nature methods | VOL.7 NO.6 | JUNE 2010 | 439

news and views

K.M. Philos. Trans. R. Soc. Lond. B Biol. Sci. 361, 2155–2172 (2006).

5. Le Bars, D., Gozariu, M. & Cadden, S.W. Pharmacol. Rev. 53, 597–652 (2001).

6. Whiteside, G.T., Adedoyin, A. & Leventhal, L. Neuropharmacology 54, 767–775 (2008).

7. Caruthers, P. Language, Thought and Consciousness (Cambridge University Press, Cambridge, 1996)

8. National Research Council Recognition and Alleviation of Pain in Laboratory Animals. (The National Academies Press, Washington, DC, 2009).

9. Kadosh, K.C. & Johnson, M.H. Trends Cogn. Sci. 11, 367–369 (2007).

ourselves but also as a means of improving the welfare of the animals we use?

ComPetinG FinanCiaL interests The author declares no competing financial interests.

1. Ranger. M., Johnston, C.C. & Anand, K.J.S. Semin. Perinatol. 31, 283–288 (2007).

2. Langford, D.J. et al. Nat. Methods 7, 447-449 (2010)

3. Flecknell, P.A. & Roughan, J.V. Anim. Welf. 13 (Suppl.), S71–S75 (2004).

4. Tate, A.J., Fischer, H., Leigh, A.E. & Kendrick,

be markers of pain in this model. One long-standing debate has been the nature of pain in animals. In people, pain has both sen-sory (what type of pain, where it is and how intense it is) and emotional (how it feels) components. It is the emotional component that makes pain unpleasant and distress-ing. In animals, it has been suggested that this emotional component is either greatly reduced or completely absent7. However, a recent consensus view was that animal pain does have an emotional component, but it was acknowledged that measuring this component would prove difficult8. It is pos-sible that assessment of facial expressions, which may measure the emotional com-ponent, could be of value in this context. Assessing ‘pain faces’ in animals may also be easier for humans to learn than other behavioral measures because we have spe-cialized neural apparatus for attending to and processing faces9.

The methodology2 is at an early stage of development, but Mogil and col-leagues have already completed a broad range of investigations using different pain assays to substantiate their initial studies. It remains to be seen whether the technique proves more sensitive and better able to detect changes in pain state than other assays used in pain research. It may prove to be an adjunct rather than a replacement for current approaches, or it may primarily be measuring a different dimension of pain. The authors2 point out that other behavioral statesfor example, sleephave in common some, but not all, facial expressions with pain faces, and other measures may be needed to differentiate between other situations that could produce some of these changes in expression.

Despite these caveats, this study could lead to a radical change in approach to assessing pain, not only in animal models in pain research but more generally in ani-mal husbandry. Changes in legislation in Europe will require more detailed evalua-tion of the welfare of laboratory animals, and assessment and alleviation of pain will be of central importance. Similarly, the last decade has seen an upsurge in interest in the welfare of other animals, notably farm and companion animals. If mice and human infants have a pain face, do rats, rabbits, dogs, cattle and the other species that are used by society? If they do, can we use this not only as a research tool to develop better pain therapies for

Knock it down, switch it onJean-Louis Bessereau

The arsenal of methods to investigate gene function in Caenorhabditis elegans continues to grow—with new approaches to generate targeted deletion mutants and to control gene expression.

The genetic toolbox for a model organ-ism (Figure 1) should contain efficient methods for targeted gene inactiva-tion and conditional gene expression. Surprisingly, these two major tools remain relatively unsatisfactory for the nematode Caenorhabditis elegans. In recent papers in Nature Methods1,2, researchers in the laboratories of Erik Jorgensen and Martin Chalfie independently report new strat-egies that should expand the ability to control gene expression in the worm and could even be a source of inspiration for developing techniques in other systems.

A standard and powerful strategy to get at the function of a gene is to inactivate its expression and analyze the resulting phenotype, a so-called ‘reverse genetics’ approach. The sequencing of the C. ele-gans genome and the discovery of RNA interference (RNAi) opened up the excit-ing perspective of being able to inactivate every worm gene3. RNAi is easily achieved in the worm and has thus become widely used for C. elegans research, but it has a few intrinsic limitations. First, RNAi effi-ciency is very sensitive to the experimen-tal conditions, and output can be variable. Second, residual gene expression persists to an extent that is difficult to predict for a given gene. Third, some tissue types such

as neurons are partially resistant to RNAi. Fourth, genes sharing sequence similar-ity with the primary target can sometimes be undesirably downregulated. Obtaining strains containing heritable null mutations in every gene therefore remains comple-mentary to RNAi-based analysis.

In C. elegans, most strategies used so far to isolate gene knockouts relied on randomly mutagenizing large worm populations and then screening by PCR for the presence of a specific deletion. Thanks to the efforts of the C. elegans Gene Knockout Consortium in the United States and Canada and the National BioResource Project in Japan, deletion alleles have been obtained for about 5,500 out of 20,000 predicted genes4. These strains are extremely useful but also have a few limitations. First, deletions are usu-ally small and are sometimes not molecu-larly null mutations. Second, strains have been heavily mutagenized, and mutations are ineluctably present in the background, sometimes tightly linked to the deletion allele. Third, deletions can be associated with complex chromosomal rearrange-ments. Fourth, some genes are difficult to target because they are small or because null alleles cause lethality and sterility, and are therefore difficult to recover from the

Jean-Louis Bessereau is at Institut National de la Santé et de la Recherche Médicale U1024, Institute of Biology of the Ecole Normale Supérieure, Paris, France. e-mail: jlbesse@biologie.ens.fr

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

440 | VOL.7 NO.6 | JUNE 2010 | nature methods

news and views

and at different developmental stages, especially when complete loss of func-tion causes early lethality or sterility. In C. elegans, which grows between 15 °C and 25 °C, one of the first efficient ways to achieve conditional gene expression relied on temperature-sensitive mutations iso-lated in forward genetic screens. However, these mutations are rare. In an alternative strategy, loss-of-function alleles can be complemented with transgenes driving conditional expression of the gene prod-uct. This was initially achieved using heat-shock promoters, but such systems are not tightly controlled. More recently, tissue-specific expression of Cre or FLP recombinases has been used to control gene expression from transgenes with lox sites or FLP recombinase target sites8–10. Such strategies provide increased flexibil-ity for time- and cell type–specific expres-sion, but the constructs to be recombined are typically present as multiple copies in repetitive transgenes, which can poten-tially cause uncontrolled recombination among distant copies or incomplete rear-rangements in the transgenes.

Calixto et al.2 report an elegant system built upon decades of genetic dissection of mechanosensation in the worm. They demonstrate that the RNA-processing factor MEC-8 regulates the alternative splicing of mec-2, a gene encoding a com-ponent of the mechanosensory trans-duction complex. MEC-8 is required to remove the ninth intron of the mec-2 pre-mRNA, which otherwise contains a stop codon and causes premature termination of the message. Critically, a fragment con-taining this mec-2 intron retains MEC-8–dependent splicing properties, so that expression of any open reading frame pre-ceded by this fragment can be turned on and off posttranscriptionally by MEC-8. This strategy is especially interesting because of a temperature-sensitive mec-8 allele: worms with this allele express func-tional MEC-8 at 15 °C but not at 25 °C. In turn, an expression vector containing the mec-2 intron 9 drives expression of a gene of interest at 15 °C but not at 25 °C in these mutant worms.

Because MEC-8 regulates the splicing of genes other than mec-2, it will be impor-tant to determine that the process of inter-est does not depend on mec-8 expression before using this system. With this restric-tion in mind, this conditional expression system is extremely appealing. First, the

Second, deletion endpoints are specified in the repair template, which provides a way to design molecularly null mutant alleles. Third, the risk of background mutations is minimized. Putative reinsertion of the excised transposon can easily be checked by PCR using Mos1-specific primers. Fourth, the presence of a positive selection marker in the deletions permits lethal or sterile mutations to be maintained.

An obvious drawback of this tech-nique is the necessity for a Mos1 inser-tion in the relevant region of the genome. The Nematode Gene-Tagging Tools and Resources (NemaGENETAG) consor-tium has generated about 14,000 strains with identified Mos1 insertions7. Frøkjær-Jensen et al.1 calculated that 99.4% of worm genes fall within 25 kilobases of at least one of these insertions and that 44% of genes (8,833) can be selectively deleted without perturbing the coding sequence of other genes. Obtaining strains with additional transposon insertions would certainly expand the possibilities of cus-tomizing the C. elegans genome. It also remains to be evaluated how large dele-tions and genomic insertions of a positive selection marker might affect the expres-sion of nearby loci.

Last month in Nature Methods, Calixto et al. described a clever strategy to achieve conditional gene expression in C. elegans2. Such techniques are essential to analyze genes that function in different tissues

mutagenized population. The MosDEL method reported by Frøkjær-Jansen et al. in this issue1 provides an interesting complement to previous strategies for generating gene knockouts.

MosDEL relies on Mos1 excision–induced transgene-instructed gene conversion (MosTIC), a gene-targeting technique recently developed in C. ele-gans5. It starts with a strain containing an insertion of the Drosophila mauriti-ana transposon Mos1 in the locus to be manipulated. Expressing the Mos trans-posase triggers Mos1 excision and causes a DNA double-strand break. A transgene containing sequence homologous to the broken locus can then be used as a repair template. Sequence variations contained in the transgene will be copied in the genome. The MosTIC technique was pre-viously used to engineer point mutations or to knock in GFP within one kilobase of the breakpoint5,6. In MosDEL, the repair template is slightly different: it contains on one side sequence adjacent to the bro-ken locus and on the other side sequence located up to 25 kb away, with a positive selection marker in between1. Thus, its use results in deletion of the interven-ing sequence and insertion of the marker gene.

MosDEL has numerous advantages. First, the technique is simple and efficient because the deletion strains are identified within a week using simple phenotypic screens.

Figure 1 | More tools for worm research.

Kim

Cae

sar

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

nature methods | VOL.7 NO.6 | JUNE 2010 | 441

news and views

ComPetinG FinanCiaL interests The author declares no competing financial interests.

1. Frokjaer-Jensen, C. et al. Nat. Methods 7, 451-453 (2010).

2. Calixto, A., Ma, C. & Chalfie, M. Nat. Methods 7, 407–411 (2010).

3. Fraser, A.G. et al. Nature 408, 325–330 (2000).4. Moerman, D.G. & Barstead, R.J. Brief. Funct.

Genomics Proteomics 7, 195–204 (2008).5. Robert, V. & Bessereau, J.L. EMBO J. 26, 170–183

(2007).6. Gendrel, M. et al. Nature 461, 992–996 (2009).7. Bazopoulou, D. & Tavernarakis, N. Genetica 137,

39–46 (2009).8. Macosko, E.Z. et al. Nature 458, 1171–1175

(2009).9. Davis, M.W. et al. PLoS Genet. 4, e1000028

(2008).10. Voutev, R. & Hubbard, E.J. Genetics 180, 103–119

(2008).

minimally overlapping expression pat-terns to independently drive mec-8 and the mRNA containing the mec-2 intron 9 sequence. It is even conceivable that heat-shock promoters or recombinase systems could be used to achieve condi-tional expression of MEC-8 and in turn to increase the flexibility and accuracy of transgene expression control.

The clever techniques developed by the Jorgensen and Chalfie groups add to the ingenious tools already in place to manip-ulate the C. elegans system. Combining several of these techniques should make it possible to address gene function in vivo with unprecedented accuracy.

control of mec-2 intron 9 splicing by MEC-8 seems very tight, thus avoiding expression leakage at the nonpermissive temperature. Expression can be turned on within about an hour, but notably, MEC-8 activity seems to perdure for a prolonged period after shifting to restrictive tem-perature, which is not ideal to analyze ‘off switches’. Second, this system can be used in most cell types because MEC-8 is broadly expressed from early embryo to adulthood. Precisely defining the cells that may not contain MEC-8 will be interest-ing in the future. Third, it may be possible to restrict the expression of a transgene to a very few cells by using promoters with

© 2

010

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.