In manufacturing bacterial-derived biologics, it is essential to eliminate live parental bacteria and free endotoxins before in vivo administration to minimize the risk to patients of infection and septic shock. Purification currently relies on complex combinations of physical and biological methodologies such as differential centrifugation, filtration, irradiation, antibiotic treatment, and heating. Currrent purification processes can reduce yields and reproducibility, inactivate the biologics produced, increase production costs, and lead to contamination with undesirable products such as free endotoxins. Ideally, contaminating bacteria could be killed by introducing a simple, inducible, genetic suicide mechanism that would eliminate many problems of conventional purification.

Bacterial suicide mechanisms based on conditional cell lysis (1,2,3) and nucleic acid degradation have been described previously (4,5). Although these systems are effective in killing bacteria, they are problematic for purification of biologics. Cell lysis releases free endotoxins that can cause septic shock, thereby compromising the safety of bacterial-based biologics. Although nuclease-based suicide systems do not cause cell lysis (4), restriction endonucleases (e.g., EcoRI) and nonspecific nucleases for killing bacteria can interfere with production of nucleic acid-based biologics such as plasmids and interference RNA (RNAi) molecules. An ideal suicide system would consist of a target-specific endonuclease that kills bacteria by introducing irreparable double-stranded DNA (dsDNA) breaks only within chromosomes to ensure minimal cell lysis and maximal killing eff iciency. It would need to be easily used in a range of bacterial host strains.



PRODUCT FOCUS: ALL PRODUCTS OF BACTERIAL EXPRESSION
PROCESS FOCUS: PRODUCTION
WHO SHOULD READ: PRODUCT AND PROCESS DEVELOPMENT
KEYWORDS: GENETIC ENGINEERING, CELL LINE DEVELOPMENT, ENDOTOXINS, FERMENTATION
LEVEL: ADVANCED

The I-CeuI homing endonuclease from the green alga Chlamydomonas moewusii (also known as C. eugametos) (6) is ideal for such a system. This endonuclease recognizes and creates a staggered dsDNA break in a naturally occurring sequence (7,8) within the highly conserved 23s rRNA in rrn operons (7,8,9,10,11,12,13, sites) on the chromosomes of many bacterial species used in biologics production. For example, Escherichia coli and Salmonella enterica each have seven I-CeuI recognition sites on their chromosomes (9,10). I-CeuI will not damage plasmid DNA within bacterial cells because the long recognition sequence limits its target to 23s rRNA genes. In addition, the I-CeuI–based suicide system is versatile not only because of its effective host range, but also because it is active in a wide range of pH, temperature, and salt concentrations (11).

In this study, we show that a thermoinducible I-CeuI–based suicide system can facilitate elimination of viable, contaminating bacterial cells in the manufacture of bacterial-based biologics. Because our commercialization goals are centered on the clinical-grade manufacturing of bacterial minicells (12), we have successfully incorporated and used this novel system into a panel of minicell-producing strains (13,14). As demonstrated here, we used the system to eliminate contaminating parent cells in processing clinical-grade minicell preparations. The same system also has broad applicability that would benefit many bacterial-based production schemes.

Materials and Methods



Culture Methods and Bacteria Strains: We routinely cultivated bacterial strains in Luria-Bertani (LB) medium supplemented with 0.2% D-glucose at 30 °C, providing aeration unless otherwise indicated. LB medium was supplemented with 20 µg/mL chloramphenicol (Cam), 50 µg/mL kanamycin (Kan), and/or 100 µg/mL ampicillin (Amp) when appropriate, as detailed below. We used noncarbon E (NCE) medium with 0.2% succinate and E medium containing 0.2% D-glucose and 0.05 mM thiamine as minimal media (15).

The thermoinducible I-CeuI suicide system was built into an antibiotic resistance marker–free, pir⁻, E. coli MG1655 derivative with a deletion in the closely linked lactose use and proline biosynthesis genes (▵lac-pro XIII). We constructed the background strain (VA X-6C3) by transducing the ▵lac-pro region from E. coli JMX Ac into E. coli MG1655 with P1vir transduction. Transductants that inherited the deletion were selected by demanding growth on medium containing 2 mM of the toxic lactose analog 2-nitrophenyl 1-thio-β-D -galactopyranoside (TONPG) from Sigma (www.sigmaaldrich.com) (16,17), then screened for proline auxotrophy. The proline-auxotrophic phenotype allows antibiotic resistance gene– free selection by demanding stable restoration of Pro⁺ in minimum media lacking proline.

The integration plasmid pVX-66 (Figure 1) containing the I-CeuI suicide system and minicell-producing construct was stably integrated into the attBλ site on the chromosome of VAX-6C3 (Pir⁻ and Pro⁻) to construct a stable, antibiotic resistance marker–free E. coli strain, VAX-8I3. Chromosomal integration of pVX-66 was mediated by λ integrase expressed from plasmid pKD16 as described elsewhere (18), and we confirmed construction of the desired strain using P1vir transduction to backcross with VAX-6G3 (pACTAK2 kan^R integrated at attBλ) (19). After transduction, we screened the kan^R transductants for proline auxotrophy and loss of the integrated pVX-66. Because pACTAK 2 has a pir-dependent R6K origin of replication (20), any unintegrated plasmid is unable to replicate in a pir⁻ strain.

Polymerase Chain Reaction (PCR) Primers: Table 1 lists PCR primers we used during strain construction. The table also shows restriction sites and an optimized version of the E. coli Shine-Dalgarno (SD) sequence and the E. coli trp operon transcriptional termination site engineered in primer sequences. We conducted PCR reactions with error-proof Pfx DNA polymerase from Invitrogen (www.invitrogen.com). All PCR products were blunt-end ligated into a pCR-Blunt II-TOPO cloning kit, also from Invitrogen, before subsequent restriction digestion and subcloning. Table 1: Oligonucleotide PCR Primers

Construction of Inducible ftsZ Minicell Production System: We PCR-amplified an optimized isopropyl β-D-1-thiogalactopyranoside (IPTG) inducible ftsZ-based minicell-producing cassette (T4, lacI^q, _Ptac⁻ ftsZ20) from genomic DNA of MPX1B9 using primers 1B1 and 1B2 (13). We then cloned a minicell-producing cassette with a natural ClaI site downstream of ftsZ into NotI-ClaI sites of pir-dependent pKD3 (21) following restriction digestion to construct pVX-37. Functional copies of the proB⁺A⁺ genes with native promoters were PCR-amplified using genomic DNA from E. coli strain MG1655 and primers 1B3 and 1B4. The PCR-amplified construct (SphI-proBA-KpnI) was digested and directionally subcloned into SphI-KpnI sites adjacent to attPλ in pLacZYattP (pir-dependent) to construct pVX-42. Next, pVX-37 and pVX-42 were digested with HindIII. We ligated together the fragment of pVX-37 containing the minicell-producing construct and a piece of pVX-42 with attPλ and proBA to create pVX-43 (pir-dependent and antibiotic resistance marker–free). Because HindIII cleaves R6K origin of replication in pVX-37 and pVX-42, we directionally cloned the minicell-producing construct into pVX-42 to restore the R6K origin of replication. Clones were selected for loss of proline auxotrophy and maintained using VA X-6C9 strain (Pir⁺ and Pro⁻).

Construction of Inducible I-CeuI Suicide System: We PCR-amplified the I-CeuI gene with primers 1B9 and 1C1, then cloned it into SalI-XbaI sites of pRHA-67 to express I-CeuI from the RhaRS-regulated pRHA promoter system, which can be induced with l-rhamnose (22). The rhamnose-inducible I-CeuI suicide system was then PCR-amplified with primers 1C1 and 1C2, and the PCR product was blunt-end ligated into a pCR-Blunt II-TOPO kit to construct pVX-55.

We constructed the integrated I-CeuI plasmid (pVX-66) as follows: After selection and validation of pVX-43, we PCR-amplified the I-CeuI endonuclease gene from purified C. moewusii whole-genome using primers 1G1 and 1G3. The PCR product (XhoI-SD-I-CeuI-BamHI) was restriction digested and cloned into XhoI-BamHI sites of pCGV1 to place I-CeuI expression under control of p_R and p_L dual promoters regulated by the phage λ cI857(Ts) (23,24). We then PCR-amplified the heat-inducible suicide gene construct — XbaI-cI857(Ts), λ p_R-p_L-I-CeuI-BamHI — using primers 1G4 and 1G5 and cloned it into XbaI-BamHI sites of pVX-43 to construct the pir-dependent plasmid pVX-66. Clones were selected for loss of proline auxotrophy on E-media at 30 °C and maintained using the VA X-6C9 strain. The PCR-amplified suicide cassette was also blunt-end ligated into a pCR-Blunt II-TOPO cloning kit to construct pVX-65.

TUNEL Assays: We used a terminal deoxynucleotidyl transferase (TdT)–mediated dUTP nick-end labeling (TUNEL) assay as described elsewhere (25,26) to detect I-CeuI– dependent dsDNA breaks in the E. coli chromosome. After thermoinducing I-CeuI, we harvested cells and fixed them in 4% paraformaldehyde, permeabilizing them in 0.1% Triton X-100 containing 0.1% sodium citrate. The cells were then treated with dUTP-fluorescein isothiocyanate (FITC) from Roche Applied Science (www.roche-applied-science.com) to enzymatically label 3′-OH DNA ends using TdT and counterstained with the nucleic acid–specific TOTO-3 iodine dye and the cell membrane–specific FM4-64 dye, both from Invitrogen.

We analyzed the TUNEL-labeled and counterstained cells using a f low cytometer with a BD FACSAria cell sorter from Becton-Dickinson (www.bdbiosciences.com) and examined a total of 50,000 events per sample. The total percentage of dsDNA-damaged (TUNEL-positive) cells was counted as a fraction of the membrane-enclosed (FM4-64 positive) or DNA-containing (TOTO-3 positive) cells. We treated noninduced controls (cultured continuously at 30 °C to suppress I-CeuI expression) identically to the I-CeuI induced cultures. And we substituted TOTO-3 dye with 4′,6-diamidiono-2-phenylindole (DAPI) from Sigma for fluorescent microscopic analysis.

Growth Inhibition Assays: An overnight culture of VA X-8I3 grown at 30 °C in 2 mL of LB was subcultured into 2 mL LB at a 1/200 dilution and grown to an optical density (OD₆₀₀) of 0.1 or 1.0 at 30 °C. At those OD₆₀₀ values, we added 20 µM IPTG to induce minicell production and raised the incubation temperature to 42 °C to activate expression of I-CeuI. Control cultures were grown continuously at 30 °C to suppress I-CeuI expression. We measured the number of corresponding colony forming units (CFU) by spot plating tenfold serial dilutions of each culture (0–10⁻⁶) onto LB agar with 0.2% D–glucose at various time points up to 24 hours after induction. Following incubation at 30 °C, total CFU were counted so we could calculate the CFU/mL of each original culture. We calculated relative viability by dividing CFU/mL of the suicide-activated cultures (grown at 42 °C) by that without activation (grown at 30 °C).

The growth inhibition experiment for E. coli strains with pVX-55 and pVX-65 was performed as described for the VAX-8I3 strain, but these strains were grown in LB-Kan. We induced I-CeuI in pVX-55 with _L-rhamnose (10 mM) at 37 °C.

Minicell Purification: For minicell production and purification, VAX-8I3 was cultured as described for growth inhibition assays, but we increased the culture volume to 200 mL LB. Minicells were harvested from primary cultures 24 hours after induction and purified by differential centrifugation and linear sucrose gradients as described elsewhere (12,14). We examined growth inhibition by counting the CFU at 24 hours postinduction and quantified minicells by measuring OD₆₀₀ and applying this equation (14):

To quantify contaminating viable parent cells, we plated 1010 minicells on LB agar with glucose and grew them at 30 °C overnight.

Statistical Analysis: Data are presented as the mean ± standard errors of the mean (SEM). We used a two-tailed, paired Student's t-test to calculate the statistical significance of differences between experimental groups.

Results



Using the I-CeuI endonuclease gene from the alga C. moewusii, we constructed an inducible suicide system capable of introducing irreparable chromosomal damage to a panel of Gram-negative bacteria, including an inducible minicell-producing E. coli strain. This suicide system consists of the C. moewusii I-CeuI suicide gene expressed from the phage λ p_R-p_L promoter system cloned into a plasmid-based integration construct, pVX-66 (Figure 1). The cI857(Ts) heat-sensitive repressor is tightly regulated, allowing high-level induction of I-CeuI expression at 42 °C and complete repression at 30 °C.

To determine the effect of I-CeuI expression on the viability of VAX-8I3, we compared viable populations from I-CeuI induced cultures with those of noninduced cultures by CFU analysis under a range of conditions (Figure 2). Activation of the chromosome-based I-CeuI suicide system immediately decreased the relative number of viable bacteria, and viable cells continued to decrease over a 24-hour period to a significant overall reduction (p<0.005) of ~10⁶-fold (Figure 2, TOP). All cultures tested for minicell production showed identical levels of parental cell growth inhibition, regardless of the culture volume (≤1.2 L, highest volume tested). Delaying activation of the suicide system until the culture reached an OD₆₀₀ of 1.0 was also bactericidal 24 hours after activation (Figure 2, TOP).

Based on the OD₆₀₀ readings, VAX-8I3 continued to grow after suicide system activation (Figure 2, BOTTOM). Although activation resulted in cell death, it did not cause cell lysis — as indicated by the OD₆₀₀ readings of suicide-activated cultures (Figure 2, bottom) and the presence of intact but uniformly filamented cells (Figure 3). In addition, I-CeuI induction alone led to filamented cells (Figure 3, MIDDLE).

When I-CeuI expression was induced from high–copy-number plasmid constructs, pVX-55 and pVX-65 (both with pUC origin), 10⁵- to 10⁶-fold reduction in viable cell populations occurred three to six hours after induction (Figure 2, MIDDLE). Thermoinduction of the I-CeuI from pVX-65 caused ~10 × more cell death than the L-rhamnose–induced I-CeuI expressed from pVX-55 at the postinduction time point of six hours. However, initial killing efficiency was higher with I-CeuI expression from pVX-55.

TUNEL assays verified that induction of I-CeuI caused dsDNA breaks in the chromosomal DNA (Figure 4). Fluorescent microscopic observations revealed that most cells had dsDNA damage, as represented by green fluorescent TUNEL-positive cells 12 hours after induction of the suicide system (Figure 4, TOP). By contrast, TUNEL-positive cells were rare in uninduced cultures (Figure 4, MIDDLE). We determined the proportion of bacteria with DNA damage (TUNEL labeled) among membrane-enclosed (FM4-64 positive) or DNA-containing (TOTO-3 positive) populations by flow cytometric analysis. The proportion of membrane-enclosed bacteria with dsDNA damage accumulated over time and became saturated by 12 hours postinduction (Figure 4, BOTTOM). The number of TUNEL-positive cells slightly increased in the bacteria that had not been induced for the suicide system (Figure 4, BOTTOM, black bars), which may indicate DNA damage caused by FtsZ overexpression during minicell production (27).

Because our goal is to develop methods to manufacture bacterial minicells as clinical-grade biologics, we used the thermoinducible suicide system to eliminate contaminating E. coli cells from minicells in batch production runs. To demonstrate the effectiveness of the I-CeuI–based suicide system for this process, we produced minicells with and without induction of the suicide system, then purified them by successive differential and sucrose density gradient centrifugation. We then compared minicell purity, determined by live parental cell contamination and yield.

Induction of the suicide system reduced the contaminating viable bacteria population by 10⁵-fold (p < 0.05), giving a purified preparation with a mean contamination level of <1 viable bacterium per 10¹⁰ minicells. When the I-CeuI suicide system was activated, minicell yields increased by sevenfold (p < 0.005) from 5.6 × 10⁷ ± 3.8 × 10⁶ minicells/mL in control cultures (inactivated) to 4.1 × 10⁸ ± 1.9 × 10⁷ minicells/mL in suicide-system active cultures. Larger-scale minicell production runs (≤1.2 L) gave similar minicell yields and contamination levels (data not shown).

Discussion



To eliminate viable parental E. coli from purified minicell preparations, we constructed a bacterial suicide system based on thermoinducible I-CeuI endonuclease. This system offers several genetic advantages: