Worm Breeder's Gazette 10(2): 89

These abstracts should not be cited in bibliographies. Material contained herein should be treated as personal communication and should be cited as such only with the consent of the author.

Sense and Antisense Disruption of Muscle Genes

A. Fire and S. Harrison

Figure 1

J.  Izant and H.  Weintraub originally proposed the use of antisense 
RNA as a technique to disrupt expression of specific genes in vivo.  
This approach appears to work for C.  elegans muscle proteins.
We have used the plasmid 'pPD#12.1' containing sequences from the 
unc-54 gene to provide the 'promoter' in these experiments.  In fact 
pPD#12.1 contains the first four exons and three introns of pUNK54 as 
well as 3' flanking DNA.  The reason for including these sequences 
stems from the complicated requirements for unc-54 gene expression (
see previous article).  Into this vector we have inserted segments 
from a variety of muscle genes.  The resulting constructs are injected 
into wild type (N2) animals, and we look in the next generations for 
animals that are odd or uncoordinated.  Another set of experiments has 
been done with the whole unc-54 gene with and without the internal 2.
8kb BamHI-B fragment in inverted orientation.
[See Figure 1]
The 'odd phenotypes' described occur at a fairly high frequency, 
particularly with the antisense unc-54 constructs, for which as many 
as 20% of the injected oocytes give rise to unc worms with about 10% 
of these giving rise to some unc progeny and eventually heritable 
lines.  The unc-54 sense construct (i.e.  the whole gene) has about 
the same frequency of heritable transformation, but only rarely gives 
the transient-only expression.
These data seemed complex but mildly encouraging.  Injection of the 
antisense constructs was giving a phenotype, and the phenotype was 
roughly correlated with that expected for the fragment inserted into 
the antisense vector.  The odd result was that some, but not all, of 
the 'sense' constructs also gave rise to uncoordinated phenotypes 
correlated with the insert.  To see what was going on we have further 
characterized the transformed lines that came from these experiments (
to date about thirty such lines have been generated).  The lines have 
the injected DNAs present at very high copy number (several hundred) 
in the type of tandem arrays originally described by Stinchcomb et al. 
In most cases the array behaves genetically as though it is an 
extrachromosomal element, although in two cases the large arrays 
appear to have integrated.  The lines are phenotypically variable, 
with some of the extrachromosomal lines exhibiting mosaicism by 
polarized light microscopy (The latter observation thanks to R.  
Waterston).  Although very severe in some cases, it is unlikely that 
the mutants are completely 'null' for the function of respective 
endogenous genes, since some of the unc animals can lay eggs.
The most striking results came from staining these lines with 
antibodies: our usual mixture of monoclonals from D.  Miller (
Fluorescein-5.8[anti unc54] + Rhodamine-5.6[anti myo3).  The twitcher 
lines transformed with unc-22 antisense constructs have disorganized 
filament structure but have ratios of myo-3 to unc-54 that are roughly 
equivalent to wild type.  In order to look at the unc-54 sense and 
antisense transformed lines, it was advantageous to first look at myo-
3 staining (using the Rhodamine channel).  In general there were two 
types of cells: cells with a well organized myo-3 pattern, and cells 
in which the myo-3 pattern was disorganized.  The former group of 
cells generally exhibit a wild type staining pattern with unc-54 
antibody.  Such cells are probably the normal (untransformed) cells in 
these mosaic animals.
An obvious difference between the 'sense' and 'antisense' unc-54 
lines came from looking at the distribution of unc-54 protein in the 
cells with a poorly organized myo-3 pattern.  The lines transformed 
with the whole unc-54 gene that exhibit an unc-54 like phenotype 
actually seem to overexpress the unc-54 product, so that the abnormal 
body wall muscle cells are full of aggregated, nonfilamentous unc-54 
protein.  In contrast, in lines transformed with the unc-54 antisense 
constructs the disorganized muscle cells seem to have almost no unc-54 
protein as observed by staining [these cells actually stain more 
brightly with the anti-myo3 antibody, a phenomenon often seen (R.  
Barstead p.c.) in mutants with disorganized muscle.
The over-expression phenotype in transformed lines carrying a high 
copy numbers of an intact unc-54 gene was actually not new.  In our 
experiments to map out determinants required for unc-54 expression, we 
have occasionally obtained lines with a high copy number of the 
injected DNAs.  In general these lines behave as very unstable free 
duplications and give an odd partially rescued unc-like phenotype.  
Excess unc-54 protein is present in the body wall muscle of these 
animals.  It seems likely that what happens in these lines is too much 
rescue, with the result being the somewhat unc phenotype.  The 
consequent suggestion is that our assay for unc-54 function in an e190 
background is actually picking up expression only within a certain 
fixed range.
There are three ways that the various 'sense' and 'antisense' 
constructs could be causing their unc phenotypes.
1) A peptide which interferes with muscle function or gene 
expression is being produced by the transgene.
2) A factor which is involved in normal gene expression is being 
specifically titrated out by the large tandem array of sequences from 
the gene.
3) Antisense RNA is being transcribed from the transgene DNA and 
interfering with expression of the wild type gene.
Possibility 2) can be readily ruled out as the primary cause of 
these observations since the sense orientation of the unc-54 SstIB 
fragment has exactly the same DNA sequences as the antisense 
orientation but does not lead to unc worms (or to massive lethality).
It is difficult to completely rule out the first possibility.  All 
of the constructs so far tested have the ability to make some 
polypeptide.  In the case of the unc-54 antisense constructs, the 
argument could be made that the synthesis of these aberrant proteins 
could actually interfere specifically with the synthesis of the 
endogenous unc-54 product (but not with myo-3).  It is not clear just 
how this specificity would arise, and we don't consider this 
possibility very likely.  Making some 'antisense' constructs with lots 
of frameshift mutations in the different reading frames might help 
resolve this.
Given that we think that the apparent disruption by the 'antisense' 
constructs indeed proceeds through the production of antisense RNA, it 
is somewhat surprising that in the case of unc-22, the apparent gene 
disruption is not completely dependent on the orientation of the 
insert.  Avid WBG readers may remember an article by A.  Fire and D.  
Moerman in vol. 9#2 in which it was reported that certain (but not all)
fragments from the inside of the unc-22 gene can cause twitching 
behavior when injected.  In those experiments there was no attempt to 
provide promoter sequences, but given that disruption was occurring 
either by peptide production or by antisense, the tandem arrays 
present in these animals must have been transcribing in the absence of 
any obvious promoter sequence.  This suggests that the long tandem 
arrays, either because of their copy number or because of some aspect 
of chromatin structure can give rise to 'spurious' transcripts not 
starting in a physiological promoter region.  If this is true then it 
is quite possible that the unc-22 'sense' constructs could actually be 
disrupting as a result of some relatively low level transcription on 
the negative strand.  This effect may be much less prevalent for unc-
54 as a result of the greater abundance of the endogenous unc-54 
message, hence requiring a greater amount of antisense RNA for 
disruption.  Of course it is also quite reasonable for the unc-22 
'sense' constructs that disruption could be occurring by expression of 
an unc54-unc22 fusion peptide, since the junction in the constructs 
probably occurs in frame.
To conclude, it seems remarkably easy to specifically disrupt the 
function of specific genes.  We believe that unc-54 and unc-22 
expression can be disrupted by expression of corresponding segments of 
antisense RNA using an unc-54 derived vector.  In addition, muscle 
function can apparently be disrupted by over-expressing the unc-54 
protein.  In any case the plasmids described should be useful as 
dominant markers for high copy-number tandem array transformation.

Figure 1