{"paper_id":"1ecdb7fc-19cd-43b1-9c7e-282006df30ff","body_text":"1 \nNanocellulose hydrogels as bio-interface analogs for studying nanomaterial transport and 1 \naccumulation 2 \nJoshua Prince1,a, Darryl Taylor2,b and A-Andrew D. Jones, III1,2,3,c,* 3 \nAffiliations: 4 \n1. Department of Civil & Environmental Engineering, Pratt School of Engineering, Duke 5 \nUniversity, Durham, NC 27708  6 \n2. University Program in Materials Science & Engineering, Duke University, Durham, NC 7 \n27708,  8 \n3. Thomas Lord Department of Mechanical Engineering & Materials Science, Pratt School 9 \nof Engineering, Duke University, Durham, NC 27708 10 \na. Present address: joshuaprince58@gmail.com, ICF ORCID: 0000-0001-6050-9544 11 \nb. darryl.taylor@duke.edu, ORCID: 0000-0001-6525-857X 12 \nc. Corresponding author: andrew.jones3@duke.edu, ORCID: 0000-0003-3840-8039 13 \n 14 \nAbstract 15 \n 16 \nNanomaterials have been proposed as drug delivery vehicles to enhance targeting and efficiency 17 \nof traditional and novel therapeutics and have subsequently been studied for potential ecotoxicity. 18 \nPrevious studies have identified size, surface charge, and volume exclusion as factors that 19 \ninfluence nano material diffusion and retention. However, there is little accepted or successful 20 \nquantification of how these parameters influence nano material penetration relative to biological 21 \nadaptation and biological response . Part of the challenge is the response of living biological 22 \ninterfaces to many of these nanomaterial delivery vehicles and nanosized drugs. This study aimed 23 \nto emulate key physicochemical barriers to diffusion found in living biomaterials by developing a 24 \ntunable, synthetic hydrogel. Through the controlled exposure of 150 kDa and 2 MDa nanodextrans 25 \nwith neutral and negative surface charge , we evaluated the system’s ability to emulate three core 26 \nphysicochemical features often implicated in biofilm -associated transport resistance: size 27 \nexclusion, charge interactions, and volume exclusion. We demonstrated a 30% statistically 28 \nsignificant decrease in partition coefficients for 2 MDa nanodextran from 150 kDa nanodextran, 29 \nconfirming the ability of the nanocellulose-based microcaps to mimic the permeability of hydrated 30 \nbiomaterial matrices. These findings reflect patterns observed in , for example,  living biofilm 31 \nstudies, where size-based diffusion hinderance is commonly reported, but charge-based interaction 32 \nand volume exclusion are more context-dependent. This controllable system can be coupled with 33 \nin silico  modeling to understand interfacial transport phenomena for nanomaterial -biomaterial 34 \ninteractions. 35 \n 36 \nKeywords: Biointerface materials, nanoparticles, antibiotic recalcitrance, transport phenomena 37 \n 38 \nGraphical Abstract 39 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n2 \n 40 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n3 \n1. Introduction 41 \nBiological interfaces, like microbial cells and biofilms, are more complex than 42 \nengineered porous membranes[1] and subsequently the transport relationships are more complex. 43 \nFor example, the liquid-surface biointerface typically contains soft-layers comprised of fats, 44 \nsugars, and proteins of various charges [2]. The interior of the biointerface exhibits heterogenous 45 \nporosity and tortuosity [3], heterogeneous mechanical compliance, heterogenous hydrophilicity, 46 \nand heterogeneous charge distribution due to the presence of cells and extracellular components 47 \n[4]. These complex structures increase the difficulty of decoupling the impacts of charge, size, 48 \ngeometry, and chemistry on the transport of novel drug and drug delivery vehicles through these 49 \nbiological interfaces. In microbial biointerfaces, like bacterial biofilms [5-8], this complexity is 50 \nincreased by the dynamic response that can occur at similar time scales to the time for total 51 \npenetration. The dynamic response from bacteria can include extracellular matrix production of 52 \npolymers like alginate, nanocellulose, proteins like flagellin [9, 10], and secretion of extracellular 53 \nDNA. These properties contribute to enhanced virulence [11] and treatment failure [12]  in 54 \ndiseases such as cystic fibrosis. Consequently, new therapeutic strategies, including the use of 55 \nnanomaterials as drug delivery vehicles, have been proposed to enhance penetration and efficacy 56 \nwithin biofilms[13-15]. 57 \nThese new therapeutic strategies will come with their eventual environmental 58 \ndissemination. While engineered nanomaterials cannot be studied as a uniform class[16], 59 \nexperimental and computational platforms that account for their fate and transport must be 60 \ncreated [17-21]. Periphyton biofilms are the primary sink for nanomaterials in estuary 61 \nenvironments [22].  The primary factors which effect ENM ecotoxicological parameters such as 62 \ntoxicity and bioaccumulation for a given ENM-biofilm-environment system can be evaluated. 63 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n4 \nWhat emerges is a set of four pairwise factors of the biofilm-ENM system of interest which 64 \nshapes bioaccumulation: size (particle/pore) [23-29], charge (surface/ECM) [24, 25, 27, 30-32], 65 \nchemistry (surface/ECM) [25, 31, 33], and hydropathy (surface/ECM) [27, 30, 34]. 66 \nPrevious studies on the impact of mechanical stress on biointerfaces have used alginate 67 \nsince it is an essential component of some biofilms and has facile preparation [35-38]. However, 68 \nalginate has 1 -2 orders of magnitude higher storage and loss moduli than natural biofilms [37]. 69 \nAdditionally, alginate has different chemical absorption properties, specifically it is less able to 70 \nabsorb divalent metallic cations than natural biofilms [37]. We have previously shown the 71 \nnanocellulose preparation used here has mechanical and certain chemical properties in range 72 \nwith natural biofilms [39]. Nanomaterial exclusion from biofilms has been proposed to be a 73 \nfunction of the various heterogeneous matrix components alginate, nanocellulose, proteins; the 74 \nmatrix size; commensal phage; bacterial cells, particularly their charge, and the channels in the 75 \nbiofilm [4, 30, 40]. As heterogeneity is difficult to achieve synthetically, this may require 76 \nsystematic study. For example, the importance of alginate on antibody binding in P. aeruginosa 77 \nbiofilms was discovered through adding alginate back alginate deficient mutants. Recent work 78 \non diffusion in alginate showed the impact of matrix cross-linking on size-exclusion [41].  79 \nWe propose the biofilm matrix and cells with their respective charges can be studied with 80 \nsynthetic systems. We synthesized microcaps from nanocellulose that are modified with divalent 81 \ncalcium ions as the biofilm matrix. We added neutral and charged microspheres to represent cells 82 \nand their respective binding affinities. We tested the ability of charged and neutral nanodextran 83 \nof different molecular weights to accumulate into the defined nanocellulose matrix.  84 \nThis study tested the ability of a microcap synthesis method to replicate the important 85 \neffects of a biofilm on species diffusion: (a) the size-effect, (b) the volume-exclusion effect, and 86 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n5 \n(c) the attachment effect. These important biofilm effects were mapped on to three hypotheses. 87 \nWe hypothesize that if the size-effect is replicated that larger diffusing species would accumulate 88 \nat lower concentrations in the matrix. We hypothesize that if the volume-exclusion effect is 89 \nreplicated, that diffusing species would accumulate at a lower concentration in a microcap with 90 \nimpermeable particles compared to one without. We hypothesize that if the attachment effect is 91 \nreplicated that diffusing species would accumulate at a higher concentration in the microcap 92 \nwhen attachment sites are present. Biofilms have distinct porosity profiles based on 93 \nenvironmental conditions [42, 43], distinct matrix components based on nutrient conditions and 94 \nexternal threat [36, 43], and distinct cell populations based on age and nutrient conditions. This 95 \nstudy and platform may aid in understanding how each component contributes to overall 96 \naccumulation. 97 \n2. Methods 98 \nThe goal of this study is to test the ability of a synthetic biomaterial to replicate the 99 \nimportant effects of a biofilm on species diffusion. We leveraged our recently developed 100 \nnanocellulose hydrogel that has closer mechanical stiffness to natural biofilms. We embedded 101 \nthe biofilm with charged and non-charged microspheres to test volume-exclusion and attachment 102 \nand charge based exclusion. We used two sizes of nanodextrans to test size exclusion. We used 103 \nstandard optical tools for examining biofilms to measure the effects of the tested parameters, 104 \nFigure 1. 105 \n2.1 Microcap Design and Nanodextran Information 106 \n To generate the microcaps, hydrogels were used as the hydrated mesh domain to achieve 107 \na size-effect. Hydrogels have been used in prior studies to simulate the extracellular matrix of the 108 \nbiofilm [41]. The hydrogel platform chosen for our system was nanocellulose hydrogels. We 109 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n6 \nrecently developed a nanocellulose that better matched the mechanical storage and loss moduli 110 \n[39], that is also easily chemically modifiable for which future studies could choose to utilize for 111 \na method of generating attachment sites independent of microspheres.  112 \nFluorescent polystyrene microspheres (Spherotech |  Lake Forest, IL) ranging from 0.7-113 \n0.9 microns in diameter were used to simulate bacteria and achieve a volume-exclusion effect. 114 \nThree distinct microspheres, each with distinct microsphere surface chemistry, were used: no 115 \nsurface modifications (Plain) (Prod# FP-0862-2), carboxylate-modified (Carboxyl) (Prod# FP-116 \n0862-2), and amine-modified (Amino) (Prod# FP-0862-2). The carboxylate and amine-modified 117 \nmicrospheres were coated with these functional groups, which are either negatively or positively 118 \ncharged at neutral pH. These charged microsphere surfaces were designed to only act as 119 \nattachment sites for the diffusing species if the diffusing species was oppositely charged. The 120 \nmicrospheres were fluorescently labelled with a vendor proprietary fluorescent tag. 121 \n Nanodextrans conjugated to FITC (fluorescein isothiocyanate) were used as the diffusing 122 \nspecies for the system (TdB Labs, Uppsala, Sweden). Two nanodextran molecular weights used 123 \nwere 150 kDa and 2,000 kDa, on the premise these molecules would have different sizes, to test 124 \nthe size-exclusion effect. The fluorescently labelled nanodextrans were chemically modified to 125 \nhave either carboxymethyl (CM) groups with negative charge (Prod# FITC-CM-dextran-150), 126 \ndiethylaminoethyl (DEAE) groups with positive charge (Product # FITC-DEAE-dextran-150), or 127 \nno modification at all (Dx) (Prod# FITC-Dx-150 and Prod# FITC-Dx-2000). The use of charged 128 \nnanodextrans and charged microsphere surfaces was designed to produce the attachment effect.  129 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n7 \n \nb \n \nc\n \nFigure 1: A. Experimental protocol for testing diffusion into hydrogel. B. Experimental Apparatus Diagram \nC. Artistic representation of microcap. Hydrogels were made as described in [39]. After centrifugation, \nmicrospheres were added. Resultant , A flow cell (b) from Biosurface Technology was seeded with hydrogel inserts \nalong the coupon recess/imaging window. Red circles represent microspheres. Microspheres with amine-coating \nrepresented. Nanodextran with no modification shown.    \n 130 \n2.2 Nanocellulose Solution Synthesis 131 \nA nanocellulose synthetic biofilm was generated as described in [39]. Briefly, 4.00 ± 0.05 132 \ng of dry cellulose powder (CAS#9004-34-6 | Sigma-Aldrich | Burlington, MA) and slowly 133 \nmixing it with an ionic liquid solution composed of 7.00 ± 0.05 g sodium hydroxide (CAS#1310-134 \n73-2 | Sigma-Aldrich | Burlington, MA), 12.00 ± 0.05 g urea (CAS#57-13-6 |  Sigma-Aldrich | 135 \nBurlington, MA), and 81.00 ± 0.05 g of reverse-osmosis filtered (RO) water and adding it to a 136 \n125 mL Erlenmeyer flask. The resulting suspension was mixed using an inert stir bar at 500 rpm 137 \nuntil it was homogenous and consistently cloudy. The suspension was then submerged in an ice-138 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n8 \nisopropyl alcohol bath with a temperature ranging between -10 and 0 OC while still being stirred. 139 \nAfter more than 1 hour of stirring and the dissolution of the cellulose, the liquid was separated 140 \ninto two 50 mL centrifuge tubes. The tubes were centrifuged at 10,000 RPM (rcf=13,751) for 10 141 \nminutes while maintaining a temperature of 22 OC. After centrifugation, the supernatant was 142 \ndecanted and saved as nanocellulose solution, with any settled solids discarded.  143 \n2.3 Nanocellulose-Microsphere Solutions 144 \nFor each batch of nanocellulose solution made, four different nanocellulose-microsphere 145 \nsolutions were made: one with Plain microspheres, one with Carboxyl microspheres, one with 146 \nAmino-microspheres, and one with no microspheres. These nanocellulose-microsphere solutions 147 \nwere prepared by mixing 5.0±0.1 mL of nanocellulose solution with 50±0.1 µL of 1wt% 148 \nmicrosphere solution in the well of a 6-well culture plate. Once all four nanocellulose-149 \nmicrosphere solutions were prepared for each nanocellulose batch on the same plate, the plate 150 \nwas covered, wrapped in aluminum foil, and mixed for 24 hours on a shaking plate at 250 RPM 151 \nat room temperature ranging from 22-26 ºC. After 24 hours of mixing, the resulting 152 \nnanocellulose-microsphere solutions were stored at 4 ºC.  153 \n2.4 Microcap and Flow Cell Preparation 154 \nWith the nanocellulose-microsphere solutions prepared, a 5 mL syringe affixed with a 155 \n27G nozzle was used to fill the nozzle with nanocellulose-microsphere solutions. The 156 \nnanocellulose solution was then extruded through the nozzle using the syringe. Once a small 157 \namount of nanocellulose solution was extruded, priming the nozzle, the nozzle would then “leak” 158 \nthe solution semi continuously. The nozzle tip was then repeatedly (between 10-30 times) 159 \npressed onto the surface of a polycarbonate coupon, with a microscopic amount of hydrogel 160 \nbeing left behind on the coupon as residue with each press. The coupon then sat at room 161 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n9 \ntemperature in a plastic petri dish for greater than 48 hours to dry with no plastic cover but 162 \naluminum foil cover, which prevented contamination and fluorescent overexposure, but allowed 163 \nfor air flow. This let the hydrogel residues to dry onto the coupon and form microcaps. Before an 164 \nexperiment, the microcap covered coupon was loaded into a flow cell (Prod# FC 275-AL | 165 \nBioSurface Technologies | Bozeman, MT, USA), where the flow cell chamber containing the 166 \ncoupon was filled with reverse-osmosis water first to rehydrate the microcap. Almost 167 \nimmediately after, the chamber was then filled with a 5 µM solution of Calcofluor White (CW) 168 \nstain dissolved in water (Cat# 29067 | Biotium | Fremont, CA, USA). The fluorophore would 169 \nthen slowly diffuse into the microcaps, allowing for fluorescent imaging of the hydrogel domain 170 \nof the microcap.  171 \n2.5 Nanodextran Exposure and Imaging Procedure 172 \nOnce the flow cell was loaded with coupons covered with CW-stained microcaps, the 173 \nflow cell was then connected to a microfluidic apparatus with two inlets: reverse-osmosis filtered 174 \n(RO) water and 30 mg/L fluorescently labelled-nanodextran dissolved in Dulbecco’s phosphate 175 \nbuffer solution (PBS) (Figure 1). The flow cell was placed on a confocal laser scanning 176 \nmicroscope, which allowed for real-time 3-D fluorescent imaging of the microcaps.   177 \nEach experiment involved a two-phase procedure: characterization and accumulation. To 178 \nperform an experiment, the flow chamber was flushed with 1 mL/min RO water during the 179 \ncharacterization phase. While the flow chamber containing the microcap-covered coupon was 180 \nflushed with 1 mL/min RO water, the microcaps on the coupon were checked to determine if 181 \nthere were three microcaps less than 250 microns in diameter (all tests had at least three). A z-182 \nstack, or a series of microscope images taken at close, regularly spaced focal lengths to 183 \nreconstruct a 3-D image of a sample, was taken of three microcaps on each coupon (with the 184 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n10 \nthree smallest microcaps usually chosen). Field of view size was adjusted for each microcap to 185 \ntake up the entire image. Nyquist sampling was used to select z-section thicknesses and image 186 \nresolution. These z-stacks were set up to image three features in each microcap: the hydrogel 187 \nmesh by imaging the CW stain (Laser: 405 nm/Detector Range: 410-546 nm), the impermeable 188 \nmicrospheres (Laser: 561 nm/Detector Range: 585-700 nm), and the presence of FITC (Laser: 189 \n488 nm/Detector Range: 410-546 nm). In addition to measuring the presence of FITC, this same 190 \nsignal was also used to determine the location of the coupon within the image, as the coupon 191 \nreflected green light at its surface. This first image before any FITC was added to the system will 192 \nbe referred to as the characterization image.  193 \nAfter the characterization image of each of the microcap of interest, the microcaps were 194 \nexposed to a continuous stream of nanodextran during the accumulation phase of the experiment. 195 \nThis was done by flowing 1 mL/min of the 30 mg/L nanodextran dissolved in PBS into the flow 196 \nchamber continuously for 24 minutes. After this continuous exposure to a constant concentration 197 \nof nanodextran, the flow in the chamber was stopped, and the microcaps re-imaged using the 198 \nsame parameters as those for the characterization image, producing an accumulation image for 199 \neach microcap. This was done to quantify two things: the concentration of nanodextran (via the 200 \nproxy measurement of FITC-signal) in the water immediately surrounding each microcap, and 201 \nthe concentration of nanodextran within each microcap.  202 \nDuring accumulation phase of each experiment, time-lapse imaging of a single-z-plane 203 \nwithin one of the imaged microcaps was taken to monitor if the system reached equilibrium 204 \nduring each phase (which all were confirmed to reach).  205 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n11 \n2.6 Experimental Design 206 \nTo test the functional properties of these microcaps, three predicted experimental effects 207 \nwere tested to determine the capacity of these microcaps to replicate the biofilm features of 208 \ninterest, Figure 2. The size-exclusion effect was tested to determine if a hydrated mesh was 209 \nformed. The volume-exclusion effect was tested to determine if the embedded microspheres 210 \nwere diffusing species impermeable. The attachment effect was tested to determine if the 211 \ndiffusing species were immobilized via opposite charge interactions with the surface of the 212 \nmicrospheres, indicating diffusing species attachment sites.   213 \na  \nb  \nc \n  \n \nFigure 2: (a-c)Diagrams of hypothesized size-exclusion effect, charge based attachment effects, \nand volume exclusion effect. d-e Hypothesized outcomes of nanodextran concentration in \nhydrogels based on the given effect. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n12 \n 214 \nTo test for the size-exclusion effect, the two different sized nanodextrans were used: 150 215 \nkDa and 2,000 kDa. The 2,000 kDa nanodextran was expected to have a larger particle size than 216 \nthe 150 kDa case. This was tested and confirmed using Nanoparticle Tracking Analysis and 217 \nDynamic Light-Scattering measurements, shown in Table 1. While the measured sizes were 218 \nsimilar sizes via NTA and DLS, we used them as it well known that while designed for specific 219 \nsizes, environmental nanomaterials are never pristine. These nanodextan molecular weights were 220 \nchosen because they have been used in biofilm penetration studies as both small and large sized 221 \nnanomaterials [23, 25, 44, 45]. Additional details on nanoparticle characterization are provided 222 \nin the Supplementary Information.   223 \nTable 1: Nanodextran Characterization. All nanodextrans dispersed in Dulbecco’s Phosphate Buffer \nSolution (PBS). NTA error reported as standard error from measured particle size distribution (n=5). *- \nStandard error not reported on instrument. FITC-Dx-2000 size measured twice on NTA, both sizes reported. \nDLS/ELS error reported as standard deviation of three triplicate measurements. \nNanodextran Nanodextran Molecular \nWeight [kDa] \nSize via NTA \n[nm] \nSize via DLS \n[nm] \nCharge via ELS \n[mV] \nFITC-Dx-150 150 226 ± 14 115 ± 24 -1.2 ± 0.4 \nFITC-DEAE-150 150 178 ± 5 107 ± 16 -0.4 ± 0.4 \nFITC-CM-150 150 173* 771 ± 1000 -5.7 ± 3.2 \nFITC-Dx-2000 2000 297 ± 18 \n306 ± 24 \n81 ± 0.1 -2.4 ± 0.7 \nSince hydrated meshes such as a biofilm extracellular matrix and a hydrogel are 224 \ncomprised of disorganized, overlapping biopolymer chains, they can allow particles to diffuse 225 \nthrough them, but only up to a certain size. As the particle reaches a characteristic “mesh size”, 226 \nless of the volume within the hydrated mesh network is available for it to occupy.  Once the 227 \nparticle reaches a critical size, no amount of particle can accumulate within the matrix. Thus, as 228 \nparticle size increases, the equilibrium concentration it reaches within a hydrated matrix 229 \ndecrease, as less volume is available for the particle to permeate into. Hence, a decrease in 230 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n13 \ndiffusing species concentration with particle size within the matrix at equilibrium shows the 231 \npresence of a size-exclusion effect in the microcap, which in our system is likely indicative of 232 \nthe formation of a hydrogel. Microcaps prepared using identical conditions were used in size-233 \nexclusion experiments.   234 \nTo test for the volume-exclusion effect, hydrogels were prepared either with 235 \nmicrospheres within them or without microspheres. The microcaps without microspheres within 236 \nthem would be expected to reach a higher concentration of nanodextran if the volume-exclusion 237 \neffect was observed. The reasoning for this is similar to the reasoning for the size-exclusion 238 \neffect: certain portions of the microcap volume, specifically the volume occupied by the 239 \nmicrospheres, would be unavailable for the nanodextran to occupy, leading to a lower 240 \nconcentration overall in the microcap. Identical nanodextrans were used in volume-exclusion 241 \nexperiments.  242 \nTo test for the attachment effect, hydrogels embedded with three different microspheres 243 \nwere prepared and tested against three different nanodextrans. The different microspheres tested 244 \nwere no surface functionalization (Plain, neutrally charged surface), amine-functionalization 245 \n(Amino, positively charged surface), and carboxyl-functionalized (Carboxyl, negatively charged 246 \nsurface). The different nanodextrans tested were no chemical modification (None, neutrally 247 \ncharged), carboxymethyl-modified (CM, negatively charged) and diethylaminoethyl-modified 248 \n(DEAE, positively charged). Since we only expect oppositely charged combinations to lead to 249 \nattachment between microsphere surface and nanodextran, we expect to see significant increases 250 \nin nanodextran concentration within the microsphere only for the combination of CM-modified 251 \nnanodextran accumulating in microcaps with amine-modified microspheres, and DEAE-252 \nmodified nanodextran accumulating in microcaps with carboxyl-modified microspheres. Two 253 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n14 \ndifferent control cases are considered for each of these combinations: (a) charged nanodextran 254 \nand uncharged microspheres, and b) uncharged nanodextran and charged microspheres. 255 \nDecreases in nanodextran concentrations within the microcap would be expected in both cases 256 \ncompared to the oppositely charged nanodextran/microspheres case.  257 \n2.7 Image Analysis 258 \nTo determine nanodextran concentration for each z-stack image taken, each pixel was 259 \nassigned to be in one of four spatial domains in the image: (1) the water domain (liquid domain 260 \nΩ!), (2) the hydrogel domain (interstitial domain Ω\") (3) the microsphere domain (Ω#), and (4) 261 \nthe coupon domain (solid domain Ω$).  262 \n \nFigure 3: Labelled Microcap from Characterization Image.  Red channel: fluorescent polystyrene \nmicrospheres (carboxyl-modified). Blue channel: CW-stained nanocellulose hydrogel. Green-channel: \npolycarbonate coupon-microcap interface.  \n 263 \nThis was done by segmenting the hydrogel using the CW signal, the microspheres using 264 \nthe AF594 signal, and the coupon using the FITC-signal in each image, and assuming all 265 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n15 \nremaining pixels were within the water domain. The total microcap domain was the union of the 266 \nmicrocap and microsphere domains.  267 \nThe microspheres were segmented using an Otsu threshold on the AF594 signal in 268 \nMATLAB. The hydrogel domain was segmented in ImageJ using a Otsu threshold, followed by 269 \na region-filling algorithm (CW-stain only penetrated ~10 microns into hydrogel), followed by an 270 \nalgorithm for discarding very small filled regions. The coupon was segmented using an edge-271 \ndetection algorithm along the z-direction.  An important note on this process was that coupon 272 \nsegmentation was not perfect. The coupon segmentation algorithm was designed to favor a pixel 273 \nas coupon as opposed to water or hydrogel on purpose, since the accuracy of the water and 274 \nhydrogel signal was more important than the coupon signal. 275 \nWith each pixel assigned to a domain, the average value of the nanodextran/FITC signal 276 \nin each domain in the microcap was quantified. Rather than use average FITC-signal in the 277 \nmicrocap, [𝐹𝐼𝑇𝐶]%&'()'*+ , directly, the ratio of the FITC-signal in the microcap to the FITC-278 \nsignal in the water, [𝐹𝐼𝑇𝐶],*-.( , or the nanodextran-microcap partition coefficient, 𝐾/, was 279 \ncalculated for each microcap’s accumulation image, Equation 1. This ensured any variation in 280 \nFITC concentration and signal within the flow cell was normalized between microcaps.  281 \n𝐾/ = [𝐹𝐼𝑇𝐶]%&'()'*+\n[𝐹𝐼𝑇𝐶],*-.(\n Equation 1: Microcap Partition Coefficient \nThe MATLB code for image analysis is included as Supplementary Information and is 282 \navailable via Github.  283 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n16 \n2.8 Statistical Analysis  284 \n \nFigure 4: Design of Experiments and Replicates \nFor each nanodextran-hydrogel combination tested, two types of replicates were used, all 285 \non the hydrogel side: batch replicates and microcap replicates. Three batches of each hydrogel 286 \ntested (no microspheres, plain microspheres, carboxyl-microspheres and amine-microspheres) 287 \nwere generated, leading to 12 different hydrogels with three batch replicates each. For each 288 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n17 \nbatch, one coupon covered with microcaps was generated for each nanodextran tested, leading to 289 \n33 different coupons. On each coupon, three microcap replicates were tested on each. This led to 290 \na total of 99 microcaps tested.  291 \nTo compile the data for each experimental condition (nanodextran-microcap 292 \ncombination), the nanodextran-microcap partition coefficient for each microcap tested was first 293 \nfound (see Image Analysis for how this was calculated).  This value was then averaged for the 294 \nthree microcaps on each coupon tested, as these replicates were determined to be technical 295 \nreplicates, which warrant averaging together for the purposes of statistical analysis. This left 296 \nthree nanodextran-microcap partition coefficients for each experimental condition, each one 297 \nrepresenting a different hydrogel batch. These three nanodextran-microcap partition coefficients 298 \ncomprised the sample for each experimental condition.  299 \nFor each set of experimental conditions compared for each hypothesis, the three 300 \nnanodextran-microcap partition coefficients for each batch were compared to each other using 301 \nWelch’s one-way t-tests using an alpha of 0.05. Welch’s test was used since variances were not 302 \nassumed to be equal. One-way tests were used since only an effect in one direction would 303 \nconstitute evidence for each hypothesis. All samples passed the Shapiro-Wilks test for normality.  304 \n3. Results 305 \nThe microcaps generated using this synthesis procedure were tested for their ability to 306 \nreplicate the core features of a synthetic biofilm. These included size-exclusion effects, volume 307 \nexclusion effects and attachment effects. Different combinations of microcaps and diffusing 308 \nnanodextran were used to test for these effects through equilibrium nanodextran partition 309 \ncoefficients in the microcaps.  310 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n18 \nTesting for Hydrated Mesh Property via the Size Effect  311 \nTo test for the size-exclusion effect, nanodextran-microcap partition coefficients for 312 \nmicrocaps with plain microspheres were compared between a 150 kDa nanodextran and a 2,000 313 \nkDa nanodextran. The results, as shown in Figure 5, show a statistically significant size-314 \nexclusion effect in the hydrogel mesh was observed.  315 \n 316 \n \nFigure 5: Size-Exclusion Effect Test for Verifying the \nHydrogel Mesh Property of the Synthetic Biofilm \nSystem.  Error bars represent 95% confidence intervals \non the sample. P-value represents results from one-tailed \nWelch’s t-test. Experiments used 4 wt% hydrogels loaded \nwith 0.01 wt% plain polystyrene microspheres with \nneutrally charged nanodextrans. Partition coefficient \nrepresents nanodextran-microcap partition coefficient.  \n \n 317 \nTesting for Impermeable Sub-volume via the Volume-Exclusion Effect 318 \nTo test for the volume-exclusion effect, nanodextran-microcap partition coefficient for 4 319 \nwt% hydrogel microcaps embedded with either no microspheres or 0.01 wt% plain microspheres 320 \nwere compared using a neutrally charged, 150 kDa nanodextran. The results, as shown in Figure 321 \n6, show no volume-exclusion effect due to the presence of the microspheres was observed.  322 \n 323 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n19 \n \nFigure 6: Volume-Exclusion Effect Test for \nVerifying the Impermeable Subdomain Property of \nthe Synthetic Biofilm System.  Error bars represent \n95% confidence intervals on the sample. P-value \nrepresents results from one-tailed Welch’s t-test. \nExperiments used 4 wt% hydrogels loaded with \nvariable wt% plain polystyrene microspheres and \nneutrally charged, 150 kDa nanodextran. Partition \ncoefficient represents nanodextran-microcap partition \ncoefficient. \n \n 324 \nTesting for Attachment Sites via the Attachment Effect 325 \nTo test for the attachment effect, 4 wt% hydrogels embedded with 0.01 wt% neutrally, 326 \nnegatively, and positively charged microspheres were tested against neutrally, negatively, and 327 \npositively charged 150 kDa nanodextrans, shown in Figure 7. Opposite charge combinations 328 \nwere tested for increases in nanodextran-microcap partition coefficients compared to both the 329 \nsame charged nanodextran and uncharged microspheres and the uncharged nanodextran and 330 \nsame charged microspheres. No statistically significant attachment effect was observed across all 331 \ncomparisons. A non-statistically significant attachment effect was observed for the microcaps 332 \nembedded with amine-modified microspheres with a carboxymethyl-modified nanodextran for 333 \ncomparisons to both control cases. 334 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n20 \n \nFigure 7: Attachment Effect Test for Verifying Presence of Attachment Sites in the Synthetic Biofilm \nSystem.  Error bars represent 95% confidence intervals on the sample. P-values represent results from one-tailed \nWelch’s t-test. Experiments used 4 wt% hydrogels loaded with variable 0.01 wt% plain-, carboxyl- and amine-\nmodified polystyrene microspheres and neutrally, positively or negatively charged, 150 kDa nanodextrans. \nPartition coefficient represents nanodextran-microcap partition coefficient. \n4. Discussion 335 \nThis research sought to create a synthetic biofilm system for use studying species 336 \ntransport in biofilms. To accomplish this, nanocellulose hydrogel microcaps loaded with 337 \npolystyrene microspheres were developed to proxy bacterial biofilms in the features of a 338 \nhydrated mesh, an impermeable subdomain, and presence of attachment sites. The replication of 339 \nthese features in the generated microcaps were tested for by quantifying an important 340 \nexperimental effect for each feature: the size-exclusion effect for verifying the hydrogel mesh, 341 \nthe volume-exclusion effect for verifying the impermeable subdomain, the attachment effect for 342 \nverifying presence of attachment sites.  343 \nExperimental results indicated that the size-exclusion effect was verified, and therefore 344 \nthe hydrogel mesh feature of the microcaps. However, the attachment effect and the volume 345 \nexclusion effect were not observed. This indicates that the microcaps as currently designed are 346 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n21 \nunable to replicate the key biofilm properties needed to perform transport studies of an 347 \nimpermeable subdomain and presence of attachment sites.  348 \nSize-effect agrees with prior literature on chemical species diffusion in single bacterial 349 \nspecies biofilms [23, 46]. However, evidence of this effect is limited in multispecies biofilm 350 \nliterature on nanoparticles [47]. This is most likely due to effects of agglomeration, dissolution 351 \nand changes to the nanoparticle biopolymer corona in multispecies biofilms studies having 352 \nstrong effects on effective nanoparticle size during diffusion [48, 49]. This shows the need for 353 \nstudies with more controls over nanoparticle stability in multispecies biofilm transport studies.  354 \nA few distinct hypotheses could explain why the presence of microspheres and the use of 355 \nchemically modified microsphere surfaces and nanodextrans did not elicit the effects of volume-356 \nexclusion and attachment. One is the use of a 0.01 wt% microsphere percentage within the 357 \nhydrogels was too low to observe the anticipated effects. If little volume is made impermeable 358 \ndue to low concentration of microspheres, then the volume-exclusion effect would be anticipated 359 \nto be small and below any statistically significant threshold. In addition, too low of a 360 \nconcentration of microspheres would mean there are very few sites for nanodextran attachment, 361 \nwhich would also predict a small attachment effect again below any statistically significant 362 \nthreshold.  Another possibility is that nanodextran can penetrate and diffuse into the polystyrene 363 \nmicrospheres. While there is evidence for adsorption onto the surface of polystyrene [50], there 364 \nis less evidence for penetration and diffusion into polystyrene microsphere. However, no 365 \nexperimental evidence is offered in this research proving or disproving this hypothesis. 366 \nFor the attachment effect, another possibility is that the charged microsphere surfaces do 367 \nnot act as attachment and immobilization sites for its oppositely charged nanodextran. This could 368 \nbe due to a variety of reasons including the charge-charge interactions not being strong enough 369 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n22 \nto immobilize the nanodextran, the microsphere or the nanodextran not being charged species as 370 \nanticipated, or even that the immobilization of the nanodextran does not lead to higher levels of 371 \nnanodextran to accumulate into the hydrogel to balance the chemical potential gradient. While it 372 \nis possible attachment does occur in these systems but too slowly to observe in this experiment, 373 \ndirect charge-charge interactions are the strongest intermolecular forces in chemistry, which 374 \nwould make for quick attachment, making this possibility seem unlikely. This lack of a clear 375 \neffect of charge is common in the scientific literature on diffusion of chemical species in 376 \nhydrogel matrices such as biofilms [41, 44, 45, 51, 52]. Future experiments on this system could 377 \ninclude testing the accumulation and desorption of ions instead of nanomaterials as these have 378 \nbeen shown to matter in the biofilm matrix [53]. 379 \nAssuming the hypothesis that the non-observed effects were primarily due to a low 380 \nconcentration of microspheres within the microcaps, further experimental work could replicate 381 \nthese experiments with higher concentrations of microspheres and look for the same effects on 382 \nattachment and volume-exclusion. Further evidence for the size-exclusion effect could be found 383 \nby comparing current results to higher weight percentage nanocellulose hydrogels, which 384 \ntheoretically lead to smaller critical mesh size, which would be anticipated to cause lower 385 \nnanodextran concentrations in these higher weight percentage microcaps [23, 54].   386 \nThe data presented here does not lend itself to diffusivity calculations common in the 387 \nliterature. However, Bryers and Frummond showed lumped parameter diffusivity calculations 388 \nare inaccurate for describing transport in biofilm [55]. The data can be compared to 389 \nbioconcentration factors (BCFs) which have been widely reported in ecotoxicological studies of 390 \nengineered nanoparticles effects on environmental biofilms. Since estimates for BCFs in 391 \necotoxicological studies are usually only order of magnitude estimates, logarithmic BCF values 392 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n23 \n(pBCF) will be used here to compare BCFs between different experiments [56-58]. pBCFs will 393 \nbe calculated from dimensionless BCFs on a mass-by-mass basis (µg ENM/g biofilm divided by 394 \nµg ENM/g water) using the wet weight of the periphytic biofilms when possible. Since 395 \nperiphytic biofilms are hydrated in the environment, wet weights are used to give a more 396 \nintuitive sense of the extent of the increase in concentration of ENM that would be seen in an 397 \nenvironmental system. If only dry-weight/AFDM BCFs are reported, wet-weight BCFs will be 398 \nestimated by assuming a 100:1 ratio of biofilm wet mass to biofilm dry mass. Finally, many 399 \nstudies for the bioaccumulation of ENM quantify total metal uptake, not ENM specific uptake. 400 \nUnless controls for dissolution or specific quantification of ENM is stated, all BCFs reported 401 \nhere will reflect this, limiting their representation of \"true\" ENM BCFs. Since most studies do 402 \nnot report detailed kinetic analysis of ENM bioaccumulation, characteristic times, 𝑡0, of 403 \nbioaccumulation will be estimated when possible. Estimations of these are mostly qualitative, 404 \nrepresenting approximately the time necessary to reach concentrations 50% of equilibrium 405 \nconcentrations.  406 \nThe data in this study are 2 – 10 times less than the pBCF reported in literature, Table 2. 407 \nWhile this hydrogel matches mechanical properties better than alginate, many biofilms contain 408 \nboth alginate and nanocellulose polymers. Lastly, biofilms are much more than cells or beads 409 \ndistributed in a hydrogel mesh, they also contain channels filled with pore water [55]. Equation 2 410 \nshows that if the amount of nanoparticles in the pore water is in equilibrium with the surrounding 411 \nwater, the BCF in a porous biofilm will be less than an equivalent amount of nanomaterials being 412 \nadded to a fixed amount of biofilm mass without. Equivalently, the pBCF of a biofilm with pores 413 \nwill be greater than one without for the same amount of substance added to the biofilm, by a 414 \nfactor of log2 for biofilm of equal weight to water. 415 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n24 \n𝑚!\"#,% \t + \t 𝑚!\"#\n𝑚% \t + \t 𝑚&\n𝑚!\"#/𝑚&\n=\n𝑚!\"#,% \t + \t 𝑚!\"#\n𝑚% \t + \t 𝑚&\n𝑚!\"#/𝑚&\n=\n(𝑚!\"#,% \t + \t 𝑚!\"#)/𝑚%\n1 + \t 𝑚&/𝑚%\n𝑚!\"#/𝑚&\n<\n𝑚!\"#,% + 𝑚!\"#\n𝑚%\n𝑚!\"#/𝑚&\n Equation 2 \n 416 \n 417 \nTable 2: Summary of literature on bioaccumulation and bioconcentration kinetic parameters of ENMs in \nperiphytic biofilms. n.d. no data reported. *Some hydropathies are inferred. Titanium dioxide hydropathy \ndepends on crystal phase which also modifies its toxicity [59]. \nType Biofilm Size (nm) Hydropathy Charge tC pBCF Study \nLatex \ncarboxylate \nWetland 100 Hydrophilic (-) < 24 h ~2 [60] \nGold Estuary 65 Hydrophobic (+) < 12 d 2.18 [61] \nGold polystyrene \nsulfonate \nEstuary 50 Hydrophobic* -53 mV ~12 h 3.46 [22] \nCitrate capped \nsilver \nEstuary 30, 115 Hydrophobic (-) ~12 h 1.4 – 1.5 [62] \nCitrate capped \nsilver \nMarine 100 – 700 Hydrophobic -18 mV < 24 h 2.05 [47] \nPVP capped \nsilver \nBenthic 117 Hydrophilic -17 mV >10 min , < 4 d 4.6 [63] \nTitanium dioxide Algal 150 n.d.* -17 mV < 28 d 4 - 5.6 [64] \nCopper oxide Periphyton 100 – 800 Hydrophilic* -35 – +15 mV ~50 min 3.6 – 4.25 [65] \nCitrate capped \nsilver \nAquabacteri\num \ncitratiphilum \n30, 70 Hydrophobic -60 mV < 20 h 0.8 [66] \nFITC-Dx-150 Hydrogel 92 – 240 Hydrophilic -1.2 ± 0.4 mV 24 min 0.153  \nFTIC-Dx-2000 Hydrogel + \nneutral \nspheres \n81 – 330 Hydrophilic -2.4 ± 0.7 mV 24 min 0.319  \nFITC-Dx-150 Hydrogel + \nneutral \nspheres \n92 – 240 Hydrophilic -1.2 ± 0.4 mV 24 min 0.164  \nFITC-DEAE-150 Hydrogel + \nneutral \nspheres \n91 – 183 Hydrophilic -0.4 ± 0.4 mV 24 min 0.132  \nFITC-CM-150 Hydrogel + \nneutral \nspheres \n173 – 1000 Hydrophilic -5.7 ± 3.2 mV 24 min 0.160  \nFITC-Dx-150 Hydrogel + \ncarboxyl (-) \nspheres \n92 – 240 Hydrophilic -1.2 ± 0.4 mV 24 min 0.145  \nFITC-DEAE-150 Hydrogel + \ncarboxyl (-) \nspheres \n91 – 183 Hydrophilic -0.4 ± 0.4 mV 24 min 0.179 \n \n \nFITC-CM-150 Hydrogel + \ncarboxyl (-) \nspheres \n173 – 1000 Hydrophilic -5.7 ± 3.2 mV 24 min 0.135 \n \n \nFITC-Dx-150 Hydrogel + \namino (+) \nspheres \n92 – 240 Hydrophilic -1.2 ± 0.4 mV 24 min 0.125 \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n25 \nFITC-DEAE-150 Hydrogel + \namino (+) \nspheres \n91 – 183 Hydrophilic -0.4 ± 0.4 mV 24 min 0.158  \nFITC-CM-150 Hydrogel + \namino (+) \nspheres \n173 – 1000 Hydrophilic -5.7 ± 3.2 mV 24 min 0.068  \n 418 \nIn summary, a systematic analysis through a synthetic biofilm model adds to the toolkit 419 \nof biointerface studies. Size exclusion has been replicated in alginate [41] and now in 420 \nnanocellulose. While attachment and volume-exclusion were not replicated, future work using 421 \nsimilar matrices should also consider concentration of microspheres within the nanocellulose 422 \nmatrix. For example, there may be a critical concentration of microspheres or bacterial cells 423 \nnecessary for some of the absorption effects. While these can be studied with non-toxic 424 \nnanomaterials, these affects are often very difficult to uncouple in living systems [67]. 425 \n5. Conclusion 426 \nThis study aimed to emulate key physicochemical barriers to diffusion found in natural 427 \nbiofilms using tunable synthetic microcap biofilm matrix system. Through the controlled 428 \nexposure of nanodextrans with varying size and surface charge, we evaluated the system’s ability 429 \nto emulate three core physicochemical features often implicated in biofilm-associated transport 430 \nresistance: size exclusion, charge interactions, and volume exclusion. The results demonstrated a 431 \nstatistically significant size-exclusion effect, confirming the ability of the nanocellulose-based 432 \nmicrocaps to mimic the selective permeability of hydrated biofilm matrices. However, the 433 \ndesigned system did not display statistically significant volume-exclusion or attachment effects, 434 \nsuggesting that the current microsphere concentration and charge configurations were 435 \ninsufficient to replicate these additional features. 436 \nThese findings reflect patterns observed in natural biofilms studies, where size-based 437 \ndiffusion hinderance is commonly reported, but charge-based interaction and volume exclusion 438 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n26 \nare more context-dependent. The absence of strong attachment or volume exclusion effects may 439 \nbe due to insufficient microsphere loading, incomplete charge immobilization, or the dynamic 440 \nbehavior of the nanodextrans in the hydrated mesh.  441 \n Future studies should explore increased microsphere loading, use of covalently bound 442 \nattachment sites or incorporation of more biologically relevant surface chemistries to better 443 \nrecapitulate these additional transport-limiting features. Ultimately, refining this synthetic 444 \nbiofilm platform will enhance its utility in advancing the ecotoxicology of engineered 445 \nnanomaterials. 446 \n 447 \nAcknowledgements 448 \n 449 \n• DT was supported in part by the National Science Foundation under Grant No. DGE-450 \n2022040.  451 \n• Research reported in this publication was supported by the National Institute of General 452 \nMedical Sciences of the National Institutes of Health under Award Number NIH R35 453 \nGM142898 The content is solely the responsibility of the authors and does not 454 \nnecessarily represent the official views of the National Institutes of Health. 455 \n 456 \nReferences 457 \n 458 \n1. Vattulainen, I. and O.G. Mouritsen, Diffusion in Membranes, in Diffusion in Condensed 459 \nMatter, P. Heitjans and J. Kärger, Editors. 2005, Springer: Berlin. 460 \n2. Dingari, N.N. and C.R. Buie, Theoretical investigation of bacteria polarizability under direct 461 \ncurrent electric fields. Langmuir, 2014. 30(15): p. 4375–84. 462 \n3. Van Wey, A.S., et al., Anisotropic nutrient transport in three-dimensional single species 463 \nbacterial biofilms. Biotechnol Bioeng, 2012. 109(5): p. 1280–92. 464 \n4. Prince, J. and A.D. Jones, 3rd, Heterogenous biofilm mass-transport model replicates periphery 465 \nsequestration of antibiotics in Pseudomonas aeruginosa PAO1 microcolonies. Proc Natl Acad 466 \nSci U S A, 2023. 120(47): p. e2312995120. 467 \n5. Flemming, H.C. and J. Wingender, The biofilm matrix. Nat Rev Microbiol, 2010. 8(9): p. 468 \n623–33. 469 \n6. Miller, M.B. and B.L. Bassler, Quorum sensing in bacteria. Annu Rev Microbiol, 2001. 55: p. 470 \n165–99. 471 \n7. Xu, K.D., et al., Spatial physiological heterogeneity in Pseudomonas aeruginosa biofilm is 472 \ndetermined by oxygen availability. Appl Environ Microbiol, 1998. 64(10): p. 4035–9. 473 \n8. Martin, M., et al., Laboratory Evolution of Microbial Interactions in Bacterial Biofilms. J 474 \nBacteriol, 2016. 198(19): p. 2564–71. 475 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n27 \n9. Tremblay, P.L., et al., A genetic system for Geobacter metallireducens: role of the flagellin and 476 \npilin in the reduction of Fe(III) oxide. Environ Microbiol Rep, 2012. 4(1): p. 82–8. 477 \n10. Hochvaldova, L., et al., E. coli and S. aureus resist silver nanoparticles via an identical 478 \nmechanism, but through different pathways. Commun Biol, 2024. 7(1): p. 1552. 479 \n11. Secor, P.R., et al., Staphylococcus aureus Biofilm and Planktonic cultures differentially impact 480 \ngene expression, mapk phosphorylation, and cytokine production in human keratinocytes. BMC 481 \nMicrobiol, 2011. 11: p. 143. 482 \n12. Ciofu, O., et al., Tolerance and resistance of microbial biofilms. Nat Rev Microbiol, 2022. 483 \n20(10): p. 621–635. 484 \n13. Campos, J.V., et al., Advancing Nanotechnology: Targeting Biofilm-Forming Bacteria with 485 \nAntimicrobial Peptides. BME Front, 2025. 6: p. 0104. 486 \n14. Mohanta, Y.K., et al., Nanotechnology in combating biofilm: A smart and promising 487 \ntherapeutic strategy. Front Microbiol, 2022. 13: p. 1028086. 488 \n15. Choi, V., et al., Drug delivery strategies for antibiofilm therapy. Nat Rev Microbiol, 2023. 489 \n21(9): p. 555–572. 490 \n16. Klaper, R.D., The Known and Unknown about the Environmental Safety of Nanomaterials in 491 \nCommerce. Small, 2020. 16(36). 492 \n17. Suhendra, E., et al., A Review on the Environmental Fate Models for Predicting the 493 \nDistribution of Engineered Nanomaterials in Surface Waters. International Journal of 494 \nMolecular Sciences, 2020. 21(12). 495 \n18. Patil, S.S. and U.M. Lekhak, Toxic effects of engineered carbon nanoparticles on environment. 496 \nCarbon Nanomaterials for Agri-Food and Environmental Applications, 2020: p. 237–260. 497 \n19. Nguyen, M.K., J.Y. Moon, and Y.C. Lee, Microalgal ecotoxicity of nanoparticles: An updated 498 \nreview. Ecotoxicology and Environmental Safety, 2020. 201. 499 \n20. Lekamge, S., et al., The Toxicity of Nanoparticles to Organisms in Freshwater. Reviews of 500 \nEnvironmental Contamination and Toxicology, Vol 248, 2020. 248: p. 1–80. 501 \n21. Salieri, B., et al., Fate modelling of nanoparticle releases in LCA: An integrative approach 502 \ntowards \"USEtox4Nano\". Journal of Cleaner Production, 2019. 206: p. 701–712. 503 \n22. Burns, J.M., et al., Surface charge controls the fate of Au nanorods in saline estuaries. 504 \nEnvironmental Science & Technology, 2013. 47(22): p. 12844–12851. 505 \n23. Peulen, T.O. and K.J. Wilkinson, Diffusion of Nanoparticles in a Biofilm. Environmental 506 \nScience & Technology, 2011. 45(8): p. 3367–3373. 507 \n24. Morrow, J.B., C. Arango, and R.D. Holbrook, Association of Quantum Dot Nanoparticles 508 \nwith  509 \n Biofilm. Journal of Environmental Quality, 2010. 39(6): p. 1934–1941. 510 \n25. Guiot, E., et al., Heterogeneity of diffusion inside microbial biofilms determined by fluorescence 511 \ncorrelation spectroscopy under two-photon excitation. Photochemistry and Photobiology, 512 \n2002. 75(6): p. 570–578. 513 \n26. Tripathi, S., D. Champagne, and N. Tufenkji, Transport Behavior of Selected Nanoparticles 514 \nwith different Surface Coatings in Granular Porous Media coated with  515 \n Biofilm. Environmental Science & Technology, 2012. 46(13): p. 6942–6949. 516 \n27. Desmau, M., et al., Dynamics of silver nanoparticles at the solution/biofilm/mineral interface. 517 \nEnvironmental Science-Nano, 2018. 5(10): p. 2394–2405. 518 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n28 \n28. Sahle-Demessie, E. and H. Tadesse, Kinetics and equilibrium adsorption of nano-TiO 519 \n particles on synthetic biofilm. Surface Science, 2011. 605(13-14): p. 1177–1184. 520 \n29. Thuptimdang, P., T. Limpiyakorn, and E. Khan, Dependence of toxicity of silver 521 \nnanoparticles on  522 \n biofilm structure. Chemosphere, 2017. 188: p. 199–207. 523 \n30. Li, X.N., et al., Control of nanoparticle penetration into biofilms through surface design. 524 \nChemical Communications, 2015. 51(2): p. 282–285. 525 \n31. Ikuma, K., et al., Deposition of nanoparticles onto polysaccharide-coated surfaces: implications 526 \nfor nanoparticle-biofilm interactions. Environmental Science-Nano, 2014. 1(2): p. 117–122. 527 \n32. Liang, Y., et al., Interaction forces between colloidal particles in liquid: Theory and experiment. 528 \nAdvances in Colloid and Interface Science, 2007. 134-35: p. 151–166. 529 \n33. Nevius, B.A., et al., Surface-functionalization effects on uptake of fluorescent polystyrene 530 \nnanoparticles by model biofilms. Ecotoxicology, 2012. 21(8): p. 2205–2213. 531 \n34. Habimana, O., et al., Diffusion of Nanoparticles in Biofilms Is Altered by Bacterial Cell Wall 532 \nHydrophobicity. Applied and Environmental Microbiology, 2011. 77(1): p. 367–368. 533 \n35. Schambeck, C.M., et al., Chemical and physical properties of alginate-like exopolymers of 534 \naerobic granules and flocs produced from different wastewaters. Bioresour Technol, 2020. 312: 535 \np. 123632. 536 \n36. Leid, J.G., et al., The exopolysaccharide alginate protects Pseudomonas aeruginosa biofilm 537 \nbacteria from IFN-gamma-mediated macrophage killing. J. Immunol., 2005. 175(11): p. 7512–538 \n7518. 539 \n37. Felz, S., et al., Impact of metal ions on structural EPS hydrogels from aerobic granular sludge. 540 \nBiofilm, 2020. 2: p. 100011. 541 \n38. Moraes Schambeck, C., R.H. Ribeiro da Costa, and N. Derlon, Phosphate removal from 542 \nmunicipal wastewater by alginate-like exopolymers hydrogels recovered from aerobic granular 543 \nsludge. Bioresour Technol, 2021. 333: p. 125167. 544 \n39. Taylor, D.W. and A.A.D. Jones, Synthesis of Metal-Modified Nanocellulose as a Biofilm 545 \nAnalogue for Biofilm Mimicry in Biomedical and Environmental Applications. Biopolymers, 546 \n2025. 116(4). 547 \n40. Wang, L.S., A. Gupta, and V.M. Rotello, Nanomaterials for the Treatment of Bacterial 548 \nBiofilms. Acs Infectious Diseases, 2016. 2(1): p. 3–4. 549 \n41. Rodriguez-Suarez, J.M., et al., Heterogeneous Diffusion of Polystyrene Nanoparticles through 550 \nan Alginate Matrix: The Role of Cross-linking and Particle Size. Environ Sci Technol, 2020. 551 \n54(8): p. 5159–5166. 552 \n42. Jones, A.A.D. and C.R. Buie, Continuous shear stress alters metabolism, mass-transport, and 553 \ngrowth in electroactive biofilms independent of surface substrate transport. Scientific Reports, 554 \n2019. 9(1): p. 2602. 555 \n43. Herbert-Guillou, D., B. Tribollet, and D. Festy, Influence of the hydrodynamics on the biofilm 556 \nformation by mass transport analysis. Bioelectrochemistry, 2001. 53(1): p. 119–25. 557 \n44. Sankaran, J., et al., Single microcolony diffusion analysis in Pseudomonas aeruginosa biofilms. 558 \nNPJ Biofilms Microbiomes, 2019. 5(1): p. 35. 559 \n45. Zhang, Z., E. Nadezhina, and K.J. Wilkinson, Quantifying diffusion in a biofilm of 560 \nStreptococcus mutans. Antimicrob Agents Chemother, 2011. 55(3): p. 1075–81. 561 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n29 \n46. Stewart, P.S., Diffusion in biofilms. J Bacteriol, 2003. 185(5): p. 1485–91. 562 \n47. Fabrega, J., et al., Impact of silver nanoparticles on natural marine biofilm bacteria. 563 \nChemosphere, 2011. 85(6): p. 961–966. 564 \n48. Gil-Allué, C., et al., Long-term exposure to silver nanoparticles affects periphyton community 565 \nstructure and function. Environmental Science-Nano, 2018. 5(6): p. 1397–1407. 566 \n49. MacCuspie, R.I., et al., Challenges for physical characterization of silver nanoparticles under 567 \npristine and environmentally relevant conditions. J Environ Monit, 2011. 13(5): p. 1212–26. 568 \n50. Carmonaribeiro, A.M. and B.R. Midmore, Synthetic Bilayer Adsorption onto Polystyrene 569 \nMicrospheres. Langmuir, 1992. 8(3): p. 801–806. 570 \n51. Singh, R., et al., Penetration barrier contributes to bacterial biofilm-associated resistance against 571 \nonly select antibiotics, and exhibits genus-, strain- and antibiotic-specific differences. Pathog 572 \nDis, 2016. 74(6). 573 \n52. Prince, J., Theoretical, Simulation, and Experimental Approaches to Understanding Periphery 574 \nSequestration of Diffusing Chemical Species in Bacterial Biofilms as a Mechanism for 575 \nAntibiotic Recalcitrance. 2024, Duke University: Durham, NC. 576 \n53. Kurniawan, A., et al., Characterization of the internal ion environment of biofilms based on 577 \ncharge density and shape of ion. Colloids Surf B Biointerfaces, 2015. 136: p. 22–6. 578 \n54. Amsden, B., Solute Diffusion within Hydrogels. Mechanisms and Models. Macromolecules, 579 \n1998. 31(23): p. 8382–8395. 580 \n55. Bryers, J.D. and F. Drummond, Local Macromolecule Diffusion Coefficients in Structurally 581 \nNon-Uniform Bacterial Biofilms Using Fluorescence Recorvery After Photobleaching, in 582 \nBiotechnol Bioeng. 1998. p. 462–473. 583 \n56. Park, R.A., J.S. Clough, and M.C. Wellman, AQUATOX: Modeling environmental fate and 584 \necological effects in aquatic ecosystems. Ecological Modelling, 2008. 213(1): p. 1–15. 585 \n57. Hobbs, W.O., et al., Toxic Burdens of Freshwater Biofilms and Use as a Source Tracking Tool 586 \nin Rivers and Streams. Environmental Science & Technology, 2019. 53(19): p. 11102–11111. 587 \n58. Luoma, S.N. and P.S. Rainbow, Why is metal bioaccumulation so variable? Biodynamics as a 588 \nunifying concept. Environmental Science & Technology, 2005. 39(7): p. 1921–1931. 589 \n59. Bolis, V., et al., Hydrophilic/hydrophobic features of TiO2 nanoparticles as a function of crystal 590 \nphase, surface area and coating, in relation to their potential toxicity in peripheral nervous 591 \nsystem. J Colloid Interface Sci, 2012. 369(1): p. 28–39. 592 \n60. Flood, J.A. and N.J. Ashbolt, Virus-sized particles can be entrapped and concentrated one 593 \nhundred fold within wetland biofilms (Reprinted from Advances in Environmental Research, vol 594 \n3, pg 403-411, 2000). Advances in Environmental Research, 1999. 3(4). 595 \n61. Ferry, J.L., et al., Transfer of gold nanoparticles from the water column to the estuarine food 596 \nweb. Nature Nanotechnology, 2009. 4(7): p. 441–444. 597 \n62. Cleveland, D., et al., Pilot estuarine mesocosm study on the environmental fate of Silver 598 \nnanomaterials leached from consumer products. Science of the Total Environment, 2012. 421: 599 \np. 267–272. 600 \n63. Kroll, A., et al., Mixed messages from benthic microbial communities exposed to nanoparticulate 601 \nand ionic silver: 3D structure picks up nano-specific effects, while EPS and traditional endpoints 602 \nindicate a concentration-dependent impact of silver ions. Environmental Science and 603 \nPollution Research, 2016. 23(5): p. 4218–4234. 604 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint \n\n \n30 \n64. Wright, M.V., et al., Titanium dioxide nanoparticle exposure reduces algal biomass and alters 605 \nalgal assemblage composition in wastewater effluent-dominated stream mesocosms. Science of 606 \nthe Total Environment, 2018. 626: p. 357–365. 607 \n65. Miao, L.Z., et al., Effects of pH and natural organic matter (NOM) on the adsorptive removal of 608 \nCuO nanoparticles by periphyton. Environmental Science and Pollution Research, 2015. 609 \n22(10): p. 7696–7704. 610 \n66. Grün, A.Y., et al., Sublethal concentrations of silver nanoparticles affect the mechanical stability 611 \nof biofilms. Environmental Science and Pollution Research, 2016. 23(23): p. 24277–24288. 612 \n67. Fabrega, J., J.C. Renshaw, and J.R. Lead, Interactions of silver nanoparticles with 613 \nPseudomonas putida biofilms. Environ Sci Technol, 2009. 43(23): p. 9004–9. 614 \n 615 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703274doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}