Our idea and design - REE

Our idea

Our idea to tackle the critical problem of lacking efficient and sustainable REE recycling systems is by using synthetic biology to engineer a modular protein display system capable of selectively binding and capturing specific REE's (La, Ce, Gd, Dy, Nd and Pr) dissolved in waste solutions. The aim is to then release those captured metals, separating the metal ions from the protein-system, leading to a complete recovery of the metal ions in the remaining supernatant. As a final step, it would be possible to convert the metal ion solution to metal salts. Further, salts could be converted into REE-oxides and provided as circular feedstock for electronic modules manufacturing, thus promoting sustainable recovery of e-waste-sourced REE.

Our system is made of two modules: a protein (curli) fiber biomass produced by an optimized E.coli strain, that acts as a scaffold, and REE-specific binding proteins. The two modules are covalently bound thanks to peptide tags that lead to a covalent protein-protein ligation, i.e. the DogTag-DogCatcher system, whereby a DogTag is fusing to the curli fibers’ proteins (CsgA), and a DogCatcher to the REE-binding proteins (Keeble et al., 2022).

Figure 1: Engineered Curli-Based Biofilm System for Lanthanide Capture. The illustration depicts E. coli Curli producing strain and a curli nanofiber. CsgA subunits, nucleated by the CsgB base protein on the outer membrane, assemble into extracellular fibers. Each CsgA is engineered to display a DogTag domain. Through the DogTag-DogCatcher covalent interaction, lanthanum binding protein is anchored onto the curli nanofibers. The binding protein domains specifically capture La³⁺ ions from the environment, enabling efficient metal sequestration via the engineered biofilm matrix. Created in https://BioRender.com.

Design

Curli fibers

The system is made up of a biomass of curli fibers produced by an optimized strain of E.coli which the curli protein production controlled by a constitutive promoter.

Curli fibers are amyloid protein fibers normally produced by some bacteria in biofilms. In E. coli , curli fibers are encoded by the csgBA and csgDEFG operons, which include six proteins needed for the assembly of the fibers themselves and their export machinery (Bahrnart & Chapman, 2010). The csgBA operon encodes two different beta-sheet proteins - CsgA and CsgB, which are the main components of the extracellular amyloid fiber. The CsgB protein is needed as a "pedestal" for the CsgA subunits to polymerise in the extracellular environment - without CsgB, the curli will not be assembled (Bahrnart & Chapman, 2010). The advantage of using curli in our design is twofold: (1) firstly, the presence of multiple CsgA subunits allows us to maximise metal-binding capability of our system: for such purpose, we engineered each CsgA subunit to carry a DogCatcher, so that each monomer of the curli will hybridise with a REE-binding protein. (2) Secondly, the density of the REE-binding proteins alone would not allow for pelleting, as they would remain in the supernatant due to their low molecular weight: the presence of the curli fiber acting as a scaffold brilliantly resolves this issue.

Figure 2: Schematic representation of curli fiber formation and functionalization. The E.coli cell produces CsgA and CsgB subunits from the csgBA operon. CsgB anchors to the outer membrane and serves as a nucleator, enabling the polymerization of CsgA into extracellular curli fibers. In our engineered system, each CsgA subunit is fused to a DogCatcher domain. Created in https://BioRender.com.

To this end, the Schaerli lab provided us with two different curli producing E.coli strains:

• csgA– strain (negative control): this strain has the csgABCDEFG operon deleted from the genome. This strain does not produce curli fibers.
• csgA+ DogTag strain: this strain of curli-producing E. coli carries a DogTag expected to efficiently interact with the DogCatcher.

A schematic representation of the gene encoding the curli-DogTag protein module can be found in Table 1 below.

Binding proteins

For each REE we wanted to be able to capture, we started with the raw sequences of each individual binding protein found in the literature (Table 1):

Table 1: Schematic representation of the binding protein and curli gene modules. The table lists the collection of binding proteins employed in our REE-binding system, with the respective PDB identifier and a link to the iGEM parts repository. A schematic representation of the gene for our curli-DogTag module is also displayed in the last row. All REE-binding protein genes are preceded by a T7 promoter and an IPTG-inducible Lac operator, as well as a ribosome binding site (not shown).

Protein Metal (Target Metal)	Graphical DNA representation	Link iGEM parts	Original Publications
Lanmodulin 8DQ2 (La/Ce)		BBa_25K7UGOJ.	Mattocks et al. (2023)
Lanmodulin 8FNR (Dy)		BBa_25WS2YVO.	Mattocks et al. (2023)
GLamouR GLAM (Gd)		BBa_25ODY87T.	Lee et al. (2025)
Lanmodulin 8FNS (Nd)		BBa_25IUKDSB.	Harris & Wesley, (1986)
Pr (PedH 6ZCW)		BBa_25L0PQDR.	Wehrmann et al. (2020)
CsgA-Tag (curli-DogTag)		BBa_25BU4MXM.	Tay et al., (2018)

Here we have 3 different Lanmodulin binding protein variants for binding La/Ce, Dy and Nd.

To better illustrate the structural features of each REE-binding protein, the section below presents their 3D models.

Structural alignment of AlphaFold 3 predictions with experimental PDB structures

▼

This section presents ChimeraX structural alignment visualizations showing the superposition of AlphaFold 3-predicted structures (blue/light green) with corresponding experimental PDB structures (light brown) for each REE-binding protein analyzed.

The curli model was also generated in ChimeraX: the DogTag-DogCatcher domain is represented in pink, the REE-binding protein in yellow, and the polymeric CsgA nanofibers in blue.

Figure 3: 8FNR protein binding Dysprosium

Figure 4: 8DQ2 protein binding Lanthanum and Cerium

Figure 5: 8FNS protein binding Neodymium

Figure 6: GLAM (3WLC) binding Gadolinium

Figure 7: 6ZCW binding Praseodymium

Figure 8: Structural model of the DogTag–DogCatcher–binding protein system on curli fibers

❮

×

❯

Broader implementation of our system

Since our system can selectively capture individual REE, we conceptualised a broader implementation strategy aimed at scaling up the process to enable REE extraction from e-waste sludge on an industrial scale (Figure 9).

In our results section, we present a proof-of-concept of our technology, which enables REE capturing from solution. Moreover, our Dynamic Modeling page showcases the different affinities of the binding proteins to the REE, thus paving the way for an extremely efficient and selective REE separation technology.

Our idea is based on the fact that existing e-waste recycling facilities are nowadays able to recover some precious metals from e-waste streams, such as silver, gold, and copper (UMICORE, 2025). This recovery system is based on hydrometallurgical methods, such as liquid-liquid extraction and leaching or ion separation - following other steps such as shredding and pyrometallurgical treatments (Ashiq et al., 2019). This technology however, does not allow the recovery of REE, being the lanthanides too similar to each other from a physico-chemical standpoint. We conducted further research, and discovered that after recovery of the precious metals, the recycling facilities are left with a liquid e-waste sludge, which is generated to dissolve the precious metals in the first place, but which also dissolves the REE contained in the waste into ions - similarly to how we treated the REE in our experimental procedure. The REE themselves are though not recovered from the sludge, which is directly sent to water treatment facilities.

Starting from this knowledge, we therefore conceptualised a system that could be incorporated directly into the e-waste recycling facilities’ existing apparatus, thus allowing us to offer a protein-based REE-recovery service that could potentially be launched as a venture. Our idea is illustrated in Figure 9.

Figure 9. Conceptual framework for REE extraction scale-up. In our design, a series of chemostats each containing the curli/REE-binding protein system specific for a single REE are sequentially connected. The e-waste liquid sludge retrieved from other recycling facilities is made to flow through the chemostat system, to allow selective REE capturing in each of the cylinders. The chemostats are separated by a filter that only allows the flow of the liquid sludge and the REE contained within, thus preventing the REE-binding protein biomass to contaminate the following chemostat.

Within such a framework, a series of chemostats are sequentially interconnected. Each chemostat contains one type of curli/REE-binding protein system, so as to make each of the cylinders specific for a single REE. The order of REE-binding proteins in the chemostats is arranged according to the lanthanide contraction series, meaning larger ions are targeted first. This design enhances selectivity between different lanmodulins (La&Ce, Nd, and Dy) by applying a “size exclusion” principle. By starting with a binding protein that prefers the largest atomic radius, the system ensures that smaller ions (like Nd) do not bind at this initial stage.

After the e-waste sludge passes through all the chemostats, the cycle can be repeated to fully saturate the curli/REE-binding proteins and maximize the extraction of REE. Once several extraction cycles are complete, the sludge is returned to the wastewater treatment process, now free of REE, which have been efficiently recovered by our system. Next, we thought that a low-pH buffer could be used to denature the curli/REE-binding proteins, causing them to release the REE ions into the solution for recovery. These recovered REE ion solutions can then be heated to form salts, which are converted into oxides and returned to the market for electronics manufacturers to use, thus forming a circular REE-recovery process.

Thus, our system has the potential for working on a broader scale, allowing us to implement REE extraction in the existing e-waste recycling stream. We strongly believe that such a framework has the potential to be incorporated in current e-waste recycling systems, with the objective to promote a sustainable and circular recovery of these elements for the benefit not only of the e-waste recycling sector, but of the electronics manufacturing market as well.

References

• A. Ashiq, J. Kulkarni and M. Vithanage; “Electronic Waste Management and Treatment Technology”, edited by M. N. V. Prasad and M. Vithanage Butterworth-Heinemann; 2019. doi: Electronic Waste Management and Treatment Technology 2019, Pages 225-246
• Barnhart, M. M., & Chapman, M. R. (2006). Curli biogenesis and function. Annu Rev Microbiol, 60, 131-147.
• Kan, A., Birnbaum, D. P., Praveschotinunt, P., & Joshi, N. S. (2019). Congo Red Fluorescence for Rapid In Situ Characterization of Synthetic Curli Systems. Appl Environ Microbiol, 85(13).
• Keeble, A. H., Yadav, V. K., Ferla, M. P., Bauer, C. C., Chuntharpursat-Bon, E., Huang, J., Bon, R. S., & Howarth, M. (2022). DogCatcher allows loop-friendly protein-protein ligation. Cell Chemical Biology,29(2), 339-350.e310.
• Umicore. (n.d.). Precious metals. Retrieved October 5, 2025