AL-PHA beads: bioplastic-based protease biosensors for global health applications

Proteases are multi-functional proteolytic enzymes that have complex roles in human health and disease. Therefore, the development of protease biosensors can be beneficial to global health applications. To this end, we developed Advanced proteoLytic detector PolyHydroxyAlkanoates (AL-PHA) beads – a library of over 20 low-cost, biodegradable, bioplastic-based protease biosensors. Broadly, these biosensors utilise PhaC-reporter fusion proteins that are bound to microbially manufactured polyhydroxyalkanoate beads. In the presence of a specific protease, superfolder green fluorescent reporter proteins are cleaved from the AL-PHA beads - resulting in a loss of bead fluorescence. The Tobacco Etch Virus (TEV) AL-PHA biosensor detected the proteolytic activity of at least 1.85 pM of AcTEV. AL-PHA beads were also engineered to detect cercarial elastase from Schistosoma mansoni-derived cercarial transformation fluid (SmCTF) samples, as well as cancer-associated metalloproteinases in extracellular vesicle and cell-conditioned media samples. We envision that AL-PHA beads could be further developed for use in resource-limited settings.


Introduction
Synthetic biology is an established scientific field based upon engineering design principles, that has led to innovations in the development of biosensors and bioreporters geared towards global health applications [1,2]. These diverse applications have inspired a multitude of biosensor designs that have innovated beyond typical electrochemical formats, towards wholecell bioreporters (WCBs) and cell-free biosensors [3]. More recently, convergences between the materials sciences and synthetic biology are opening up new opportunities for global health biosensor applications [4]. In particular, we envisage that functionalised biomaterials may enable the emergence of novel strategies for detecting biomedically important proteases [5].
Proteases are multi-functional proteolytic enzymes, that have complex roles in human health and disease [6]. Protease functions are diverse and can be broad, such as aiding food digestion, or highly evolved and specialised targeting more specific substrates [6]. Exemplars of proteases that have evolved to serve complex biological functions can be found within the matrix metalloproteinase (MMP), the A Disintegrin And Metalloproteinase (ADAM) and A Disintegrin and Metalloproteinase with Thrombospondin motifs (ADAMTS) protease families [7]. Members of these protease families contribute to an array of biological processes including: cellular metabolism, cell-signalling, cell-migration, immunomodulation and tissue remodelling [8,9]. Changes in human metalloproteinase gene expression and/or their proteolytic activities can lead to cardiovascular or inflammatory pathologies, neurodegenerative diseases, changes in immunoregulation and cancer [10,11]. Proteases also have important roles in communicable diseases, whereby infectious microorganisms and parasites employ proteases to support pathogenesis [12]. In the case of schistosomiasis (also known as bilharzia or snail fever), a neglected tropical disease that affects over 250 million people worldwide [13][14][15], the invasive Schistosoma cercariae release a cocktail of proteases, including elastase, that help the parasite invade their host through the skin [16,17].
Understanding the activities of proteases can lead to important insights into communicable and non-communicable diseases [6]. Indeed, novel protease detection strategies, especially those intended for field use, may be beneficial to many different clinical, biotechnological, environmental and epidemiological global health applications [5,[18][19][20][21]. For example, in our previous study we used a synthetic biology approach to engineer Escherichia coli and Bacillus subtilis WCBs that can detect the elastase activity from the cercariae of the parasite Schistosoma mansoni [22]. Importantly, our study demonstrated the detection of the proteolytic activity of a specific protease (i.e. cercarial elastase) within complex biological samples. However, the implementation of WCBs within global health settings is challenging and many other complex practical, cultural, societal, data protection and regulatory concerns must also be addressed [23]. Understandably, amongst those concerns, the accidental release of living engineered WCBs is commonly cited [24]. In response to these challenges, the development of WCBs has led to important innovations in physical (e.g. sealing WCBs within devices) and genetic (e.g. genetically encoded kill switches or auxotrophy) containment strategies, that help mitigate the risk of accidental release [23]. Whilst these biological containment strategies are impressive, we anticipate that non-living biosensors may be desirable in certain contexts.
To this end, we developed modular, functionalised, polyhydroxyalkanoates (PHAs)-based, bioplastic beads for protease detection. We termed these biosensor beads -Advanced proteoLytic detector PHAs (AL-PHA) beads and initially optimised their design using a commercially available Tobacco Etch Virus (TEV) protease. As a proof-of-concept for a global health applications, AL-PHA biosensors were assayed against S. mansoni derived samples containing soluble cercarial antigens, termed cercarial transformation fluid (SmCTF). AL-PHA biosensors successfully detected cercarial elastase activity within these samples. We also showed that AL-PHA beads engineered to detect cancer-associated metalloproteinases including: MMPs, ADAMs and ADAMTSs were functional. Most notably, AL-PHA beads detected recombinant MMP14 and extracellular vesicle (EV)-associated ADAM10 derived from an in vitro model of non-small cell lung cancer (nsclc). Furthermore, we also demonstrate the potential use of AL-PHA beads in a high-throughput screening context -whereby an entire library of metalloproteinase biosensors was tested in parallel. To the best of our knowledge, this proof-of-concept study is the first to demonstrate the use of functionalised PHAs-based protease biosensors for health care applications in resource-limited environments.

Functionalised polyhydroxyalkanoates (PHAs)-based beads for protease detection
PHAs represent a diverse family of biopolymers with different mechanical characteristics, thermal properties, biodegradabilities and biocompatibilities [25]. Poly-3-hydroxybutyrate (P(3HB)) is one of the most well studied PHAs polymers and can be used within food packaging or for other industrial applications, including medical and tissue engineering applications [26]. Polymerised P(3HB) naturally forms into spherical granules (beads) within suitably engineered E. coli [27]. Interestingly, these PHAs beads can also be functionalised in vivo with engineered fusion proteins enabling novel protein purification strategies [28], as well as the development of PHAs-based vaccines [29]. To further extend the applications of functionalised PHAs materials we engineered our own PhaC-fusion proteins, which could act as protease biosensors, on the surfaces of these PHAs beads -which we termed AL-PHA beads ( Fig. 1).
To achieve this, we adapted a highly active C104 phaCAB biosynthetic operon from our previous studies, to develop AL-PHA producing operons in engineered E. coli [30,31]. These AL-PHA operons contain PhaC-fusion proteins have been designed to incorporate a flexible and modular amino acid linker that comprises protease-specific cleavage sites, and a GFP  Table 1). These PhaC-fusion proteins were designed such that specific proteases can be detected via the proteolytic cleavage of sfGFP from the surface of the AL-PHA beads.
AL-PHA beads were produced in E. coli cells, isolated using a sonication-based method and tested using a simple assay (Fig. 1C). Essentially, control or protease specific AL-PHA beads were incubated with proteases, with optimal reaction conditions tailored to the specific protease being tested (See materials and methods). Post-incubation, AL-PHA beads were centrifuged allowing the supernatant and pellet to be analysed separately. This allows proteolytically released sfGFP in the supernatant to be measured using a plate reader, whilst the associated concomitant loss in AL-PHA bead fluorescence was measured using flow cytometry.
To assess whether our PhaC-sfGFP fusion proteins were correctly localised on the AL-PHA bead surface, we used the non-specific protease trypsin. As expected, trypsin treatment (1 µg) significantly decreased AL-PHA bead fluorescence compared to untreated controls (0 µg; Supplementary Fig. 3). Usefully, this reduction in bead fluorescence was observable to the naked eye when the beads were pelleted and placed onto a transilluminator ( Supplementary   Fig. 3C). This initial data supported the notion that the engineered PhaC-fusion proteins are accessible and susceptible to proteolytic activity on the surface of the AL-PHA beads.
We next chose three different proteolytic linker designs specific for TEV protease, which differed only in length, as follows: 12 amino acids (12L), twenty-two amino acids (22L) and one hundred and twelve amino acids (112L) (  Table 2). We previously identified that longer linker lengths can increase proteolytic sensitivity, likely through improving cleavage site access [22]. AL-PHA beads with the relevant TEV linker designs were assayed with either 0 or 10 U of AcTEV protease and incubated at 30 o C for 2 hours with shaking. Post-assay supernatants and AL-PHA beads were assessed, separately, using either plate reader (supernatant) or flow cytometry (AL-PHA bead) workflows ( Supplementary Fig. 4). In comparison to untreated controls (0 U AcTEV), supernatant fluorescence levels increased for all AcTEV treated (10 U) AL-PHA bead samples indicating that all three linker designs can specifically detect AcTEV protease ( Fig. 1D-F).
Interestingly, the most sensitive linker was 112L (PhaC-112L-T-G), with a ~10-fold relative increase in supernatant fluorescence levels compared to controls (Fig. 1F), which we quantified as being equivalent to 1.87 µM sfGFP being released ( Supplementary Fig. 5). By analysing flow cytometry data ( Fig. 1F; Supplementary Fig. 5) we estimated that ~187 pmoles of GFP is being released from ~300,000 AL-PHA beads, which is within the range of previous studies in terms of surface bead coverage [28,32]. Furthermore, the concomitant reduction in AL-PHA bead fluorescence was also significantly more pronounced for 112L (~49% decrease; Fig. 1F) and this reduction was observable to the naked eye using a transilluminator ( Supplementary   Fig. 6). We also observed that after 2 hours of treatment the PhaC-112L-T-G AL-PHA beads are sufficiently sensitive to detect 0.5 U of AcTEV activity ( Fig. 2A, B), which is ~1.85 pM of AcTEV protease. Taken together, our data shows that AL-PHA beads with the longest linker  Table 3). Several batches of these different AL-PHA beads were generated in engineered E. coli and purified as described above. AL-PHA bead sizes were characterised using dynamic light scattering (DLS) and were typically 1.1 ±0.02 µm in diameter ( Supplementary Fig. 8A). Flow cytometry was also used to confirm that that these AL-PHA bead biosensors were fluorescent. Essentially, 112L AL-PHA beads were typically ~30-48 fold more fluorescent than the non-functionalised control PHAs beads (C104), which indicates correct functional assembly and surface localisation of the PhaC-fusion protein in vivo.
( Supplementary Fig. 8B). In contrast, the MMP9 and elastase specific AL-PHA beads were only ~4 fold more fluorescent than control beads suggesting that certain protease recognition motifs (cleavage sites) within the linker region appear to affect the correct assembly of the PhaC-sfGFP fusion protein on the beads. However, flow cytometry data showed that all 112L AL-PHA beads were functionalised and suitable for application testing despite the variability described above. of AcTEV protease. Proteolytically released sfGFP in supernatant samples were analysed using a CLARIOstar plate reader (483-14 nm/530-30nm) and these fluorescence data were normalised against untreated controls of the same biosensor batch. AL-PHA beads were analysed using flow cytometry and AL-PHA bead geometric mean (BL1-A, 488nm/530-30nm) of AcTEV treated beads were normalised against untreated controls of the same biosensor batch. Error bars denote standard error of the mean, n=4-8, Student t-test **P<0.01, ***P<0.001, ****P<0.0001or not statistically significant (ns). The neglected tropical disease schistosomiasis is of increasing burden to global health, with estimates suggesting that 779 million people are at risk of infection, leading to an annual mortality upwards of 280,000 people in sub-Saharan Africa alone [13,14,33]. Therefore, the ability to detect this parasite is of great importance. Indeed, several trap systems have been proposed that can capture and concentrate schistosoma cercariae ready for downstream sample processing and testing [23]. To investigate whether the AL-PHA assay is applicable to the detection of S. mansoni -one of the principal causative agents of human schistosomiasis, we engineered AL-PHA beads specific for this parasite. Previously, we targeted the S. mansoni cercarial elastase activity as our marker for detection using WCB's [22]. The cercariae utilise this elastase activity to penetrate the skin barrier, thereby enabling invasion and infection of their definitive hosts, in this case humans [16,17].
We first replaced the TEV protease recognition motif with that of S. mansoni cercarial elastase via inverted PCR. The elastase specific recognition motif (-SWPL-) used here and in our previous study [22], was identified using positional scanning and synthetic combinatorial library screening [17]. To test whether our 112L elastase specific AL-PHA design could detect elastase, we tested three biologically distinct S. mansoni-derived extracts containing soluble cercarial antigens, termed cercarial transformation fluid (SmCTF; Fig. 3A) [34]. These samples were obtained by mechanically transforming cercariae released from the intermediate snail host, Biomphalaria glabrata (Fig. 3A) [34]. When the elastase specific AL-PHA beads were exposed to the SmCTF samples, the beads detected elastase in SmCTF samples 2 and 3, shown by an increase in supernatant sfGFP fluorescence and a reduction in sfGFP bead fluorescence (Fig. 3B). Although SmCTF1 caused a slight reduction in bead fluorescence, no corresponding increase in supernatant fluorescence was detectable (Fig. 3B), suggesting that this sample has lower amounts of elastase. We did observe some off-target cleavage for the three SmCTF samples against the TEV protease control beads (Fig. 3B), which is likely due to non-specific proteases within the cercarial gland extract [22]. However, for two of the SmCTF samples, the elastase specific AL-PHA beads are cleaved significantly more compared to controls (Fig. 3B).
This demonstrates the ability of AL-PHA beads and the appropriate controls to identify samples significantly enriched with S. mansoni cercarial elastase. Applying AL-PHA beads to detect matrix metalloproteinase MMP14 The matrix metalloproteinases (MMPs) are a family of proteases that have important physiological roles in extracellular matrix (ECM) turn-over, tissue homeostasis, immunomodulation, and cell signalling [18]. Interestingly, differential MMP14 (also called MT1-MMP) expression, protein levels or proteolytic activities are implicated in several diseases including neurodegenerative disorders and several cancers (e.g. breast, gastric, lung and ovarian cancers) [35][36][37]. Therefore, the detection of MMP14 proteolytic activity might be highly informative as a prognostic or diagnostic biomarker for several different biomedical applications.
To examine this, we tested MMP14-specific 112L AL-PHA beads using pre-activated recombinant MMP14 (Fig. 4). as well as residual AL-PHA bead fluorescence, we were able to mitigate any off-target effects (Fig. 4B) [35], are expected to be below the observed AL-PHA bead detection limit, appropriate patient sample processing steps (e.g. protein concentration or exosome isolation protocols) will be required to concentrate MMP14 to within detectable AL-PHA biosensor levels. To test this concept, we next tested AL-PHA beads against EV-associated metalloproteinases.
Detecting membrane associated metalloproteinases in extracellular vesicles using AL-PHA beads A number of physiologically important MMPs are membrane associated (e.g. MT-MMPs and ADAMs) and one potential strategy for detection would be through the isolation of extracellular vesicles (EVs). EVs, including exosomes, are readily accessible from patient liquid biopsies and recent studies indicate that EV-associated metalloproteinases have complex roles in cancer metastasis [39,40], including lung cancer where early detection could positively impact patient outcomes [41]. Indeed, exosomes isolated from in vitro models of lung cancer (e.g. A549 cells) or the blood of nsclc patients exhibit increased ADAM10 activity [37,42].
For A549 cells, ADAM10 is strongly expressed and secreted within A549 EVs [42]. Therefore, ADAM10 specific AL-PHA beads may be a useful tool for lung cancer biomarker research.
After constructing ADAM10-specific 112L AL-PHA beads, we tested their ability to detect for ADAM10 proteolytic activity from purified A549 EVs (Fig. 5). Prior to the AL-PHA bead assays, A549 EVs were characterised using nanoparticle tracking analyses (EV mode diameter 88.8 ±4.8 nm; Supplementary Fig. 9), an Exo-Check dot blot array (CD81, CD63, ICAM, ANXA5, TSG101 positive; Supplementary Fig. 10), and flow cytometry surface marker characterisation (CD9, CD63 and CD81 positive; Supplementary Fig. 11). Control and ADAM10-specific AL-PHA beads were treated with either 0 µg or 50 µg (total protein) of A549 EVs using an EDTA-free protease inhibitor cocktail to inhibit a broad array of serine and cysteine proteases, but not metalloproteinases (see materials and methods). ADAM10 AL-PHA and control assays were incubated at 37 o C for 4 h and then analysed, post-assay, using a plate reader and flow cytometry (Fig. 5A). Relative supernatant fluorescence levels increased by ~0.5-fold (~46% increase) for A549 EV-treated samples compared against untreated control samples (Fig. 5B). Furthermore, A549 EV treatment also caused a significant decrease in AL-PHA bead fluorescence (~16% decrease), whilst control beads and supernatant fluorescence levels were unaffected (Fig. 5B). Taken together, our data show a positive detection of ADAM10 proteolytic activity from A549 EVs (50 µg total protein) which is in agreement with a previous study using lysed nsclc-patient EVs (60 µg -total protein of exosome lysates) [37].
However, in comparison to the aforementioned study [37], we envision that AL-PHA assays may enable a more simplified approach for detection. Future studies will be required to investigate if ADAM10 AL-PHA bead assays are sensitive enough to detect EV-associated ADAM10 within liquid biopsy samples or isolated patient exosomes.
To support these future studies and in order to optimise AL-PHA bead screening efficiency, we devised and tested a simple, semi high-throughput assay for screening AL-PHA metalloproteinase-specific bead libraries ( Fig. 6A; Supplementary cells and their secretomes (>20 kDa proteins and EVs) within ~20 ml harvest volumes.
Therefore, as expected, this high-throughput screening approach enabled the successful detection of the proteolytic activities of several metalloproteinases ( Fig. 6B; Supplementary

Discussion
Protease structure-function studies continue to highlight the biomedical importance of proteases in an array of communicable and non-communicable diseases [6]. Therefore, novel strategies for detecting proteolytic activity are desirable. Classical protease activity assays typically incorporate fluorogenic small molecule, peptide or nanoparticle substrates, Fluorescence Resonance Energy Transfer (FRET)-probes, electrochemical components or zymography methods [46][47][48][49]. Whereas, recent synthetic biology approaches have led to the development of more sophisticated modelling-led design strategies, the embedding of protease biosensors within smart materials and also increasingly complex whole-cell bioreporters [4,5,19,22,23]. These classical and synthetic biology approaches each have their own advantages and limitations, several of which we briefly summarise and compare alongside AL-PHA beads (Supplementary Table 6). Protease biosensor implementations, especially those intended for field or point-of-care use, must also consider responsible research and innovation requirements (e.g. implementation costs, political, regulatory and societal contexts) [23].
Beneficially, AL-PHA beads are non-living and are biodegradable [4,31] which further extends the flexibility of their implementation and safe disposal. Moreover, their modular design, ability to be measured using different approaches (visual, plate reader and flow cytometry) and capacity for use within high-throughput screening workflows, greatly extends their usefulness in terms of the different contexts where AL-PHA biosensors could potentially be implemented in the future (e.g. diagnostic labs or in the field/point-of-care). We also estimate that the raw material cost to microbially produce enough AL-PHA beads for a typical AL-PHA assay, is around 2-4 pence (GBP). Even when sample processing costs, labour and other miscellaneous costs are taken into account, AL-PHA assays are likely to remain cost competitive against competing technologies. In conclusion, we have developed a protease-detection assay using AL-PHA beads and have shown that a library of low-cost, biodegradable, AL-PHA bead protease biosensors can be utilised to detect specific protease activities. We suggest that AL-PHA beads could be a platform for implementation as low-cost global health biosensors for use in resource-limited settings.

Bacterial strains and general growth conditions
Plasmid constructs and strains used in this study are listed in Supplementary Table 3. E. coli JM109 was used for both cloning and production of AL-PHA protease biosensors. For plasmid recovery E. coli strains were grown in Luria-Bertani (LB) medium supplemented with 34 µg/ml Chloramphenicol (final concentration) and cultured at 37°C with shaking (220 rpm).

Construct assembly
Empty vector plasmid EV104 was originally sourced from the 2013 distribution of the iGEM Registry of Standard Biological Parts (partsregistry.org; BBa_K608002) and was transformed during this study into E. coli JM109 to create strain EV104-JM109. Plasmid C104-JM109, encoding a constitutively expressed, engineered phaCAB operon was originally derived from our previous study [31] and was transformed during this study into E. coli JM109 to create strain C104-JM109. C104-JM109 was used to create wild-type, non-functionalised PHAs beads.
For 12L AL-PHA biosensor construction, the PhaC fusion (12L-G) region was ordered as a Oligonucleotide primers used for generating biosensor constructs and DNA sequencing are shown in Supplementary Table 4.

Production, and characterisation of AL-PHA biosensor beads
Glycerol

SmCTF sample preparation
The three SmCTF samples tested (SmCTF1-3) were produced by BioGlab Ltd. (Nottingham, UK) as previously described [22,34]. Freeze-dried SmCTF samples were reconstituted in sterile distilled water and stored at -20°C until required.  Fig. 12). Cell growth rate was indirectly evaluated through daily monitoring of the glucose level of the cell culture medium (#GC001000, FiberCell Systems, Inc., MD, USA). The cell culture medium was changed once glucose levels significantly decreased to less than half the original level. Once glucose consumption levels significantly increased (e.g. 50% glucose consumed within 24 h) the cell culture media was

GFP Calibration curve
Super folder green fluorescent protein (sfGFP) samples were purified from AL-PHA beads and used to create an AL-PHA GFP calibration curve ( Supplementary Fig. 5). Briefly, four 1.

Statistics
Statistical analysis (standard error of the mean (s.e.m.) and unpaired t-test) was carried out on at least three experimental replicates using GraphPad Prism 8.4.2 (GraphPad Software Inc., La Jolla, California) and flow cytometry data analysis was performed using FlowJo (vX 10.5.3) software.

Declaration of conflicts of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.