Chapter 4: Individual gene function One of the most important types of experiments in genetics is to infer the normal function of a gene from mutations, by analyzing the relationship of the mutant gene to the wild-type gene (the nature of the mutation) and its phenotype. Additional information is gained by examining the relationship among alleles of the same gene by complementation analysis. Moreover, it is crucial to learn where and when each gene acts by site-of-action and time-of-action experiments.

4A. Inferring gene function from mutations The function of an individual gene can be assessed by perturbing its activity and observing the consequences for the phenotype of the organism.

Loss-of-function are the gold standard for inferring gene function The strongest inference of gene function can be made by completely eliminating the function of the gene. Other types of alterations in gene activity can be highly informative but occasionally misleading, as discussed below. If we remove a gene A from the genome and the organism is longer than the wild-type organism (that with the a+ allele), we infer that A is necessary to limit the size of the organism. Since geneticists often name genes based on their mutant phenotypes, such a gene might be called long1. Indeed, genes in C. elegans named lon for long are required to limit body size. Going in the other direction, the wee1 gene in Schizzosacharomyces pombe is necessary for appropriate cell size. Because loss-of-function is the gold standard for genetic inference, great care is taken to asses whether a particular allele is loss-of-function. Figure 1. Loss-of-function phenotypes. examples.. We use the term gene activity as an abstract concept of function. The concept of activity includes any function of a gene including the ability of its product to catalyze a reaction, to bind to another molecule, or to act as a nutrient reservoir. How do we know if an allele lacks activity?. A molecular definition of loss-of-function is that the DNA corresponding to the gene is absent from the organism. Of course, this is not typically the case, and thus other criteria are used. A DNA null lacks DNA. An RNA null lacks detectable mRNA. A protein null lacks detectable protein product. One important caveat to using gene products (RNAs or proteins) as evidence is that a gene could have activity at a low level of product and that with available reagents or methods we cannot detect that product but there is nonetheless enough present in the right place and right time to provide sufficient gene activity. Furthermore, we might not know what activities are present in a protein fragment. Another molecular criterion is that the gene lacks the ability to make the crucial functional domain of its product. For example, a gene encoding an ion channel, which lets ions pass through a cell membrane usually in a regulated fashion, might have an allele that would not be able to make the trans-membrane pore. Or, a gene encoding a protein tyrosine kinase might have an allele that lacks the ATP-binding region of the kinase domain. Of course, these interpretations make the assumption that we know the functional domains of the protein. There are many cases in which the known functional domain(s) of a protein comprise only a small portion of the mass of the protein. Proteins domains are discussed in more detail in Chapter 8. Genetic tests can indicate that a mutation is not loss of function    

 

2011 P.Sternberg & L Huang    ch. 4A    4/11/11   

1 

If a diploid is constructed such that it has one copy of the mutant allele and lacks the other copy of the gene (mutation is placed in trans to a deletion of the locus), one can test whether the mutation behaves like a deletion of the gene. Specifically, if the heterozygote b1/deletion does not have the same phenotype as the homozygous b1/b1, we can conclude that that the b1 allele is not complete loss-of-function. Other ways of reducing gene activity can also be used to test the “null hypothesis.” For organisms for which RNAi or equivalent knock-down method is available, that method can be used. For example, if an untreated homozygous strain [b1/ b1] does not have the same phenotype as the same homozygous strain b1/ b1 treated with RNAi against gene B, then b1 is not complete loss-of-function. This sort of analysis is easiest to interpret if by strength of phenotype we mean penetrance or expressivity, both of which are quantitative measures (Figure 2). A minor caveat to the use of deletion is that by definition, the deletion likely removes more than the locus and thus might remove one copy of a negative regulator (or other suppressor; see Chapter 5). (c1 +/ Df[clf dsup]). Specifically, if the there is a dominant suppressor (dsup) of the loss-of-function phenotype of gene c, then a deletion that removes both c and d would alter the observed phenotype. BOX: genetic definition of deletion Null mutations can be constructed Given a genome sequence a definitive way of obtaining a null mutation is to construct one, and this is possible in some organisms (Figure 3). In budding yeast this is straightforward. In mouse it is standard practice, but the locus size often means only part of the gene is removed (hence the discussion about functional domains above). The known sequence is used to design and generate reagents, in particular to obtain a knockout of the gene. This process can be laborious, but the quality of the inference makes it worthwhile. Figure 3. Knock-out strategies in various organisms. Mouse, yeast, C. elegans, Drosophila, dictyostelium. Chapter 8 describes finding alleles by re-sequencing the genomes of mutant strains. ZFNs Frequency of obtaining mutations can suggest the nature of the mutation. In general, there are more ways to reduced or disrupt the activity of a gene by mutation than to alter or increase its activity. Therefore, the frequency of obtaining loss-of-function mutations is typically higher than the frequency of obtaining other types of alleles. A number of codons can mutate to stop codons with a single nucleotide substitution. For example, mutation of tryptophan codon TGG in the DNA to stop codon TAG would terminate translation. Insertion or deletion of 1-2 base pairs will alter the reading frame (frameshift mutation) deleting all downstream codons, inserting some number of abnormal codons and leading to a premature stop. The effect of a frameshift mutation will thus depend on the protein and location. For example, in Figure 1052 B, a frameshift in the first half of the proteins would likely disrupt activity, but one later in the gene would disrupt the negative regulatory domain. In many cases, a long untranslated region after a stop codon or the stop codon in the altered reading frame will result in the mRNA being degraded by the Nonsense Mediated Decay (NMD) pathway, and could thus result in a loss-of-function even though the conceptual translation product would be a gain-of-function! Figure 1052. The frequency of obtaining classes of mutations. A. standard protein. B. Particularly pernicious protein with big negative regulatory domain! [Figure SGF-1052.]

   

 

2011 P.Sternberg & L Huang    ch. 4A    4/11/11   

2 

Figure 4. Example of mutation to stop and to frameshift.

Mutations can be classified by their effect on gene activity More generally, we want to infer the wild-type activity of a gene from the properties of the mutation. We thus need to know what the mutation does to the activity of the gene. H. Muller in 1932 defined classes of alleles. An amorph completely eliminates activity of the gene. A hypomorph has less than the wild-type level of activity. He defined three classes of gain-of-function mutations: hypermorph, neomorph and antimorph. lf, loss-of-function (amorph) rf, reduction-of-function (hypomorph) gf, gain-of-function (anti-, hyper-, neo-morph) dn, dominant negative (antimorph) Muller’s definitions: A hypomorph can be detected if the phenotype is enhanced in trans to a deletion of the locus. If h1/h1 is less severe (less penetrant or less highly expressed) than is a1/deletion, we infer that only partially decreases gene activity. If multiple copies of h1 leads to a less severe phenotype than does two copies of h1, we infer that h1 has partial activity and this can be increased by additional alleles. Genotype +/+ h1/+ h1/h1 h1/deletion h1/h1/h1

Activity 100% 100% 30% 15% 60%

Interpretation wild-type control allele is recessive mutant control allele is enhanced in trans to a null increased copy number has a less severe phenotype

A neomorph has altered activity that is qualitatively distinct from wild type. The wild-type allele acts like a null with respect to the neomorph. Neomorphs are often dominant. One way that neomorphs are recessive is if the wild-type product competes with the neomorph for interacting molecules. An antimorph encodes a poison product. These are often known as “dominant negatives,” an evocative name, but strictly speaking, antimorphs do not have to be dominant.

   

 

2011 P.Sternberg & L Huang    ch. 4A    4/11/11   

3 

Dosage analysis alters the numbers of copies of an allele. For example, a hypermorph can suppressed in trans by Df or enhanced in trans by extra copy of wild-type or mutant allele. In a simple example (Table Hypermorph), assume that a wild-type allele provides 50% activity, and thus the diploid with two wild-type alleles has 100% activity. A hypermorphic mutant allele H provides 150% activity (three times the wildt-type level). Homozygous H has 300% activity. A strain that is hemizygous for H (H/Df where Df is deficiency) has 150% activity, and thus mutant in trans to a deficiency is less severe than the homozygote, by contrast to the hypomorph discussed above. Additional copies o either H or wild-type allele exacerbate the phenotype. Three copies of H would have 450% activity, while H/H/+ would have 350% activity compared to 300% for H/H. Multiple copies of wild-type allele would have more activity and could confer a phenotype. We have not assumed any specific relationship between gene activity and phenotype, except that more activity has a more severe phenotype. In practice the relationship of gene activity to phenotype does not have to be linear or even monotonic (steadily increasing or decreasing). Table Hypermorph. Genotype Activity H/H/H 450% H/H/+ 350% H/H 300% H/+ 200% +/+/+ 150% H/Df 150% +/+ 100%

Interpretation increased copy number has a more severe phenotype increased copy of wild-type has a more severe phenotype mutant control allele is dominant increased copy of wild-type has a more severe phenotype allele is suppressed in trans to a null wild-type control

An allelic series can imply gene activity. A series of alleles each with different level of activity can be ordered and the hypothesis that they fall along a single monotonic activity curve tested by examining the phenotypes of heterozygotes. The simple prediction is that the activity of a heterozygote would lie between the activities of the two homozygotes. genotype +/+ j1/j1 j2/j2 j3/j3 j4/j4 j5/j5 j6/j6

activity 100% 95% 70 40 20 10 3

j1/j5 j5/j6 j2/j6 j3/j4 j2/j5

50% 5% 36% 30% 40%

Mutations that cause gain-of-function can be constructed. Gain of function mutations are of a number of types: some increase an essentially wild-type activity, some have a wild-type activity at the wrong place, time or under wrong conditions; yet    

 

2011 P.Sternberg & L Huang    ch. 4A    4/11/11   

4 

others confer a novel activity. Addition of an inappropriate localization signal to a protein or RNA can make it a neomorph. Mulitiple copies of the wild-type gene can increase the amount of wild-type gene product made. Similarly, expression of a cDNA under the transcriptional control of a strong enhancer/promoter can also make a gain-of-function genotype. Mutations that are dominant negative mutations can be constructed Dominant negative alleles exert their effect often by non-productive interactions. For example, the DN protein binds a target but does not have catalytic activity, and prevents the wild-type protein from accessing the target. Site-specific DNA-binding proteins such as transcriptional regulators often have separable DNA-binding and regulation domains. A protein with intact DNA-binding but no activation domain can prevent the wild-type protein from functioning. figure with various DNs (eg. from Herskowitz 1985) Loss-of-function mutations can be dominant As discussed in Chapter 3, alleles can be dominant, recessive or in between (co-dominant). While there is a correlation of the nature of the allele and its dominance, it is by no means absolute. Loss-of-function alleles are typically recessive to the wild-type allele, because the wild-type allele exerts its effect. However in some cases a loss-of-function allele is dominant because halving gene dosage causes defects. Such dominant or semi-dominant loss-offunction mutations are referred to as haploinsufficient. gain-of-function mutations can be recessive Figure with subunit constitution Infection with engineered viruses can perturb gene activity. as a way of perturbing gene activity Multiple alleles of genes can help define gene function. Having more than one class of allele in hand can more accurately indicate gene function. Genes with opposite loss- and gain- of function phenotypes. Having both lf and gf alleles can help define gene function. One of the simplest cases is that lf alleles cause one phenotype while hypermorphic alleles cause the opposite phenotype. For example, C. elegans lin-14(lf) cause a precocious phenotype in which early developmental events are skipped, while lin-14(gf) causes a retarded phenotype in which early developmental events are reiterated (Figure …). The lin-14(gf) alleles make mRNA that lack a negative regulatory site in their 3’UTR. This regulatory site is the taret of the lin-4 microRNA (miRNA)

   

 

2011 P.Sternberg & L Huang    ch. 4A    4/11/11   

5 

[from TIGs]

Genes can have similar loss- and gain- of function phenotypes. For some genes loss- and gain- of-mutations have the same phenotype. Yeast SEC4 is a GTPase of the ras superfamily. G proteins undergo a cycle of guanine nucleotide (the G) exchange and hydrolysis Figure ). They exist in three states: empty, bound to GTP, and bound to GDP. GTP is at higher concentration than GDP in the cell, so the empty protein binds GTP mre often than GDP. GTP binding changes conformation of the protein (the active state). The G protein has an intrinsic GTPase activity that releases Pi leaving GDP bound. The GDP then falls off the G protein. The rates of GTPase and GDP exit are different in different proteins. They are often slow and need to be accelerated by regulatory proteins: GTPase Avctivating Proteins (GAPs), and Guanine Nucleotide Exchange Factors (GEFs). These proteins have well defined gf mutations: inactivation of the GTPase activity leaves the protein in an “active” conformation. However, sec4(gf), with an activated GTPase, and sec4(lf), with no functional protein, have the same defect in secretion. The simplest interpretation of this apparently paradoxical result is that it is crucial for SEC4 to complete its GTPase cycle in order for secretion to proceed.

Figure . G protein cycle of guanine nucleotide exchange and hydrolysis. Intrinsic GTPase is stimulated by GAP. Release of GDP is stimulated by EF. Since cellullar [GTP] >> [GDP], release of GDP allows GTP binding and hence activation.

 

  BOX: dn and gf RAS in C. elegans dominant negative alleles can interfere with either activation of protein or its effect. gf/gf Multivulva gf/+ Multivulva dn/dn Vulvaless dn/+ Vulvaless dn/gf Multivulva    

 

2011 P.Sternberg & L Huang    ch. 4A    4/11/11   

6 

  Additional Reading Veitia RA. Dominant negative factors in health and disease. J Pathol. 2009 Aug;218(4):409-18. Review. PubMed PMID: 19544283. Herskowitz I. Functional inactivation of genes by dominant negative mutations. Nature. 1987 Sep 1723;329(6136):219-22. Review. PubMed PMID: 2442619. Muller, H. J. in Proceedings of the Sixth International Congress of Genetics (ed. Jones, D.F.) 213−255 (Brooklyn Botanic Gardens, Menasha, Wisconsin, 1932). Levchenko, A., Bruck, J. and Sternberg, P. W. (2001). Combination of biphasic response regulation and positive feedback as a general regulatory mechanism in homeostasis and signal transduction. In H. Kitano, ed. Foundations of Systems Biology. Cambridge, Massachusetts: M. I. T. Press, pp. 263-278. [is this available online???]

   

 

2011 P.Sternberg & L Huang    ch. 4A    4/11/11   

7