• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of plntcellLink to Publisher's site
Plant Cell. Dec 2007; 19(12): 3843–3851.
PMCID: PMC2217636

Exploring the Molecular Etiology of Dominant-Negative Mutations[W]

Herskowitz (1987) provided the classical definition of dominant-negative (DN) mutations as those leading to mutant polypeptides that disrupt the activity of the wild-type gene when overexpressed. Although Herskowitz' definition referred basically to intralocus interactions, it is now recognized that interlocus (e.g., trans-acting) interactions can also lead to dominance (Omholt et al., 2000) and to DN effects (DNEs; Veitia 2002). DNEs are thought to have an impact on plant genome evolution. Indeed, Gibson and Spring (1998) have suggested that paralogous genes encoding multidomain proteins might be overretained in polyploids because mutations in such genes could produce DN phenotypes. From a practical viewpoint, DN mutations have been widely used to study gene function in plants, for example, in situations where null mutations of functionally redundant genes result in no phenotype. Here, I explore the mechanistic foundations for DNEs and some of their quantitative aspects to describe the general features of their effects at the molecular level, improved understanding of which should help in the design of DN proteins.

Students are accustomed to simplified concepts of dominance and recessivity. These extreme terms stem from the analysis of qualitative characters for which only a few states are scored. More quantitative studies of the phenotypic effects of a mutation have led to proposals for a more exhaustive classification (Müller, 1932). The terms amorph, hypomorph, and hypermorph have been applied to alleles (and gene products) displaying no activity, lower, or higher activity than the wild-type allele, respectively. As noted above, an important source of dominant phenotypes is the production of a mutant protein that interferes with the action of the wild-type one, and alleles producing these DNEs are called antimorphs. The effects of strong antimorphs, such as those reviewed by Herskowitz (1987), can be easily understood. However, the quantitative dimensions of more complex DNEs require further examination in the era of high-throughput genetics and genomics.


The simplest examples of DN mutations come from factors that oligomerize, wherein the mutant polypeptide can poison the complex. For instance, in the context of a dimer AA, if the mutant protein a renders dimers Aa and aa inactive (which is easily conceivable, as we will see below), in the heterozygote there will be production of 25% AA, 50% Aa, and 25% aa, leading to only 25% of activity with respect to the wild type. This is to be contrasted with the case of a heterozygous A/- organism (carrying one wild-type and one loss-of-function allele), which is expected to have 50% activity relative to the wild type. This kind of reasoning has led to the proposal that heterozygous DN mutations cause more severe effects than simple null alleles of the same gene (Strachan and Read, 2004). Moreover, if a is overexpressed, the proportion of dimer containing only normal subunits will decrease. Thus, the current shorthand definition of a DN implies that (1) the phenotype of the heterozygote A/a is worse than that of A/- and (2) overexpression of a leads to a stronger phenotype in the heterozygote than when [A] = [a].

A textbook example of a DNE at the level of signal transduction involves a membrane receptor whose cytoplasmic portion contains a protein kinase (PK) domain. These receptors often dimerize upon interaction with an extracellular ligand, and the PK activity of one monomer phosphorylates the cytoplasmic side of the other monomer (i.e., cross-phosphorylation), which triggers signal transduction (Figure 1). In a heterozygous individual expressing one normal allele and one mutant allele encoding a protein lacking the cytoplasmic portion, only 25% of the dimeric receptor will work properly. Often 25% receptor activity is not enough to effectively transduce the signal, and so the actual impact of a DN allele may be greater than that of a knockout allele, where 50% activity might elicit a response (depending on the settings of the particular system and probably on the environmental conditions). The Leu-rich repeat receptor-like Ser/Thr kinase (RLK) ERECTA represents such an example in Arabidopsis. This protein is involved in the regulation of organ shape and inflorescence architecture. A truncated ERECTA protein lacking the cytoplasmic PK domain (Δ-kinase) produces a DNE when expressed under the control of the native ERECTA promoter and terminator. Transgenic plants expressing the mutant Δ-kinase display compact inflorescences and short, blunt siliques. In addition, the Δ-kinase protein enhances the phenotype of null erecta plants, and it cannot be excluded that it associates with other RLKs (and/or ligands shared by other RLKs) and inhibits their functions (Shpak et al., 2003).

Figure 1.
A Typical Example of a DN Mutation.

Other examples involve mutations in structural proteins that form polymers (see a general quantitative treatment for multimeric proteins in the Supplemental Materials online). Mutations in such proteins are frequently dominant, and heterozygous individuals carry a mixture of normal and abnormal protein molecules that copolymerize. For example, a missense mutation in the actin 2 gene in Arabidopsis (act2-2D) leads to defects in the initiation and elongation of root hairs, the elongation of root epidermal cells, and growth in aerial portions. Not surprisingly, in transgenic plants overexpressing the mutant gene, actin filament bundles in root epidermal cells are shorter than in the wild type or in the null mutants (Nishimura et al., 2003).

DNEs can also arise in newly formed polyploids and hybrids. During allopolyploid and hybrid formation, the parental genomes do not contribute equally to the transcriptome and the proteome of the resulting individual. The alleles contributed by each parent can be expressed at different levels and often encode different isoforms. Thus, a portion of the changes reported in newly formed polyploids and hybrids, either transcriptional or at another phenotypic level (Adams et al., 2003), could be due to DNEs in the context of multimeric complexes. If we consider a protein working as a homodimer, the orthologous proteins in the two parental lineages might have accumulated mutations that change their primary sequences but not the quaternary structure and the activity of the dimer. However, if the parental lines contribute incompatible forms (i.e., A and a), a tetraploid will have 50% of functionally impaired Aa dimers (i.e., here AA and aa are functional), which might have a phenotypic impact depending on the particular metabolic or signaling system and the environmental conditions.


DN mutations also can appear in homodimeric ligands (Figures 2A and 2B). Indeed, additional quantitative insights into DNEs can be obtained by examining the case of a ligand AA that interacts with a homodimeric protein receptor or a bipartite DNA or RNA sequence (RR). For simplicity, we will assume that dimers AA (and also Aa and aa in a heterozygote) are irreversibly formed and recognize the target RR with the affinity constants KAA, KAa, and Kaa, respectively. Based on this assumption, the activity of AA and its variants, Aa and aa, can be studied quantitatively by assessing the degree of saturation (Y) of their target: Y = [occupied RR]/[total RR]. For simplicity, we will further assume that the biological response is proportional to Y.

Figure 2.
DNE Mutations Can Also Appear in Homodimeric Ligands.

The curve representing Y versus protein concentration is a hyperbola ranging from 0 (no occupation of RR) and 1 (saturation) (formal treatment in the Supplemental Materials online). An analysis of this system allows us to explore some quantitative subtleties of DN mutations. We will consider two possibilities: (1) when a has completely lost its binding capacity and (2) when it has retained residual binding.

When there is coexpression of A and a and the latter has completely lost its binding capacity, according to the example of the dimer given in the introduction, intuition suggests that Y, at any point, will be ~25% of the level in the homozygote A/A. However, this tends to be so only for concentrations of A and a <1/KAA. Thus, a strong DNE is expected if the system works at low concentrations of A and a. As their concentrations increase, AA becomes saturating and easily displaces Aa, which binds the receptor much more weakly. In agreement with the textbook definition of DNEs, the curve for the heterozygous individual A/- is steeper and reaches saturation faster than that of A/a (Figure 3A). This example shows how the same mutation can behave differently according to the level of protein expression: it could be DN in one type of cell/tissue (low expression of A and a) and recessive in another (high expression). Moreover, differences in the levels of expression might explain why the phenotypes induced by the same DN mutation can differ between two species.

Figure 3.
Binding Curves for a Dimer AA (Aa or aa) Interacting with a Target RR.

This simple system is also instructive for understanding the interplay between hypomorphy and negative dominance. Let us consider that a has kept residual intrinsic binding activity (i.e., by definition, a is hypomorphic). This means that aa is able to bind the receptor, although more weakly than the wild-type dimer. One interesting hypomorphic allele encodes a monomer a, which has retained a binding capacity of ~1/3 of the wild-type activity (according to the model underlying the results presented in Figure 3). In this case, the curves for A/- and A/a are virtually superimposable. In other words, the mutant a, which is clearly hypomorphic, behaves as an apparently null allele in the presence of A. This might be explained by a molecular DNE in which part of the binding impairment of Aa is compensated by the residual binding of aa, so that total binding is similar to that of the heterozygote A/-. Of course, alleles encoding proteins with lower intrinsinc (i.e., per monomer) binding capacity will behave more like the traditional DN mutations.

DNEs are frequently studied by overexpressing the relevant allele, with the expectation that its overexpression will strongly affect the activity of the wild-type protein. This is always so for the extreme DN alleles (i.e., a is completely inactive). However, when a retains residual activity (i.e., slightly below one-third normal according to the model in Figure 3), a typical DNE is obtained when there is expression of A and a at similar concentrations. Interestingly, depending on the extent of the defect, when a is overexpressed, there may be conditions (two- or fivefold overexpression) that lead to a situation where there is higher binding activity than in A/- or in A/a when [A] = [a] (Figure 3B).

Some DN mutations lead to even more complex interactions, as observed in the case of a DN mutation of the PITX2a homeodomain protein that leads to a dominant eye disorder in human (Axenfeld-Rieger syndrome). PITX2a can form homodimers in the absence of DNA, but in its presence, binding is cooperative. Cooperativity in this context refers to the facilitating effect that one monomer bound to DNA (i.e., a form of the receptor RR) exerts on the binding of another monomer. The observation of cooperativity for PITX2a excludes the possibility that all the protein binds DNA as preassembled dimers. PITX2a monomers have a well-documented tendency to form dimers in solution, but when their concentration is low, most of them will recognize DNA as monomers. Then, an incoming monomer can be attracted concertedly by the bound monomer and by a nearby free DNA binding site (Figure 2C). This effect is pronounced at low protein concentrations and generates cooperativity, while at higher concentrations, preassembled dimer will bind DNA preferentially (by definition, in a noncooperative fashion).

The DN mutation K88E, which occurs in the DNA binding homeodomain of PITX2a, hinders the interaction of the protein with DNA but not its dimerization. Not surprisingly, the coexistence of the K88E variant (a) with the wild type (A) greatly reduces cooperativity. Indeed, at low protein concentration, only the normal monomer A will recognize DNA and attract preferentially, in concert with DNA, another normal monomer (diluted by the presence of the mutant). The attraction of a monomer a is much weaker as only the monomer A bound to DNA can interact with it (i.e., a cannot see DNA). At higher equimolar concentrations, abnormal dimers (Aa and aa) appear in solution leaving in the end 25% of normal dimers that strongly bind to DNA (Saadi et al., 2003). This example will be discussed further below to see how a defect of one monomer affects not only DNA binding but transcription itself.


Stoichiometric imbalances within a protein complex can further blur the distinction between DN and other alleles such as hypermorphic alleles (Veitia, 2002). Consider, for instance, the heterotrimer A-B-A, formed through a random assembly pathway allowing the existence of intermediates A-B and B-A (Figure 4). The key point is that B is a bridge between two molecules. A and B are not necessarily proteins; they could be RNA subunits as well. For the purpose of this example, we will assume that the normal amount of A is twice that of B and that the association reactions to form the intermediate dimers and trimers are equally fast and irreversible. When there is overproduction of B, the excess of B sequesters a fraction of A monomers as intermediate dimers (i.e., there is insufficient production of A to complete the formation of heterotrimers; Figure 4). The effect of overexpression of B is even worse in a tetramer A3B (where A monomers are linked only to B). Such a strong B allele behaves as an overexpression trans-dominant negative. Indeed, at first sight it is difficult to tell whether a phenotype stems from a classical DNE or from overexpression of a wild-type protein leading to a kinetic trap as described here. One example in yeast involves mLc1p, a light chain, which stabilizes the unconventional myosin Myo2p. A heterozygous inactivating mutation of mLc1 is lethal. However, reduced amounts of the interactor Myo2p help overcome lethality. Conversely, overexpression of Myo2p is toxic, and this toxicity is suppressed by overexpression of mLc1p (Stevens and Davis, 1998). A myriad of other examples of abnormal phenotypes due to overexpression of gene products involved in macromolecular complexes in yeast have been reported (Papp et al., 2003).

Figure 4.
Trans-DNEs Due to Overexpression.

More classical trans-DN mutants have been described for Gα subunits of heterotrimeric G-proteins involved in signal transduction. Such mutants have been used extensively to study G protein–coupled receptor (GPCR)–dependent signal transduction. Mutations in different regions of Gα subunits can give rise to trans-DNEs by three different mechanisms: sequestration of the Gβγ subunits, sequestration of the activated GPCR by the heterotrimeric Gαβγ complex, and sequestration of the activated GPCR by nucleotide-free Gα (Barren and Artemyev, 2007).

Negative transdominance may also arise in oligomers sharing common subunits. For example, if AA, AB, and BB have different functions, overexpression of A may shift the balance toward a molar excess of AA and AB, with negative consequences for the function of BB (Figures 5A and 5B). This happens for the mammalian Id helix-loop-helix (HLH) transcription factors (TFs), which lack a DNA binding domain. Id proteins interact with other basic helix-loop-helix (bHLH) proteins that contain both DNA binding and dimerization domains. Not surprisingly, constitutive expression of Id-1, containing only the dimerization domain, strongly deregulates bHLH function (Sun, 1994). In Arabidopsis, overexpression of phytochrome B has been demonstrated to interfere with phytochrome A, suggesting that both phytochromes might interact with a common reaction partner (Short, 1999). More classical transdominance, by complex poisoning, is thought to be induced by DN mutations in the animal signaling molecule BMP15 (Figure 5C). This DN protein might interfere with the formation of BMP15 and GDF9 homodimers and/or GDF9/BMP15 heterodimers (Yan et al., 2001).

Figure 5.
Other Trans-DNEs.


DNEs also can arise in the context of transcriptional regulation. For example, a DNE can be obtained by removing the transactivation domain of a modular TF, leaving only the DNA binding domain. This truncated factor can behave as a competitive inhibitor of transcription. This is known to occur in nature. For example, the C/EBP protein in mammals is expressed as three alternative polypeptides. The longer polypeptides contain an N-terminal transcriptional activation domain, while the short form lacks it. As the long and short isoforms assemble into homo- and heterodimers, the latter behaves as a natural DN (Zahnow et al., 1997). This is also the case for Stats 5 and 6 in vertebrates, which are able to dimerize. Their proteolytic processing in response to physiological signals leads to the removal of the C-terminal activation domain and converts them into powerful inhibitors that negatively regulate signal transduction (Nakajima et al., 2003). In plants, a large number of MYB proteins function as transcriptional regulators. In Arabidopsis, proteins containing a single MYB DNA binding repeat but lacking the transactivation domain are involved in specifying aspects of epidermal cell fate. These proteins interact with other TFs, including bHLH proteins, and due to the absence of a transactivator domain, they behave as DN and trans-DNs by forming inactive complexes (Ramsay and Glover, 2005).

Removal of the DNA binding domain can also lead to DNEs. This happens in bHLH TFs. As noted above, the Id-1 gene encodes a naturally occurring DN inhibitor of this family of TFs. Complete bHLH proteins (with DNA binding and dimerization domains) can be expressed constitutively. However, the regulated expression of Id-1, containing only the dimerization domain, imposes regulation on bHLH protein activity (Sun, 1994). A similar phenomenon is also expected in plants. For instance, the genome of Arabidopsis encodes ~120 bHLH proteins predicted to bind DNA and 27 proteins with a less basic region than required for binding (Toledo-Ortiz et al., 2003). These non-DNA binding HLHs may function like animal Id proteins, as negative regulators of bHLH proteins through the formation of heterodimers unable to bind DNA (Toledo-Ortiz et al., 2003). Similar effects are expected in TFs belonging to the basic domain/leucine zipper (bZIP) family, which contain a basic DNA binding motif, a leucine zipper dimerization domain, and domains for transactivation. Arabidopsis encodes 67 bZIP proteins, all of which are predicted to function as homo- and/or heterodimers (Deppmann et al., 2004). Some of these are very small and may lack activation domains. In a classic example in plants, Fukazawa et al. (2000) elucidated the function of the bZIP TF REPRESSION OF SHOOT GROWTH (RSG) in gibberellin signaling through the use of a DN form of RSG, which lacked a transcriptional activation domain and therefore acted to repress the function of the wild-type protein when expressed in transgenic tobacco. Moreover, in line with what has been said in the section dealing with trans-DNs by overexpression, DNA–protein transcription complexes are also sensitive to gene dosage balance (Birchler et al., 2001; Veitia, 2002). Alterations of this balance by decreased or increased expression of one TF relative to others involved in the same complex can induce abnormal phenotypes.

A simple model of transcriptional activation can be used to explore some quantitative subtleties of DN mutations in this context. Studies of viral systems and Drosophila have shown that transcription often displays a sigmoidal relationship with respect to the concentration of a TF. In the case of a system responding to a single type of activator (A), this sigmoidal response can be dissected into two main components: cooperative binding of A to the promoter (p) of a target gene and synergy (Figure 6). Synergy is the result of the concerted interactions between the molecules of A already bound to the promoter and the transcriptional machinery (Carey, 1998; Veitia, 2003).

Figure 6.
DNEs in Transcription.

Consider a promoter, p, containing two binding sites for A. The same binding sites are also recognized by the variant a, which might act as a competitive inhibitor. We assume that there may be cooperativity during interaction between A molecules with the promoter. This might also be so for the interactions between A, a, and the promoter. One possible source of cooperativity was mentioned above (i.e., A tends to form dimers in solution, but this is enhanced during DNA binding). Another possibility is that monomers are unable to interact in solution and that interaction of one monomer with DNA leads to an allosteric change that increases the affinity of bound A for an incoming monomer. It is also possible, though less likely, that there are no A–A interactions and that binding of one monomer to DNA leads to a change in the neighboring site that increases its affinity for a newcomer monomer. Whatever the case, cooperativity means that the reaction pA + A = pAA occurs more readily than p + A = pA.

Due to the existence of synergy, the molecular species that contributes the most to transcription is the promoter occupied by two molecules of activator: pAA. This also means that if the affinity constant for the association between the complex pA and the polymerase is KpolA, the K for the association of pAA and the polymerase will be much higher than 2K (of the order of K2polA; see Zlotnick, 1994). To accommodate partial transactivation activity in this model, we will use Kpola (for the reaction pa + pol) and Kpola2 (for paa + pol). Under these assumptions, an equation for the transcriptional response (TR) as a function of the concentration of A (and a) can be derived as described by Veitia (2003) and Veitia and Nijhout (2006) (see Supplemental Materials online).

With a rather simple equation in hand, several conditions can be explored: (1) the wild-type situation A/A, (2) when there is a missing allele (A/-), and (3) when there is coexpression of A and a truncated version a lacking the transactivation domain. In the latter case, we can distinguish two different situations: (3a) when the mutation of a abolishes cooperativity or (3b) when A and a interact cooperatively. Finally, we can also explore a situation (4) where the transactivation capacity of a is normal and cooperativity absent, and (5) when cooperativity is normal but transactivation capacity is partial.

Figure 7 shows that a TR, relative to the maximum output of the promoter versus [A] exhibits a sigmoidal relationship, ranging between 0 and 1. Saturation reflects the maximum response of the system, but this does not imply that the promoter functions only at saturation. According to the figure, and in general, the values on the curve for the heterozygote A/- at any point are lower than in A/A (for each value of relative [A], the heterozygote A/- has two times less [A] in absolute terms than the wild type). Interestingly, at low relative concentrations of A, the shift between the curves is very pronounced Y(A/-) is ~25% of Y(A/A). However, as intuitively expected, for high values of A, saturation is also reached in A/-. If this system were normally functional at low concentrations of A, an individual A/- would display a typical haploinsufficient phenotype.

Figure 7.
TR of a Promoter (with Two Sites) to the Activator A Alone or Coexpressed (in Equimolar Amounts) with Its DN Form a.

What happens in A/a when a lacks the transactivation domain in the absence of cooperativity? According to the classical DN definition, the curve is lower at any point than that of A/-. However, there is a tendency to reach saturation with increasing concentrations of A and a. In fact, A tends to occupy preferentially the promoter as it ensures cooperative interactions with incoming A monomers. However, it is apparent that promoter recognition at low protein concentration occurs less readily in A/a than in the wild-type condition. In practice, a truncated monomer unable to ensure cooperative interactions will lead to a weak DNE. The situation is completely different at the other extreme, when cooperativity between A and a is fully maintained. Indeed, the plateau of the curve for A/a is reached at TR = 0.25. This is expected because pAA, which drives transcription (i.e., the contributions of pAa and paa are negligible), represents only 25% of the occupied promoter species at saturation.

A potential example is provided by an artificial mutation in the TF FOXL2. This TF represses the promoter of the human steroidogenic acute regulatory gene, which contains multiple putative binding sites. A version of FOXL2 containing the DNA binding domain but lacking the C-terminal domain is capable of inducing a DNE that impairs transcriptional repression. However, this effect is obtained only when the DN version is much more strongly expressed (5× and 10×) than the wild-type protein (Pisarska et al., 2004). As outlined above, this is may be due to absence of cooperative interactions between FOXL2 molecules on this promoter.

A more telling example is provided in Figure 8, which represents the response of two different promoters containing one or two binding sites for the TF PTX2a and its DN version, as described earlier (Saadi et al., 2003. At low amounts of transfecting DNA (0.05 μg in Figure 8), the response of the promoter with two sites is more than two times stronger (i.e., 3×) than the response of the promoter with only one site. This is the combined signature of cooperativity and synergy. Moreover, at high concentrations of transfecting constructs WT+DN, the TR of the promoter with two binding sites is ~25% of response of the wild type alone. This is expected because at high protein concentration dimers can be preassembled even before reaching the target DNA. In this case, only 25% of dimers will be normal. The drop in TR is less dramatic for the promoter with only one binding site (expected 50%). From a practical viewpoint, to avoid overlooking a potential DNE in in vitro experiments, low amounts of the WT+DN constructs should be transfected with an excess of reporter promoter to avoid its saturation by the wild-type form. More generally, response curves for different TF concentrations should be provided for such transfection experiments.

Figure 8.
Response of Two Different Artificial Promoters (p) Containing One or Two Bicoid-Like Binding Sites for TF PITX2a and Its DN Version (K88E).

As expected intuitively, when the transactivation capacity of a is normal and cooperativity absent, a very mild DNE appears that leads to a behavior close to a null allele in a heterozygous state. Isolation of this kind of mutant is possible using an elegant yeast genetic screen described by Burz and Hanes (2001). A mutation can affect the levels of cooperativity less dramatically. An interesting case arises when cooperativity drops to approximately one-tenth of the normal level (according to the parameters underlying the results presented in Figure 7) and transactivation capacity is normal. In these conditions, the variant a behaves as a hypomorphic allele in the homozygote a/a and as a null allele in A/a (i.e., A/a = A/-; data not shown). This highlights again the lack of distinct boundaries between hypomophic, DN, and null alleles.

When there is residual transactivation (i.e., 1<Kpola<KpolA) and cooperativity is normal, the plateau in A/a is not reached at TR = 1 but at a lower level. Alleles with partial activation capacity can be easily produced in some cases. The paradigm is provided the yeast TF Gal4, which contains two activating regions (ARI and ARII) involved in the recruitment of the transcriptional machinery. Deletions in the acidic region ARII lead to a decrease in transactivation capacity (Ptashne and Gann, 2002; Ptashne, 2007). A combination of such an allele with the wild-type Gal4 should behave as described. As shown in Figure 7, the curve for A/a crosses that of A/-. Thus, the same allele can be hypomorphic if the system works at low saturation levels (before the curves intersect each other) and can be DN in molecular terms at higher protein concentrations. An intuitive explanation can be provided. Consider, for instance, a protein a that interacts with the polymerase with, say, 90% of wild-type strength. For a range of concentrations the heterozygote A/a will tend to behave like A/A. However, at saturation only 25% of the promoter species will be pAA, which interacts with the polymerase with maximum strength. By contrast, in a heterozygote A/-, at low protein concentrations it is more difficult to occupy the promoter, whereas at saturation 100% of the promoter species will be pAA. Thus, the curves for A/a and A/- must intersect each other at some point.

All target promoters within a cell are not equally sensitive to a DN TF. When there is strong cooperativity and synergy, the sensitivity of a promoter to a DN protein should depend on the number of binding sites on the DNA. The simplest case to visualize is when a lacks a transactivation domain but interacts cooperatively with A. If the promoter species that drives transcription is the one fully loaded with wild-type protein, as assumed above, the maximal TR can be calculated using the formula (of the binomial probabilities) given in the Supplemental Materials online. For a promoter with two identical binding sites, maximum TR will be 25% with respect to the wild-type condition output: for three binding sites, 12.5%, and for four binding sites, 6.25% (when A and a are expressed at equimolar concentrations). For more complex situations, the answer is not intuitive and requires the analysis of models not dealt with here.


The various theoretical and experimental examples described above illustrate how DNEs can arise from both intra- and interloci interactions. The exploration shows how the boundaries between DNs and other classes of mutants are not clear. For instance, two different models demonstrate how DN effects can lead a hypomorphic allele to behave as an apparent loss-of-function allele in a heterozygote. Moreover, overexpression of a normal allele can induce trans-DNEs if the gene product interacts with other (normal) subunits. Finally, an exploration of DN mutants in the context of transcriptional control shows the importance of cooperativity to elicit strong negative effects and that the response of target promoters to a DN TF critically depends on the structure of the formers. A better understanding of the mechanistic foundations for DNEs should help in the design of DN proteins to study gene function.

Supplemental Data

The following material is available in the online version of this article.

  • Supplemental Materials. Formalisms of the Models Discussed in Text.

Supplementary Material

[Supplemental Data]


I thank A. Russo, I. Saadi, and two anonymous referees for their helpful comments on this article and N. Eckardt for comments, editing, and for providing some interesting references.


[W]Online version contains Web-only data.



  • Adams, K.L., Cronn, R., Percifield, R., and Wendel, J.F. (2003). Genes duplicated by polyploidy show unequal contributions to the transcriptome and organ-specific reciprocal silencing. Proc. Natl. Acad. Sci. USA 100 4649–4654. [PMC free article] [PubMed]
  • Barren, B., and Artemyev, N.O. (2007). Mechanisms of dominant negative G-protein alpha subunits. J. Neurosci. Res. 85 3505–3514. [PubMed]
  • Birchler, J.A., Bhadra, U., Bhadra, M.P., and Auger, D.L. (2001). Dosage-dependent gene regulation in multicellular eukaryotes: implications for dosage compensation, aneuploid syndromes, and quantitative traits. Dev. Biol. 234 275–288. [PubMed]
  • Burz, D.S., and Hanes, S.D. (2001). Isolation of mutations that disrupt cooperative DNA binding by the Drosophila bicoid protein. J. Mol. Biol. 305 219–230. [PubMed]
  • Carey, M. (1998). The enhanceosome and transcriptional synergy. Cell 92 5–8. [PubMed]
  • Deppmann, C.D., Acharya, A., Rishi, V., Wobbes, B., Smeekens, S., Taparowsky, E.J., and Vinson, C. (2004). Dimerization specificity of all 67 B-ZIP motifs in Arabidopsis thaliana: A comparison to Homo sapiens B-ZIP motifs. Nucleic Acids Res. 32 3435–3445. [PMC free article] [PubMed]
  • Fukazawa, J., Sakai, T., Ishida, S., Yamaguchi, I., Kamiya, Y., and Takahashi, Y. (2000). Repression of shoot growth, a bzip transcriptional activator, regulates cell elongation by controlling the level of gibberellins. Plant Cell 12 901–915. [PMC free article] [PubMed]
  • Gibson, T.J., and Spring, J. (1998). Genetic redundancy in vertebrates: Polyploidy and persistence of genes encoding multidomain proteins. Trends Genet. 14 46–49. [PubMed]
  • Herskowitz, I. (1987). Functional inactivation of genes by dominant negative mutations. Nature 329 219–222. [PubMed]
  • Müller, H.J. (1932). Further studies on the nature and causes of gene mutations. In Proceedings of the Sixth International Congress on Genetics, D.F. Jones, ed (Menasha, WI: Banta Publishing Company), pp. 215–255.
  • Nakajima, H., Suzuki, K., and Iwamoto, I. (2003). Lineage-specific negative regulation of STAT-mediated signaling by proteolytic processing. Cytokine Growth Factor Rev. 14 375–380. [PubMed]
  • Nishimura, T., Yokota, E., Wada, T., Shimmen, T., and Okada, K. (2003). An Arabidopsis ACT2 dominant-negative mutation, which disturbs F-actin polymerization, reveals its distinctive function in root development. Plant Cell Physiol. 44 1131–1140. [PubMed]
  • Omholt, S.W., Plahte, E., Oyehaug, L., and Xiang, K. (2000). Gene regulatory networks generating the phenomena of additivity, dominance and epistasis. Genetics 155 969–980. [PMC free article] [PubMed]
  • Papp, B., Pal, C., and Hurst, L.D. (2003). Dosage sensitivity and the evolution of gene families in yeast. Nature 424 194–197. [PubMed]
  • Pisarska, M.D., Bae, J., Klein, C., and Hsueh, A.J. (2004). Forkhead l2 is expressed in the ovary and represses the promoter activity of the steroidogenic acute regulatory gene. Endocrinology 145 3424–3433. [PubMed]
  • Ptashne, M. (2007). Words. Curr. Biol. 17 R533–R535. [PubMed]
  • Ptashne, M., and Gann, A. (2002). Genes and Signals. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press).
  • Ramsay, N.A., and Glover, B.J. (2005). MYB–bHLH–WD40 protein complex and the evolution of cellular diversity. Trends Plant Sci. 10 63–70. [PubMed]
  • Saadi, I., Kuburas, A., Engle, J.J., and Russo, A.F. (2003). Dominant negative dimerization of a mutant homeodomain protein in Axenfeld-Rieger syndrome. Mol. Cell. Biol. 23 1968–1982. [PMC free article] [PubMed]
  • Short, T.W. (1999). Overexpression of Arabidopsis phytochrome B inhibits phytochrome A function in the presence of sucrose. Plant Physiol. 119 1497–1506. [PMC free article] [PubMed]
  • Shpak, E.D., Lakeman, M.B., and Torii, K.U. (2003). Dominant-negative receptor uncovers redundancy in the Arabidopsis ERECTA Leucine-rich repeat receptor-like kinase signaling pathway that regulates organ shape. Plant Cell 15 1095–1110. [PMC free article] [PubMed]
  • Stevens, R.C., and Davis, T.N. (1998). Mlc1p is a light chain for the unconventional myosin Myo2p in Saccharomyces cerevisiae. J. Cell Biol. 142 711–722. [PMC free article] [PubMed]
  • Strachan, T., and Read, A. (2004). Human Molecular Genetics. (Oxford, UK: Oxford University Press).
  • Sun, X.-H. (1994). Constitutive expression of the Id1 gene impairs mouse B cell development. Cell 79 893–900. [PubMed]
  • Toledo-Ortiz, G., Huq, E., and Quail, P.H. (2003). The Arabidopsis basic/helix-loop-helix transcription factor family. Plant Cell 15 1749–1770. [PMC free article] [PubMed]
  • Veitia, R.A. (2002). Exploring the etiology of haploinsufficiency. Bioessays 24 175–184. [PubMed]
  • Veitia, R.A. (2003). A sigmoidal transcriptional response: cooperativity, synergy and dosage effects. Biol. Rev. Camb. Philos. Soc. 78 149–170. [PubMed]
  • Veitia, R.A., and Nijhout, H.F. (2006). The robustness of the transcriptional response to alterations in morphogenetic gradients. Bioessays 28 282–289. [PubMed]
  • Yan, C., Wang, P., DeMayo, J., DeMayo, F.J., Elvin, J.A., Carino, C., Prasad, S.V., Skinner, S.S., Dunbar, B.S., Dube, J.L., Celeste, A.J., and Matzuk, M.M. (2001). Synergistic roles of bone morphogenetic protein 15 and growth differentiation factor 9 in ovarian function. Mol. Endocrinol. 15 854–866. [PubMed]
  • Zahnow, C.A., Younes, P., Laucirica, R., and Rosen, J.M. (1997). Overexpression of C/EBPbeta-LIP, a naturally occurring, dominant-negative transcription factor, in human breast cancer. J. Natl. Cancer Inst. 89 1887–1891. [PubMed]
  • Zlotnick, A. (1994). To build a virus capsid. An equilibrium model of the self assembly of polyhedral protein complexes. J. Mol. Biol. 241 59–67. [PubMed]

Articles from The Plant Cell are provided here courtesy of American Society of Plant Biologists
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...