Comparison of Fixatives, Fixation Time, and Severity of Infection on PCR Amplification and Detection of Mycobacterium marinum and Mycobacterium chelonae DNA in Paraffin-Embedded Zebrafish (Danio rerio)

Abstract

Mycobacteriosis is a common disease of laboratory zebrafish (Danio rerio). Different infection patterns occur in zebrafish depending on mycobacterial species. Mycobacterium marinum and M. haemophilum produce virulent infections associated with high mortality, whereas M. chelonae is more wide spread and not associated with high mortality. Identification of mycobacterial infections to the species level provides important information for making management decisions. Observation of acid-fast bacilli in histological sections or tissue imprints is the most common diagnostic method for mycobacteriosis in fish, but only allows for diagnosis to the genus level. Mycobacterial culture, followed by molecular or biochemical identification is the traditional approach, but recently it has been shown that DNA of diagnostic value can be retrieved from paraffin blocks. We investigated effects of the following parameters on the ability of our qPCR test for the hsp gene (primer set HS5F/hsp667R) to retrieve specific DNA from paraffin-embedded zebrafish: type of fixative, time in fixative before processing, species of mycobacteria, and severity of infection. Whole zebrafish were experimentally infected with either M. chelonae or M. marinum, and then preserved in 10% neutral buffered formalin or Dietrich’s fixative for 3, 7, 21 and 45 days. Subsequently, fish were evaluated by H&E and Fite’s acid-fast stains to detect mycobacteria within granulomatous lesions. The PCR assay was quite effective, and obtained PCR product from 75% and 88% of the M. chelonae and M. marinum infected fish, respectively. Fixative type, time in fixative, and mycobacterial species showed no statistical relationship with the efficacy of the PCR test.

Introduction

Piscine mycobacteriosis is a well-characterized disease, with several mycobacterial species infecting multiple genera and species of fish (Decostere et al. 2004, Lewis and Chinabut 2011). Zebrafish (Danio rerio) are now a widely used vertebrate animal model in biomedical research (Lele and Krone 1996, Dooley and Zon 2000, Grunwald and Eisen 2002, Rubinstein 2003, Alestrom et al. 2006, Allen and Neely 2010) with an ever-increasing number of academic and private laboratories using zebrafish. Since the emergence of the zebrafish as a pre-eminent animal model, there has been a corresponding interest in diseases that may occur in this species within a laboratory setting. The Zebrafish International Resource Center (ZIRC) in Eugene, Oregon (http://zebrafish.org/zirc/home/guide.php) has provided a diagnostic service to the zebrafish research community since 2000. Mycobacteriosis, based on observation of acid-fast bacteria in histological sections, has been detected at 41% of submitting facilities in about 150 cases with over 500 fish (http://zebrafish.org/zirc/health/diseaseManual.php). Several species of Mycobacterium have been reported to cause disease in zebrafish, including M. chelonae (Astrofsky et al. 2000, Kent et al. 2004, Murray et al. 2011), M. peregrinum (Kent et al. 2004), M. haemophilum (Whipps et al. 2007, 2012), and M. marinum (Ostland et al. 2007). Based on observations reported in diagnostic cases (Astrofsky et al 2000, Kent et al. 2004, Whipps et al. 2008, Murray et al. 2011, Whipps et al. 2012) and laboratory transmission studies (Watral and Kent 2007, Whipps et al. 2007, Ostlander al. 2008), the severity of mycobacteriosis is usually related to the Mycobacterium species causing the infection. Mycobacterium chelonae is relatively widespread and causes chronic infections but minimal mortalities (Whipps et al. 2008, Murray et al. 2011). We have seen about six outbreaks of M. haemophilum, which caused fulminating infections with high but chronic mortality (Whipps et al. 2007). Mycobacterium marinum is relatively rare, but whe n infection occurs it is associated with acute disease and high mortalities (Watral and Kent 2007, Ostland et al. 2008). Histology is the primary diagnostic method that we use with zebrafish, but diagnosis of mycobacterial infections by histology only allows for identification of the bacteria to the genus level. Of the over 150 cases diagnosed by histology by ZIRC, identification of the bacteria to the species level using culture or molecular methods has been achieved on less than 20 cases (Watral and Kent 2007, Whipps et al. 2007, 2008, 2012). Several studies have previously demonstrated that mycobacterial DNA can be amplified from human and animal (including fish) tissues from paraffin blocks (Ghossein et al. 1992, Miller et al. 1997, Marchetti et al. 1998, Zink and Nerlich 2004, Pourahmad et al. 2009a). Efforts to develop more reliable PCR assays that would reduce the time required for diagnosis as well as increase both the specificity and sensitivity of detecting mycobacteria in formalin-fixed, paraffin-embedded (FFPE) tissues have been ongoing, and are mostly focused on human mycobacteriosis (Pao et al. 1988, Pao et al. 1990, Fiallo et al. 1992, Hardman et al. 1996, Rish et al. 1996, Osaki et al. 1997, Salian et al. 1998, Whittington et al. 1999, Li et al. 2000, Singh et al. 2000, Baba et al. 2008) and to a lesser extent, mycobacteria infections of animals including fish (Gyimesi et al. 1999, Puttinaowarat et al. 2002, Pourahmad 2009 a,b, Zerihun et al. 2011). Results have been mixed with this approach, and time in fixative before processing into paraffin blocks appears to be an important factor for successfully obtaining mycobacterial DNA from tissues (Tokuda et al. 1990, Greer et al. 1991). Fixative formulations, especially those containing acids or alcohol at various concentrations can also influence the ability to retrieve DNA (Eltoum et al. 2001). Most human and veterinary laboratories use 10% neutral buffered formalin as a fixative, which is also used for preservation of zebrafish and tissues from other fishes (Ferguson 2006, Harper and Lawrence 2011). However, zebrafish have been traditionally preserved in Dietrich’s fixative, a mixture of fixatives that contains both acetic acid and 95% alcohol in addition to formaldehyde. The ZIRC zebrafish diagnostic program is largely based on histologic evaluation of whole fish specimens; however, there is a pressing need to identify mycobacterial infections to the species level in order to allow for more informed management of these infections. Subsequently, we developed a PCR test for retrieval of mycobacteria DNA and evaluated the influence of time and fixative (either 10% neutral buffered formalin or Dietrich’s) on the ability to recover DNA for PCR assays. This was achieved by experimentally infecting zebrafish with either M. chelonae or M. marinum, and preserving the infected fish in both fixatives for various time points up to 45 days.

Materials and Methods

Mycobacteria cultures used for inoculation were prepared in the following manner. Stock cultures from the Kent laboratory reference collection of Mycobacterium marinum OSU 214 (M. marinum) and Mycobacterium chelonae H1E2 (M. chelonae) were incubated on Columbia CNA (colistin and nalidixic acid) 5% sheep blood and Middlebrook 7H10 agar (Remel, Lenexa, KS) for 14 days at a temperature of 27.6 °C. Fresh culture material from agar plates with luxuriant growth was loop inoculated into sterile PBS and washed three times to eliminate chemical contamination from the agar plates and then adjusted in sterile PBS to an optical density of 1.0 based on the MacFarlane scale, yielding an approximate dose of 1 × 10⁶ bacteria/25 µl for the final injection amount.

AB strain adult zebrafish of mixed sex were obtained from the Sinnhuber Aquatic Research Laboratory at Oregon State University as experimental animals. Fish were housed in a biosafety-level 2 (ABSL-2) laboratory and maintained for twelve weeks after inoculation, allowing for progression of infection, in triplicate flow-through tanks at 25–27 °C with supplemental aeration and a 14/10 light/dark photoperiod schedule. For inoculation with M. marinum and M. chelonae, fish were anesthetized with 100 ppm tricaine methanosulfonate (MS-222) then aseptically given intraperitoneal injections in the right flank, to avoid accidentally lancing the spleen, with 25 µl of either sterile PBS (sham-inoculated control group) or the prepared M. marinum and M. chelonae inocula. At the end of 12 weeks, moribund fish and all surviving fish were euthanized by overdose of MS-222, the operculae removed and fish incised along the ventrum in order to expose viscera to the fixative solutions, then immediately placed into 15 ml of either Dietrich’s fixative or 10% neutral buffered formalin (10% NBF). Fish were held in Dietrich’s fixative or 10% NBF for 3, 7, 21 and 45 days (Table 1).

Table 1.

Proportion of PCR positives (pr imer set HS5F/hsp667R) and mean granuloma intensity in paraffin-embedded zebrafish (Danio rerio). Fish were preserved in either 10% neutral buffered formalin (NBF) or Dietrich’s fixative for various times before processing. There was no statistical significance in prevalence between mycobacteria species, fixatives or duration of fixation (all p ≥ 0.3, Fisher’s exact test). Granulomas/fish are reported as the mean for the total group, followed by the mean of PCR negative and positive fish in parentheses

Group	PCR positive/ No. tested	Granulomas (−/+)
M. chelonae
3 days NBF	5/5	14.4
7 days NBF	4/5	6.2 (9.0, 5.5)
21 days NBF	3/5	15.4 (19.5, 12.7)
45 days NBF	1/5	5.6 (5.5, 6.0)
3 days Dietrich’s	4/5	8.4 (14.0, 7.0)
7 days Dietrich’s	4/5	8.0 (1.0, 9.75)
21 days Dietrich’s	4/5	11.8 (15.0, 11.0)
45 days Dietrich’s	5/5	8.6
M. marinum
3 days NBF	5/5	14.6
7 days NBF	5/5	35.0
21 days NBF	5/5	38.0
45 days NBF	5/5	20.2
3 days Dietrich’s	4/5	36.2 (26.0, 38.75)
7 days Dietrich’s	4/5	33.0 (25.0, 35.0)
21 days Dietrich’s	4/5	20.8 (16.0, 22.0)
45 days Dietrich’s	3/5	29.0 (22.5, 33.3)

Histology and PCR

Decalcification was performed on the fixed zebrafish samples as follows: Fish preserved in Dietrich’s fixative were decalcified in 5% Trichloroacetic acid (TCA) overnight (following the Zebrafish International Resource Center protocol), while fish preserved in 10% NBF were placed in CalExII (Fisher Scientific, Fair Lawn, New Jersey) for 48 h. After decalcification and rinsing for 30 min with de-ionized water, all fish were transferred into 70% ethanol and held no longer than 48 h before processing. Following fixation, zebrafish were processed for paraffin embedding by routine procedures. Sections examined for histology were cut at 5 µm and stained with hematoxylin and eosin as well as Fite’s acid fast stain for detection of acid-fast bacilli (Luna 1968).

PCR scrolls were made from dry tissue sections cut from blocks at 5 µm thickness after decontamination and prepared for DNA extraction in the following manner. Prior to sectioning the scrolls, each tissue block was wiped down with one DNA AWAY™ wipe (Molecular BioProducts, Inc. San Diego, CA), then faced with a new microtome blade. The block face was wiped with a fresh DNA AWAY™ wipe, the microtome blade changed and twelve 5 µm sections cut in a scroll. Two scrolls, handled with clean wooden toothpicks, were placed into a sterile, pre-labeled 1.5 mL microfuge tube. Microtome blades were discarded after cutting scrolls for each tissue block, in order to avoid cross-contamination. The entire microtome was then wiped down with xylene (to remove excess paraffin waste) and 100% ethanol, allowed to air dry then wiped down with a DNA AWAY™ wipe and air-dried. Gloves were changed between microtome cleaning, blade changes and tissue block handling. After every five blocks, scrolls were obtained from a block containing a negative zebrafish from the Sinnhuber Aquatic Research Laboratory SPF colony. All tissue blocks and PCR scrolls were prepared in an identical manner.

Paraffin-embedded tissue sections were washed twice with 1.2 ml of xylene and then twice with the same volume of 100% ethanol following the recommended protocol for the Qiagen DNeasy Blood & Tissue Kit (Qiagen Inc., Valencia, California). The tissue pellet was suspended in 375 µl of ATL buffer and Antifoam A (Sigma-Aldrich) was added to a final concentration of 1%. This was transferred to a 0.5ml screw cap tube containing 300µl of 0.1mm Zirconia/Silica Beads (BioSpec Products Bartlesville, OK) and placed on a Mini-BeadBeater 16 (BioSpec Products) for 3 minutes. We incorporated the bead beating step as our preliminary studies showed that this yielded more positive results. Thirty µl of 20 mg/ml Proteinase K was added to this disrupted tissue and digested overnight in a 45 °C water bath. The sample was pulsed down in the centrifuge, then 200µl removed and DNA extracted following the manufacturer’s instructions. PCR was carried out with primers HS5F (GTC ATC ACC GTC GAG GAG) and hsp667R (Selvaraju et al. 2005), yielding a product 156 bp of hsp65 sequence. Amplifications were performed on a C1000™ Thermal Cycler (BioRad Laboratories, Hercules, CA) with initial denaturation at 95° C for 3 min, followed by 35 cycles of 94° C for 30s, 54° C for 45s, 68° C for 60s, and a final extension at 68° C for 7 min. Product amplification was evaluated by observation on a 2% agarose gel. To confirm the specificity of the PCR reactions, products from a subset of three positives each of M. chelonae and M. marinum samples were sequenced. Amplification products were purified using the E.Z.N.A. Cycle Pure Kit (Omega Bio-Tek, Norcross, GA) and direct sequencing was performed using primer HS5F on the ABI BigDye Terminator Cycle Sequencing Ready Reaction Kit v3.1, using the ABI3730xl Genetic Analyzer (Applied Biosystems, Foster City, CA).

Severity of Infection

Individual infected fish were scored as light or heavy infections by enumerating granulomas containing acid-fast bacteria in two whole body sagittal sections per fish/slide. For both M. marinum and M.chelonae, light and heavy infections were classified as 1–6 granulomas and greater than 6 granulomas, respectively.

Statistics

All statistical analyses were conducted with the program R, version 2.7.2 (R Development Core 315 Team). Significance was set at p ≤ 0.05 and p-values are 2-tailed. Differences in PCR prevalence between Mycobacterium species, fixative types and fixation duration was tested with the Fisher’s exact test. A generalized linear regression model was used to evaluate the influence of fish sex, Mycobacterium species, fixative type, fixation time and number of granulomas (indicating level of infection) detected by histology had on the ability of the PCR (primer set HS5F/hsp667R) to successfully detect Mycobacterium DNA. Logistic regression was the most appropriate technique to evaluate these effects because the response of PCR detection is a binary variable. This model had a logit link function and an interaction effect of fixative and fixation time was explored. The likelihood ratio test was used to determine how well the model fit the data compared to the null model of just the intercept. Individual variables and combinations thereof were also tested using a backwards elimination technique.

Results

We were capable of obtaining DNA from a large number of the histology-positive samples, irrespective of total fixation time, type of fixative, or Mycobacterium spp. (Table 1). There was no significant interaction effect of fixative and fixation time (p > 0.99) so this variable was removed and the resulting model that contained the variables of interest did not fit the data as well as the null model (p = 0.37; chi-square of 7.5 with 7 df), meaning that none of these variables explained the variation in the detection of mycobacteria DNA by PCR. Similar results were obtained by evaluating individual variables and combinations thereof. Fixation time of 45 days showed a trend towards fewer PCR positives, and one of the four groups at this time period (M. chelonae in 10% neutral buffered formalin) had only one of five positive samples. Although this time in fixative was not a statistically significant influence, it was suggestive of having an effect (p = 0.08). None of the histology negative control fish that were included throughout the evaluations were positive by PCR. Sequencing of three M. marinum and three M. chelonae PCR positive samples confirmed their identity.

We also evaluated severity of infection as an explanatory variable. Mortality occurred in the M. marinum group, and overall these fish exhibited about three times the number of granulomas containing acid fast bacteria than the M. chelonae-infected fish (29.6 versus 9.4 granulomas/fish, respectively) that were positive by PCR (Table 1). The M. chelonae groups yielded fewer positive fish by PCR, but there was no significant relationship with PCR compared to either Mycobacterium species or number of granulomas/fish. There was also no significant difference between fixative types.

Discussion

Evaluation of DNA sequences has become a cornerstone in bacterial species identification. This is certainly the case for Mycobacterium species, as they are often fastidious, grow slowly and typically provide few culture-specific or biochemical traits by conventional methods that facilitate species-level identification (Daniel 1990, Cousins et al. 1992). It is often more useful to obtain sequences for diagnoses directly from infected tissues, particularly for Mycobacterium species that are slow growing or require specialized media. This approach has been used to diagnose mycobacteria infections directly from frozen or fresh infected fish (Kaattari et al. 2005, Poort et al. 2006, Whipps et al. 2003, 2007). Often only formalin preserved tissues embedded in paraffin are available, which has led to attempts to obtain mycobacterial DNA from paraffin-embedded tissues following observation of either chronic lesions (i.e., granulomas) or the presence of acid-fast bacteria in tissues that are indicative of mycobacterial infection (Ghossein et al. 1992, Marchetti et al. 1998, Zink and Nerlich 2004, Miller et al. 2007, Pourahmad et al. 2009a,b).

Fixatives used to preserve tissues (regardless of whether it is a coagulant or non-coagulant fixative) and the total time the tissues are held in a fixative can result in degradation of DNA and RNA within tissues (Dubeau et al. 1986, Fiallo et al. 1992, Foss et al. 1994). Exposure of mycobacterial DNA to formalin causes production of Schiff bases on the free amino groups of nucleotides (Fraenkel-Conrat 1954, Dubeau et al. 1986), and subsequent cross-linking between tissue proteins and DNA (Jackson and Chalkey 1974). Therefore, time in formalin is considered to be one of the most important criteria for successful DNA retrieval and PCR amplification following tissue processing for histology (Ben-Ezra et al. 1991, Greer et al. 1991). Once the tissues are embedded in paraffin wax, this degradation process is slowed and subsequent amplification of short DNA sequences is successful (Shibata et al. 1988). Fish are often preserved at a research laboratory, aquaculture facility or in the field, and then shipped many days later to another laboratory for histology processing. Hence, we were particularly interested in the effects of time in fixative. We consistently obtained positive results with fish held in fixative up to 21 d, and the same occurred at 45 d with the exception of one group. This positive result at the later time points was somewhat surprising. Mycobacterium spp. has a unique, waxy cell wall comprised of lipids and fatty acids (Kolattukudy et al. 1997), which may exclude aldehydes to some extent. Perhaps this characteristic plays a role in preserving DNA more than in non-acid fast bacteria exposed to formalin-based fixatives.

Dietrich’s, Davidson’s and Bouin’s fixatives are commonly used with fish tissues. These contain acid and alcohol, both which cause denaturation and coagulation of proteins and nucleic acids within tissue specimens by dehydration and disruption of electrostatic and hydrogen bonding (Fournie et al. 2000, Eltoum et al. 2001). Tissues preserved in Bouin’s solution (which contains both picric and acetic acid) are particularly problematic for DNA retrieval (Greer et al. 1991). Interestingly, we saw no reduction in positive samples with Dietrich’s compared to 10% buffered formalin.

Some samples were negative in both fixatives at some of the early time points. The inability to retrieve DNA logically would be influenced by the amount of bacterial DNA. Incorporation of a bead beating enhances retrieval of DNA from formalin-preserved samples to liberate more DNA (Tripathi and Stevenson 2012). We included this with our protocol because from our previous experiences we have found that it consistently yields more positive results with PCR from paraffin-embedded tissues. Nevertheless, there was no significance in the ability to detect mycobacterial DNA based on severity of infection. Our results here were similar to previous transmission studies regarding the virulence of M. chelonae compared to M. marinum in zebrafish (Watral and Kent 2007, Ostland et al.2008, Whipps et al. 2008). The M. marinum isolate was highly virulent and several fish became moribund or died over the 8 wk period, and these fish were included in 45 d samples. In contrast, all fish injected with M. chelonae became infected, but none exhibited morbidity or mortality. The M. marinum-infected fish had three times the numbers of granulomas with acid-fast bacteria, but the ability of retrieving DNA was not different than with fish infected with M. chelonae.

Most zebrafish cases submitted to the ZIRC and our laboratory for health screening and diagnosis consist of multiple fish, and usually they are processed and embedded within 3 wk. Hence, even with only about 50–70% ability to obtain mycobacterial DNA, we are confident that diagnosticians can use our approach to identify these infections to the species level for many of the fish within a particular case. This is important as the severity and distribution of these various Mycobacterium spp. found in zebrafish are quite different, and fish health managers and clinicians with species identifications in hand would be able to make more informed decisions. For example, facilities with M. haemophilum infections often euthanize infected populations, disinfect the aquaria, etc., and repopulate (Kent et al. 2009, Kent et al. 2011, Whipps et al. 2012), and we recommend the same for M. marinum outbreaks. In contrast, M. chelonae infections are presently managed by cleaning tanks more often (Murray et al. 2011) or using different wild type strains of zebrafish (Whipps et al. 2008). We often observe acid-fast bacteria in the intestinal lumen, with and without infections in the visceral organs. Therefore, a concern with amplification of DNA using whole fish sections may be confusion of mycobacteria in granulomas versus intraintestinal mycobacteria that may be associated with other constitutive microbiota in the intestinal tract. A solution for the potentially confounding problem of intraintestinal bac terial contaminants would be to use histologic sections as a template to guide in the removal of specific cores from the tissue block that correspond to confirmed areas of infected organs (Sfanos et al. 2008) or, even more precisely, by using laser capture microdissection to select individual intralesional granulomas containing mycobacteria for PCR assays (Ryan et al. 2002, Zhu et al. 2003, Selva et al. 2004). Both methods would allow for accurate selection of infected tissue sites within the infected zebrafish prior to PCR analysis for mycobacterial identification and PCR primer-specific molecular identification of Mycobacterium spp.