Designed by: William Briand   Group: iGEM18_GO_Paris-Saclay   (2018-08-01)


Executive summary : This BioBrick is a full translation unit encoding CPG2, an enzyme that converts the anticancer drug methotrexate (MTX) into less toxic compounds. When expressed in E. coli, we have shown that it can rapidly remove MTX from the culture medium. This is proven conclusively by HPLC, and a bioassay that measures residual toxicity.

The Pseudomonas carboxypeptidase G2 (CPG2 or ‘glucarpidase’) is an hydrolase that cleaves the (poly)glutamate tail off folates and analogs, leaving behind a pteroate ring system[1]). There is broad substrate specificity, including endogenous folates, and notably the chemotherapeutic drug methotrexate (MTX). Its metabolite DAMPA (2,4-diamino-N10-methylpteroic acid) (Figure 1) has little antifolate activity in vitro, and norelevant clinical effects compared to MTX[2].

This has led to its use as both an antidote in case of MTX poisoning[3], and as a general purpose platform for novel drug delivery methods, where CPG22 would activate polyglutamated, soluble prodrugs in situ[4].

Figure 1 : The MTX biotransformation by Pseudomonas CPG2

Sequence and Features

Assembly Compatibility:
  • 10
  • 12
  • 21
  • 23
  • 25
    Illegal NgoMIV site found at 132
    Illegal NgoMIV site found at 820
  • 1000
    Illegal SapI.rc site found at 919

Expression vector design

Figure 2: Genetic organization of the folC-cpg2 encoding vector (BBa_K2688009)

We engineered a two-component pathway for efficient MTX degradation with folC and cpg2 under inducible ara (AraO2 BBa_R0081 + AraC BBa_R0080) promoter and a phage lambda pL promoter with lacO sites (BBa_R0011) lac, respectively. The expression of their respective repressors (AraC and LacI) by E. coli K12 strains allows the inducible expression in presence of arabinose (0.2 %) and IPTG (0,5 mM).

The CPG2 protein encoding sequence from Pseudomonas strain RS-16 was modified in two ways:

  • The periplasmic localisation signal was removed
    • EMBOSS software detects hydrophobic signal at amino acid 13->25 or 12->24
    • Decision was taken to remove AA 1 to 22 to achieve protein sequence identity with the commercially available CPG2 (drug INN glucarpidase)
  • The sequence was codon-optimzed for an optimal expression in E. coli (%GC, illegal restriction site and codon bias)

The synthetic cpg2 gene was cloned within pSB1C3, alone or associated the folC expression cassette (Figure 2).


Summary of methods: We aimed to evaluate the activity of our degradation pathway, by incubating a degrader strain containing cpg2 encoding plasmid in LB medium containing presence of 512 µM MTX. The culture supernatant was harvested and filtered before assessing the residual MTX either by a bioassay (bacteria particularly sensitive to MTX) or HPLC.

Results summary: This characterization was successful, with both the bioassay and HPLC reporting almost complete MTX degradation. After only 5 h of incubation with bacteria harbouring a cpg2 encoding plasmid, the MTX was almost completely removed from the medium (HPLC analysis).

Materials and methods

Strains and plasmids

Two categories of strains were used: degrader strains (‘chassis’ harbouring cpg2 encoding plasmid), and indicative strains with high MTX sensitivity, as biosensor for the biological effects of MTX. All strains were derived from E. coli K12 BW25113 (Δ(araD-araB)567, ΔlacZ4787(::rrnB-3), λ-, rph-1, Δ(rhaD-rhaB)568, hsdR514) parental strain and obtained from the Keio mutant collection[5].

For characterization purpose, two degrader strains were selected: WT and ΔtolC (Keio mutant JW5503). The latter has an intrinsic growth defect, but its lower MIC (minimum inhibitory concentration) for MTX[6] made it a possibly better degrader chassis. We used ΔacrA mutant (Keio JW0452) as an indicative strain since it is clearly MTX-sensitive.

We analysed MXT biotransformation by WT and ΔtolC strains harbouring pSB1C3-cpg2 (BBa_K2688003) or pSB1C3-folC-cpg2 (BBa_K2688009). Strains containing pSB1C3-tet (BBa_R0040) or pSB1C3-folC (BBa_K2688008) were used as controls.

MTX-biotransformation assays

The MTX degradation experiments were performed in LB medium containing 0.5 mM IPTG and 0.2% (m/v) arabinose supplemented with: i) 2% v/v DMSO (which is the solvent we used to prepare MTX stock solution) for the control condition, or ii) with512 µM MTX (kindly provided by Raphaël Labruère).

The degradation experiments were conducted in 15 mL Falcon tubes, with 1 mL of either control or MTX medium, inoculated with the strains at an initial OD of 0.2. The tubes were incubated for 5 to 20 h at 37°C in ambient air. Afterward, the cultures were centrifuged 15 min at 5400 g. The supernatants were filtered (0.22 µM) to eliminate cells and obtain an evaluable supernatant filtrate.

To assess the quantity of MTX remaining, two assays were conducted: HPLC and MTX bioassay.

MTX-detection bioassays

The MTX bioassay was a viability experiment using the ΔacrA indicative strain cultured in 96-wells plate with 50 µL of filtrate and 50 µL ΔacrA (OD 0.04 obtained from an overnight culture in LB).

This viability experiment was conducted in a 96-wells plate over 700 min in a Clariostar® plate reader (BMG LABtech GmbH), with orbital shaking, temperature control at 37°C and ambient air atmosphere, measuring OD600 every 20 min. The transparent plates were sealed with an adhesive transparent top. Duplicate or triplicate were performed for each condition.

HPLC analysis

The HPLC analysis was performed using a reverse phase C18 column. Detection was assessed by a UV spectrophotometer at 303 nm (peak absorption 302-303 nm[7]) with peak-triggered UV-vis spectrum (diode array). The eluting solvent was a gradient of acetonitrile and water (with 0.1% v/v HCOOH). This mode of operation is very similar to the assay and impurities protocols of reference pharmacopeia for MTX. [8][9]


We focused on two aspects to evaluate the success of our project: firstly, whether the degradation process removed from the medium the toxicity of MTX; and secondly whether MTX was actually biotransformed as predicted. While related, these two goals are not identical. Indeed MTX toxicity could be removed from the medium either by sequestration within chassis cells or by biotransformation. The HPLC analysis directly allows the direct detection of MTX biotransformation, allowing a distinction between these two scenarii.

MTX Bioassay

We compared the growth of the ΔacrA indicative strain in presence of supernatants from cultures of various chassis incubated with MTX.

Figure 3: Final opacimetry (OD600nm) reached by culture of the ΔacrA1 indicative strain incubated with various supernatants from cultures initially containing or not MTX and strains harbouring plasmid encoding cpg2. The chassis strains (WT or ΔtolC) are indicated in abscissa. The MTX was initially at 512 µM and incubated 20h with the chassis strains. The supernatants were 2-fold diluted during their inoculation in the 96-plate containing the indicative strain so that the maximal MTX concentration is 256 µM during the bioassay. The WT strain containing pSB1C3-tet was used as a negative control.

When pre-treatment of the medium is done using a control strain which does not contain cpg2 transgene (“WT pSB1C3-tet”), the growth of the indicative strain is greatly diminished in presence of MTX. This indicates that the medium contained a high concentration of residual MTX: it is expected for this control condition.

However, when considering the other pre-treatment conditions, we observed that the indicative strain culture reached the same growth yield regardless of the initial presence of MTX in the chassis culture medium. This means the strains harbouring cpg2 encoding plasmid have successfully removed MTX from the medium.


These results demonstrate the action of the degradation pathway on the MTX toxicity.

In terms of effectiveness of the various degrader strains, it seems that the co-expression of cpg2 and folC is associated to an better MTX removing from the medium only in the ΔtolC strain. Possibly, disabling the efflux pump mechanism allows a better intracellular accumulation of MTX following glutamylation by FolC.

Nonetheless all degrader strains produce similar results in our bioassay. It may be that some are faster than other, but we could not assessed this property since our earliest sampling (at 5 h) has been shown by HPLC to be “too late” (almost all the MTX is already gone from the medium at this time point).

Side notes

In the course of running our bioassay, we noticed that the filtrates from 5 h of incubation with the chassis strains were more toxic than either the initial MTX containing solution or the filtrates obtained after 20 h of incubation. This observation is puzzling since the HPLC assay reports the MTX to have been quasi-completely eliminated at the 5 h. Indeed the ΔacrA indicative strain is DAMPA-sensitive (data not shown) whereas its growth is not impacted by glutamate, the other degradation product (Figure 4). We hypothesize this effect may be related to availability of DAMPA in solution: it may crystallize out of solution slowly, unavailable for bacteria but still detectable by HPLC.

Figure 4 : Growth curve (OD600nm) of ΔacrA culture with varying concentration of glutamate

HPLC assay

While our bioassay results prove that there is no bacteriostatic activity left in the filtrate, this does not conclusively prove that the MTX has been bio-transformed. We endeavoured to show this by HPLC analysis, directly measuring MTX and metabolite concentration.


At the selected detection wavelength, there was no interference from either LB medium or DMSO (Figure 5).

Figure 5 : LB + DMSO 2% v/v (no MTX, no bacteria in original medium) after 5 h wait

In LB + DMSO, the MTX peak was well defined. The spectrum and retention time were similar to the MTX standard. In both cases, the retention time is 14.6 min (Figure 6).

Figure 6 : MTX in LB + DMSO 2% v/v after 20 h wait
Figure 7: UV-Vis spectra of peaks in figure 6. Below: spectrum of the main peak

The chassis had no intrinsic effect without the degradation pathway, and metabolic by-products of the culture did not cause interferences. Indeed, MTX-filtrate from control culture (E. coli WT strain harbouring pSB1C3-tet) exhibited a peak with the same retention time (14.6 min) as the MTX standard, and no other compound were detected (Figure 10).

Figure 10: Filtrate from the control culture (WT pSB1C3-tet) after 5 h incubation
Figure 11: UV-Vis spectra of peaks in figure 10. Below: spectrum of the main peak

‘MTX-Cleaning Factories’

When considering filtrates from WT chassis strain harbouring folC-cpg2 encoding plasmid culture incubated 5 h with MTX, the MTX peak (14.6 min) disappears almost completely and a new peak appears at 15.8 min (Figure 12).

Figure 12: Comparison of MTX standard, filtrate from control culture (WT pSB1C3-tet) and ‘MTX-Cleaning factory’ culture (WT pSB1C3-folC-cpg2) after 5 h incubation
Figure 13: UV-Vis spectra of peaks after degradation. Below: spectrum of the new peak (not MTX)

No further change is apparent after 20 h of incubation: the MTX-biotransformation is complete in at most 5 h.

Figure 14: Filtrate from ‘MTX-Cleaning factory’ culture (WT pSB1C3-folC-cpg2) after 20 h incubation

Quantitatively speaking, using the standard we can calculate the residual MTX concentration according to peak area, for this particular HPLC system. This allows us to calculate residual MTX concentration for the various conditions :

T--GO Paris-Saclay--9 hplc quant area.png
T--GO Paris-Saclay--10-hplc quant residual.png

We have also performed this HPLC analysis with a modified ‘MTX-Cleaning factory’ chassis ΔtolC (deletion of the outer membrane efflux channel TolC). The results obtained are similar to the WT chassis

Figure 15: Filtrate from ‘MTX-Cleaning factory’ culture (ΔtolC pSB1C3-folC-cpg2) after 5 h incubation

The entirety of our HPLC chromatograph are available here: []


Our HPLC assay protocol proved well suited to our purpose, considering the remarkable lack of interference from either LB components, metabolic by-products or DMSO at the chosen wavelength. This validates the method as an easy to use assay for MTX in bacterial cell culture.

Bacteria without cpg2 transgene do not biotransform MTX, excluding an intrinsic MTX-degradation ability of E. coli.

In contrast, the expression of CPG2 has a profound effect: the MTX peak almost completely disappears, giving way to a new peak, likely corresponding to a reaction product that might be DAMPA. This process is finished by 5 h, and no further change is seen at 20 h.

While the identity of this new peak has not been formally ascertained, it is plausible that it is DAMPA for a few reasons:

  • As a more hydrophobic molecule, DAMPA is expected to have a longer retention time in reverse phase HPLC, which is displayed by the unknown compound;
  • Given the pteridine cycle is the main chromophore of folate-like molecules, we expect both MTX and DAMPA to have similar UV-Vis spectra. The unknown compound has a very similar spectrum, with slightly sharper peaks;
  • For the same reason, we also expect the compounds to have a similar response factor, which is verified (relative response factor 0.8 to 1.0).

The MTX biotransformation by tolC deletion mutants is quite similar to that of the WTchassis strain. This modification may still have an effect on the transformation kinetic. However, because degradation is complete at the time of our first sample (5 h), our HPLC data can neither confirm nor rule out such effect.

New characterization by 2019 GO_Paris-Saclay team

Group and author of the new contribution: <a href=""> GO_Paris-Saclay </a> 2019 team added quantitative experimental characterization data to an existing Part from the Registry. Hugues Herrmann from the 2019 GO_Paris-Saclay team carried out the experiments.

Summary CPG2 activity has been previously demonstrated by HPLC analysis performed after 5 h and 20 h incubation of MTX with bacteria harbouring pSB1C3-cpg2. Since all the MTX was already gone from the medium after 5h, we wanted to gain new insights into the kinetics of MTX biotransformation by bacteria expressing cpg2. Here we report that within 1h bacteria expressing cpg2 are able to biotransform 70 % of the MTX present in the media into DAMPA. Incubation for 3h was enough to efficiently remove most of the MTX from LB media.

Materials and methods

E. coli BW25113 strains harbouring pSB1C3-cpg2 (BBa_K2688003) or pSB1C3-tet (BBa_R0040) were grown in LB containing 0.5 mM IPTG and 0.2 % (m/v) arabinose (same conditions than previously used by the first team that characterized BBa_K2688003). The biotransformation experiments were conducted in 15 mL Falcon tubes, with 1 mL of the same medium supplemented with 100 µM MTX, inoculated with the strains at an initial OD600nm of 0.2. The tubes were incubated under agitation at 37°C. Two hundred µl were harvested after 1 h and 3 h incubation with bacteria. Samples were filtered (0.2 µm) and submitted to HPLC analysis performed using a reverse phase C18 column as previously described (see the detailed protocol of HPLC analysis method above). Pure 100 µM MTX and DAMPA were used as controls for HPLC analysis.


Figure 16: Kinetic analysis of MTX biotransformation. Bacteria were incubated in LB medium containing 100 µM MTX (see the M&M for details). LB media without bacteria containing MTX or DAMPA was used as a control. DAMPA control was performed at the beginning of the experiment (time point 0). Results were normalized on the percentage of MTX control condition.

We observed that bacteria expressing cpg2 are able to remove efficiently 100 µM MTX from LB media in 3 h (Figure 16). We managed to catch intermediate conditions in which MTX is still detected. After 1 h, 70 % of the MTX molecules were biotransformed into DAMPA. Therefore, we could show that bacteria expressing cpg2 are able to remove rather rapidly (as little as 3 hours) 100 µM MTX for the media.

  1. Rowsell, S., Pauptit, R. A., Tucker, A. D., Melton, R. G., Blow, D. M., & Brick, P. (1997). Crystal structure of carboxypeptidase G2, a bacterial enzyme with applications in cancer therapy. Structure (London, England : 1993), 5(3), 337–347.
  2. Widemann, B. C., Sung, E., Anderson, L., Salzer, W. L., Balis, F. M., Monitjo, K. S., … Adamson, P. C. (2000). Pharmacokinetics and metabolism of the methotrexate metabolite 2, 4-diamino-N(10)-methylpteroic acid. The Journal of Pharmacology and Experimental Therapeutics, 294(3), 894–901.
  3. Widemann, B. C., Balis, F. M., Kim, A. R., Boron, M., Jayaprakash, N., Shalabi, A., … Adamson, P. C. (2010). Glucarpidase, leucovorin, and thymidine for high-dose methotrexate-induced renal dysfunction: Clinical and pharmacologic factors affecting outcome. Journal of Clinical Oncology, 28(25), 3979–3986.
  4. Masterson, L. A., Spanswick, V. J., Hartley, J. A., Begent, R. H., Howard, P. W., & Thurston, D. E. (2006). Synthesis and biological evaluation of novel pyrrolo[2,1-c][1,4]benzodiazepine prodrugs for use in antibody-directed enzyme prodrug therapy. Bioorganic & Medicinal Chemistry Letters, 16(2), 252–256.
  5. Baba, T., Ara, T., Hasegawa, M., Takai, Y., Okumura, Y., Baba, M., … Mori, H. (2006). Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: The Keio collection. Molecular Systems Biology, 2, 2006.0008.
  6. Kopytek, S. J., Dyer, J. C. D., Knapp, G. S., & Hu, J. C. (2000). Resistance to methotrexate due to AcrAB-dependent export from Escherichia coli. Antimicrobial Agents and Chemotherapy, 44(11), 3210–3212.
  7. Merck. (2007). The Merck Index: An Encyclopedia of Chemicals, Drugs, and Biologicals, 14th. Journal of the American Chemical Society, 129(7), 2197–2197.
  8. Council of Europe. European Pharmacopoeia. 6th ed. Strasbourg: Council of Europe; 2014. Methotrexate; p. 2735.
  9. United States Pharmacopeia and National Formulary (USP 30-NF 25). Vol 2. Rockville, MD: United States Pharmacopeial Convention; 2007:1553-1554.
ligandsfolates, folate analogues, various polyglutamated small molecules, various short peptides