Introduction     Homologous recombination

Hydrogen production

E. coli naturally produces hydrogen through the mixed acid fermentation. Our goal in this part of the project is to increase the H2 production by modifying the carbon flow through the pathway of the mixed acid fermentation (MAF). The pathways of this fermentation are quite simple. Phosphoenolpyruvate (PEP), which is produced by the glycolysis, is transformed into pyruvate or into oxaloacetate. Those two molecules are then transformed into several products after a chain of reactions. Hydrogen is one of those products. It is produced from formate to regulate the intracellular pH of the bacteria [1].

Here is a simplified global overview of the mixed acid fermentation (fig.1):

Fig. 1 : Mixed acid fermentation pathway [2].

 Each reaction is controlled by one or several enzymes. By overexpressing or deleting specific genes of the MAF pathway, we expect to increase the hydrogen production. The part of MAF pathway in which we are interested is indicated by red lines (fig. 2).

Fig. 2: The interesting pathway for hydrogen production via the mixed acid fermentation.

All the reactions crossing the red lines should be avoided or limited, and the three main reactions which lead to hydrogen production should be increased. The hydrogen is produced from formate by the formate hydrogenase lyase system (FHL), an intracellular membrane-bound complex composed of the formate dehydrogenase (FDHH) and the hydrogenase 3 [1]. This complex catalyzes the following reaction:

Formate + H+ <=> H2 + CO2

Hydrogen is not excreted outside of the cell but it is immediately consumed by two uptake hydrogenases, the hydrogenase 1 and 2 [1][2]:

A menaquinone + H2 + 2H+ <=> A menaquinol + 2H+

The expression of the FHL system is activated at the level of transcription by the FhlA activator (formate hydrogen-lyase transcriptional activator). Expression of fhlA is positively regulated by FNR and repressed by HycA [1]. FNR is the main transcriptional regulator that mediates the transition between aerobic and anaerobic growth through the regulation of hundreds of genes [2]. All the regulations linked to the transformation of formate into hydrogen are shown here (fig.3):

Fig. 3 : Regulation of hydrogen production in E.coli [1].


MAF pathway modification: genes to delete

Several steps of the MAF pathway are irrelevant for hydrogen production (fig. 2). To inactivate these reactions, the genes encoding enzymes that catalyze these reactions have to be deleted using the technique described in the homologous recombination module. After exhaustive analyzes of the literature, we decided which genes we would overexpress and which we would disrupt. Here is the list of all the genes that we planned to delete. By using the KEIO Collection database, we checked that none of those genes were essential [4].

  • ppc

The ppc gene encodes the phosphoenolpyruvate carboxylase enzyme (Ppc). This enzyme catalyzes the following reaction [2]:

Phosphoenolpyruvate + CO2 + H2O <=> Oxaloacetate + Phosphate + H+

Deletion of the ppc gene will therefore prevent the PEP to be transformed into oxaloacetate which is not interesting for our application. As shown on figures 1 and 2, the left part of the MAF pathway does not lead to hydrogen production. Thus, deletion of ppc will inactivate this part of the pathway and all the PEP will be transformed into pyruvate.

  • ldhA

Pyruvate is transformed into (R)-lactate or into formate + acetyl-CoA. The second reaction is interesting for our application. The first reaction is mediated by the D-lactate dehydrogenase (LdhA) which catalyzes the following reaction:

NADH + Pyruvate + H+ <=> NAD+ + (R)-lactate

Deletion of ldhA should ensure that all the pyruvate is transformed into formate + acetyl-CoA.

  • focA

The focA gene encodes for a formate transporter located in the plasmic membrane. FocA transports formate outside the cell. It has been shown that disabling this transporter increases the intracellular formate concentration [5]. A higher formate intracellular concentration is obviously interesting for hydrogen production.

  • hyaB and hybC

To prevent hydrogen uptake by the two hydrogenases, HyaB and HybC, we have to inactivate them. Both hydrogenases-encoding genes are located in operons. The hya operon is composed of 6 genes (from hyaA to hyaF). The hyaB gene encodes the largest subunit of the hydrogenase [2]. This subunit is required for the hydrogenase to be functional [2]. Deleting this gene should inactivate the hydrogenase 1 [1].

The second hydrogenase-encoding gene is located in an operon composed of 7 genes (from hybA to hybG). The hybC gene encodes the largest subunit of the second hydrogenase. As for hyaB, deleting the gene encoding the largest subunit should inactivate it [1].


Materials and methods


  1. Gene deletion by the λ phage Red system
  2. In order to delete the five genes listed above, we used a method based on the λ phage Red system. As this method is described in great details in the ‘homologous recombination module’, we will briefly summarize it here. This method consists in the production of a gene-specific PCR fragment which contains an antibiotic resistance gene and site-specific recombination sites (FRT) flanking this gene. This PCR fragment is then electroporated in a strain containing the plasmid helper λ (PH λ) (see homologous recombination module). This plasmid encodes the genes of the Red system. The target gene is replaced by the resistance gene through homologous recombination using the Red system. We worked with two different antibiotic resistances: chloramphenicol (cm) and kanamicine (kan).
    The antibiotic resistance cassette is eventually removed by site-specific recombination using the Flp recombinase. This method leads to the ‘clean’ deletion of the gene of interest without any traces of antibiotic resistance gene. It is also adapted to delete genes in operons since the resistance-FRT cassettes used in this work also contain translation signal (RBS). This allows a proper expression of the genes downstream of the deletion (no polar effect). The primers used deleting the 5 genes of interest are listed in Table 1. 

  3. Phage P1 transduction
  4. P1vir stock :
    Overnight cultures of the donor strain are diluted 100 fold in 20 ml of LB medium supplemented with MgSO4 10-2M and of CaCl2 5.10-3M and grown at 37°C. At an optical density at 600nm of 0.2, 10µl of a P1 vir stock is added (approximately 109 pfu/ml). Once the bacteria are lysed, 1/20 (v/v) of chloroform is added and the culture is vigorously shacked. The mix is centrifuged at 4°C at 3,700 g and the supernatant is conserved at 4°C. 1/20 of the total volume of chloroform is added and vigorously shacked, then centrifuged again in the same conditions. The surnageant containing the P1 particles is stored at 4°C.
    P1vir transduction :
    Overnight culture of the recipient strain is pelleted and resuspended in the same volume of MgSO4 10-2M and of CaCl2 5.10-3M. To 100µl of the resuspended pellet, 100µl and 10µl of the P1 lysat is added. The mix is conserved 30 min at 37°C. 3 ml of agar 0.7% is added as well as 100µl of Na citrate 1M. The entire mix is plated on selective plates at 37°C for 16h. The candidates are streaked 3 times on selective plates and a PCR is then performed to confirm the transduction of the marker of interest. 




  • Deletion of the ldhA gene by homologous recombination

  • The amplification of the resistance cassettes with the primers specific to our 5 genes of interest by PCR took us a little more than a week. Because of the size of the primers used, we had to try several times the PCR and vary the parameters.
    Here is the picture of the electrophoresis of those PCR for the 5 genes:

    For each genes we tried to obtain both chloramphenicol (about 1100bp) and kanamicine (about 1500bp) resistance cassettes. From left to right we have the genes ldhA (1 and 2), focA (4 and 5), ppc (7 and 8), hybC (10 and 11) and hyaB (13 and 14). The sinks 3, 6, 9, 12 and 15 are the five negative controls.
            Once we obtained these cassettes, it took us about 1 month to obtain our first deletion candidate. The electroporation efficiency of the MG1655/PHl strain was not good. We obtained a lot of false-positive deletion candidates (i.e. clones growing on the selective medium). As explained in the homologous recombination module, we had to purify at least three times each candidate (this process takes therefore at least 3 days) before checking the deletion/insertion of the resistance cassette. Using primers complementary to the flanking regions of the genes of interest, PCR were performed on candidate colonies (data shown below). Unfortunately, the candidates were most of the time wild-type for the gene of interest. Our hypothesis is that the resistance cassette was inserted at another location in the genome. The reason for that is unknown since the genes of interest are not essential for E. coli viability, at least in the conditions we used. 
    After several weeks of unsuccessful attempts, we finally obtained the MG1655 DldhA::cm strain. Here is the picture of the electrophoresis where we tested different deletion candidates:

    The sinks we are interested in are the 4 first ones. The wild type version of the gene ldhA is in the third sink. The two first sinks are the deletion candidates for ldhA with the cm resistance cassette. The fourth sink is the negative control for this gene. For information, the other sinks are the tests for the 4 other genes: 5 and 6 are the deletion candidates for ppc; 9 and 10 are the deletion candidates for focA; 17 and 18 are the deletion candidates for hybC: 21 and 22 are the deletion candidates for hyaB. Unfortunately, the cm resistance cassette (about 1100 bp) and the gene ldhA (990 bp) have approximately the same length. In order to testify if we really had the MG1655 DldhA::cm strain, we did a test with a restriction enzyme which cut the gene ldhA.
    The result of this test is shown here:

    After the purification process, we had 4 candidates for the MG1655 DldhA::cm strain. For those 4 candidates, we did a PCR on the gene ldhA (not knowing if it really was ldhA or the cm resistance cassette that we amplified) and we digest the product of the PCR with an enzyme which cut ldhA. The digested gene ldhA is in the sink 7, the digested cm resistance cassette is in the sink 6. The four digested candidates are in the 4 first sinks. The result is therefore a success, because the candidates are not digested like ldhA. The wild-type version of the gene has been successfully replaced with the cm resistance cassette. We have our MG1655 DldhA::cm strain.

    • Construction of the MG1655 Dppc::kan, MG1655 DfocA::kan, MG1655 DhyaB::kan, MG1655 DldhA::cm Dppc::kan, MG1655 DldhA::cm DfocA::kan andMG1655 DldhA::cm DhyaB::kan mutant strains by P1 transduction

    By that time, we obtained the KEIO strains deleted for our 5 genes of interest. We were thus able to make a P1 stock on these strains (see materials and methods) and to transduce the Dppc::kan, DfocA::kan, DhyaB::kan and DhybC::kan markers in the MG1655 wild-type strain and in the MG1655 DldhA::cm strain. We obtained 6 mutant strains: the single MG1655 Dppc::kan, MG1655 DfocA::kan and MG1655 DhyaB::kan mutant strains as well as the double MG1655 DldhA::cm Dppc::kan, MG1655 DldhA::cm DfocA::kan andMG1655 DldhA::cm DhyaB::kan mutant strains.  These strains were checked by PCR. 
    The results are shown here:


    For each strain tested, we have: four candidates, one negative control and the wild type version of the gene that is supposed to have been deleted. If the amplified DNA segment of a candidate has the size of the reistance cassette instead of the size of the wild type version of the gene, the candidate is a mutant. Here are the results:

    • From 1 to 6 is the test for the MG1655 DhyaB::kan mutant: the 4 candidates are mutants.
    • From 8 to 13 is the test for the MG1655 DldhA::kan mutant: the candidates in the sinks 10 and 11 are mutants.
    • From 17 to 22 is the test for the gene ldhA of the double MG1655 DldhA::cm Dppc::kan mutant: the 4 candidates are mutants.
    • From 24 to 29 is the test for the gene ldhA of the double MG1655 DldhA::cm DfocA::kan mutant: the 4 candidates are mutants.
    • From 33 to 38 is the test for the gene focA of the double MG1655 DldhA::cm DfocA::kan mutant: the candidates in the sinks 35 and 36 are mutants.
    • From 40 to 45 is the test for the gene ppc of the double MG1655 DldhA::cm Dppc::kan mutant: the candidates in the sinks 40, 42 and 43 are mutants.

            We managed to obtain different double mutant strains, ready for another transduction. Unfortunately, due to the lack of time we were unable to continue our progression and we were also unable to characterize the double mutant strains for growth in different medium, at different temperature, …


    MAF pathway modification: Genes to overexpress


    In order to enhance hydrogen production, we proposed to over-express some genes involved in the hydrogen production. Three steps in the mixed acid fermentation pathway are of interest:

    • The transformation of phosphoenolpyruvate into pyruvate:

    Phosphoenolpyruvate + ADP + 2H+ <=> Pyruvate + ATP

    This reaction is present in other pathway. We thus decided not to improve it assuming it will not be the limiting step.

    • The transformation of pyruvate into formate:

    Pyruvate + Coenzyme A <=>  Formate + Acetyl-CoA

    This reaction is catalyzed by two pyruvate formate lyase: PflB and TdcE. The formate is the key molecule for hydrogen production and intracellular elevated formate concentration is needed for an efficient hydrogen production [6].

    • The transformation of formate into hydrogen:

    Formate + H+ <=>  CO2 + H2

    This reaction is under the control of the FHL complex (formate hydrogenase lyase composed of the formate dehydrogenase (FDHH) and the hydrogenase 3). As stated above, the FHL complex is regulated by FNR, a transcriptional regulator that mediates the transition between aerobic and anaerobic growth, and repressed by HyaC. It has been showed that FHL activity is not the limiting step for hydrogen production. We decided not to over-express the two enzymes of the FHL complex [3].
    On the basis of these three reactions, we decided to overexpress the following genes:

    • fnr

            fnr is quite an obvious choice for the first gene to overexpress. FNR is the main transcriptional regulator that mediates the transition from aerobic to anaerobic growth [2]. The concentration of the FNR protein is similar under aerobic and anaerobic conditions. The regulation occurs at the level of FNR activity: in the presence of O2, FNR is inactivated. Under anaerobic conditions, FNR is functional and activates expression of hundreds of genes involved in anaerobic metabolism. In these conditions, FNR represses genes involved in aerobic metabolism [2]. Over-expressing the fnr gene should increase FNR concentration and should thus enhance the anaerobic metabolism. The three reactions of the mixed acid fermentation pathway leading to hydrogen described above should therefore be increased.

    • pflB

            The pflB gene encodes one of the two pyruvate formate lyases which catalyze the transformation of pyruvate into formate [2]. This reaction is the most important one for hydrogen production. Over-expressing pflB should increase formate concentration and therefore hydrogen production. Although pyruvate is the substrate for many other reactions, high level of PflB should direct pyruvate to formate synthesis and therefore hydrogen production.

    • tdcE

    The tdcE gene encodes the second pyruvate formate lyase [2]. The reasons to overexpress this gene are the same as for pflB.

    Materials and methods

    To over-express these three genes of interest, we decided to insert them into plasmid pSB1C3 with a strong RBS, a constitutive promoter and a terminator. We decided to work in the BioBrick standard assembly 10 because it is compatible with pSB1C3. 
    First of all, we had to obtain the three genes to ligate them with the RBS, the promoter and the terminator. Unfortunately, the fnr gene contains an EcoR1 restriction site. Therefore, we had to introduce a silent mutation in the sequence of this restriction site and ordered the modified fnr on
    The tdcE gene was obtained by PCR using specific primers and the wild-type genome of E. coli.
    The pflB sequence contains a restriction enzyme present in the standard assembly 10: Pst1. We could have also introduced a silent mutation in this sequence and ordered it on However, due to the size of the pflb gene (2283 bp) and our limited budget, we were not able to afford it.
    The RBS, the promoter and the terminator that were selected are the same for the three genes and are described as being the ‘strongest’ ones received in spring 2010 distribution.
    The promoter is: BBa_J23100
    The RBS is: BBa_B0034
    The terminator is: BBa_B0024
    After obtaining the gene sequences either by PCR or by in vitro synthesis, the plan was to ligated them into the plasmid containing the RBS opened with Spe1 and Pst1 in order to clone the genes downstream of the RBS. The genes were also digested by Xba1 and Pst1. The second step was to ligate the part RBS + gene with the promoter. The third step consisted in extracting the construction promoter + RBS + gene and inserting upstream of the terminator. The last step was to extract the promoter+RBS+gene+terminator and insert it into the pSB1C3 plasmid.
    Unfortunately, due to lack of time, we were unable to do all these constructs.
    We managed to obtain the gene tdcE by PCR as we can see here:

    The gene tdcE is in the sink 3 and 5. The two sinks correspond to two different PCR programs. The sinks 2 and 4 are the negative controls for those PCR.
               We digested tdcE and inserted it in the pSB1C3 to send it as a BioBrick: BBa_K348000.

    Genes of the MAF pathway that we did not consider


    Several other genes of the MAF pathway could have been potential candidates for modifications (deletion or overexpression). However, after exhaustive literature readings, we did not consider them. 

    1. hycA
    2.         This gene encodes HycA, a regulator of the hydrogenase 3 in the FHL system [2]. This regulator limits the amount of hydrogenase 3 in the cell. Deleting hycA should therefore lead to an increase of the concentration of hydrogenase 3. Note that as stated above, the hydrogenase 3 catalyzes the transformation of formate into hydrogen. We did not consider this candidate since transformation of formate into hydrogen do not appear to be the limiting step for hydrogen production and normal concentration of hydrogenase 3 is fully adequate[3].

    3. aceE
    4.         This gene encodes a subunit of the pyruvate dehydrogenase complex. This complex catalyzes the following reaction [2]:
      Pyruvate + NAD+ + coenzyme A <=> Acetyl-CoA + CO2 + NADH
      For hydrogen production, transformation of pyruvate into formate is needed. The AceE enzyme catalyzes a variety of reactions, and deleting the corresponding gene reduces E. coli viability [1]. We therefore decided not to delete this gene. We focused on the transformation of pyruvate into formate and to overexpress the tdcE gene (see above).

    5. poxB
    6.         The poxB gene encodes a pyruvate oxydase, which catalyzes the transformation of pyruvate into acetate and CO2 [1]. It has been described that deleting poxB does not increase hydrogen production [1].

    7. focB
    8.         FocB is a formate transporter (the other formate transporter is FocA, see above). Unlike FocA, FocB is less efficient and less specific to formate. We could have deleted focB as well but deleting the two formate transporter would have reduced the growth rate[1].

    9. fdoG and fdnG

      fdoG encodes the formate dehydrogenase-O and fdnG the formate dehydrogenase-N. Those two formate dehydrogenase catalyze the transformation of formate into CO2 + H+. The formate dehydrogenase-O is active under aerobic conditions. We can therefore avoid deleting fdoG simply because our bacteria will be used under anaerobic conditions. The formate dehydrogenase-N needs a source of nitrate to be active. In the laboratory, it is easy to work on a medium which does not contain nitrogen and we can avoid deleting this gene. In reality, in a sewage treatment facility, this could be much more problematic. We can propose that our bacteria will be used to produce hydrogen after the nitrogen removal step. Deleting fdnG would be necessary if we want to produce hydrogen in a medium that contains nitrogen. Moreover, the formate dehydrogenase-N is a much more effective enzyme than the FHL system. Most of the formate will thus be consumed by the formate dehydrogenase-N instead of being transformed into hydrogen.

    Table 1

    All the sequences are listed in the directionality 5' -> 3'.
    For FocA, we didn't try to delete the entire gene because FocA is located next to pflB (one of the genes useful for hydrogen production, see above). The RBS and the promoter of pflB are included into the sequence of FocA. We decided to remove only the half of FocA, leaving the RBS and promoter of pflB untouched.



    [1] Maeda, T.; Sanchez-Torres, V.; K. Wood, Thomas. - Enhanced hydrogen production from glucose by metabolically engineered Escherichia coli - Appl Microbiol Biotechnol (2007) 77:879–890 – September 2007.
    [2] The database Ecocyc for the bacterium Escherichia coli K-12 MG1655, consulted online in July, August and September 2010:
    [3] Node, K.; Watanabe, M.; Makimoto, H.; Tomiyama, M. - Effect of hydrogenase 3 over-expression and disruption of nitrate reductase on fermamentive hydrogen production in Escherichia coli

    [4] The KEIO Collection consulted online in July, August and September 2010:

    [5] Suppmann, B.; Sawers, G. - Isolation and characterization of hypophosphite--resistant mutants of Escherichia coli: identification of the FocA protein, encoded by the pfl operon, as a putative formate transporter - Lehrstuhl für Mikrobiologie der Universität München, Germany. - Mol Microbiol. 1994 Mar;11(5):965-982. – March 1994.

    [6] Sawers, R.G. - Formate and its role in hydrogen production in Escherichia coli - Biochemical Society Transactions (2005) Volume 33, part 1 – Pages 42-46 – September 2004.

    Introduction     Homologous recombination