Mark Dorris PhD Thesis 1999
Chapter 3 Chapter 4 Appendix 1
REVERSING THE EFFECTS OF FORMALIN FIXATION
Introduction
Much of this research involves the extraction, amplification, cloning and screening of genetic material from fixed specimens. Sequence information is obtained by automated sequencing of the PCR amplified marker sequence ( Appendix A1). From fresh, live tissue the procedure is relatively straightforward and is outlined in figure 3.1. The formalin derivatives or ethanol traditionally used to store these specimens inhibits many molecular techniques such as the protease digestion of the sample and the PCR amplification of marker sequences, an essential first step in the study of molecular phylogeny and systematics. These problems are compounded when dealing with minute amounts of material such as nematodes, which typically are 1-5 mm in length. Formalin is the aqueous solution of formaldehyde, a highly reactive colourless gas, the structure of which is shown in figure 3.2a. Formalin is phosphate buffered, contains around 37% formaldehyde by weight and usually 7% methanol to prevent formaldehyde polymerisation. Nematodes are usually immersed in hot formalin (70o-80o) before transfer into 30% glycerol. After complete dehydration they may be mounted onto slides. In addition to the problems in amplification of fixed material, the samples will have been previously handled and are almost invariably contaminated with host, human, bacterial, or fungal material.
3.1 Mechanism of formaldehyde fixation
The interactions of formaldehyde with macromolecular structures are unclear to this day. It was traditionally thought that DNA is progressively destroyed altered by hydroxymethylation or irreversibly cross-linked by the action of formaldehyde which is known to react with proteins via thiols, phenolic groups and terminal amino groups . It is perhaps more likely that aggregates are formed between inactivated proteins which when bound to DNA render the nucleic acid inaccessible to the thermopolymerases in the PCR reaction or inhibit effective denaturation of the DNA to allow binding and/or extension of the primer. No more than 100-200bp has been amplified from material fixed for more than a small period of time and the length of PCR product is inversely correlated with time of fixation . To enable fixed samples to be protease digested and PCR amplified the mechanism of formaldehyde interaction with proteins and DNA must in part be elucidated.
The functional group of formaldehyde (and other aldehydes) is the carbon-oxygen double bond of the carbonyl group, the properties of which are illustrated in figure 3.2b. Oxygen is much more electronegative than carbon. Thus the electrons in the C=O bond are attracted to the oxygen, and the bond is highly polarised (see fig. 3.2b). Consequently, many reactions of carbonyl compounds involve attack of a nucleophile (supplier of electrons) on the carbonyl carbon atom. It is the extreme polarity of the carbonyl group on such a small molecule as formaldehyde, which causes polymerisation in the absence of intra-polymeric hydrogen bonding. However, carbonyl compounds readily form hydrogen bonds with other OH or NH compounds and are thus soluble in water.
The carbon of the carbonyl group has a partial positive charge. The ¹ electrons of the C=O bond move to the oxygen atom, which because of its electronegativity can easily accommodate the negative charge that it acquires. When these reactions are carried out in an aqueous environment, they are usually completed by addition of a proton to the negative oxygen, forming a hydroxyl group. Hydroxymethylation (the covalent modification of a hydroxyl group) seems unlikely as this reflects the opposite of what is observed. Rather, the combination of hydroxyl group and aldehyde leads to a highly unstable combined ether-hydroxyl group, which is rarely isolated. Even if hydroxymethylation did occur it would be easily (and naturally) reversible.
3.2 Targets for formaldehyde
The primary amino groups of lysine residues and the N-terminus of proteins are probably the main targets for formaldehyde. Amines have an unshared electron pair on the nitrogen atom and act as nitrogen nucleophiles toward the carbonyl carbon atom . The reaction mechanism of formaldehyde with a primary amine is illustrated in figure 3.3. The first product formed is a dipolar ion. The positive nitrogen loses a proton and the negative oxygen gains a proton, thus forming a tetrahedral addition product. The elimination of water then gives the observed imine or 'Schiff"s base" product (see fig. 3.3). Schiff"s bases are important intermediates in some biochemical reactions, particularly in the capacity illustrated here of binding carbonyl compounds to the free amino groups of proteins. Thus, in vivo they are normally reversible. Since nematodes are fixed in hot formalin it also seems likely that some denaturation of DNA takes place and given the high surface area to volume ratio of nematodes it is probable that formaldehyde will form imines with the free amino groups of adenine, cytosine, and guanine within internal cellular structures before renaturation can take place. Localised renaturation will be then be inhibited due to the steric hindrance and loss of H-bonding potential of the imine groups. DNA modification by formaldehyde will presumably take place at other single stranded sites e.g. during transcription etc. at the time of fixation. The amino targets for formaldehyde modification are shown in figure 3.4.
3.3 The effects of ethanol on biological material
Ethanol preserves specimens by the inhibition of cellular enzymes. This process is assumed to be complete and irreversible and thus samples must be rehydrated in increasing concentrations of H2O before protease digestion and DNA extraction. DNA is readily extracted from mammalian tissue previously stored in ethanol . In ethanol stored nematodes (personal observations) and other invertebrates however, DNA has been notoriously difficult to extract and the nematodes themselves have been difficult to digest with protease K in comparison with either fresh or formalin-fixed specimens (personal observations). DNAses are not irreversibly inhibited by ethanol and some marine invertebrates have highly stable DNAses with up to 18 disulphide bonds, which show 50% activity in 1M urea, 0.1% SDS, 30% ethanol, and no reduction of activity in 0.8% dithiothreitol (DTT) or ß-mercaptoethanol (BME) . In addition, nucleases with high activity and the presence of protease K inhibitors have been shown for isopods (H. Shulenberg, pers. comm.). A minimum of 95% ethanol has been recommended for long term storage .
Methods
All Strongyloides specimens subjected to phylogenetic analysis with the exception of S. ratti and S. stercoralis had been previously fixed in formalin. Other nematode samples had been stored in 70% ethanol. Some had been mounted on slides after dehydration in glycerol and others had been subjected to DNA extraction procedures. None had proved amenable to PCR amplification by standard procedures (data not shown). In consideration of the amount of sample available (in some cases less than a single worm) optimisation of DNA extraction and amplification was essential. To this end >50 C. elegans adult hermaphrodites were fixed in formalin and subsequently dehydrated in glycerol ( Appendix A1). After 14 days in glycerol, 8 nematodes were rehydrated ( Appendix A1) and incubated overnight in GTES buffer (100mM glycine, 0.05% SDS, 10mM Tris/Cl, 1mM EDTA). EDTA (ethylenediaminetetraacetate) serves a dual function. Firstly, it is a weak carboxylic acid. Carboxylic acids can be formed by the oxidation of aldehydes thus the addition of EDTA forces the reaction to the left and thus helps convert formaldehyde to primary alcohol (methanol). In addition, EDTA has two primary nitrogen groups, which will help titrate free formaldehyde. The free amino acid glycine is the smallest amino acid and is present to titrate free formaldehyde in the parts EDTA can"t reach e.g. within pockets in DNA- protein aggregation. The structures of EDTA and glycine are shown in figure 3.5. An aggressive detergent (sodium dodecyl sulphate; SDS) is used to permeate the nematode cuticle prior to protease digestion. Worms were then washed in 10 mMTris/Cl and subjected to digestion and direct PCR as with normal unfixed nematodes (see fig. 3.1 and Appendix A1). One of the eight samples fragmented during washing and was omitted from the analysis. Different universal primer sets were used to determine the maximum size of product that could be amplified from the formalin fixed samples. These primer sets are described in figure 3.6. For primer sequences and PCR cycling parameters see Appendix A1.
20 live adult C. elegans hermaphrodites were immersed in 70% ethanol and stored at room temperature for two weeks. During rehydration of 5 of the samples, 10mM EDTA was added as a chelating agent to combat the action of DNAses. ß-mercaptoethanol was added to another 5 samples to help denature DNAses rich in disulphide bonds. Both EDTA and ß-mercaptoethanol was added to a further 5 samples. To eliminate any possibilty of DNAse activity during rehydration, another 5 ethanol-stored samples were desiccated in a vacuum centrifuge. All samples were then subjected to protease K digestion and PCR amplification using the A/22R primer sets known to routinely amplify from C. elegans.
Results
Eight PCR products from single C. elegans digests are illustrated in figure 3.6. Seven of these were from formalin fixed products (lanes 1-3 and 5-8) and another was from fresh tissue (lane 4). Full length PCR products (1700bp) were unobtainable (data not shown) but products up to 1000bp were readily obtained (see fig. 3.6). One of these products (fig. 3.6 lane 3) was subjected to automated sequencing ( Appendix A1) using a single forward primer (DF). There were no apparent differences to the published C. elegans SSUrDNA sequence (data not shown). A positive control is shown in lane 4, fig. 3.6 in which a product is amplified from a fresh non-fixed C. elegans digest, using the same primers as in reaction 1 (lane 1). PCR products were obtained from the four ethanol-stored subsets (EDTA, BME, EDTA+BME, and dessicated) All of these products were contaminants from a laboratory clone (data not shown).
Discussion
For nematode material fixed in 70% ethanol, the processes of fixation render amplification of products impossible. It is probable that nematodes possess DNAses which are active in 70% ethanol. No PCR amplification of any sized fragment has been possible from nematode material stored at room temperature in 70% ethanol. It is perhaps significant that contaminants amplified from ethanol fixed samples were from a laboratory clone and thus introduced during the PCR amplification procedure. This supports the view that DNA, including any contaminating DNA within the sample, has been a substrate for DNAse activity.
For material recently fixed in formalin, the processes of fixation appear to be reversible at least for products up to 1000bp. The inability to amplify greater sized products suggests that formaldehyde is effecting some mechanism of irreversible fixation. Direct amplification of rehydrated recently fixed C. elegans nematodes, without GTES incubation produced no products (data not shown). This shows that the elements of GTES buffer have a direct effect in the reversal of formaldehyde fixation. The contribution of individual GTES components to the reversal process was not assessed. This procedure provides strong evidence that formaldehyde fixation is reversible and does not reduce the fidelity of the DNA sequence to a degree previously attested . It remains to be seen how applicable this method is for different tissue types and storage conditions. For Strongyloides specimens, which were stored for over 10 years, this method, in itself, is insufficient. Direct sequencing of PCR products from fixed Strongyloides specimens gave low yield mixed populations of sequence. Cloning and screening (described later) of products detected only contaminants. This data supports the inverse relationship shown for length of PCR product as a function of fixation time . The restrictive effects of formalin to molecular techniques and growing concerns about the safety aspects of formaldehyde usage has led to a search for alternative fixatives for field specimens. Some are currently commercially available but none are found to be as effective as formalin, both in terms of maintaining the integrity of tissue structure and ease of application. A common alternative to formalin fixation is to store samples in 70% ethanol. For most tissues for which ultrastructural study is not required, ethanol storage does not appear to have a detrimental effect on the application of molecular techniques. This is not the case with nematodes, which may have DNAses active in 70% ethanol, in common with some other marine organisms .
EXTRACTING
QUALITY DNA FROM FORMALIN-FIXED SAMPLES
Introduction
Nematodes fixed for long periods appear to be refractory to protease digestion; they are still structurally intact after 24 hours incubation with protease K. Formaldehyde fixation of the nematode cuticle seems extensive and irreversible even in the presence of an aggressive detergent (SDS). Apart from a lipid-containing epicuticle and carbohydrate-rich surface coat the nematode cuticle is mainly proteinaceous. The cuticle of C. elegans is extensively cross-linked by covalent disulphide bonds, tyrosine bonds and possibly glutamyl-lysine bonds. That the effect of formaldehyde on the cuticle seems irreversible may lend weight to a proposed thiol- and hydroxy-methylation action on cysteine thiol groups and tyrosine hydroxyl groups respectively, although the additional presence of the terminal lysine amine group provides a natural substrate for formaldehyde. Whatever the actions and targets for formaldehyde it is clear that the nematode cuticle, already extensively cross-linked and containing all the proposed substrates for formaldehyde, provides every opportunity for effective fixation. It is this effectiveness that has resulted in formalin as the nematologists" preferred choice of fixative. Enzymatic disruption of the fixed cuticle remains impossible for all but very recently fixed nematode specimens.
Methods
The cuticle clearly has to be physically disrupted. To this end, after GTES incubation, six fixed parasitic females of Strongyloides fuelleborni were picked from a slide (on which they were mounted in glycerol after formalin fixation 10 years previously), freeze-fractured in liquid N2 and ground with a mini mortar and pestle. Whole genomic DNA was then extracted for subsequent PCR. The DNA extraction procedure employs the Nucleon HT extraction method from Scotlab ( Appendix A1). This differs from other commercially available kits for DNA extraction in that the column resin binds the cell impurities rather than the DNA. The significance of this difference is illustrated in figure 4.1.
Four Strongyloides samples; S. suis, S. papillosus, S. robustus, and S. westeri were supplied by Tom Moore as pre-extracted genomic DNA. Products greater than 150bp could not be PCR amplified from these samples which had been extracted from formalin-fixed material. S. suis genomic DNA was diluted 1:10 with GTES, incubated overnight and subjected to Nucleon HT extraction. Due to the constraints in length of PCR product from fixed samples, partial SSUrDNA sequence was amplified. The 5" 400bp of the SSU sequence represents around 25% of the gene but contains almost 50% of the phylogenetically informative sites of a dataset sampled across Secernentea. In addition, the primer set A/22R (see fig. 3.6), used to amplify the 5" ~400bp, is known to routinely amplify from all nematodes tried. This region was chosen for phylogenetic analysis of fixed specimens. In addition, the primer set A/18p which amplifies the full length SSUrDNA gene was used to test the upper limit of product amplifiable using this extraction protocol. Primer sequences and PCR cycling parameters are described in Appendix A1.
Results
Figure 4.2 shows the result from PCR amplification. PCR from unfixed C. elegans forms the primer controls. A high yield of product is visible for the partial 5" sequence (lanes 1 and 2) but very poor yield of the full-length product from both fixed specimens (lanes 4 and 5). Origin and sequence of the full-length products could not be determined. Multiple products are evident from amplification of both the fixed samples (lanes 1 and 2).
Discussion
GTES incubation coupled with Nucleon HT extraction results in genomic DNA of sufficient quality for high yield PCR of at least 400bp in length. Since the starting material had been fixed for ten years this protocol should be applicable for most fixed material. Even previously extracted DNA (fig. 4.2 lane 2) can be 'cleaned" and rendered accessible to PCR. There appears to be some amplification of full length (~1700bp) sequence (fig. 4.2 lanes 4 and 5) but the yield is extremely low and may represent either a contaminant or leaching of the C. elegans positive control sample in the gel. DNA yield is very low using this extraction method. Not only is the volume of starting material extremely limited, but a large proportion of the genetic material will be withheld in the matrix with aggregated proteins. Ribosomal DNA sequences provide ideal markers for amplification as they are present in many copies per genome and thus at least some of the copies in some of the cells will provide a template for PCR. Invariably, the PCR product contains a mixed population of sequence (note multiple bands in lanes 1 and 2, fig. 4.2). This is partially due to frameshift effects caused by the insertion of non-specific nucleotides at damaged or inaccessible regions of the template resulting in heterogeneous products of the same size but also due to contaminating sequences, reflected in products of different sizes. Band excision from the gel can eliminate the latter, but not the former. PCR products are therefore cloned, screened, and at least three clones sequenced in both directions to build a consensus sequence ( Appendix A1). Although not a trivial procedure, extraction of quality genomic DNA from long-fixed specimens is now a reality.
Qiagen miniprep for purification of plasmid DNA
Spin purification of PCR products using Spin-X UF100 filter
Qiaquick gel extraction
Hybaid DNA purification from solutions
PCR conditions for obtaining partial sequence from formalin fixed tissue
Target is 1ml of Nucleon extracted DNA.
2.5ml 2mM nucleotide mix
2.5ml 10X buffer
1.5ml 25mM MgCl
1 U Taq
0.8ml 100ng/ml primer 1
0.8ml 100ng/ml primer 2
1ml target
xml ddH2O to a final volume of 25ml
Typical PCR parameters
1 cycle 95o, 3 min
35 cycles 94o, 1 min
54o, 1 min 30 sec
72o, 2 min
1 cycle 72o, 10 min
Conditions were varied with respect to cycling times (94o, 30 sec-1 min, 72o, 30 sec-2 min) and Mg (1-3 mM) concentration to optimise each reaction
Buffers and solutions
LB Tryptone 10g
Yeast extract 5g
NaCl 5g
H2O to 1L
TE(pH 8.0) 10mM Tris/Cl (pH8.0)
1mM EDTA
GTES 100mM glycine
10mM Tris/Cl (pH 8.0)
1mM EDTA
0.05% SDS
3M Na acetate pH 5.2
Dissolve 40.824g in little H2O, adjust pH to 5.2 with glacial acetic acid and add H2O to 100ml
5X TBE Tris base 54g
Boric acid 27.5g
50mM EDTA 20ml
H2O to 1L
ABI loading buffer
50mg/ml Dextran blue
25mM EDTA
Worm lysis buffer (WLB) (Peter Hunt)
NaCl 100mM
Tris/Cl,pH 8.5 100mM
EDTA 50mM
SDS 1%
2-mercaptoethanol 1%
Reagent B (Nucleon)
Tris/Cl 400mM
EDTA 60mM
NaCl 150mM
SDS 1%
Adjust pH to 8.0 using 40% NaOH
Formalin solution 8% formalin
2% glycerin