key: cord-1011542-ioxkqv31
authors: Esposito, Dominic; Gillette, William K.; Miller, David A.; Taylor, Troy E.; Frank, Peter H.; Hu, Renqui; Bekisz, Joseph; Hernandez, Jessica; Cregg, James M.; Zoon, Kathryn C.; Hartley, James L.
title: Gateway cloning is compatible with protein secretion from Pichia pastoris
date: 2005-01-05
journal: Protein Expr Purif
DOI: 10.1016/j.pep.2004.12.006
sha: a589104bbf112c6cde97ed650049fce68d04d1b2
doc_id: 1011542
cord_uid: ioxkqv31

Secretion of a recombinant protein from the yeast Pichia pastoris requires the presence of a signal peptide at the amino terminus. Maintaining the full amino acid sequence of the signal peptide is thought to be important for proper signal processing and protein secretion. We show that at least for one protein, a synthetic human interferon, the presence of a Gateway recombination site within the signal peptide is fully compatible with high levels of protein secretion. The amino termini of the secreted interferon proteins cloned with Gateway and cloned with restriction enzymes and ligase are identical, and the proteins were highly active in biological assays. Compatibility with Gateway cloning simplifies construction of plasmids directing secretion of recombinant proteins from P. pastoris.

Interferons (IFNs) are widely used as human therapeutics [1] . A new class of hybrid IFNs [2] is being investigated for possible human use against cancer and viral pathogens such as the coronavirus that causes severe acute respiratory syndrome (SARS). One of these, a chimeric human IFN-called Hy3 (GenBank AF085805), was insoluble when expressed in Escherichia coli (unpublished data), so the yeast Pichia pastoris was tested as an expression host.

Pichia pastoris is a widely used host for expressing recombinant proteins (http://faculty.kgi.edu/cregg/Pichia 2004.htm). Because it can attain very high cell densities in chemically deWned medium, and because it naturally secretes few proteins of its own, P. pastoris can be an eYcient producer of secreted proteins. Secretion of the protein of interest is often directed by fusing the gene to be expressed to the Saccharomyces cerevisiae mating factor (AMF) signal sequence [3] . The AMF signal sequence comprises the amino terminal 89 amino acids of the AMF protein. During secretion of the AMF from S. cerevisiae the signal peptide is removed by two processing enzymes. The enzyme Kex2 cleaves after a speciWc lysinearginine, then Ste13 clips oV the dipeptides EA and EA to yield the mature AMF [4] (Fig. 1 ). Homologous enzymes operate in P. pastoris. It is preferable to keep the amino acid sequence of the signal peptide unchanged if possible while fusing it to a gene to be expressed. The standard way to do this is to remove part of the cloning vector from the XhoI site 6 amino acids upstream of the processing sites to a point downstream, then to construct the XhoI fragment containing the gene of interest with a 5Ј

XhoI site and add a sequence encoding the missing AMF amino acids onto the gene during its construction, e.g., by adding a sequence encoding the missing amino acids in frame via PCR (Fig. 2) . These steps reconstitute the complete natural AMF signal peptide fused to the gene of interest, such that the activities of P. pastoris homologs of Kex2 and Ste13 should yield secreted protein with the desired amino terminal structure. Even with the full mating factor signal peptide, however, the extent of processing during secretion varies with the particular protein being expressed (for example, see [5, 6] ).

Executing this construction with restriction enzymes and ligase (REaL) is sometimes complicated by the presence of enzyme sites within the gene to be cloned. Spe-ciWcally, the native XhoI site within the AMF signal peptide is the only useful site for cloning protein genes fused to the signal peptide. If the gene of interest contains an internal XhoI site, as is the case with the Hy3 gene, it must be removed before cloning can proceed, or a diVerent cloning strategy (e.g., incorporation of a SalI site on the amino end of the gene of interest) must be used.

Gateway technology [7] is widely used for cloning genes for protein expression (http://web.ncifcrf.gov/rtp/ pel/pubs.asp). Gateway greatly simpliWes the planning and execution of cloning projects, but it requires the presence of Gateway recombination sites in expression vectors. Here we show that Gateway cloning is compatible with secretion of proteins from P. pastoris. This improvement simpliWes cloning for secretion from this yeast.

The P. pastoris secretion vector pPICZ (Invitrogen) was converted into a Gateway destination vector by cutting with XhoI, Wlling in the ends with Klenow polymerase, and ligating in the Gateway reading frame B cassette (Invitrogen). The new destination vector was named pDest-910.

An internal XhoI site in the Hy3 gene was removed by site-directed mutagenesis (QuickChange, Strategene). The product of this mutagenesis was used as the template for PCR ampliWcation. The Hy3 coding sequence was ampliWed with primers in two stages. For Wve cycles the reaction contained the amino primer GGCCTCG AGAAAAGAGAGGCTGAAGCTTGTGATCTGCC TCAGACCC (XhoI site underlined) and the carboxy primer GGGGACAACTTTGTACAAGAAAGTTGC GGCCGCTTATTCCTTCCTCCTTAATCTTTCTTG (NotI site underlined). After 5 cycles an adapter primer for the 5Ј side of the gene was added (GGGGAC AACTTTGTACAAAAAAGTTGGCCTCGAGAAAA GAGAGGCTG) and cycling was continued for 15 more cycles (Platinum Taq HiFidelity Supermix (Invitrogen), primers at 200 M). This PCR product contained modi-Wed attB sites, attB1.1, CAACTTT GTACAAAAAA GTTG, and attB2.1 CAACTTTGTACAAGAAAG TTG. These sites give about four times as many colonies in a Gateway BP reaction as the original attB1 and attB2 sites [8] . The PCR product was puriWed with QiaQuick PCR puriWcation columns (Qiagen). An aliquot of PCR product was digested with XhoI and NotI and ligated into pPICZ cut with the same enzymes (REaL expression clone). A second aliquot was cloned into pDonr223 (spectinomycin-resistant version of pDonr221, Invitrogen) by the Gateway BP reaction according to manufacturer's instructions (Invitrogen). The inserts in both REaL and Gateway clones were fully sequenced. Then the insert in the Gateway entry clone was transferred by recombina- tion into pDest-910 to make the Gateway expression plasmid. The Gateway and REaL expression plasmids were electroporated into P. pastoris strain X33 and selected on Zeocin (100 g/ml) plates (http://www.teknova.com) at 30°C (Invitrogen "EasySelect" manual, http://www. invitrogen.com/content/sfs/manuals/easyselect_man.pdf). Colonies were picked into YPDS medium, grown at 30°C, and glycerol stocks were made according to the EasySelect manual.

For P. pastoris fermentations, seed cultures of each strain (REaL or Gateway) were grown at 28°C in BMGY medium (EasySelect manual) in shake Xasks, and used to inoculate 1 L of basal salts medium containing 4% glycerol with trace mineral (PTM) salts (EasySelect manual) in New Brunswick BioFlow 110 minifermenters. Both fermentations were transitioned to methanol feeding, with a maximum of 12 ml methanol per L per hour. Aliquots were removed and centrifuged, and supernatants were applied to an SDS-PAGE gel, or diluted in PBS and spotted on nitrocellulose, next to aliquots of puriWed IFN-2c. The nitrocellulose membrane probed with rabbit anti-human IFN-2a (1:10,000 dilution in 2% dry milk), followed by HRP-conjugated goat anti-rabbit secondary antibody (Pierce). Chemiluminescence was developed using SuperSignal West Femto reagent (Pierce). Putative interferon proteins were enriched by ion-exchange chromatography, separated by SDS-PAGE, transferred to a PVDF membrane, stained with Coomassie brilliant blue R-250, and excised for N-terminal sequence determination by automated Edman degradation. Antiviral and antiproliferative assays of Hy3 activity were performed according to [9, 10] , respectively.

We sought to preserve the proteolytic processing sites of the yeast AMF signal peptide of pPICZ while installing the Gateway cassette just upstream of them. When the Gateway destination vector pDest-910 (Materials and methods) was recombined with a Gateway entry clone of the gene of interest that contained the six amino acids lost during pDest-910 construction, the resulting expression plasmid encoded the complete AMF signal peptide fused to the protein of interest, with 9 extra amino acids encoded by the Gateway attB1 site (Figs. 3 and 4) .

The hybrid human IFN-gene Hy3 was cloned into pPICZ with restriction enzymes XhoI and NotI and ligase, and into pDest-910 by Gateway recombination. For REaL cloning, it was necessary to remove an XhoI site within the Hy3 gene of the PCR template by sitedirected mutagenesis. Since both expression vectors had lost 23 bases of the mating factor signal peptide coding sequence as a result of the XhoI digestion, the Hy3 gene was PCR ampliWed with primers that added those 23 bases to the 5Ј end of the Hy3 coding sequence. For REaL cloning the PCR primers contained XhoI (5Ј) and Shaded codons in B derive from the attB1 site, and represent amino acids that are added to the signal peptide by conversion of the secretion vector to the Gateway system. The mature Hy3 sequence is italicized, CDLPQTƒ NotI (3Ј) sites. The PCR primers also contained attB sequences for cloning by Gateway recombination into the donor vector pDonr223, i.e., the same PCR product was used for both REaL and Gateway cloning. Inserts and junctions of both types of clones were fully sequenced, then the insert of the Gateway entry clone was transferred into pDest-910 by in vitro recombination. The base and amino acid sequences of the relevant regions of the Gateway and REaL clones are shown in Fig. 4 .

The two P. pastoris expression plasmids pHy3-REaL and pHy3-Gateway were electroporated into strain X-33 and selected on zeocin plates. Single colonies were picked into broth medium and frozen glycerol stocks were made. Supernatants from shake Xask cultures induced with methanol were spotted onto nitrocellulose and probed with anti-IFN-2a antibody (data not shown), which also detects Hy3; both cultures were positive for secreted Hy3. Both REaL and Gateway strains were then grown in 1 L cultures in minifermenters with pH and dissolved oxygen control. Following initial growth with glycerol as a carbon source, the cultures were transitioned to methanol feeding and grown for 90 h after the start of methanol induction (Wnal cell densities were approximately 0.5 kg wet weight per liter). Culture supernatants were harvested every 24 h and either applied directly to an SDS-PAGE gel that was subsequently stained with Coomassie blue (Fig. 5A ) or probed on a dot blot (Fig. 5B) .

The human IFN Hy3 produced by both the REaL and Gateway clones migrated identically on an SDS-PAGE gel. Both Hy3 and the IFN-2c standard migrated as doublets, with the Hy3 doublet migrating more slowly than the 2c doublet. Both the REaL and Gateway Hy3 proteins were partially puriWed and their N-terminal amino acid sequences were determined. The upper and lower bands of the Gateway and REaL Hy3 proteins (Fig. 5) had the same sequences, NH2-EAEACDL,, showing that while the Pichia homolog of Kex2 had cleaved at the expected site (Fig. 4) , the Ste13 homolog did not. IneYcient Ste13 processing of the AMF signal peptide in Pichia is common (for example, see [5, 6] ). Migration of IFNs as doublets on SDS-PAGE is presumably due to disulWde bond formation, since extensive heating with reducing agent considerably diminished the doublets (our unpublished observations).

The Hy3 protein is a synthesis of the amino 94 amino acids of IFN-2c and the carboxy 71 amino acids of IFN-21a [2] . We used puriWed 2c protein to estimate the concentrations of Hy3 produced by the Gateway and REaL strains at time points in our fermentation (Fig. 5) . The Gateway strain was more productive early in the run, but the REaL strain ended up secreting about as much protein. Comparison to the 2c standard on both the SDS-PAGE gel and the antibody dot blot suggested concentrations of secreted Hy3 protein of 100-150 mg per liter. Since neither of these expression strains was selected for multiple copies of the expressed gene, we assume that each contains one copy per P. pastoris genome, the most common number. The Gateway and REaL culture supernatants had similar speciWc antiviral activities, on the order of 10 7 international units/mg protein (approximately 10% pure). They also had similar antiproliferative activities, approximately 2 ng/ml inhibited cell proliferation by 50%.

These results show that the presence of amino acids in the yeast mating factor signal peptide corresponding to the Gateway attB1 site does not signiWcantly aVect the secretion of recombinant Hy3 from P. pastoris cells during fermenter growth. While other proteins may behave diVerently, the ease with which genes of any size or restriction map can be cloned for secretion via Gateway make this method an attractive alternative to standard cloning methods to attempt production of secreted proteins in P. pastoris.

Interferons: alpha, beta, omega and tau

Divergence of binding, signaling, and biological responses to recombinant human hybrid

Production of mouse epidermal growth factor in yeast: high-level secretion using Pichia pastoris strains containing multiple gene copies

Enzymes required for yeast prohormone processing

High-level expression of bovine beta-lactoglobulin in Pichia pastoris and characterization of its physical properties

Functional phytohemagglutinin (PHA) and Galanthus nivalis agglutinin (GNA) expressed in Pichia pastoris correct Nterminal processing and secretion of heterologous proteins expressed using the PHA-E signal peptide

DNA cloning using in vitro site-speciWc recombination

Concerted assembly and cloning of multiple DNA segments using in vitro site-speciWc recombination: functional analysis of multi-segment expression clones

PuriWcation and characterization of multiple components of human lymphoblastoid interferon

Evidence for multiple binding sites for several components of human lymphoblastoid interferon

We thank Young Kim