Team:Imperial/Water Filtration


Imperial iGEM 2014


By attaching functional proteins to cellulose, we can expand its properties and selectively capture specific contaminants in water and use it as a filter. We used five different cellulose binding domains and fused them to different metal binding proteins. After determining the best CBDs in our possession through an assay, we proved their ability to bind cellulose as well as bind metals, both in a filter set up as well as in a stable well plate. Moreover, we serendipitously found that bacterial cellulose itself has metal chelating properties, validating its potential as a versatile, functionalisable material.

Key Achievements

  • Synthesized and characterised cellulose binding domains (CBD) and their phytochelatin fusions.
  • Proved that cellulose can be functionalised with CBD-phytochelatins.
  • Showed that bacterial cellulose in itself can successfully filter metal out of solution.
  • Successfully demonstrated filtration of turbid water.
  • Captured Nickel ions on our customisable cellulose filter.
  • Functionalised our homegrown bacterial cellulose with our fusion proteins.


Water is typically purified by passing it through layers of porous materials, each specially selected for its ability to remove specific forms of contamination. All porous materials can filter particles by size, many have extra capabilities thanks to their chemical properties. For example, charcoal - or activated carbon - is a porous component of household and industrial water filters that can also bind large or electron-rich molecules via the van der Waals forces, and catalyse the breakdown of other chemicals such as molecular chlorine. While there are many different types of filters, we can categorise and compare them using their measurable physical and chemical properties. The key physical properties are:

  • Pore size - average or maximum size of pores in the material
  • Porosity - volume of the filter not occupied by solid material
  • Tortuosity - length of paths through the filter compared with a straight line
  • Adhesion - the strength of hydrogen bond interactions between the fluid and filter
  • Kinetic rate constants - parameters defining how the filter material affects chemical reactions in the fluid

These properties contribute to the filter’s flow rate, measured in units of fluid filtered per unit surface area of the filter per unit time (e.g. Ga m-2 h-1). Altering the filter’s flow rate generally affects the filter’s efficiency at removing its targets, as increasing the flow rate typically involves increasing the pore size or reducing the surface area of the filter in contact with the fluid. This trade-off is the main hurdle in industrial water filtration, where demands for high-quality filtration require very low flow rates. Industrial processes are further complicated by blockage of the pores, fouling of the filter by organisms and the lack of a complete set of materials that will exhaustively filter all contaminants - heavy metal ions, small organic molecules and non-polar compounds remain difficult to filter without prior chemical treatments that themselves need to be filtered out.


Ultrafiltration membranes remove particulates by size exclusion. Our bacterial cellulose membranes completely removed turbidity from input water outputting clean, potable water. Input water containing 1% w/v organic solids from Imperial College Queens lawn was vortexed then 3ml passed through filters of 25cm3 area. Turbidity of flowthrough was measured in a spectrophotometer, absorbance at OD500 as recommended by Epa (1993).

Figure 1: The picture shows different amount of colouring in the water representing the debris left after filtering water mixed with dirt using different filter methods. From left to right: Unfiltered water, tissue paper filtered water, filter paper filtered water, bacterial cellulose filtered water, clean distilled water.

Cellulose sheets produced by G. xylinus iGEM were dried to thin filters around 20 g/m3. Coffee press filtration with these filters was compared to similar filtration by tissue paper and VWR qualitative medium filter paper (5-13 um pore size).

Average OD500 values from three replicates:

  • Water unfiltered: 0.072.
  • Tissue Paper filtered: 0.039.
  • Filter paper filtered: 0.023.
  • Bacterial Cellulose filtered: 0.00.
  • Distilled water: 0.00.

Cellulose functions as a water filter and completely removes suspended particles.

Phytochelatin-dCBD metal binding assay

Metal binding proteins fused to CBDs were bound onto cellulose in 96-well plates and tested against 3 different metals (Nickel, Copper, Zinc). First, the fusion protein lysate was incubated overnight in the cellulose wells. Following this, the metal salt solutions are added in excess into the wells. Finally, an EDTA step removes the bound metal ions into solution, and the metal concentration in the EDTA solution is quantified by mass spectrometer. Multiple washes with PBS and water were done between each binding step, ensuring that metal ions read were released from the metal binding proteins. Then as mentioned in Xu (2002), the assay was repeated using the same cellulose discs, to test if the phytochelatin was re-useable.

Figure 2: This table summarises the metal concentration in EDTA after washing cellulose samples exposed to different treatments: different phytochelatin-CBD fusion protein lysates added and the salt solution it was incubated in. Concentrations were obtained via Inductively coupled plasma optical emission spectrometry (ICP-OES). The lower part of the table shows secondary treatment of the cellulose after the first EDTA wash.

If the functionalised cellulose has bound the metal in question then the final EDTA solution that is measured using the mass spectrometer should be found to have a high concentration. EDTA is a powerful chelator and will bind any metals in the sample preferentially to other organic molecules. The highest concentrations were found within the cellulose only tests for all metals, however due to the large number of hydroxyl groups, it isn’t too surprising that the cellulose is capable of chelating metals. All samples that had cell lysate applied were much lower, even the non-CBD phytochelatin controls, which may be caused by the phytochelatin and other molecules in the cell lysate reducing the natural chelating power of the cellulose. For Zinc, the metal that phytochelatin most successfully binds, the PC-CBD fusion tests all resulted in roughly double the concentration of Zinc. We propose that the reduction in concentration due to blocking of the cellulose by lysate is still there and that the extra concentration recorded is purely from Zinc that was bound to PC-CBD. However, this pattern is not observed in the Copper or Nickel tests.

Figure 3: Graph to compare the different cellulose treatments and their correspondence to the average metal concentrations after the first EDTA wash.
Figure 4: Graph to compare the different cellulose treatments and their correspondence to the average metal concentrations after the second EDTA wash.

The results from the re-used functionalised cellulose assay demonstrated much lower concentrations of metals than the first wash. However the cellulose samples in this part of the experiment have been in contact with EDTA which, due to its carboxyl group, has been shown capable of binding to cellulose (Tsukinoki, 2008). Since the non-functionalised cellulose samples showed much lower concentration of all metals, it can be concluded that the cellulose binding to metals is disrupted. However, one of the PC-CBD fusions had consistently higher concentrations of metal; the PC-dCBD. This suggests that even with the interference with cellulose that is observed when adding the EDTA, the PC-dCBD stays bound to the cellulose and works as a functioning phytochelatin. In fact, PC-dCBD showed the only increase in metal concentration in the second wash compared to the first, seen in its nickel sample. The other PC-CBD fusions have results that match the PC alone samples more suggesting that they do not bind the cellulose as strongly under these conditions. This concept can be observed in the graph that shows the difference in concentration between the two washes; the PC-dCBD samples are the only samples that change little between the two experiments. The PC-CBDcipA copper samples are quite similar but the values are very low for both. Note that sfGFP-dCBD fusions showed high results in the sfGFP characterisation of CBD assay.

Figure 5: This graph compares the averages between the first and second wash for each sample type, showing that all but one of the samples reduced in metal concentration in the second wash.

Time constraints have prevented repetition of the experiment, but clearly more work is needed to fully characterise the binding of these different CBDs. The assay should ideally be repeated with different concentrations of the PC-CBDs, though same total concentration of lysate by combining with wild type lysate, to investigate the impact of other proteins blocking the binding of the cellulose. We conducted the experiment with lysate as being able to apply lysed cells expressing the protein of interest would be cheaper than the purifying the protein first. However, in the interest of characterisation of the CBDs it would be beneficial to purify the protein to remove any interaction from background molecules and proteins found in the lysate. Another adaptation would be to, after the binding of the metals has occurred and the full wash cycles completed, dissolve the cellulose membranes in Nitric acid and run those samples on the mass spectrometer. This would allow us to confirm that the EDTA does indeed fully chelate any metal found within the functionalised cellulose (be the metals bound to the cellulose or the CBD-PC). Finally, testing with different concentrations of metal is vital to show the range that the functionalised cellulose is useful for.

Nickel filtration assay

Figure 6: The set up for the nickel filtration assay. A) The coffee press used for the mounting of the cellulose and for filtration process. B) Cellulose without attached phytochelatin-dCBD protein. C) Functionalised cellulose with attached phytochelatin-dCBD protein.

The nickel ions are an example of heavy metals, poisonous even in relatively small concentrations in water and notoriously difficult to filter with current filtration methods. Therefore our filtration concept was tested against a concentration of nickel in water that far exceeds the safe limits. We have attempted to filter high amount of nickel (250 μM) through the cellulose filters grown by the G. Xylinus ATCC53582 strain (K1321305). The phytochelatin-dCBD fusion (K1321110) was coated on the surface of the cellulose to make a nickel specific functionalised ultra filtration membrane. To test the two membranes we have used coffee press. As a control measure we have also attempted to the filter the nickel solution through the cellulose that was not further functionalised. Obtained filtrates have been quantified and analysed using Mass Spectrometry.


Figure 7: Here shown is the decrease in the nickel content of the filtered nickel solution. The filtrate was obtained afer filtering with the phytochelatin-dCBD functionalised cellulose membrane. The concentration of unfiltered solution is provided as control.
Figure 8: Here shown is the difference in the decrease of the nickel content when filtered with cellulose compared with the functionalised cellulose with phytochelatin-dCBD.
Figure 9: This chart show the increase in nickel concentration with each sample, for the bacterial cellulose functionalised with PC+dCBD. The bacterial cellulose filtration showed consistent results with each sample.
Figure 10: This table summarises the readings from the IOP-EOS for the pre-filter Nickel solution, the filtrate from the bacterial cellulose only, as well as the filtrate from bacterial cellulose functionalised with PC+dCBD

From the results in figure 7, it can be seen that the cellulose functionalised with phytochelatin-dCBD (K1321110) is able to decrease the nickel content in the filtrate by 95%. This shows that the ultrafiltration method with functional protein domains on the cellulose is working and is a viable option for heavy metal removal from polluted water. Also the results observed in figure 8 shows that the non-functionalised cellulose is able to remove nickel from water, and the functionalisation of the cellulose further increases the incremental decrease of the nickel content in the filtrate by an approximate further half. The results presented suggest that bacterial cellulose is a good method for removing nickel from water and functionalisation of the cellulose provides a further decrease in the final concentration of the nickel filtrate.

This mechanism shows that the cellulose itself is able to remove contaminants from water, in this particular example Ni2+. It shows good ability to remove nickel ions, possibly due to nickel ion small size and hydrophillic nature. It is also shown that the functionalisation of the cellulose with the specific heavy metal binding chelating protein fusion with a cellulose binding domain provides an extra incremental decrease in the Ni2+ concentration found in the filtrate. Interestingly, a step-wise increase of the nickel concentrations can be seen with the functionalised cellulose filter, suggesting that the filter is getting saturated quite quickly. It is possible that depending on the nature of the contaminant, the incremental change provided by the functionalisation step can be increased. This can be of specific importance for the contaminants that cannot be caught by the inherent filtration ability of the cellulose material.


  • Tsukinoki, S (2008) Solid phase extraction of Cu, Ni and Co by EDTA type chelating cellulose. The Japan Society of Analytical Chemistry [Online] 57 (12) 1033-1038. Available from: [Accessed 17th October 2014]
  • Xu, Z (2002) Heavy metal removal by novel CBD-EC20 sorbents immobilized on cellulose. Biomacromolecules [Online] 3 (3) 462-5. Available from: [Accessed 17th October 2014]