Genetic Manipulation of Prochlorococcus Strain MIT9313: Green Fluorescent Protein Expression from an RSF1010 Plasmid and Tn5 Transposition

Abstract

Prochlorococcus is the smallest oxygenic phototroph yet described. It numerically dominates the phytoplankton community in the mid-latitude oceanic gyres, where it has an important role in the global carbon cycle. The complete genomes of several Prochlorococcus strains have been sequenced, revealing that nearly half of the genes in each genome are of unknown function. Genetic methods, such as reporter gene assays and tagged mutagenesis, are critical to unveiling the functions of these genes. Here, we describe conditions for the transfer of plasmid DNA into Prochlorococcus strain MIT9313 by interspecific conjugation with Escherichia coli. Following conjugation, E. coli bacteria were removed from the Prochlorococcus cultures by infection with E. coli phage T7. We applied these methods to show that an RSF1010-derived plasmid will replicate in Prochlorococcus strain MIT9313. When this plasmid was modified to contain green fluorescent protein, we detected its expression in Prochlorococcus by Western blotting and cellular fluorescence. Further, we applied these conjugation methods to show that a mini-Tn5 transposon will transpose in vivo in Prochlorococcus. These genetic advances provide a basis for future genetic studies with Prochlorococcus, a microbe of ecological importance in the world's oceans.

Prochlorococcus, a unicellular cyanobacterium, is globally distributed throughout the temperate ocean gyres. Densities of Prochlorococcus reach 700,000 cells ml⁻¹ of seawater (2), likely making it the most numerically abundant photosynthetic organism in the oceans (20). As a consequence of its global abundance, Prochlorococcus plays an important role in the global carbon cycle. For example, up to 79% of primary productivity in the North Atlantic is due to Prochlorococcus (11). The Prochlorococcus community is composed of a number of related strains (16, 26) that have different depth distributions in the ocean and thus occupy distinct ecological niches with respect to light and nutrients. The strain used in this study, Prochlorococcus strain MIT9313, is relatively most abundant near the base of the euphotic zone at depths around 100 m (9, 28, 30). In accordance with its depth distribution, MIT9313 grows at lower light levels than other strains (14), which are relatively more abundant in the surface waters. The nutrient physiology of MIT9313 also appears to be adapted to conditions deep in the euphotic zone. For example, MIT9313 has nitrite transport and reduction genes (21) and grows on this nitrogen source (15), whereas the high-light-level-adapted MED4 strain lacks genes for nitrite utilization (21) and does not grow on nitrite (15). The ability to grow on nitrite correlates with the depth distribution of the MIT9313 ecotype, since a well-defined nitrite maximum is often found at the base of the euphotic zone (18). Prochlorococcus strains thus have distinct nutrient physiologies that likely represent adaptations for exploitation of distinct niches in the ocean water column (reviewed in reference 7).

The global abundance of Prochlorococcus makes it an important system for the study of marine microbial ecology. The complete genome of Prochlorococcus strain MIT9313 (2.4 Mb with 2,328 genes) has been sequenced (21), along with other Prochlorococcus strains, such as SS120 (5) and MED4 (21) and the related oceanic Synechococcus strain WH8102 (19). Currently, nearly half of the MIT9313 genes are of unknown function, underlining the importance of the development of genetic methods for Prochlorococcus. Several aspects of Prochlorococcus biology have hindered the development of genetic tools in the past. While many Prochlorococcus strains are in culture, only three (MED4, MIT9313, and MIT9312) have been rendered free of contaminants and are thus suitable for genetic studies. Even under rigorously controlled culture conditions, Prochlorococcus grows more slowly (doubling times of 1 to 4 days) and to much lower densities than many other bacteria and cyanobacteria. At entry to stationary phase, Prochlorococcus cultures reach densities of 10⁸ cells ml⁻¹. Prochlorococcus strains grow either not at all or with low efficiencies on seawater agarose-based plates, and colonies require 6 weeks or more to appear. No Prochlorococcus plasmids have been isolated or identified during any of the genome projects; thus, it was uncertain whether the cell would sustain replication of foreign plasmids.

Our initial goal was to develop a protocol for DNA transfer to Prochlorococcus. To date, there is no evidence for natural competence or susceptibility to electroporation for Prochlorococcus. We thus focused on conjugation-based methods because of their high efficiency and insensitivity to species barriers. For example, conjugation has been used to efficiently transfer DNA from E. coli to many other cyanobacteria (29), including marine Synechococcus (1). Conjugation has even been extended to transfer DNA from E. coli to mammalian cells (27). We initially focused on the conjugal transfer of plasmids that might autonomously replicate in Prochlorococcus. Broad-host-range plasmids derived from RSF1010, such as pRL153, used in this study, have been shown to replicate in marine Synechococcus (1) and other cyanobacteria (13). We modified pRL153 to express a variant of green fluorescent protein (GFP) called GFPmut3.1, which is optimized for bacterial GFP expression. GFPmut3.1 expression was driven by the synthetic Ptrc promoter, which has been shown to be active in other cyanobacteria (17). We further applied these methods for conjugal transfer of foreign DNA to Prochlorococcus to show that Tn5 will transpose and integrate into the Prochlorococcus chromosome. Transposon mutagenesis has been used with other cyanobacteria to randomly inactivate gene function and study such processes as heterocyst formation (3). Recently, Tn5 has been shown to transpose in the marine cyanobacterium Synechococcus (12). In total, these data provide new opportunities to investigate Prochlorococcus genes in situ using reporter genes and tagged mutagenesis.

MATERIALS AND METHODS

Microbial growth conditions.

The microbial strains used in this study are listed in Table 1. Prochlorococcus strain MIT9313 was grown at 22°C in Pro99 medium (15) with a continuous photon flux of 10 μmol quanta m⁻² s⁻¹ from white fluorescent bulbs. Prochlorococcus strain MIT9313 grew under these conditions with a doubling time of 3.3 days (μ = 0.24 day⁻¹). Growth of cultures was monitored by chlorophyll fluorescence using a Turner fluorometer (excitation, 450 nm; absorbance, 680 nm). Cell concentrations of Prochlorococcus cultures were quantified using a FACSscan flow cytometer that was modified for the detection of picoplankton (6). Prochlorococcus was plated in seawater agarose pour plates (1). To prepare the plating medium, Pro99 liquid medium was supplemented with 0.5% UltraPure low-melting-point agarose (product 15517-014; Invitrogen Corp.) and heated to dissolve the agarose. The liquid agarose solution was then cooled to 28°C, at which time Prochlorococcus cells and antibiotic were added at the appropriate concentrations directly to the liquid agarose-Pro99 medium. This solution was then poured into the plates, where it solidified with cells embedded in the plating medium.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Description	Sourcea
E. coli
1100-2	mcrA− endA−; host for pRK24 and pRL153	Yale E. coli stock center
DH5α	Cloning strain used for all transformations	Invitrogen Corp., Carlsbad, CA
BW19851	Tra+ Tetr Smr Pir+; host for pRL27	B. Metcalf, Univ. of Illinois
Phage
E. coli phage T7	Phage to kill E. coli in Pro99 medium	D. Endy, MIT
Prochlorococcus
MIT9313	Conjugal recipient	Chisholm lab, MIT
Plasmids
pRL153	Kanr RSF1010 derivative	P. Wolk, MSU
pRK24	Tetr Ampr RP4 conjugal plasmid	D. Figurski, Columbia University
pJRC03	Cloning vector with Ptrc-driven GFPmut3.1	A. Van Oudenaarden, MIT
pRL27	Kanr; mini-Tn5 oriR6K	B. Metcalf, Univ. of Illinois
pRL153-GFP	Ptrc-driven GFPmut3.1 cloned into NheI site of pRL153	This study

^aUniv., University; MIT, Massachusetts, Institute of Technology; MSU, Michigan State University.

E. coli strains were grown in Luria-Bertani (LB) medium supplemented with ampicillin (150 μg ml⁻¹), kanamycin (50 μg ml⁻¹), or tetracycline (15 μg ml⁻¹) as appropriate at 37°C. Cultures were continuously shaken except for cultures expressing the RP4 conjugal pilus, which were not shaken to minimize the probability of shearing the conjugal pili.

Conjugation.

pRL153 was conjugally transferred to Prochlorococcus from the E. coli host 1100-2 containing the conjugal plasmid pRK24. pRL27 was transferred from the E. coli conjugal donor strain BW19851. E. coli was mated with Prochlorococcus strain MIT9313 using the following method. A 100-ml culture of the E. coli donor strain containing the transfer plasmid was grown to a mid-log-phase optical density at 600 nm (OD₆₀₀) of 0.7 to 0.8. Parallel matings were conducted under the same conditions using E. coli containing the appropriate transfer plasmid (pRL153 or pRL27) but lacking conjugal capabilities. These controls were included to confirm that the presence of E. coli containing the kanamycin-resistant transfer plasmid, in the absence of conjugation, was not sufficient for Prochlorococcus to grow under kanamycin selection. The E. coli cultures were centrifuged three times for 10 min at 3,000 × g to remove antibiotics from the medium. After the first two spins, the cell pellet was resuspended in 15 ml LB medium. After the third spin, the pellet was resuspended in 1 ml Pro99 medium for mating with Prochlorococcus.

A 100-ml culture of Prochlorococcus MIT9313 was grown to late log phase (10⁸ cell ml⁻¹). The culture was concentrated by centrifugation for 15 min at 9,000 × g and resuspended in 1 ml Pro99 medium. The concentrated E. coli and Prochlorococcus cells were then mixed at a 1:1 volume ratio and divided into aliquots as a series of 20-μl spots on HATF filters (product HATF08250; Millipore Corp.) on Pro99 plates containing 0.5% UltraPure agarose. The plates were transferred to 10 μmol quanta m⁻² s⁻¹ continuous white light at 22°C for 48 h to facilitate mating. The cells were resuspended from the filters in Pro99 medium and transferred to 25-ml cultures at an initial cell density of 5 × 10⁶ cells ml⁻¹. Kanamycin was added to the cultures after the Prochlorococcus cells had recovered from the mating procedure, such that the chlorophyll fluorescence of the culture had increased twofold. Fifty micrograms per milliliter kanamycin was added to cultures mated with pRL153, and 25 μg ml⁻¹ kanamycin was added to those mated with pRL27. After conjugation, cultures contained kanamycin at all times. The growth of the cultures was followed by measuring chlorophyll fluorescence.

Isolation of pure Prochlorococcus cultures after conjugation.

After the mated Prochlorococcus cultures were transferred to liquid medium, they were monitored by chlorophyll fluorescence until they grew under kanamycin selection in 35 to 50 days. When the mated cultures reached late log phase, cells were transferred to pour plates containing 25 μg ml⁻¹ kanamycin to isolate colonies. Colonies appeared in the pour plates 40 to 60 days after plating. Colonies were excised using a sterile spatula for transfer back to liquid medium containing either 50 μg ml⁻¹ (for pRL153) or 25 μg ml⁻¹ (for pRL27) kanamycin. When these cultures reached late log phase (30 days), a 100-μl aliquot of the culture was spread onto LB plates to determine titers of the remaining E. coli. Unfortunately, 10² to 10³ E. coli cells ml⁻¹ often remained viable in the MIT9313 cultures even after isolating MIT9313 colonies on pour plates. To eliminate the remaining E. coli, the MIT9313 cultures were infected with E. coli phage T7 (4, 25) at a multiplicity of infection of 10⁶ phage per E. coli host. The day after T7 infection, for a 100-μl aliquot of the Prochlorococcus culture, titers of E. coli were again determined on LB plates to confirm that no viable E. coli cells remained.

Plasmid isolation from Prochlorococcus.

After treatment with phage T7, cultures were transferred to fresh medium containing kanamycin and grown for 14 days to early stationary phase. Immediately before plasmid isolation, assays on LB plates were repeated to confirm that no residual E. coli remained. Plasmid DNA from MIT9313 cultures expressing pRL153 was then isolated from 5-ml stationary-phase cultures using a QIAGEN mini-prep spin column kit. Similar to the case with Synechococcus (1), the yield of pRL153 from Prochlorococcus was too low to visualize directly by gel electrophoresis. We thus transformed E. coli with the plasmids isolated from Prochlorococcus in order to compare the structure of pRL153 from MIT9313 to that of the original plasmid. pRL153 was isolated from kanamycin-resistant E. coli transformants and digested with the restriction endonucleases EcoRV and HindIII to compare its structure with that of the original plasmid. All restriction enzymes used in this study were purchased from New England Biolabs (Beverly, MA) and were used according to the manufacturer's instructions.

pRL153-GFP plasmid construction.

To determine if GFP expression could be detected with Prochlorococcus, pRL153 was modified to express GFPmut3.1 from the synthetic Ptrc promoter (Fig. 1). pRL153 contains unique sites for HindIII and NheI in the Tn5 fragment that are outside the kanamycin resistance gene. Ptrc-GFPmut3.1 was cloned into the unique NheI site to create pRL153-GFP. To this end, Ptrc-GFPmut3.1 was PCR amplified from pJRC03 using Pfu polymerase (Invitrogen Corp., Carlsbad, CA), and the following primers with 5′ NheI recognition sites: forward primer (Ptrc), 5′-ACGTAC-GCTAGC-CTGAAATGAGCTGTTGACAATT-3′; and reverse primer (GFPmut3.1), 5′-CGTACC-GCTAGC-TTATTTGTATAGTTCATCCATGC-3′. The Ptrc-GFP PCR product was then digested with NheI, treated with calf intestinal alkaline phosphatase (New England Biolabs), and ligated with NheI-digested pRL153. The ligation was transformed into E. coli DH5α, and an absence of point mutations in the Ptrc-GFP insertion was confirmed by DNA sequencing. GFP expression from pRL153-GFP in E. coli was visualized by epifluorescence microscopy.

FIG. 1.

Diagram of the RSF1010-derived plasmid pRL153, modified to express GFPmut3.1 from the Ptrc promoter. Unique restriction sites of pRL153-GFP are shown, along with the NheI sites used to clone the GFPmut3.1 gene and the HindIII and EcoRV sites used in the restriction analysis shown in Fig. 3.

GFP detection.

GFP expression was independently confirmed by Western blotting. Transconjugant MIT9313 cultures expressing pRL153-GFP, which had been treated with phage T7 as described above, were confirmed to contain no viable E. coli by LB plating assays directly before protein extraction. Total protein extracts from Prochlorococcus were made by centrifugation of 50 ml of late-log-phase cultures, resuspension in buffer (10 mM Tris-HCl [pH 8.0], 0.1% sodium dodecyl sulfate), and boiling at 95°C for 15 min. Protein concentrations were determined by Bradford assay. Ten micrograms of protein was added to each lane for resolution by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on a 4 to 15% Tris-HCl gradient gel (Bio-Rad Corp., Hercules, CA). Protein was transferred to a nitrocellulose membrane and blocked using 4% nonfat dry milk in phosphate-buffered saline (PBS) with 0.1% Tween 20. GFP was detected by incubation with rabbit polyclonal anti-GFP (Abcam Corp., Cambridge, United Kingdom) antisera diluted 1:5,000 in PBS with 0.1% Tween 20. Peroxidase-conjugated donkey antirabbit immunoglobulin G secondary antibody (Amersham Biosciences, Piscataway, NJ) was used at a dilution of 1:10,000. Chemiluminescence detection was achieved by incubation with the ECL reagent (Amersham Biosciences).

The fluorescence spectrum of transconjugant MIT9313 cultures containing pRL153-GFP was determined directly after they were treated with T7, transferred to fresh medium, and confirmed to contain no E. coli. GFPmut3.1 has maximal excitation and emission wavelengths of 501 nm and 511 nm, respectively. The fluorescence emission spectra of MIT9313 cells expressing pRL153-GFP and control cells of equal density expressing pRL153 were quantified using the Perkin-Elmer luminescence spectrometer LS50B. The cells were excited at 490 nm, and their cellular fluorescence was measured at 5-nm intervals from 510 to 700 nm. Cells from duplicate, independently mated MIT9313 cultures with GFP (+GFP) (pRL153-GFP) or without GFP (−GFP) (pRL153) were measured. We quantified fluorescence differences between +GFP cells and −GFP cells as the mean of the +GFP measurements minus the mean of the −GFP measurements.

Identification of transposon insertion sites in Prochlorococcus.

The Tn5 delivery vector pRL27 carries Tn5 transposase expressed from a broad-host-range tetA promoter from RP4 (10). The transposon itself contains a kanamycin resistance gene and the origin of replication from the plasmid R6K, which requires that the Pir protein be supplied in trans for the plasmid to replicate. Since Prochlorococcus lacks the pir gene, pRL27 does not replicate in the cell, and a stable insertion of the transposon into the genome is required for Prochlorococcus to be kanamycin resistant. Because the origin of replication is within the transposon cassette, transposon insertions along with the flanking genomic DNA can be cloned in pir⁺ E. coli and sequenced to determine the insertion site in the Prochlorococcus genome. After mating with pRL27, transconjugant Prochlorococcus cultures were grown in liquid Pro99 medium containing 25 μg ml⁻¹ kanamycin to late log phase. Genomic DNA was then isolated from 10 ml of culture using a QIAGEN DNeasy tissue kit (QIAGEN Corp., Valencia, CA). One microgram of genomic DNA was digested with BamHI. The genomic DNA was ethanol precipitated and religated using T4 DNA ligase (New England Biolabs) overnight at 16°C. Twenty nanograms of the ligated DNA was transformed into DH5α E. coli, and plasmids were isolated from 10 kanamycin-resistant E. coli transformants. EcoRI digestion of the plasmids revealed three distinct restriction patterns, which were sequenced using an outward-facing primer from within the Tn5 cassette (5′-AACAAGCCAGGGATGTAACG-3′) to determine the sites of insertion in the Prochlorococcus genome.

RESULTS

pRL153 replication in Prochlorococcus.

MIT9313 cultures mated with conjugal E. coli containing the plasmids pRK24 and pRL153 grew under kanamycin selection in liquid culture following a lag period of 35 to 50 days after conjugation (Fig. 2A). MIT9313 cultures mock mated with E. coli containing pRL153 but lacking the conjugal plasmid pRK24 did not grow (Fig. 2A). The lack of growth in these control cultures indicates that the presence of kanamycin-resistant, nondonor E. coli is not sufficient for Prochlorococcus to grow under kanamycin selection. Growth of Prochlorococcus under kanamycin selection is thus the direct result of conjugation with E. coli. No lag period preceding growth was observed when mated MIT9313 cultures were subsequently transferred again to fresh medium containing kanamycin (Fig. 2A), supporting that these cultures were fully kanamycin resistant.

FIG. 2.

Growth of MIT9313 cultures under kanamycin selection after conjugation with E. coli. A. MIT9313 cultures mated with E. coli containing the conjugal plasmid pRK24 and pRL153 (solid line and circles) grew under kanamycin selection. When the transconjugant cultures were transferred to fresh medium containing kanamycin, they grew at similar rates with no initial lag phase. Control MIT9313 cultures in mock matings with E. coli containing pRL153 but lacking pRK24 (dashed line and triangles) did not grow under kanamycin selection. B. MIT9313 cultures mated with the pir⁺ conjugal donor strain BW19851 containing pRL27 grew under kanamycin selection (solid line and circles). MIT9313 controls in mock matings with E. coli lacking conjugal capabilities but containing pRL27 did not grow (dashed line and triangles). Curves show averages for duplicate cultures; error bars show the range. The horizontal dotted line shows the minimum limit of detection of the fluorometer.

No kanamycin-resistant MIT9313 colonies were obtained when cells were plated directly after mating. We were, however, able to isolate colonies (plating efficiencies of 1 colony per 100 to 10,000 cells) in pour plates containing kanamycin after 40 to 60 days when the cells had previously grown in liquid medium for one transfer after mating. These plating efficiencies are similar to those of wild-type cells in the absence of kanamycin, suggesting that initially growing MIT9313 in liquid after mating may allow the cells to physiologically recover from the mating procedure such that they survive to form colonies.

It was not possible to use standard plating methods to calculate mating efficiencies directly, because Prochlorococcus colonies could be isolated only after the cells had first grown in liquid medium after mating. We thus estimated the conjugation efficiency using the following method. Chlorophyll fluorescence values from the log-phase cells shown in Fig. 2A were correlated to cell abundances using flow cytometry (see Methods). A linear regression correlating time to the number of transconjugant cells in culture was fit to the data points between days 25 and 55 of Fig. 2A (R = 0.044t + 4.82, where R is log₁₀[tranconjugant cells] and t is the number of days). We calculated the number of transconjugant cells immediately after mating as the intersection of the regression line with the ordinate axis. Using this value, we calculated the conjugation efficiency to be approximately 1% by dividing the initial number of transconjugants (6.9 × 10⁴ cells) by the number of cells initially transferred into the culture (6.5 × 10⁶ cells).

We found that 10² to 10³ E. coli cells ml⁻¹ persisted in the MIT9313 cultures even after the Prochlorococcus colonies had been excised from the pour plates and transferred back into the liquid medium. It was important to remove these E. coli cells, since they would have confounded experiments to isolate plasmids from Prochlorococcus. Physical separation of Prochlorococcus and E. coli by centrifugation or filtration proved ineffective. Residual E. coli was thus removed by infecting the cultures with E. coli phage T7. Infection with phage T7 had no adverse effects on Prochlorococcus viability, regardless of T7 multiplicity of infection. T7 phage infection may thus represent a general means of removing E. coli donor cells from a culture following interspecific conjugation. Plating assays after T7 treatment confirmed that no E. coli remained in the MIT9313 cultures.

Once we had obtained axenic Prochlorococcus cultures, we examined the structure of pRL153 in Prochlorococcus. The plasmid must autonomously replicate in Prochlorococcus without suffering structural rearrangements in order to stably express foreign proteins. We isolated plasmid DNA from MIT9313 cultures to compare the pRL153 structure from MIT9313 to that of the original plasmid. To this end, E. coli was transformed with plasmid DNA isolated from Prochlorococcus. We obtained approximately 100 E. coli transformants when DH5α cells competent to 10⁵ transformants μg⁻¹ DNA were transformed with one-fifth of a plasmid DNA prep from an MIT9313 culture of 5 × 10⁸ cells.

Restriction analysis of the rescued plasmid DNA by EcoRV/HindIII double digestion supports that the structure of pRL153 is generally conserved in Prochlorococcus (Fig. 3). In total, we examined the digestion patterns of 20 plasmids, 19 of which appeared identical to the original pRL153. The final plasmid (Fig. 3, lane 2) has an additional band of approximately 5 kb, and the smallest band is larger than in the other plasmid digests. Although it is not possible to determine if these changes to the plasmid occurred in E. coli prior to conjugal transfer or in Prochlorococcus, they illustrate that rearrangements may occur in pRL153 that retain the ability of the plasmid to replicate and to express kanamycin resistance in Prochlorococcus.

FIG. 3.

EcoRV/HindIII digestion of pRL153 plasmids isolated from MIT9313 cultures. The molecular marker is EcoRI/HindIII-digested phage λ DNA. Lane 1: pRL153 isolated from E. coli. Lanes 1 to 7: pRL153 derived from MIT9313 cultures. The structure of pRL153 was the same as that for E. coli in 19 of 20 total plasmid preparations from MIT9313. The restriction digest that differed from that of the original pRL153 is shown in lane 2.

GFP expression in Prochlorococcus.

pRL153 was modified to express GFPmut3.1 from the Ptrc promoter. The GFP protein was detected in mated Prochlorococcus MIT9313 cells by Western blotting. MIT9313 cells mated with pRL153-GFP expressed a protein recognized by the GFP antibody at the expected size of 27 kDa, whereas this band was absent in protein extracts of unmated Prochlorococcus (Fig. 4). We quantified GFP fluorescence in pure cultures of Prochlorococcus containing pRL153-GFP (+GFP) by comparing their fluorescence spectra to those of MIT9313 cells expressing pRL153 (−GFP cells) (Fig. 5). Emission at 680 nm corresponds to chlorophyll fluorescence. Both the +GFP and −GFP cells had the same emission at 680 nm, showing that both treatments had the same overall chlorophyll fluorescence. GFPmut3.1 has a maximum emission at 511 nm. The +GFP Prochlorococcus cells fluoresced significantly brighter, specifically in the wavelengths of GFP emission (Fig. 5). Further, the increased fluorescence of +GFP relative to −GFP Prochlorococcus mirrored the increased fluorescence in +GFP relative to −GFP E. coli cells (Fig. 5), supporting that the differences between +GFP and −GFP Prochlorococcus cells was the result of GFP expression. In addition, we examined individual Prochlorococcus cells expressing GFP by epifluorescence. Due to the small size (0.5 to 0.8 μm) and background due to chlorophyll autofluorescence, we were unable to visualize GFP in individual cells by epifluorescence.

FIG. 4.

Western blot comparing Prochlorococcus cells expressing GFP (+GFP) to −GFP Prochlorococcus controls. Prochlorococcus transconjugants express the GFP protein at the expected size of 27 kDa, whereas −GFP Prochlorococcus cells do not. Relative positions of bands from the protein molecular mass marker are shown to the left of the blot.

FIG. 5.

MIT9313 cells expressing GFPmut3.1 have a higher cellular fluorescence in the GFP emission spectrum (maximum emission, 511 nm) than cells lacking GFP. MIT9313 cells expressing pRL153-GFP and control cells lacking GFP were excited at 490 nm, and their fluorescence spectrum from 510 to 700 nm was measured. The dashed line shows the relative fluorescences of +GFP to −GFP E. coli cells, measured by the same method. The horizontal dotted line shows the zero line where the relative fluorescence of +GFP cells is equal to that of −GFP cells.

Tn5 transposition in Prochlorococcus.

MIT9313 cultures were mated with E. coli containing the Tn5-containing plasmid pRL27 using the same conjugation protocol as for the transfer of pRL153. Following conjugation, MIT9313 cultures mated with the E. coli conjugal donor strain BW19851 containing pRL27 grew under kanamycin selection (Fig. 2B). In contrast, there was no growth under kanamycin selection in MIT9313 cultures after mock matings with nondonor E. coli containing pRL27 (Fig. 2B). MIT9313 growth under kanamycin selection thus required the conjugal transfer of pRL27. DNA was extracted from the mated MIT9313 culture when they reached late log phase. Because the Tn5 cassette in pRL27 contains an origin of replication, we could clone and sequence the insertion sites of the transposon in the Prochlorococcus genome. In total, 10 plasmids were sequenced, which revealed 3 independent insertions in the MIT9313 genome. The most common transposon insertion (Fig. 6) represented 6 of the 10 clones sequenced.

FIG. 6.

Alignment of a cloned transposon insertion from MIT9313, the pRL27 plasmid, and the MIT9313 genome. The first 100 bp of the cloned insertion correspond to the Tn5 transposon cassette from pRL27. The remainder of the sequence shows the point of insertion of the transposon into the MIT9313 genome at bp 271016. This site is in a putative serine/threonine protein phosphatase, encoded by the gene PMT0236.

Sequencing of cloned Tn5 insertions revealed the sites at which the Tn5 transposon inserted into the MIT9313 genome. For example, the first 81 bp of the sequence shown in Fig. 6 corresponds to the Tn5 cassette, and the remainder matches the MIT9313 genome starting at bp 271016 of the MIT9313 genome sequence. The first 100 bp of the insertion site in the MIT9313 genome are shown in Fig. 6, but the sequencing revealed a total of 994 bp matching the MIT9313 genome. The only BLAST hit in the NCBI NR database for this cloned transposon insertion sequence is the MIT9313 genome. Specifically, this insertion occurred in the gene PMT0236, which encodes a putative serine/threonine protein phosphatase. The gene PMT0236 is a duplicated fragment of the gene PMT2127 (96% DNA sequence identity). This gene is likely part of a phage-derived fragment within the Prochlorococcus genome, since both PMT0236 and PMT2127 are located directly downstream of a gene for a typical phage integrase (genes PMT0234 and PMT2126). The other two cloned transposon sites similarly show fusions between the Tn5 transposon and the MIT9313 genome. These clones reveal insertions in the MIT9313 genes PMT1255 and PMT1538. PMT1255 has no other BLAST hits in the NCBI database. The insertion sequence in PMT1538 also has significant BLAST alignment with the Synechococcus sp. strain CC9605 gene Syncc9605_0516, which encodes a putative glutathione S-transferase.

DISCUSSION

The primary contribution of this paper is to describe the foundations of a genetic system for Prochlorococcus. We found conditions under which an interspecific conjugation system based on the RP4 plasmid family can be used to transfer DNA into Prochlorococcus MIT9313. pRL153, an RSF1010-derived plasmid, replicates autonomously in MIT9313, conferring resistance to kanamycin, and can be used to stably express foreign proteins, such as those for kanamycin resistance and for GFP. In addition, we found that Tn5 will transpose in vivo in Prochlorococcus. Once a liquid culture of kanamycin-resistant cells has been isolated, pour plating methods can be used to isolate individual colonies. These colonies can be transferred back to liquid medium for further characterization.

The methods described in this paper are highly repeatable; we isolated pure Prochlorococcus cultures containing pRL153 in four successive conjugations. The isolation of transgenic Prochlorococcus cultures does require a significant amount of time, however: 14 days to grow and mate the cells, 35 to 50 days to grow liquid transconjugant cultures after mating, 40 to 60 days to isolate transconjugant colonies on plates, 30 days to grow cultures isolated from colonies, and finally, 14 days for treatment with phage T7 and growth of pure cultures. Transgenic MIT9313 cultures can thus be isolated in 135 to 175 days (4.5 to 6 months). Experiments with Prochlorococcus are inevitably time consuming due its low growth rate, but there are other Prochlorococcus strains that grow faster than MIT9313. While MIT9313 grew with a doubling time of 3.3 days under the conditions in this study, Prochlorococcus strain MED4, another axenic strain, grows with a maximum growth rate of one doubling every 1.10 days (14). We avoided Prochlorococcus strain MED4 in these experiments because it is naturally resistant to kanamycin, even at 100 μg ml⁻¹ (data not shown). MED4 is highly sensitive to other antibiotics, however, such as 0.5 μg ml⁻¹ chloramphenicol, suggesting plasmids carrying chloramphenicol resistance may be appropriate for this strain. Prochlorococcus MIT9312, the third axenic strain, has a maximum growth rate of one doubling every 0.88 days (14) and thus may also be a good candidate for genetic manipulation. Improvement of growth on plates is an area for potentially increasing the efficiency of the isolation of Prochlorococcus mutants. MIT9313 colonies form on plates in 40 to 60 days with an efficiency of 1 colony per 100 to 10,000 cells. It is possible that yet-unidentified changes to the composition of the plating medium could improve the efficiency of MIT9313 colony formation. Further, future experiments to test the plating efficiencies of other Prochlorococcus strains may reveal strains that grow more efficiently on plates.

This study is the first report of GFP expression in oceanic cyanobacteria, which has a number of potential applications. The division cycle of cells in Prochlorococcus cultures synchronizes when entrained to a light/dark cycle (23), and global gene expression is controlled by a central oscillator, similar to the case with other cyanobacteria (reviewed in reference 8). Transcriptional fusions of Prochlorococcus promoters to GFP could be used to study the diel cycling in the expression of different genes in Prochlorococcus cultures. Although we were unable to quantify GFP fluorescence in individual cells, future studies using potentially stronger promoters or GFP variants with higher fluorescence (22) may make this possible. Direct subcellular localization of GFP expression is likely not feasible for Prochlorococcus, however, because the cell size (500 to 700 nm) approaches the lower limit of light microscopy as well as the wavelength of GFP fluorescence (maximum emission, ∼510 nm). Relative to other cyanobacteria, Prochlorococcus is a good candidate for studies to quantify GFP fluorescence on the whole-cell level. It does contain very low quantities of phycoerythrin, which in Prochlorococcus has a fluorescence maximum of 549 nm (and in some low-light-adapted strains, an additional maximum of 495 nm) (24, 24a). These maxima are close to that of GFP and could, in theory, be overlapping with it. However, the quantity of phycoerythrin in Prochlorococcus is so low that it is undetectable by direct spectroscopic measurement and thus does not occlude GFP fluorescence. In addition, GFP expression could provide a means to sort transgenic from nontransgenic cells by flow cytometry. Faced with variable and overall low plating efficiencies, flow sorting of cells is an attractive alternative in order to isolate mutants following conjugation. Alternatively, RSF1010-derived plasmids could be modified to cause Prochlorococcus to express other foreign proteins. For example, a tagged MIT9313 protein could be cloned into pRL153 and transferred into Prochlorococcus by conjugation. The ectopically expressed, tagged protein could then be immunoprecipitated to determine which proteins interact with it in vivo.

The Tn5 transposon from pRL27 can be conjugally transferred to Prochlorococcus as a means of making tagged mutations in the chromosome. Our results suggest that Prochlorococcus transconjugants do not survive to form colonies if they are plated directly after mating. Consequently, transconjugants are first grown under kanamycin selection as a liquid culture containing a diversity of transposon mutants. Because the liquid transconjugant culture represented a mixed population of transposon mutants, some competitively dominant mutants likely increased in relative abundance and were among those that we identified. These mutants were likely relatively abundant in the culture because they had transposon insertions in selectively neutral sites in the chromosome. These sites may be ideal for future studies seeking to insert exogenous DNA into selectively neutral sites in the Prochlorococcus chromosome. Collectively, the methods described in this study show that genetic methods including transposon mutagenesis are tractable for Prochlorococcus, thus providing a foundation for future genetic studies with this ecologically important microbe.