As well as attempting to find a suitable detergent for solubilisation and stabilisation of the protein, the expression system was also further optimised to improve overall yield. This included the use of protein fusions which are widely used for protein stabilisation and solubilisation, such as maltose binding protein (MBP), Glutathione S-transferase (GST) or Promega’s Halo-tag (Los et al., 2008). A screen of the expression levels of different tags was performed at the Oxford Protein Production Facility UK5 (OPPF, Harwell Science Campus, UK) (Table 7). OPPF is a high- throughput cloning and protein production facility, focused on optimising expression of difficult proteins and scale-up of protein production. Except for the C-terminal His- tag, the 3C-GFP-tag and the Halo7-tag, all tags were N-terminal. All constructs were cloned into OPPF pOPIN vectors using InFusion cloning (Clontech) and transformed into E. coli strains C41, C43, Lemo21(DE3) or RosettaTM 2. The vector abbreviations for CNI and CNI-TEV-GFP expression in Table 7 correspond to the OPPF nomenclature on their website.
Immunity proteins are known to interact with the pore-forming domains of colicins (Bishop et al., 1985; Mankovich et al., 1986; Benedetti et al., 1991a), and so ColN P- domain was periplasmically co-expressed with CNI from a separate vector in case a CNI – ColN P-domain complex was more stable (Table 8). The vector abbreviations used for P-domain expression are shown in Table 8 and correspond to the OPPF nomenclature on their website. For co-expression of CNI fusions and P-domain from two different vectors in E. coli strains C41, C43 and Lemo21(DE3), the constructs where CNI was fused to the N- or C-terminal His-tag, N-His-GST or N-His-MBP were used.
EGFP-His and pWaldo-CNI-TEV-eGFP-6His, until then the most efficient construct, were used as controls. All screens were performed in two different media, Overnight Express™ Instant TB Medium (TBO6) and Power Broth™ (PB7
), in 3 ml cultures. PB cultures were grown at 37 °C, induced with 1 mM IPTG at OD600 nm = 0.5 and
incubated over night at 20 °C. TBO is an auto-induction medium, so cultures were 5 https://www.oppf.rc-harwell.ac.uk/OPPF/ 6 http://www.emdmillipore.com/life-science-research/overnight-express-instant-tb-medium/EMD_BIO- 71491/p_iYSb.s1O34EAAAEjBRp9.zLX 7 http://www.moleculardimensions.com/shopexd.asp?id=3455
grown at 37 °C during the day and 25 °C overnight. Cultures were harvested and the fusion proteins were purified using IMAC utilising the His-tag present on all constructs. Protein eluates were analysed qualitatively with SDS-PAGE. Ten gels in total were used for analysis; an example of this kind of gel is shown in Figure 30 (for other gels see appendix). All protein fusions which contained GFP and were examined separately using in-gel fluorescence at 510 nm. An example is shown in Figure 31 (for other gel see appendix). In-gel fluorescence identifies which constructs are overexpressed, the relative level of expression in comparison to other conditions and highlights protein degradation and some higher order structures. It also helps identifying proteins without a western blot. Constructs without additional solubilisation tags and fusion protein partners seem to break at the fusion point more easily than constructs with additional proteins. It is unclear why this is the case but possible that the bulk of protein is somehow sterically hindering proteases or, possibly, the three proteins are interacting and stabilising each other. ColN P-domain co-expression does not improve CNI-TEV-eGFP expression levels or stability.
Table 7: Protein and tag combination used during the overexpression screen, including the tag and protein fusion sizes. Vector names refer to nomenclature used by OPPF8. All were cloned into pOPIN vectors with spectinomycin (pOPINCD vectors) or ampicillin selection.
Vector name Tag Tag MW in Da
MW of fusions with CNI (20 431 Da) in
Da
Fusions with CNI- TEV-GFP (49 882 Da) pOPINCDE C-His 969 21 400 x pOPINCDF N-His 2 158 22 589 x pOPINS3C N-His-SUMO3C 13 213 33 644 63 095 pOPINMSYB N-His-MSYB 16 268 36 699 66 150 pOPINCDJ N-His-GST 27 954 48 385 77 836 pOPINE-3C-eGFP eGFP-6His-C 28 645 49 076 x pOPINE-3C-HALO HALO7-6His-C 35 343 55 774 x pOPINCDM N-His-MBP 42 711 63 142 92 593
Table 8: Periplasmic target sequences linked to ColN P-domain (in 23 Da) in pOPIN vectors with ampicillin selection for co-expression with CNI fusions in pOPINCD vectors. Vector names refer to nomenclature used by OPPF.
Vector Periplasmic target sequence
pOPINO Omp A SS (co-express with CD vectors) pOPINP PelB SS (co-express with CD vectors) pOPINDsbA Dsb A SS (co-express with CD vectors) pOPINMalE PelB SS (co-express with CD vectors)
8
Figure 30: Example of a coomassie-stained gel showing CNI protein fusions purified by IMAC from E. coli Lemo21 and C41 cells, grown in TBO media. Constructs showing increased expression of correctly sized protein (Table 7) are highlighted red in the upper legend and indicated by red arrows. Where ColN P-domain (23kDa) is co-expressed, the signal sequence is given. Although ColN P-domain is coexpressed (purple square), it does not improve CNI expression. The eGFP control is indicated with a green square. See appendix for full set of expression screen gels. Numbers at the bottom indicate lane number.
Figure 31: In-gel fluorescence of purified GFP-tagged CNI fusion proteins, expressed in TBO media. SDS-PAGE gel identifies which constructs are overexpressed, the relative level of expression in comparison to other conditions and highlights protein degradation and some higher order structures. CNI-TEV-GFP is the previously used reference construct. Constructs without additional solubilisation tags seem to break at the fusion point more easily than constructs with additional proteins. The broken off GFP-tag is around 28 kDa. Most purified proteins show dimerisation, which could indicate aggregation. Where ColN P-domain (23 kDa) is co-expressed, the signal sequence is given. ColN P- domain co-expression does not improve CNI-TEV-eGFP expression levels or stability. Numbers indicate lane number. See appendix for full set of expression screen gels.
The screen intended to find a combination which showed a considerable improvement of expression and construct stability as well as ease of subsequent fusion cleavage. Some constructs were certainly better than others, most notably Halo- and MBP-tagged fusions, while others failed to express at all. No single one condition stands out as considerably better than all the others. MBP and HALO7 linked constructs express consistently in both media and across all strains and so any of these could have been taken forward. Co-expression with P-domains does not seem to improve expression significantly. It may, however, improve CNI stability in solution but this was not further investigated. Based on the expression screen (Figure 30-Figure 31), I decided to take the pOPIN-CNI-3C-HALO7-6His construct further. It expressed consistently well across most conditions and did not show any detectable degradation products. The MBP-linked constructs showed some degradation products, which indicated that this fusion might not be as stable as the fusion with HALO7-6His. The HALO7-tag is a commercial and versatile tag, commonly used in protein immobilisation, purification and solubilisation (Los et al., 2008) and in these constructs it can be cleaved by the 3C viral protease (Hedhammar et al., 2006), also known under its commercial name PreScission Protease (GE Healthcare). A western blot against the polyhistidine-tag was carried out to confirm the identity of the purified protein-fusions and also to compare all strains and media used for CNI-HALO7-6His expression in the same gel (Figure 32). Figure 32A compared the expression of CNI- HALO7-6His (ca. 55 kDa) to CNI-TEV-eGFP-6His (ca. 48 kDa), the expression construct used prior to this screen. Arrows indicate the size where the purified protein fusion is expected. CNI-HALO7-6His is clearly expressed at a higher level than CNI- TEV-eGFP-His, when expression strains and media are the same, and so this screen has achieved signifcant improvement in expression level. The lower bands detected on the western blot indicate that some of the fusion protein has broken down, resulting in a band at 35 kDa, the size of the HALO7-tag alone. However, the Coomassie stained gel indicates that the breakdown is exaggerated by the western blot. The subsequent western blot compares the expression of CNI-HALO7-6His in different strains and media. A Coomassie-stained SDS-PAGE gel (Figure 32B) and the western blot (Figure 32C) show IMAC purified protein fusions. The western blot confirms the identity of the purified proteins but also highlights some degradation and aggregation and differences in expression levels between strains and media. Based on this western blot it was decided to select CNI-HALO7-6His expressed in the C41
strain and grown in TBO media for further scale-up in Newcastle and perform cleavage with 3C PreScission Protease.
Figure 32: The protein fusion CNI-3C-HALO7-6His has a higher expression level than the previously used fusion CNI-TEV-eGFP-6His. (A) Expression comparison between CNI-3C-HALO7-6His and CNI-TEV-eGFP-6His in Coomassie-stained SDS-PAGE gel. (B, C) The CNI-3C-HALO7-6His was overexpressed in different strains and media. The Coomassie stained gel (B) and the western blot (C) show IMAC purified protein fusions. The protein fusion product is expected to be ca. 55 kDa, the HALO7- tag alone at ca. 35 kDa. Lanes: 1 - 4: PB, 5-8: TBO. 1 & 5: Lemo21, 2 & 6: C41, 3 & 7: C43, 4 & 8: Rosetta 2.