1. Suitable genes, the choice ("design") of target sites, and the appropriate controls: Many genes are rapidly deployed just before the period of utilization of their products in distinct developmental events, and then have dynamically changing expression levels and sites over the hs-long time scale of early development. Such dynamic RNA expression patterns are themselves evidence for relatively rapid message turnover (short half-life), but could only be part of developmental control mechanisms if accompanied by correspondingly rapid turnover of the protein product. These are the criteria recognized as favorable for antisense interference with gene function in cell systems, viz. protooncogenes and other cell-cycle-or cell-state-change-related genes (1,2,19), so that many genes distinctively expressed in early development are suitable candidates for the whole chick embryo approach. Low or modest transcription levels (genes whose signals take relatively long to "come up" in developing in situ specimens) are also favorable, because a relatively lower level of interference with their transcription might limit gene function. Many transcription factor and intercellular signal-encoding developmental genes are suitable for this criterion as well.
Not all such genes are suitable, however. For instance, antisense oligos to several sites in the chick Brachyury gene have been tried in our lab, at relevant stages in view of the known, vertebrate-conserved mutant phenotype of Brachyury. These are without gene-related or other effects, but on consideration, this would not be surprising even if they were acting significantly on the mRNA stability or translation. Although vital for one or a few specific functions in development, this gene's RNA (at least in chick) is transcribed very abundantly within the lineages of cells that will require it (as well as in many that never will), and well ahead of the time of the protein's necessary function in these cells as deduced from the mouse and fish mutants. This suggests, though it does not prove, long half-life RNA and/or protein and thus an expected "perdurance" of gene function in the face of antisense action. Even if antisense effects are exerted in a particular case at molecular level, the gene's early expression pattern does not firmly predict that early developmental steps will be compromised. This is a fascinating scientific matter that relates as much to targeted mutagenesis (knockout) studies of such genes in mice as it does to antisense studies. Thus, oligos targeted to several sites in both the follistatin and the noggin genes of chick have also failed to give effects within the early axial phase of development that the current technique covers. The enticing early expression sites, and corresponding overexpression effects obtained experimentally in Xenopus with these gene products, have meanwhile prompted mouse gene knockouts. From these, it appears that neither gene is required (subject to confirmation) until unexpectedly later steps in development, after induction of the basic axial body plan (20 and unpublished work).
Only well-controlled positive results of antisense studies are informative about a gene's role. In the nature of the case, unlike that of targeted mutagenesis, absence of effects is currently noninformative. Apart from the considerations just made, the main reason for this is the imperfect criteria for choice of sites, within the transcribed sequence of any gene, that are suitable targets for antisense oligos. There is an extensive literature on such "design criteria" that it is inappropriate to review here. A broad consensus is:
a. That 15-mer to 18- or 20-mer offer the best compromise between likelihood of binding to sequences unique within the transcribed genome, and ease of entry to relevant sites in cells.
b. That short oligo sequences with obvious self-pairing stretches leading to hairpin structures or stable homodimers should be avoided.
c. That the GC/AT content should not be too imbalanced in relation to that of the DNA generally.
d. That computer software can scan the relevant species database (very fragmentary in the case of chick) for possible matches within other genes, and also predict the melting temperature of the DNA/RNA hybrid for each sequence sector. These melting temperatures should obviously be somewhat above the incubation temperature.
An overriding consideration might be sequence-determined secondary structure features of RNA molecules under physiological conditions, determining accessibility as target, but there is little agreement about software that will adequately predict this. Few papers describe comparative results of reasonable numbers of targeted sequences within particular genes (e.g., 9), but perhaps somewhat less than one-third of all sequence stretches at random would be good for antisense interference. Oligo sequences to be used as mixtures should obviously not pair stably with each other. Striking interference has been reported for target sectors at all relative positions within mRNA, including untranslated 3'-stretches. In the absence of counter indications in the sequence, however, the translation start site and its neighborhood is a first choice.
The choice of the corresponding sense sequence as control oligo for a candidate antisense-interfering one is as illogical as the practice of specifically creating sense strand riboprobes for in situ hybridization controls. Apart from length, both display only random degrees of similarity or matching of possibly relevant properties, as sequences go, to the antisense one. In the case of gene interference, there is even evidence for a significant incidence of sense-specific effects (dependent on oligo chemistry, e.g., 21), exerted by binding and interference with transcription itself. With both the RNA detection and gene-interference effects of antisense, the cumulative evidence for genetic specificity overall must be the distinctive, gene-expression- related effects of sequences antisense to various genes, coupled with the general absence of effects of other sequences in the same procedure. For oligo (as opposed to translation length) antisense, this includes other sequences that happen to be antisense to genes. A complete strategy must take account of the occasional report of sequence-specific, but apparently gene-unrelated effects of S-oligos on cellular properties (e.g., 4). With time, therefore, the strategy becomes to design antisense oligos to each newly addressed gene, and when and if repeatable specific effects are identified, to think about more stringent individual control strategies for the gene-relatedness of these effects. We choose to use oligo sequences that are parallel with the experimental ones, but have short (6-base) 5'-3' inversions, and are screened for being antisense to known chick genes. They thus match for overall base composition, in case this should be relevant to chemical and not biological specificities of effect. Since most sequences do nothing, we additionally regard the finding of the same "phe-notype" with more than one sequence antisense to the same gene, and especially a more than linear cooperative effect between such sequences, as very strong evidence for gene specificity of effects.
2. Alternatives for nuclease resistance, and access into relevant cellular compartments, of oligo DNA: Chemical modifications to oligos to increase their half-life while preserving antisense properties fall into two categories: modifications to the bases themselves and modifications of the sugar-phosphate backbone. In the former category, the substitution of C5-propyne derivatives of pyrimidine bases in DNA may be an exciting breakthrough, affording greatly increased sequence-specific binding while preserving the ability of the hybrid sequence to act as substrate for RNase H so that oligo catalytically degrades message or premessage (9), but these are not yet commercially available or widely studied. In the latter category, we have experience of chimeric oligos in which the end stretches are made with p-ethoxy derivatization of the backbone, leaving a central 7- to 8- base stretch of normal DNA. It is believed that RNase H substrate properties of these oligos may be preserved, and that binding constants are not greatly lowered per sequence. Most importantly, we find that they are not precipitated by the albumen in the medium, so that longer-term persistence of gene interference in ring culture might be expected. In the case of the slug gene (see ref. 3), the effects of a particular antisense 15-mer sequence in its phosphoro-thioated form, and their degree of persistence during development after the intial off-membrane treatment, were well known to us for S-oligos. We found that these could be replicated when blastoderms were treated only briefly in protein-free medium with the corresponding p-ethoxy chimeric oligo sequences, and then returned to ring-culture with a much lower oligo concentration. Persistent effects were seen even in blastoderms treated once by subinjection through the periphery into the space between epiblast and vitelline membrane, at the start of culture. Unfortunately, these oligos, and also further new modifications available through our normal commercial source (see Subheading 2.), are currently too expensive to allow exploration of many sequences in new genes.
Particularly S-oligos are almost certainly taken up into cells mainly by a pinocytotic process leading into membrane-bound vesicles, from which passage into the truly intracellular locations (cytoplasm, nucleus) of their RNA targets is vastly inefficient owing to enzymic degradation. Lipofection procedures potentiate uptake, possibly by destabilizing these intracellular vesicles as well as increasing incorporation from outside the cell. It is even reported that lipofection releases previously taken up, fluorescently tagged oligo from vesicular sites to allow immediate concentration in the nuclear space (9). Attention has recently been paid to an alternative potentiating vehicle for passage of molecules from outside cells into truly intracellular sites; the so-called fusogenic peptides derived from the sequence of the flu virus envelope protein that is influential in its entry to the cell. Such peptides have the property of pH-dependent conformational change that destabilizes lysosomal membranes, greatly increasing their turnover of intact contents into the cell (22). They can apparently act significantly as vehicles for oligonucleotides in this way merely by copresence in the medium (23,24), and not necessarily by being covalently linked to them. In preliminary work (with S. Wharton, this institute), we have indications that a 20 amino acid fusogenic peptide of this kind moderately potentiates sequence-specific effects of particular concentrations of oligos. It is currently unclear whether the optimal effects follow from preparing oligo-bearing lipofection mixture with copresence of the peptide (on the order of 10 pM), adding peptide independently after lipofection, or adding to medium with the naked S-oligos. These effects have followed the ring-culture version of oligo treatment with brief off-membrane preincubation. Since there is no evidence for toxicity of the required concentrations of peptide, it now remains to combine this with the recent finding that continuous presence of oligo is effective in roller-bottle culture as described above.
3. Problems with embryo quality: Eggs from appropriate suppliers (see Subheading 2.) are usually 90%+ fertile, and develop normal embryos with high reliability. From late autumn to January (under UK conditions), there is a decline in fertility %, and during hot midsummer weather, one in quality of embryos, coupled with an increase in spontaneous abnormality. The latter may be owing to extensive periods spent by unincubated eggs at temperatures too close to incubation (25°C+). These times of year are therefore not recommended for starting this work, although with experience they need not stop continuation of studies.
4. Preparing to start work: The setting up of ring cultures, removal of embryos from membranes, and their incubation in dishes and replacement as described should all be practiced until good normal development is frequent after these procedures from the stages it is desired to work on. Development to heart formation and anterior somite segmentation can be achieved after following the whole proce dure with prestreak blastoderms, including brief (<1 h) off-membrane treatment, but this requires much more attention to detail than success with those from stage 4 (full streak) onward. It is hard to practice the lipofection without actual oligo, but all other aspects benefit from such practice before actual experiments are attempted. All the relevant sections of Subheading 3. should be read before starting out. The quicker the embryos pass through the stages of the method apart from the incubation itself, the better will be the biological results, though a careful experiment involving oligo treatment of say 20 blastoderms, to be ring-cultured as age-matched control and experimental samples of 10, involves most of a day. A good general guide is that embryos should be either incubating in oligos, or resting at low 20°Cs in a shallow layer of TCM:BSS, in their ring setups, or in a dish. They should not experience a sharp drop from incubation to resting temperature while being replaced on membranes after oligo treatment.
Was this article helpful?