Media Releases

Yeast Against the Machine: Bakers’ Yeast Could Improve Diagnosis

April 6, 2016

How our billion-year-old cousin, baker’s yeast, can reveal — more reliably than leading algorithms — whether a genetic mutation is actually harmful.

Toron­to, ON – It’s eas­i­er than ever to sequence our DNA, but doc­tors still can’t exact­ly tell from our genomes which dis­eases might befall us. Pro­fes­sor Fritz Roth is set­ting out to change this by going to basics — to our bil­lion-year-old cousin, baker’s yeast.

By test­ing the effects of human muta­tions in yeast, Roth’s research team at the Uni­ver­si­ty of Toronto’s Don­nel­ly Cen­tre for Cel­lu­lar and Bio­mol­e­c­u­lar Research and the Lunen­feld-Tanen­baum Research Insti­tute was able to iden­ti­fy harm­ful changes in the DNA bet­ter than lead­ing algo­rithms. The ulti­mate goal of his approach, detailed in the lat­est issue of Genome Research is to cre­ate “look-up tables” of dam­ag­ing muta­tions to help clin­i­cians diag­nose patients more accu­rate­ly.

The rea­son our genomes remain impen­e­tra­ble is the vast amount of genet­ic diver­si­ty among us. Of the three bil­lion DNA let­ters in the genome, three mil­lion are dif­fer­ent between any two peo­ple. The vast major­i­ty of these dif­fer­ences, also called genet­ic vari­ants, have no bear­ing on our lives. But some vari­ants change pro­teins, the mol­e­c­u­lar machines that do much of the work in our cells — and this could lead to dis­ease.

“If we want to inter­pret people’s per­son­al genomes, then we need a way of know­ing whether vari­ants are dam­ag­ing the gene they are in,” says Roth, who is also a pro­fes­sor in the Depart­ment of Mol­e­c­u­lar Genet­ics and co-direc­tor of the Cana­di­an Insti­tute for Advanced Research Genet­ic Net­works Pro­gram.

Cur­rent­ly the only way to pre­dict dam­ag­ing muta­tions, for most genes, is through com­pu­ta­tion­al meth­ods. For some genes, how­ev­er, dam­ag­ing muta­tions can be detect­ed using yeast.  The inter­na­tion­al team led by Roth did a head-on com­par­i­son of yeast against the machine to see which approach fared bet­ter at find­ing dis­ease-caus­ing muta­tions.

Yeast cells are sim­ple, yet their basic archi­tec­ture is sim­i­lar to human cells. Because almost half of our genes have a shared ances­try with a yeast gene, it is often pos­si­ble to study human genes in this easy-to-manip­u­late liv­ing organ­ism.

One way to test a human gene’s func­tion is to see whether it can replace a yeast coun­ter­part gene. Think of yeast as a ship — tak­ing a gene out leaves a hole in the bot­tom. Sci­en­tists then try to stop the leak by plug­ging the hole with the match­ing human gene to pre­vent the ship from sink­ing. If the nor­mal human gene can res­cue the yeast but a mutat­ed one can­not, Roth pre­dicts that the muta­tion is dam­ag­ing. Thanks to yeast’s fast rate of growth, it is pos­si­ble to know with­in days which ver­sions of human genes fail to keep the yeast afloat. These same vari­ants are also like­ly to be dam­ag­ing for human cells and could mat­ter for our health.

Roth’s team focused on 22 genes, linked to con­di­tions such as autism, men­tal retar­da­tion and heart dis­ease, and whose intact copies were able to replace their yeast coun­ter­parts. Pre­vi­ous work found these genes to be present in hun­dreds of vari­a­tions among peo­ple. Roth’s group test­ed 179 vari­ants, rough­ly half of which are report­ed to cause dis­ease.

To test vari­ant func­tion, the researchers insert­ed each human vari­ant, one by one, in place of a match­ing yeast gene, using a com­pre­hen­sive library of yeast strains cre­at­ed by Pro­fes­sors Bren­da Andrews and Char­lie Boone’s groups at the Don­nel­ly Cen­tre. They then watched how well the yeast grew and this allowed them to pre­dict whether or not a vari­ant was harm­ful. Impor­tant­ly, this sim­ple test in a bil­lion-year old machin­ery out­per­formed the best avail­able com­pu­ta­tion­al meth­ods. Based on cell-growth data, the researchers were able to iden­ti­fy 62 per cent of dis­ease vari­ants as dam­ag­ing.  By con­trast, the best cur­rent com­pu­ta­tion­al method could con­fi­dent­ly iden­ti­fy only 23 per cent of dis­ease vari­ants.

“By every mea­sure we are beat­ing the com­pu­ta­tion­al pre­dic­tions. Some might say it’s obvi­ous that an exper­i­ment beats a com­pu­ta­tion­al pre­dic­tion, but many clin­i­cians would not accept evi­dence about human vari­ants based on how they per­form in baker’s yeast. Our paper high­lights the impor­tant and direct role that mod­el organ­isms can play in inter­pret­ing indi­vid­ual human genomes,” says Roth.

For the sub­set of human dis­ease genes that will be able to fill in for their yeast coun­ter­parts, Roth believes it is pos­si­ble to test all vari­ants this way. For oth­er genes, sim­i­lar assays could be done in oth­er mod­el organ­isms or using oth­er tests in yeast. The goal is to cre­ate lists of exper­i­men­tal­ly test­ed muta­tions before they are detect­ed in the genomes of affect­ed patients.

“I think the way to go for­ward is to do all of the exper­i­ments up front before you’ve even seen the vari­ants in the clin­ic. Orga­nized net­works of researchers could test the vari­ants in a com­mon way so that we can build a resource so that clin­i­cians can go imme­di­ate­ly to the look-up table,” says Roth.

Read what Prof. Roth thinks about home-based genet­ic tests in the Toron­to Star Doc­tors’ Notes

The basic con­cept of test­ing human gene vari­ants in yeast



For more infor­ma­tion:

Jovana Drin­jakovic, PhD
Writer at the Don­nel­ly Cen­tre
Uni­ver­si­ty of Toron­to
Tel: +41 78 929 06 14 (Zurich)