGA production are in agreement with previous studies investigating the impact of various culture mediums on B. paralicheniformis growth and performance.^{[ 36} , ³⁷ , ^{38 ]}

Next, the structural formula of γ‐PGA produced by B. paralicheniformis either as free bacteria or when encapsulated in MNs, was confirmed using proton nuclear magnetic resonance (¹H‐NMR, Bruker AvanceTM 400), Fourier transform infrared spectroscopy (FTIR, ALPHA, Bruker) and gel permeation chromatography (GPC, Waters 2695). The ¹H‐NMR spectra (Figure 4F) showed characteristic peaks at peaks at 4.1 ppm (α‐CH), 1.85 and 2.0 ppm (β‐ CH₂) and 2.3 ppm (γ‐CH₂) supporting the presence of γ‐PGA structure (Figure 4F).^{[ 39 ]} FTIR spectra of the lyophilized polymer (Figure 4G) revealed prominent peaks at 3297 and 3128 cm⁻¹ corresponding to O─H and N─H stretching vibrations respectively; 2964 cm⁻¹ (C‐H stretching); 1551 cm⁻¹ (C═O stretching of carbonyl groups); and 1073 cm⁻¹ (C─O stretching), all consistent with the functional groups of γ‐PGA.^{[ 40 ]} Together, these results confirm the synthesis of γ‐PGA in both liquid and solid culture conditions. GPC analysis determined the molecular weights (MW) of the γ‐PGA to within the range of 63554–562, 960 Da, which aligns with literature values reported for γ‐PGA synthesized by Bacillus species, typically ranging from 10000 to 1000, 000 Da.^{[ 41 ]}

2.5. In Vivo Toxicity and Colonization

To assess the colonization capability and safety profile of B. paralicheniformis‐loaded MN patches, an in vivo study was carried out. Ten mice were randomly assigned to two groups (n = 4 and n = 6). Each mouse was applied with two MN patches to the dorsal skin: one group was treated with patches containing B. paralicheniformis, while the control group received bacteria‐free patches (Figure 5A). To maintain patch adherence during the experiment and ensure optimal bacterial proliferation, we applied 5 µL of LB medium around the periphery of the patch, effectively securing its edges. LB was selected for this purpose due to its superior support for B. paralicheniformis growth, as demonstrated in this study by its shortest lag phase and sharper exponential growth rate, when compared to both E medium and ATCC #21 formulations.

Figure 5.

In vivo assessment of bacterial MNs toxicity and colonization in mice. A) Schematic illustration of the experimental setup. B) Photograph of mice before and after treatment with either empty or B. paralicheniformis‐loaded MNs. C) Representative histological sections (H&E stained) of skin tissues (including higher magnification), from mice treated with empty MN or bacterial MNs, showing no signs of inflammation or tissue damage on day 7 of the experiment. D) PCR analysis of swabbed skin samples: one treated with PVA/PVP MNs and one with bacterial MNs (n = 10 mice). Amplification was performed using primers targeting the BALI_RS06210 gene, encoding a manganese catalase family protein specific for B. paralicheniformis.

Histological examinations of skin tissues following treatment with either empty or B. paralicheniformis‐loaded MN patches revealed normal skin architecture, with no evidence of inflammatory cells infiltration, edema or epidermal thickening (Figure 5B). Similar clinical observations were observed for B. licheniformis Me1 in previous studies, supporting the classification of our system as non‐irritant according to Commission Directive 2001/59/EC.^{[ 42 ]} Hematoxylin‐Eosin (H&E) staining confirmed the absence of inflammation in both groups (Figure 5C).

To verify bacterial colonization, B. paralicheniformis presence was assessed using polymerase chain reaction (PCR) targeting the BALI_RS06210 gene (with 477 bp PCR product), which encodes a manganese catalase family protein. DNA samples from dorsal skin swabs were collected daily, and PCR bands between 400 and 500 bp were consistently detected in treated mice on days 2, 4, 5, 6, and 7 (Figure 5D). These bands matched the positive control (bacteria from a fresh LA plate). No bands were observed on day 1 (pre‐treatment) or in the control group treated with empty MNs, confirming both successful delivery and the absence of background colonization in the native microbiome. These PCR results should be interpreted qualitatively, as swabbing of the dorsal area are affected by the contact area of the swab with the skin, and the strength applied to collect the sample.

To validate the functionality of B. paralicheniformis on murine skin, four mice were treated with MN patches – two with empty MNs and two with bacterial loaded MNs for 4 days, and sacrificed on day 5 for skin collection. The skin was cut into small pieces and incubated in basic phosphate‐buffered saline (PBS), followed by the mentioned γ‐PGA extraction method. ¹H‐NMR analysis of tissue extracts revealed spectra features potentially corresponding to characteristics γ‐PGA peaks (as shown in Figure 4F) in mice treated with bacterial MNs, along with additional signals indicative of impurities (Figure S1, Supporting Information). In contrast, no γ‐PGA corresponding peaks were observed in extracts from control mice. To further assess γ‐PGA presence on mice skin treated with bacterial loaded MNs, lyophilized skin extracts were reacted with tannic acid, a compound known to form insoluble gel matrices with pure γ‐PGA (Figure S2A, Supporting Information).^{[ 43 ]}

While the extraction from the control group (empty MN patches) did not yield any gel formation (Figure S2B, Supporting Information), the bacteria‐treated group produced a cloudy suspension and a visible gel‐like bead at the tip of the droplet (Figure S2C, Supporting Information). This bead remained insoluble in water, supporting the conclusion that gel formation had occurred. These observations provide further evidence for the presence of γ‐PGA in the treated tissue (Figure S2D, Supporting Information). The dermal colonization of the bacteria was then observed using spectral microscopy. We employed green fluorescent protein (GFP)‐labeled B. subtilis, a strain closely related to B. paralicheniformis in both morphology and lifestyle,^{[ 44 ]} while also being more amenable to genetic transformation. Following MN administration, GFP fluorescence clearly indicated that the bacteria (green) reached the dermis layer (magenta), confirming effective delivery and localization (Figure S3, Supporting Information).^{[ 45 ]} MNs were fully degraded in vivo after 2 h, similarly to result obtained for empty MN of the same composition (Figure S4, Supporting Information).^{[ 46 ]} In summary, B. paralicheniformis showed evidence of colonization in mouse skin following bacterial MN administration, without inducing any observable signs of toxicity or inflammation.

3. Conclusion

We have developed a living MN system in which B. paralicheniformis is evenly dispersed within the polymeric matrix. This bioactive platform enables continuous in situ production and delivery of high‐molecular‐weight γ‐PGA. The composition of the MNs, specifically the bacteria‐to‐polymer ratio and the presence or absence of bacteria, had minimal impact on their mechanical properties. Among the tested media, LB medium was found to be optimal for bacterial growth, whereas E medium enhanced γ‐PGA production, likely due to its higher concentration of L‐glutamic acid. In vivo evaluations of skin tissues following bacterial MN administration confirmed the successful colonization of B. paralicheniformis on the mouse skin and γ‐PGA production with no observed signs of inflammation or tissue trauma. Given that various polymers and proteins are currently employed as therapeutic agents for the treatment of chronic and acute skin diseases,^{[ 47 ]} this living bacterial delivery system presents a novel and potentially transformative approach for the sustained administration of bioactive polymers, minimizing the need for frequent user intervention. In future studies, advanced fabrication techniques such as Pulsating In Situ Dried Electrostretching ^{[ 48 ]} and innovative structural designs like Stretchable MN Adhesive Patches ^{[ 49 ]} could be considered for the development of living dermal microneedle systems with enhanced skin penetration capabilities.

4. Experimental Section

Materials

Polyvinylpyrrolidone (PVP, 230,000 g mol⁻¹), polyvinyl alcohol (PVA, 13000–23,000 g mol⁻¹), poly‐L‐glutamic acid (PGA, 50000–100,000 g mol⁻¹), ammonium chloride (NH₄Cl), magnesium sulfate heptahydrate (MgSO₄·7H₂O), ferric chloride hexahydrate (FeCl₃·6H₂O), calcium chloride dihydrate (CaCl₂·2H₂O), manganese sulfate monohydrate (MnSO₄·H₂O), citrate‐iron(III)‐ammonia complex (C₆H₈O₇·xFe³⁺·yNH₃), L‐glutamic acid, 4′,6‐diamidino‐2‐phenylindole dihydrochloride (DAPI), and Dulbecco's phosphate‐buffered saline (PBS) were all purchased from Sigma‐Aldrich (St. Louis, MO, USA). Sodium chloride (NaCl) was obtained from Biolab (Jerusalem, Israel). Tryptone, yeast extract, and agar were sourced from BD (Becton, Dickinson and Company, Franklin Lakes, NJ, USA). Citric acid was purchased from Frutarom (Haifa, Israel), glycerol from Gadot (Netanya, Israel), and dipotassium hydrogen phosphate (K₂HPO₄) from Carlo Erba Reagents (Emmendingen, Germany). Bacillus paralicheniformis ATCC 9945A was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The PureLink Microbiome Kit was acquired from Invitrogen via Thermo Fisher Scientific (Waltham, MA, USA), while FLOQSwabs were purchased from Copan Diagnostics (Murrieta, CA, USA). Hy‐Taq Ready Mix, 1 kb and 100 bp DNA ladders, and agarose were sourced from Hylabs (Rehovot, Israel). Primers were ordered from Integrated DNA Technologies (IDT, Lowa, USA).

Optimization of Growth and Production Mediums for B. Paralicheniformis

Proliferation was monitored in three different growth media: Luria‐Bertani (LB) medium, composed of NaCl (5 g), tryptone (10 g), and yeast extract (5 g) in 1 L water; E medium, containing L‐glutamic acid (20 g), citric acid (12 g), glycerol (80 g), NH₄Cl (7 g), K₂HPO₄ (0.5 g), MgSO₄·7H₂O (0.5 g), FeCl₃·6H₂O (0.04 g), CaCl₂·2H₂O (0.15 g), and MnSO₄·H₂O (0.104 g), in 1 L water with the pH adjusted to 7.4; And ATCC #21 medium: Composed of K₂HPO₄ (0.5 g), citrate‐iron(III)‐ammonia complex (C₆H₈O₇·xFe³⁺·yNH₃, 0.5 g), MgSO₄ (0.5 g), glycerol (20 g), citric acid (2.0 g), and L‐glutamic acid (4 g), in 1 L water with the pH adjusted to 7.4. Bacterial growth was monitored using a Synergy H1 Plate Reader (Biotek Instruments Inc., Winooski, VT, USA) at 600 nm (OD₆₀₀) over 48 h at 37 °C, with continuous shaking in a 24‐well plate (Nunc, Thermo Scientific, Denmark). Each well containing 2 ml of the tested medium was supplemented with lyophilized bacteria powder as received from ATCC.

Fabrication of the Living MNs

MNs were fabricated using the micromolding technique.^{[ 18 ]} First, B. paralicheniformis starter (1 O.D_{600 nm}) was centrifuged at 4200 rpm, and the supernatant was discarded. A polymeric solution was prepared by mixing polyvinyl alcohol (PVA, 10 wt.%) and polyvinylpyrrolidone (PVP, 10 wt.%) in a 1:2 volume ratio. The bacterial pellet was resuspended in this solution (1 mL per (5.3 ± 0.4)X10⁸ CFU) and applied onto a PDMS mold with an area of ≈64 mm², cavity height of 500 µm, and base diameter of 200 µm.

The solution was forced into the mold cavities using a spin coater (Ossila Ltd spin coater, Sheffield, UK) at 4000 rpm for 1 min. This process was repeated twice, and a third application was allowed to dry at room temperature for 24 h without spin coating. MNs were gently removed from the mold and placed in a dehumidified container until use. Control MNs (without bacteria) were fabricated and stored under identical conditions. All MNs were stored in a dehumidified container and used within 30 days of fabrication. It was important to note that γ‐PGA production was not expected to occur on the MN patch itself. As demonstrated in Figure 3B, B. paralicheniformis begins to proliferate after 3 h. Given that the MN patches were fabricated without nutrient media and undergo a drying process lasting several hours, bacterial metabolic activity was effectively halted during preparation. These conditions make pre‐application biosynthesis of γ‐PGA highly unlikely.

MNs Imaging

The morphology and appearance of the MNs were examined using a field emission environmental scanning electron microscope (FEI E‐SEM Quanta 200, Eindhoven, The Netherlands). Bacterial scattering was visualized using confocal microscopy (Nikon, spinning disc confocal, SDC) on MNs containing DAPI‐stained B. paralicheniformis. For DAPI staining, DAPI was dissolved in ultra‐pure water (1 mg mL⁻¹) and added to LB to achieve a final concentration of 10 µg mL⁻¹. B. paralicheniformis starter (grown overnight, 37 °C) was centrifuged at 4200 rpm for 5 min, and the supernatant was replaced with the LB/DAPI solution. The bacterial solution was incubated in the dark for 5 min followed by centrifugation under the same conditions. The supernatant was discarded, and excess dye was removed by washing with PBS, followed by another centrifugation step to collect the pellet. The pellet was then resuspended in a PVA/PVP mixture (1:1 ratio), and DAPI‐stained bacterial MNs were fabricated using the micromolding technique as previously described.

Encapsulation Efficiency

The number of live bacteria encapsulated during the fabrication process was determined using plate count of CFU per patch. Bacterial MN patches were prepared, and each patch was placed into 1 ml of LB in 1.5 ml Eppendorf vial (n = 3). After complete dissolution, the samples were serially diluted until 10⁻⁵ and 100 µl of each dilution was plated using sterile beads on LA plates. The plates were incubated at 37 °C overnight and colonies were counted the next day.

Mechanical Characterizations of the Bacterial MNs

For visualization of the MNs pattern after MN insertion onto the skin, samples were dried using Kimwipes and placed on metal grips using carbon stickers. B. paralicheniformis MNs patches were inserted manually into the skin and removed after 1 min. Then, skin morphology was captured with E‐SEM using high vacuum without coating.

To capture the insertion depth inside the skin layers, PVA/PVP MNs patches were inserted into mice skin. The treated skin samples were fixated in paraformaldehyde solution (PFA, 4% in water) overnight, which was later replaced with increasing concentrations of sucrose solution (10, 20, and 30%w). Lastly, samples were soaked in OCT compound and frozen at −80 °C for 2 h prior to slicing using a cryostat to 10 µm thickness (Leica, Wetzlar, Germany). The sliced tissues were captured using Leica DMI8 microscope after H&E staining.

The mechanical strength of MNs with different bacteria/polymers ratios was carried out using 12 needles patches and a Lloyd TA1 texture analyzer (Lloyd Instruments Ltd., WestSussex, UK) equipped with a 10 N load cell for compression test. The tested bacteria:polymers ratios (1:2, 1:1, 2:1, and 0:1) MNs were fixed to the metal plate from their bases (with needles facing upward) using a carbon patch (Agar scientific, UK). An aluminum plate was fixed to the upper moving crimp compressing the needles at a speed of 1 mm/min for up to 350 µm from the contact position. The compression force was monitored and presented as N/needle (n = 4). To image the needle bending pattern, 12 needles were cut from a whole patch (n = 4) and prepared in the same way. Compression was proceeded up to 100, 200, and 300 µm from contact position, and kept in dehumidified chamber until SEM imaging was performed.

Bacterial Release from MNs and Proliferation

Bacterial MNs were sectioned into quarters (each containing 5 × 5 needles) and placed in individual wells of a 24‐well plate, each containing 2 mL LB medium. Bacterial growth was dynamically monitored using a spectrophotometer (OD₆₀₀) at 20 min intervals under continuous shaking at 37 °C, until the stationary phase was reached. Bacterial growth and release from MNs were observed using Cytation microscopy (Cytation 5 BioTek). A 6‐well plate containing either LA or E+agar was prepared to compare colony morphology and growth rates across different media. Each MN patch was placed at the center of a well, and plates were scanned every hour while incubated at 37 °C with 90% humidity. Bacterial release was quantified by measuring colony area expansion (mm²) over 72 h.

PGA Production and Extraction

Fermentation of B. paralicheniformis was carried out in 250 mL flask containing 50 mL E medium. Four MNs were introduced into the flask and incubated at 37 °C with continuous shaking (200 rpm) for 72 h. For purification, the fermentation products were centrifuged at 4200 rpm for 15 min and the supernatant was mixed with ethanol absolute (4 °C) in 1:4 (v/v) ratio respectively, mixed thoroughly and refrigerated overnight (4 °C), followed by 15 min centrifugation at 15000 g and the supernatant was discarded. The resulting pellet was resuspended in double‐distilled water (DDW) and subjected to dialysis against 5 L of DDW for 24 h using a 1 kDa molecular weight cut‐off (MWCO) membrane (Spectrum Labs, USA). The purified poly‐γ‐glutamic acid (γ‐PGA) solution was subsequently lyophilized for 72 h to obtain the dried product, which was later weighed. For comparison, a parallel fermentation experiment was conducted under the same conditions using a bacterial starter culture (3 mL, OD₆₀₀ = 1.0) instead of MNs. To evaluate the polymer's extraction yield, the same procedure was performed using 3 mg of γ‐PGA in E medium (n = 3). The chemical structure of the polymer was assessed by Nuclear Magnetic Resonance (¹H‐NMR) in D₂O (Brucker Avance III 400 MHz), with characteristic signals corresponding to  α‐CH, β‐ CH₂ and γ‐CH₂. Fourier Transform Infra‐Red Spectroscopy (FTIR, ALPHA, Bruker), and Gel Permeation Chromatography (GPC, Waters 2695) were also used to validate the structure and length of the freshly synthesized γ‐PGA.

In Vivo Toxicity, Colonization and PGA Production

Sixteen 8‐week‐old female C57BL/6 mice were obtained from Harlan Laboratories Ltd. (Jerusalem, Israel) and maintained under a 12‐hour light/12‐hour dark cycle. The study was conducted in accordance with the Animal Welfare (Animal Protection) Laws (1994) and approved by the Council for Animal Experiments at the Israeli Ministry of Health (Ethics Approval No. IL1791123). Ten mice were randomly divided into two groups: ^{[ 1 ]} mice administered B. paralicheniformis MNs (n = 6) and ^{[ 2 ]} mice administered empty PVA/PVP MNs (n = 4). Mice were anesthetized using 2% isoflurane (Piramal Critical Care, Inc., PA, USA) in oxygen. The dorsal region was shaved using an Isis cordless clipper (Aesculap, Suhl, Germany) and Veet hair removal cream. Two square MN patches (64 mm²) were applied to the center of the dorsal region of each mouse, followed by the application of 5 µL of LB medium. The MN insertion process was repeated daily for 6 days until the mice were sacrificed, at which point tissues were collected for histological examination of Hematoxylin and Eosin (H&E) staining, conducted at the Biomedical Core Facility (Bruce Rappaport faculty of medicine, Technion, Israel).

In a subsequent experiment, two mice were treated with B. paralicheniformis‐loaded MNs, while two control mice received PVA/PVP MNs (n = 4 total). After four days of treatment, the mice were sacrificed, and skin from the application sites was excised and minced. Tissue samples were incubated in PBS solution (pH = 8, 37 °C, 100 rpm) for 3 h. Following incubation, the supernatant was collected and dialyzed against DDW, following the PGA extraction protocol using 13–14 kDa MWCO membrane. The lyophilized product was analyzed by ¹H‐NMR in D₂O while other portion was dissolved in tannic acid (0.2% (w/v)) to assess gel formation, as previously described.^{[ 43 ]}

Bacterial Swab Collection

Swabbing of the mice was performed prior to the first MN administration (day 1) and at multiple time points after: 24 h following the first, third and sixth administrations (Days 2,4,7, respectively), as well as 3 h after the sixth administration (Day 6). Samples were collected from the dorsal region using FLOQswab pre‐soaked in an aqueous solution of NaCl (0.9%) and Tween 20 (0.1%). The swabs were then trimmed to a length of 3 cm and stored at −80 °C until DNA extraction, following the protocol described in reference.^{[ 50 ]}

DNA Extraction and Bacteria Identification

Bacterial DNA was extracted using the PureLink Microbiome Kit according to the manufacture's protocol, with an additional 95 °C incubation to enhance yield. The samples were stored at −20 °C until Polymerase chain reaction (PCR) analysis, which was performed using Hy‐Taq ready mix following manufacturer's instructions. Primers designed for the identification of BALI_RS06210 gene specific for B. paralicheniformis were: Forward: 5′‐ GGTTCATGTGGCTTTATGAG ‐3′ and Reverse: 5′‐ GCCCGTCATTTAAATCCGG ‐3′, with anticipated PCR product size of 477 bp. 1 kb and 100 bp ladders were loaded to agarose gel (1.8%) with ethidium bromide at 90 V for 45 min and visualized using a UV camera (Essential V6 gel documentation system, UVITEC, UK).

Bacterial Dermal Release from MNs

To capture the penetration and release of bacteria from the MNs into the skin, GFP‐expressing B. subtilis was used as a model. For the transformation, pAD 43‐25 plasmid was used and transformed into Escherichia coli (E. coli) bacteria using electroporation (2 mm cuvette, 2500 V, 400 Ω, 21 µF) and a Gene Zapper (IBI, New Haven, USA). Using a plasmid purification kit (Promega, MDN, USA) the plasmid was extracted. B. subtilis bacteria were then cultured in MC media ^{[ 51 ]} (37 °C, 200 rpm) until mid‐log phase was reached, followed by centrifugation (4200 rpm, 5 min) and resuspension to achieve final O.D of 0.5 (600 nm). To 300 µl of this bacteria suspension, 2 µg of the pAD 43‐25 plasmid were added and incubated (3 h, 200 rpm, 37 °C). The transformed bacteria were plated on LA plates with chloramphenicol. MNs patches were manufactured using the mentioned micromolding technique. To visualize the penetration of these bacteria onto the inner skin layers, Balb‐C nude mice skin was administrated with GFP‐B. subtilis MNs patch ex vivo. The skin samples were frozen in OCT for 24 h, at −80 °C, followed by slicing to 14 µm thickness using a cryostat (Leica, Wetzlar, Germany). Images of the sliced samples were imaged using spectral microscopy (LSM 880‐pruight confocal with MP laser, Zeiss, Germany).

Bacterial MNs Degradation

Estimation of MNs degradation time was performed in vivo (n = 2). Each mouse was inserted with 2 bacterial MNs patches on the dorsal area, and 5 µl LB were used to seal the edges of the patches over the skin. The degradation was monitored and captured after 0, 1, 1.5, and 2 h. Finally, mice were sacrificed, and their skin was collected for E‐SEM and histology studies.

Statistical Analysis

Statistical analysis and comparisons (p‐value calculated) were performed with a Prism 5 GraphPad (La Jolla, CA, USA). The results were presented as mean values ± SD. A two‐way analysis of variance (ANOVA) was used for multiple comparisons between groups, followed by Bonferroni post‐test. Statistical significance between groups was considered for p‐value<0.05.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Supplemental Video 1

Supplemental Video 2

Supplemental Video 3

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.