Molecular characterization of the 3-phosphoglycerate kinase gene (PGK1) from the methylotrophic yeast Pichia pastoris

Strains, media and DNA procedures

The P. pastoris strains used in this study were GS115 (his4; Invitrogen) and SMD1168 (his4 pep4). Media for P. pastoris were described in the Pichia Expression kit (Invitrogen): MD (1.34% YNB, 4 × 10⁻⁵% biotin, 2% glucose), MDH (MD plus 2 × 10⁻³% histidine), MDHA (MDH plus 1% starch and 0.08% aspartic and glutamic acid buffer), MGH (1.34% YNB, 4 × 10⁻⁵% biotin, 1% glycerol and 2 × 10⁻³% histidine), MGHA (MGH plus 1% starch and 0.08% aspartic and glutamic acid buffer), MM (1.34% YNB, 4 × 10⁻⁵% biotin, 0.5% methanol), MMH (MM plus 2 × 10⁻³% histidine), MMHA (MMH 1% plus 1% starch and 0.08% aspartic and glutamic acid buffer), YPD (1% yeast extract, 2% peptone and 2% glucose) and YPDS (1% yeast extract, 2% peptone, 2% glucose, 1 M sorbitol).

Escherichia coli strains DH5α (F′ endA1 hsdR17 (r_K⁻ m_K⁺) glnV44 thi1 recA1 gyrA (Nal^r) relA1 Δ(lacIZYA-argF)U169 deoR (ϕ80dlac Δ(lacZ)M15) and XL1 Blue (F′::Tn10 proA⁺ B⁺lacI^q Δ(lacZ)M15/recA1 endA1 gyrA96 (Nal^r) thi hsdR17(r_K⁻ m_K⁺) gln V44 relA1 lac) were routinely used for cloning and plasmid manipulations. These strains were grown in LB medium (0.5% yeast extract, 1% peptone and 1% NaCl) supplied with appropriated antibiotics.

All molecular cloning techniques were carried out as described previously by Sambrook and Russel (2001). Restriction enzymes were obtained from New England Biolabs and Amersham Biosciences and used as detailed by the manufacturer. DNA sequencing analysis was performed on a MegaBace1000 automatic sequencer (Amersham Biosciences).

Cloning of the PGK1 gene by PCR

Based on sequence alignment of the PGK1 gene from the yeasts S. cerevisiae, K. lactis, Y. lipolytica, and Sz. pombe, we designed the degenerated primers C-PGK and PGK-F3 (Table 1), which were used in a PCR reaction consisting of: 10 ng genomic DNA from P. pastoris GS115, 10 pmoles each primer, 2.5 mM MgCl₂, 2 U Taq DNA Polymerase (Cenbiot, Brazil) in a final reaction volume of 50 µl. This system was submitted to 30 amplification cycles (30 s/94 °C, 30 s/50 °C and 30 s/72 °C), followed by an elongation cycle (5 min/72 °C). A 689 bp PCR product was cloned into pGEM-T vector (Invitrogen) and the sequence obtained was submitted to Blast analysis.

The 5′ sequences of the PGK1 gene were cloned by PCR using only primer PGK-R4 (Table 1) which anneals within a sequence present in the 689 bp fragment and upstream of it. The reaction conditions were as described above. The 3′-region of the PGK1 gene was obtained through a 3′-RACE experiment (Frohman et al., 1988). Briefly, total RNA from P. pastoris cells grown in MDH was isolated and the cDNA was synthesized using the Super Script Preamplification System Strand cDNA Synthesis kit (Life Technologies), using primer AP (Table 1) following the supplier's recommendations. PGK-F3 and AUAP primers (Table 1) were used to amplify an 800 bp fragment, which was cloned into pGME-T Easy vector (Invitrogen) following DNA sequencing.

Sequence analysis of the P. pastoris PGK1 gene

The nucleotide sequence obtained was analysed through the software programs PHRED (Ewing et al., 1998), PHRAP and CONSED (Gordon et al., 1998). The web interface of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/) was used to conduct the search for databank similarity by BLAST search tools. Multiple alignments were performed using the program ClustalW (http://www.ebi.ac.uk/clustalw). TATA-like elements prediction was made using the program HCtata (Hamming-Clustering Method for TATA Signal Prediction in Eukaryotic Genes) (http://125.itba.mi.cnr.it/∼webgene/wwwHC_tata.html). Putative transcription factor motifs were identified using the TFSEARCH (http://www.cbrc.jp/research/db/TFSEARCH.html) and MatInspector (http://www.genomatix.de/cgi-bin/matinspector/matinspector.pl) tools.

Southern and Northern Blot analysis

P. pastoris DNA isolation was performed as described by Burke et al. (2000). Approximately 10 µg genomic DNA digested with BamHI, BglII, EcoRI and PvuII were separated on 1% agarose gels and blotted onto nylon membranes. The P. pastoris PGK1 coding sequence, amplified by PCR using primers PGK-F8 and PGK-R9 (1.25 kb), was used as a probe (Table 1). The probe was labelled using the AlkPhos Direct kit (Amersham Biosciences). Hybridization (50 °C) and washing were performed as described by the suppliers. Signal detection was made with the CDP-Star reagent (Amersham Biosciences), according to the manufacturer's instructions.

Total RNA isolation from P. pastoris grown in MDH and MGH media was as described by Stateva et al. (1991). The RNA was analysed by electrophoresis in 1% agarose-formaldehyde gels and transferred to nitrocellulose filters, as described by Sambrook and Russel (2001). The 511 pb DNA probe containing the PGK1 gene fragment from P. pastoris was amplified by PCR with primers PGK-F3 and 3PGK-INT (Table 1), as described for PGK-F3 and C-PGK. Probe labelling, hybridization, washing and signal detection were carried out following the Southern blot protocol. Analyses of RNA abundance were performed using NIH image program and normalized against ribosomal RNA.

Construction of α-amylase expression vectors using P_AOX1 and P_PGK1

In order to express the B. subtilis α-amylase gene under the control of either the inducible AOX1 promoter (P_AOX1) or P_PGK1 from P. pastoris, vectors pAOXAMY and pPGKAMY were constructed, respectively (Figure 1). The B. subtilis α-amylase gene lacking the coding sequence for the C-terminal 33 amino acids residues of the enzyme was described previously (Moraes et al., 1995). This truncated α-amylase gene was obtained from the vector pGEM-T/αAMY (Amaral, 2003) by digestion with EcoRI and NotI and inserted into the P. pastoris expression vector pPICZαA (Invitrogen) digested with the same restriction enzymes. The resulting plasmid was named pAOXAMY.

Plasmid pPGKAMY was constructed by replacing P_AOX1 in pAOXAMY with P_PGK1 as follows: pAOXAMY was double-digested with BglII and BstBI, and ligated to a ∼2.0 kbP_PGK1 fragment digested with the same enzymes. This promoter fragment was obtained by PCR using primers PGK-FZ and PGK-RZ (Table 1), which were designed to introduce a BstBI site immediately upstream of the ATG translation initiation codon and a BglII site at the 5′ end of the P_PGK1 sequence, respectively. The resulting plasmid was named pPGKAMYB. Since P_PGK1 did not have suitable restriction sites for plasmid linearization, we decided to add sequences that would allow integration in the AOX1 locus. To do this, pPGKAMYB was digested with BglII and ligated to a ∼4.0 kb DNA fragment harbouring the HIS4 gene and 3′AOX1 sequence from pPIC9B digested with BamHI and BglII, creating plasmid pPGKAMY (pPIC9B was constructed in our laboratory by introducing a BamHI site downstream of the HIS4 gene in pPIC9; Invitrogen). Prior to transformation, plasmids pAOXAMY and pPGKAMY were digested with BglII and BglII/BamHI, respectively, prior to transformation in order to allow their integration at the AOX1 locus.

P. pastoris SMD1168 cells were transformed with the linearized vectors and α-amylase activity was observed through plate assay: pAOXAMY and pPGKAMY transformants were plated either in MDHA, MGHA or MMHA media for 48 h/30 °C and the plates were exposed to iodine vapour to reveal starch hydrolysis halos, as described by Moraes et al. (1995).

Isolation of the P. pastoris PGK1 gene

In order to clone the PGK1 gene from P. pastoris we used primers based on the sequence alignment of several yeast PGK1 sequences. Using P. pastoris GS115 genomic DNA as template, a PCR with primers PGK-F3 and C-PGK resulted in the amplification of a 694 bp fragment whose sequence showed high identity with the coding region of the PGK1 gene from different yeasts and filamentous fungi. This fragment corresponded to approximately 55% of the predicted PGK1 gene from P. pastoris (Figure 1). A PCR using only PGK-R4 primer amplified a 3.2 kb fragment, which added 495 bp to the coding sequence plus 2.0 kb to the PGK1 5′ upstream regulatory sequences. The 3′ end of the gene was obtained by a 3′ RACE experiment using the PGK-F3 primer, which resulted in the amplification of an 800 bp fragment. The sequence derived from this fragment was confirmed to correspond to the 3′ end of the PGK1 gene by DNA sequencing (Figure 2).

Sequence analysis of P. pastoris PGK1 gene

Analysis of the approximately 3.3 kb DNA sequences from the P. pastoris PGK1 gene revealed an open reading frame of 1251 bp with the potential to encode a polypeptide of 416 amino acid residues (Figure 2). Although the 3′ end of the gene was derived from cDNA, no introns were observed when the entire PGK1 sequence was obtained by PCR using genomic DNA as template (data not shown). Comparison of the nucleotide and deduced amino acid sequences of the PGK1 gene from P. pastoris and other yeasts showed an overall identity of ∼70% (Table 2).

The PGK1 flanking regions presented several features frequently found in other yeast genes. A TATA consensus sequence (TATAAA), which is responsible for correct transcription initiation of the S. cerevisiae PGK1 gene (Rathjen and Mellor, 1990), is present at position −71 relative to the ATG translation initiation codon. Another relevant motif present in the S. cerevisiae P_PGK1 is the Gcr1p target sequence. This sequence is essential to achieve high-level activation of glycolytic genes genes in S. cerevisiae (Baker, 1991; Zeng et al., 1997; Deminoff and Santangelo, 2001). Two putative target sequences for Gcr1p, which is composed by three repeats of the CTTCC motif separated by 10 nucleotides from the CAAG motif, were observed in the P_PGK1 from P. pastoris (Figure 2). In yeast, several glycolytic genes have their transcription increased in response to heat shock (Mager et al., 1995). In S. cerevisiae the levels of the PGK1 mRNA are enhanced by a 25–38 °C shift in growth conditions (Piper et al., 1986). This increase in mRNA level is mediated by a heat-shock element (HSE) present in the P_PGK1 (Piper et al., 1988). In the P. pastoris PGK1 promoter, a putative heat-shock element is also seen at positions −425 to −440 (Figure 2), suggesting that the P. pastoris PGK1 gene is also subjected to thermal regulation. In the downstream sequence, we detected the tripartite element 5′-TAG–TAGT–TTT-3′ (Figure 2) proposed by Zaret and Sherman (1982) as being important for transcription termination in S. cerevisiae. This sequence is observed at the 3′ end of many Y. lipolytica genes, including PGK1 (Dall et al., 1996).

The deduced amino acid sequence alignment of the PGK1 gene from P. pastoris, S. cerevisiae, K. lactis, Y. lipolytica, Sz. pombe and C. maltosa showed highly conserved amino acids (Figure 3). The yeast PGK protein folds into two domains, the N-domain and C-domain, which are composed by the residues 1–185 and 201–394, respectively (McPhillips et al., 1996). The remaining 415 amino acid residues constitute the interdomain linkage (Harlos et al., 1992; Figure 3). Crystallografic studies of porcine PGK confirmed arginines 22, 39, 66, 122 and 170 and histidines 63, 169 and 172 to constitute the binding site for 3-phosphoglycerate in the N-domain (Harlos et al., 1992) and are very similar in several yeast PGK proteins (Figure 3). Conserved residues K219, N336, D374 and T375 form the nucleotide-binding site at the hydrophobic cleft in the C-domain (Fleming and Littlechild, 1997) (Figure 3).

Codon usage in P. pastoris

The codon usage (CU) analysis of the P. pastoris PGK1 gene revealed a non-random codon choice, where 11 of the 61 possible amino acid codons are not used (Table 3). The nucleotide of choice in the third position is usually a pyrimidine (61%; Table 3), as was also observed in the PGK1 genes from P. citrinum (Nara et al., 1993) and Trichoderma reesei (Vanhanen et al., 1989). When PGK1 genes from P. pastoris, C. maltosa, Y. lypolitica, K. lactis, S. cerevisiae and Sz. pombe are compared according to CU, they appear very similar (Table 3). The CU of the P. pastoris highly expressed genes PGK1, GAP1 (Waterham et al., 1997) and AOX1 (Koutz et al., 1989) were compared and the biased codon preference stood out (Table 3). Moreover, it was shown that the PGK1, GAP1 and AOX1 genes contain 47%, 50% and 48% G + C, respectively.

PGK1 gene is represented by a singly copy in P. pastoris genome

In order to determine the number of copies of the PGK1 gene in the P. pastoris genome, a Southern blot experiment was performed (Figure 4). The probe was amplified by PCR and corresponded to the entire PGK1 coding sequence (1.25 kb). A single band was observed with genomic DNA digested with BamHI, BglII and EcoRI, and two bands with genomic DNA digested with PvuII (Figure 4A). The double signal obtained with PvuII-digested DNA can be explained by the presence of an internal site for this restriction enzyme in the coding sequence, which is depicted in Figure 4B. These results strongly suggest that, unlike T. reesei (Vanhanen et al., 1989) the PGK1 gene is represented by a single copy in P. pastoris.

Expression pattern of PGK1 gene

In order to examine the regulation of P. pastoris PGK1 gene expression, a Northern blot experiment was carried out. Total RNA from P. pastoris was obtained from glucose- and glycerol-grown cells (Figure 5A) and hybridized against a 511 bp DNA fragment of the P. pastoris PGK1 gene sequence. As shown in Figure 5B, a ∼1.5 kb transcript was detected, based on the relative position of rRNA. This size is in accordance to the predicted size of the mRNA encoded by the PGK1 gene. In S. cerevisiae and A. oryzae, low variation of the mRNA levels was observed when the carbon source was glucose or glycerol (Stanway et al., 1987; Nakajima et al., 2000). The same pattern was observed here, since the P. pastoris PGK mRNA is expressed approximately 2-fold more in glucose- than in glycerol-grown cells (Figure 5B).

Analyses of P_PGK1 activity by expression of α-amylase gene

To analyse P_PGK1 activity, P. pastoris cells transformed with either the pPGKAMY or the pAOXAMY vectors were grown on different carbon sources and the heterologous α-amylase activity assay was performed (Figure 1). As expected, when using glucose or glycerol as carbon sources, the pAOXAMY transformant did not express the B. subtilis α-amylase gene, whereas methanol-grown cells exhibited a starch hydrolysis halo (Figure 6). This occurs due to the repression/induction mechanism of P_AOX1 by the carbon source (Tschopp et al., 1987). In glucose- and glycerol-grown cells P_AOX1 is fully repressed, while in methanol-grown cells the promoter is maximally induced (Tschopp et al., 1987; Waterham et al., 1997).

When analysing expression patterns of glycolytic genes during S. cerevisiae exponential growth in glucose or lactate, several mRNAs showed different levels of induction in glucose, with regulation occurring at the transcription level (Moore et al., 1991). Moreover, in S. cerevisiae the presence of glucose resulted in glycolytic pathway acceleration, decreased respiratory activity, increased ribosome biogenesis and repression of alternative carbon source assimilation pathways (Yin et al., 2000). Therefore, promoter activity analysis of the P. pastoris PGK1 gene in different carbon sources is important for successful heterologous expression when using this promoter. The capability to secrete α-amylase of P. pastoris cells transformed with pPGKAMY, grown in either glucose, glycerol or methanol, was analysed (Figure 6). Cells grown in glucose exhibited the largest hydrolysis halo. This observation is consistent with the Northern blot results (Figure 5B), which showed higher levels of PGK expression in glucose-grown cells. This behaviour was also observed in S. cerevisiae, C. maltosa and A. oryzae, where the highest level of PGK1 gene expression was obtained with glycolytic substrates (Stanway et al., 1987; Masuda et al., 1994; Nakajima et al., 2000). In contrast, in A. nidulans and Y. lipolytica the PGK1 gene shows higher expression levels on gluconeogenic substrates than on glycolytic ones (Streatfield et al., 1992; Dall et al., 1996).

Similarly to the GAP1 promoter (Waterham et al., 1997), P_PGK1 from P. pastoris showed high efficiency of expression on the three different carbon sources analysed here. It is worth noting that the size of the α-amylase hydrolysis halos produced by the pPGKAMY transformants were larger than those produced by the pAOXAMY transformants (Figure 6). However, a more detailed promoter strength comparison analysis should be carried out, since the number of copies of the α-amylase gene integrated in the different transformants was not determined here. Nonetheless, based on the preliminary results presented in this work, the P_PGK1 promoter isolated in this work may be considered fully functional and constitutive, although its strength varies depending on the carbon source used. Therefore, the utilization of this promoter in the construction of new vectors for constitutive heterologous expression in P. pastoris should be considered.