An Alternative Efficient Procedure for Purification of the Obligate Intracellular Fish Bacterial Pathogen Piscirickettsia salmonis

Abstract

Piscirickettsia salmonis is an obligate intracellular bacterial pathogen of salmonid fish and the etiological agent of the aggressive disease salmonid rickettsial syndrome. Today, this disease, also known as piscirickettsiosis, is the cause of high mortality in net pen-reared salmonids in southern Chile. Although the bacteria can be grown in tissue culture cells, genetic analysis of the organism has been hindered because of the difficulty in obtaining P. salmonis DNA free from contaminating host cell DNA. In this report, we describe a novel procedure to purify in vitro-grown bacteria with iodixanol as the substrate to run differential centrifugation gradients which, combined with DNase I digestion, yield enough pure bacteria to do DNA analysis. The efficiency of the purification procedure relies on two main issues: semiquantitative synchrony of the P. salmonis-infected Chinook salmon embryo (CHSE-214) tissue culture cells and low osmolarity of iodixanol to better resolve bacteria from the membranous structures of the host cell. This method resulted in the isolation of intact piscirickettsia organisms and removed salmon and mitochondrial DNA effectively, with only 1.0% contamination with the latter.

Piscirickettsia salmonis, the first rickettsia-like organism to be isolated from an aquatic poikylotherm, is the etiological agent of salmonid rickettsia septicemia an aggressive disease inflicting heavy losses to the Chilean salmonid aquaculture (3, 5, 8). Since the early 1970s several reports have documented infection of fish by rickettsia-like organisms in several latitudes, which means that the syndrome is slowly surpassing natural geographic boundaries (14; Ronald P. Hedrick, personal communication).

P. salmonis is an strict intracellular parasite which requires fish host cells to replicate, and as a consequence, it can be successfully grown within different established fish cell lines (9, 12). In general, most obligate intracellular species have been proven difficult to isolate and research progress on their molecular biology has been very slow (17). P. salmonis, as a recently reported rickettsia-like (9) is no exception, and its biology and behavior are poorly understood.

Cultivation of Piscirickettsia salmonis is inherently expensive and quite laborious. As examples, low yields are obtained in high-cost media; bacteria are incompletely released from infected cells; they compartmentalize and remain in close association with host-cell components; and more than half of the total infectious yield obtained is heavily contaminated with host cell debris (9).

Several techniques have been used to purify animal and human rickettsial (6, 15, 16). Particularly, density gradient centrifugation has been preferred to recover these organisms from in vitro-grown cells. Among them, Renografin and Percoll gradients have been the most efficient (2, 11). Nonetheless, when these techniques have been assayed with Piscirickettsia salmonis, the low bacterial yield obtained and the impurity of the samples with either nuclear or mitochondrial host-cell DNA contaminating, reliable molecular studies have been hampered. Indeed, a recent study (18) reports viable bacteria and in reasonable numbers, although nothing is said about the purity of the isolated DNA.

In here, we describe an alternative highly efficient semipreparative purification procedure to recover pure Piscirickettsia salmonis from an established fish cell line. The procedure relies on the high density and low osmolarity of iodixanol, an alternative iodinated compound to Renografin, which was used to generate isoosmotic density gradient centrifugation. A single band with high purity of viable bacteria was obtained, and its specificity confirmed by immunofluorescence microscopy and semiquantitative PCR analysis. At present, we are in the process of optimizing the procedure to make it applicable to bacterial purification from naturally in vivo infected fish organs.

MATERIALS AND METHODS

Bacterial strain and experimental growth conditions.

The type strain of P. salmonis LF-89 (ATCC VR-1361) was continuously propagated in the Chinook salmon embryo cell line CHSE-214 (ATCC CRL-1681) with antibiotic-free Eagle's medium as described (8, 12). Monolayers of CHSE-214 cells were grown at 17°C in sealed culture flasks containing minimal essential medium, supplemented with 7.5% heat-inactivated fetal bovine serum and buffered to pH 7.2 with 10 mM sodium bicarbonate and 15 mM HEPES. Escherichia coli JM109 was routinely propagated at 37°C in Luria broth.

Purification of P. salmonis.

CHSE-214 monolayer cells grown on eight plastic culture triple flasks (3.600 cm²) were infected with P. salmonis (8) and incubated at 17°C for 8 to 12 days, when 80 to 90% cytopathic effect was reached. Cell culture supernatants were collected and centrifuged at 200 × g for 10 min at 4°C to pellet large host cell debris. This new and partially clean supernatant was centrifuged at 10,000 × g for 45 min at 4°C to collect bacteria. The pellet was resuspended in 6 ml of Tris-NaCl buffer (10 mM Tris-Cl, 1 mM EDTA, 150 mM NaCl, 12 mM MnCl_2, pH 7.6) and incubated with 20 U of DNase I (Boehringer Mannheim) at 30°C for 60 min. Enzymatic activity of DNase I was inhibited by addition of 1/10 volume of 0.2 M EDTA.

Density gradient centrifugation.

Solutions of 26, 24, and 22% Optiprep (iodixanol; Nycomed, Oslo, Norway), an iodinated compound (7, 10) were prepared in buffer TEN (10 mM Tris-Cl, 1 mM EDTA, 150 mM NaCl, pH 7.6) from a 40% Optiprep stock master solution. A three-step discontinuous gradient (22%, 24%, and 26%) was set into clear ultracentrifuge tubes (16 by 102 mm) to form a continuous gradient within 60 min at room temperature. Sample suspensions were layered on top of two parallel gradients and centrifuged at 25,000 × g for 3 h at 4°C in an SW28 swinging-bucket rotor (Beckman Instruments). Bacterial bands were collected from each gradient and diluted 10-fold in TEN buffer and centrifuged at 11,000 × g for 30 min at 4°C. The resulting pellet was suspended in TEN buffer.

Direct immunofluorescence staining.

P. salmonis thin smears were fixed with 3% paraformaldehyde for 15 min and permeabilized with 0.1% Triton X-100 for 10 min, followed by three washes with phosphate-buffered saline alone and a fourth with PBSA (phosphate-buffered saline plus 0.5% bovine serum albumin). Bacteria were labeled in PBSA buffer for 1 h with a 1:75 dilution of a commercially available rabbit fluorescein isothiocyanate (FITC)-conjugated anti-P. salmonis oligoclonal antibody (SRS Immuno Test; BIOSChile) and/or with a 1:50 dilution of a mouse FITC-conjugated anti-P. salmonis (iodixanol purified) antibody prepared in our laboratory. All procedures were carried out at room temperature in the dark. Coverslips were mounted onto glass slides with fluorescent mounting medium (Dako Corporation) and observed on a Zeiss LSM laser-scanning confocal microscope.

DNA isolation.

Bacterial genomic DNA was purified from banded bacteria and from Escherichia coli by the cetyltrimethylammonium bromide-NaCl procedure (1). Preparation of eukaryotic genomic DNA from a CHSE-214 monolayer culture was done by the tissue culture cell procedure in the presence of proteinase K (1). DNA concentrations were determined in a GeneQuant DNA/RNA calculator (Pharmacia Biotech).

Semiquantitative PCR amplification.

PCR amplification was performed by a serial dilution experiment in a total reaction volume of 10 μl. P. salmonis DNA was amplified with the specific primer pair RTS1 and RTS4, complementary to the intergenic transcribed spacer (ITS) region as described previously (13).

In order to estimate mitochondrial DNA contamination, primer pair P2 and Sphe, directed to the D-loop of the fish mitochondrial genome, were used (C. Orrego, unpublished data). All amplifications were performed in a Techne Cyclogene model FPHC3HT thermal cycler, and conditions were set in accordance with the annealing temperatures of the primer pair used. Amplicons were analyzed by agarose gel electrophoresis and visualized after ethidium bromide staining in a Gel Doc 1000 (Bio-Rad).

RESULTS AND DISCUSSION

Because of the rather recent appearance of Piscirickettsia salmonis as a highly aggressive obligated intracellular pathogen of salmonid fish as well as the intrinsic difficulties in purifying it to homogeneity, the molecular definition of the agent is still very limited. P. salmonis is indeed difficult to grow to high yields and very difficult to separate from host cell components due to the absolute compartmentalization into vacuoles that the replicating bacteria performs inside host cells. In spite of a number of attempts and the use of different strategies to isolate the agent, most of them continue to be unsuccessful.

Here, we describe a simple, straightforward, and effective procedure to separate, concentrate, and purify P. salmonis from its host. The procedure is based on differential sedimentation (200 × g, 10 min) followed by low-osmolarity, high-density gradient centrifugation on iodixanol (22 to 24 to 26%), as the resolving medium.

Infected supernatants from eight plastic culture triple flasks of CHSE-214-infected cells (1.1 × 10⁸) were collected at 8 to 12 days postinfection, sedimented, and treated with DNase I to minimize host cell contamination, since infected cells remained as a monolayer. The effectiveness of the gradient was demonstrated by the absolute separation of the bacteria from host cell contaminants.

Figure 1 shows the whitish single band containing bacteria, consistently found in the upper third of the gradient corresponding to a buoyant density of 1.13 g ml⁻¹. The gradient also resolved a diffuse, barely visible high-density band in the lower third (1.14 g ml⁻¹), which mostly corresponded to cell mitochondria and microsomal debris. An aliquot of the upper band obtained was observed under the microscope, confirming the presence of Piscirickettsia salmonis when exposed to a commercial rabbit FITC-conjugated anti-P. salmonis antibody (Fig. 2). Next, and in order to confirm these observations, parallel samples were assayed by semiquantitative PCR amplification. Figure 3 shows the successful amplification of the ITS region of the bacterial rDNA operon (13), confirming the identity of the material banded as Piscirickettsia salmonis.

FIG. 1.

Equilibrium iodixanol preformed step gradient. Whitish band in the upper third is purified P. salmonis.

FIG. 2.

Microscopic characterization by direct immunofluorescence of the material banded on the iodixanol gradient. Magnification, ×100. (A) FITC-labeled anti-P. salmonis (iodixanol purified). (B) Commercial FITC-labeled anti-P. salmonis.

FIG. 3.

Specific PCR amplification for P. salmonis rDNA. Primers against the ITS region of the rDNA operon were used. Lanes: 1, 160 ng/μl; lanes 2 to 6, serial dilutions from 10⁻¹ to 10⁻⁵, respectively. L, 100-bp ladder.

Next, to evaluate putative cellular DNA contamination in our purification procedure, we carried out different semiquantitative PCR amplification reactions with fish mitochondrial DNA primers against different DNAs: CHSE-214 DNA, P. salmonis DNA, and E. coli DNA at different ratios, and competitive reactions combining two of these DNAs to validate our approach (Table 1). Figure 4 shows the results of all reactions. When CHSE-214 DNA was the target, amplification occurred up to a dilution of 10⁻⁶ (lanes 1 to 7) while P. salmonis DNA did up to dilution 10⁻⁴ (lanes 8 to 14), suggestive of a minor contamination of the P. salmonis DNA with mitochondrial cell DNA. When P. salmonis DNA was mixed with CHSE-214 DNA in a 1:1 genome size ratio, amplification occurred equivalently (lanes 15 to 20), confirming a low degree of contamination in our P. salmonis DNA preparation. To validate this interpretation, we mixed E. coli DNA with CHSE-214 DNA (lanes 28 to 32), observing results similar to those described for the mix of cell DNA with that of P. salmonis.

TABLE 1.

DNAs in Fig. 4

Lane	DNA (amt, ng)	Dilution
1	CHSE-214 (15)	10−1
2	CHSE-214	10−2
3	CHSE-214	10−3
4	CHSE-214	10−4
5	CHSE-214	10−5
6	CHSE-214	10−6
7	CHSE-214	10−7
8	P. salmonis (15)	10−1
9	P. salmonis	10−2
10	P. salmonis	10−3
11	P. salmonis	10−4
12	P. salmonis	10−5
13	P. salmonis	10−6
14	P. salmonis	10−7
15	CHSE-214/P. salmonis	10−4/10−1
16	CHSE-214/P. salmonis	10−5/10−2
17	CHSE-214/P. salmonis	10−6/10−3
18	CHSE-214/P. salmonis	10−7/10−4
19	CHSE-214/P. salmonis	10−8/10−5
20	CHSE-214/P. salmonis	10−9/10−6
21	JM109 (15)	10−1
22	JM109	10−2
23	JM109	10−3
24	JM109	10−4
25	JM109	10−5
26	JM109	10−6
27	JM109	10−7
28	P. salmonis/JM109	10−4/10−1
29	P. salmonis/JM109	10−5/10−2
30	P. salmonis/JM109	10−6/10−3
31	P. salmonis/JM109	10−7/10−4
32	P. salmonis/JM109	10−8/10−5

FIG. 4.

Semiquantitative and competitive PCR amplifications. Primers against fish mitochondrial DNA were used to evaluate the purity of P. salmonis DNA. Lanes 1 to 7, CHSE-214 DNA; lanes 8 to 14, P. salmonis DNA; lanes 15 to 20, P. salmonis DNA mixed with CHSE-214 DNA in a 1:1 genome size ratio; lanes 21 to 27, E. coli DNA (JM109); lanes 28 to 32, E. coli DNA mixed with CHSE-214 DNA. L, 100-bp ladder.

How could we quantify the observed contamination? For equivalent DNA concentrations (Table 1), amplification occurred up to a 10⁻⁶ dilution for CHSE-214 DNA and up to 10⁻⁴ for P. salmonis DNA. Assuming that the latter did not inhibit PCRs and that the proportion of mitochondrial DNA in CHSE-214 DNA is equivalent to the proportion of mitochondrial DNA in our preparation, we could use the end dilution ratio to estimate the contamination. This means 1% contamination end dilution for CHSE-214 DNA (0.0015) over that of P. salmonis DNA (0.15). We should add at this time that equivalent results were obtained when a housekeeping gene primer set was used instead of the fish mitochondrial primers (data not shown).

Next, to estimate the efficiency of our procedure, we applied the following rationale: We recovered 16 μg of DNA from the upper band of the iodixanol gradient. This corresponds to approximately 10⁷ genomic units of an estimated 1.1 × 10⁶ Da per genome of intracellular bacteria (5). Considering that we started with 1.1 × 10⁸ infected host cells and that minimal effective infection is 10%, we are recovering 10 to 100 bacteria per CHSE-214-infected cell, which appears to be highly efficient. This theoretical calculation is consistent with kinetics of infection experiments run in our laboratory (data not shown). In addition, the confocal microscopy shown in Fig. 5 confirms that an average of 10 to 20 bacteria were clearly observed per infected CHSE-214 cell at 8 to 12 days postinfection, the time of our DNA isolation procedure. These estimations are in full agreement with those reported by others (11), where DNA obtained by an alternative procedure yielded numbers in the same log range.

FIG. 5.

Bacterial number estimation by confocal microscopy of CHSE-214 infected cells. (A) Phase contract microscopy. (B) Fluorescence microscopy, commercial FITC-labeled anti-P. salmonis antibody.

In summary, the described purification procedure yielded bacterial DNA close to 99% pure, a figure not achieved by any of the reported alternative procedures. In addition, the efficiency of bacterial recovery at early to middle host cell infection seems to be high, and inclusion of a DNase I digestion step prior to density gradient centrifugation was of great help in removing all any trace of genomic salmon DNA. Finally, we would like to mention that iodixanol does not affect viability or the infective potential of the isolated bacteria (unpublished data). In conclusion, the procedure described yields enough pure DNA to start reliable genomic studies on this important fish pathogen.