The mechanism that specifies olfactory sensory neurons to express only one odorant receptor (OR) from a large repertoire is critical for odor discrimination but poorly understood. Here, we describe the first comprehensive analysis of OR expression regulation in Drosophila. A systematic, RNAi-mediated knock down of most of the predicted transcription factors identified an essential function of acj6, E93, Fer1, onecut, sim, xbp1, and zf30c in the regulation of more than 30 ORs. These regulatory factors are differentially expressed in antennal sensory neuron classes and specifically required for the adult expression of ORs. A systematic analysis reveals not only that combinations of these seven factors are necessary for receptor gene expression but also a prominent role for transcriptional repression in preventing ectopic receptor expression. Such regulation is supported by bioinformatics and OR promoter analyses, which uncovered a common promoter structure with distal repressive and proximal activating regions. Thus, our data provide insight into how combinatorial activation and repression can allow a small number of transcription factors to specify a large repertoire of neuron classes in the olfactory system.
Our nervous system has a daunting diversity; it contains 100 billion neurons that all have defined functions and connections. To address how neuronal diversity is produced, we have turned to a complex but defined set of neurons, the Drosophila olfactory system located in the antenna. This system contains 34 neuron classes with different functions and connections, each defined by the unique odorant receptor they express. We set out to identify the transcription factors (regulatory genes) that are required for each class to express the correct odorant receptor. We find that seven transcription factors are continuously required in different combinations for the expression of all tested 32 odorant receptors. We also show that these transcription factors can both turn on and turn off odorant receptor genes, making the expression regulation more economical. We conclude that dual use of a small set of factors, which are always on in the neuron, can define its functional class and thereby produce diversity in the nervous system.
Citation: Jafari S, Alkhori L, Schleiffer A, Brochtrup A, Hummel T, Alenius M (2012) Combinatorial Activation and Repression by Seven Transcription Factors Specify Drosophila Odorant Receptor Expression. PLoS Biol 10(3): e1001280. https://doi.org/10.1371/journal.pbio.1001280
Academic Editor: Leslie B. Vosshall, The Rockefeller University, United States of America
Received: December 29, 2011; Accepted: January 26, 2012; Published: March 13, 2012
Copyright: © 2012 Jafari et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: M. A. was supported by Marie Curie Actions (European Commission's Fifth and Seventh Frameworks), the Swedish Research Council, and the Swedish Strategic Research Foundation. Basic research at the Research Institute of Molecular Pathology was funded by Boehringer Ingelheim GmbH. T.H. was supported by the DFG (Heisenberg-Program) and the Schram-Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: IR, inverted repeat; OR, odorant receptor; OSN, olfactory sensory neuron
The external world is perceived by peripheral neurons that each expresses only one or a stereotyped set of receptors from a large genomic repertoire –. The selective receptor expression ensures the specific function of each sensory neuron and produces a daunting diversity of sensory neuron classes. However, little is known about how the neuron class-specific receptor expression is controlled.
In the mouse olfactory system, each olfactory sensory neuron (OSN) chooses to express one odorant receptor (OR) out of approximately 1,200 OR genes . OR choice in mammalians is in part a stochastic process restricted by the developmental context, which is manifested as restricted zonal expression patterns of each OR . The zonal patterns can be resembled by the expression of transgenic OR promoters , and raises the possibility that there are transcription factors (TFs) that in combinations or in gradients specify mouse OR expression. Two TFs, Lhx2 and Emx2, have been identified as general regulators of OR expression –, but the identities of the TFs that regulate specific mouse ORs are unknown, because the large size of the OR repertoire makes systematic analysis of TF phenotypes cumbersome and specific defects difficult to detect.
In similarity to mammals the Drosophila ORs are expressed in a salt and pepper pattern within domains of the antenna OSNs (Figure S3) . Drosophila OR expression create 34 OSN classes with a stereotype neuronal number and location –, suggesting a strictly predetermined process. The large number of OSN classes and precise OR regulation makes the Drosophila antenna an extraordinary system to study how ORs are regulated and how a large number of neuron classes are specified. To date, only two TFs, Acj6 and Pdm3, has been shown to specify a subset of Drosophila ORs ,. However, no systematic approach has yet been undertaken to address the regulatory mechanism of OR expression.
To address how the olfactory system specifies the unique OR identity of a large number of sensory neurons we have performed the first systematic genetic (directed RNAi) screen for direct regulators of Drosophila OR expression. Hereby, we have identified a set of only seven TFs that regulate the complete OR collection of the adult Drosophila olfactory system. We provide a systematic analysis to demonstrate how these TFs employ multiple strategies to specify OR class identity.
A Systematic RNAi Screen Identifies Seven TFs That Regulate OR Expression
In mammals and insects, the majority of OSNs each express a single OR gene out of a large genomic repertoire. To identify the TFs that are necessary for proper OR expression in Drosophila we used the transgenic UAS-driven inverted repeats (IRs) from the Vienna Drosophila RNAi Center (VDRC)  to interfere with the 753 annotated putative TFs in Drosophila (www.flytf.org) . The TF-IRs were expressed in postmitotic OSNs by pebbled-Gal4 , and OR expression was visualized by direct OR promoter fusions with CD8::GFP (Figure 1A). In two separate rounds we analyzed the RNAi effect on the expression of four representative OR classes: Or92a and Or98a for basiconic OSNs in the distal and central antennal region, Or23a and Or47b for trichoid OSNs in overlapping proximal antennal domains (Figure 1A). We found 611, 81.1% of the TFs, to be available as RNAi lines in the VDRC library and expression of which lead to lethality of another 14.2% (Figure 1B). Of the remaining 504 gene knock downs (TF-IRs), we identified seven that resulted in a strong and highly penetrant loss of OR expression: acj6-, E93 (Eip93f)-, Fer1-, onecut-, sim-, xbp1-, and zf30-IRs (Figure 1C and 1D).
(A) Whole mount preparations of antenna from the two screening rounds (GFP in black). In the first round, expression of Or98a-CD8::GFP and Or23a-CD8::GFP in two mid-antennal domains (light blue and orange oval) were analyzed. In the second round, Or92a-CD8::GFP expression in the most proximal (dark blue oval) antenna domain and Or47b-CD8::GFP expression in the most distal (red oval) antenna domain were analyzed. (B) Statistics from the screen is depicted as a graph, summarizing the number of IR lines that did not affect OR expression (Wt, white), led to lethality (Lethal, grey) or lost OR expression (Loss of OR expression, Green). (C) Phenotype summary for the seven TF-IRs and the analyzed OSN classes, wild-type OR expression (grey dots) and loss of OR expression (black dots). (D) Antenna from each TF-IR with representative OR expression phenotypes. (E) Whole mount antennal lobe with the Or92a-CD8::GFP OSN projections shown in green and the synaptic marker, nc82, delineating the glomeruli of the antennal lobe, in magenta. The boxed region indicates the antennal lobe area in the right panel, which compares the RNAi and mutant phenotypes of acj6, sim, xbp1, zf30c. Note the loss of Or92a in both the mutant and RNAi lines.
To exclude false positives caused by off-targeting and insertion mutagenesis, multiple IR lines from VDRC (http://stockcenter.vdrc.at), National Institute of Genetics (NIG-Fly, http://www.shigen.nig.ac.jp/fly/nigfly), and Transgenic RNAi Project (TriP, http://www.flyrnai.org) that corresponded to each of the seven genes were analyzed. All constructs gave rise to identical phenotypes (Table S1), supporting the specific knock down of each TF. In addition, mutant analysis for the TFs with available defined mutant alleles (acj66, xbp1k13803, simH9, and zf30ck02506) gave similar phenotypes compared to the RNAi lines (Figure 1E; Table S1). Finally, direct expression analysis on each IR background demonstrated knock down of the corresponding TF (Figures 2 and S1). These results indicate that these seven TFs are critical regulators of OR expression.
(A) The identified TFs belong to different protein families as indicated by their protein domain organization. (B) In situ hybridizations and immunohistology on wild-type antenna sections showing the expression pattern of each TF (red) counterstained with the nuclear marker DAPI (blue). (C,D) RNAi-mediated reduction of Acj6 (C) and Xbp1(D) does not affect the overall expression pattern of the other six TFs. (E) Expression of the TFs (magenta) in either Or47b-CD8::GFP or Or92a-CD8::GFP (green) expressing OSNs. Note, that the Or47b expressing OSNs lack expression of onecut and zf30c (arrows).
Overlapping Expression of the Seven TFs
All identified TFs belonged to different protein families (Figure 2A): Acj6 (POU-Homeobox; Hox), E93 (Psq like helix-turn-helix; HTH), Fer1 (basic helix loop helix; bHLH), Onecut (cut-Hox), Sim (PAS-bHLH), Xbp1 (bZIP), and Zf30c (C2H2; zinc finger). Two of the genes, zf30c (zinc finger at 30c), which encodes a protein with ten C2H2 zinc finger domains and Fer1 (Forty eight related 1), encoding a bHLH factor, had not previously been characterized.
We next asked whether expression of the seven TFs correlated to the OR expression domains or sensilla groups. There are three main groups of sensilla, basiconic, coeloconic and trichoid, differently distributed across the antenna . In situ hybridization and immunohistochemistry demonstrated that all seven TFs were expressed in the adult antenna in various patterns (Figure 2B). Each pattern showed little restriction to domains or sensilla groups and none of the TFs were expressed in only one or only a few OSN classes (Figure 2B). Two of the TFs, acj6 and xbp1, were ubiquitously expressed and might regulate OR expression more indirectly via any of the other five TFs. When analyzed, no obvious differences in strength or distribution of E93, Fer1, onecut, sim, and zf30c expression were found in the acj6- and xbp1-IRs, indicating that the seven TFs might be directly required for OR expression (Figure 2C and 2D).
To address the extent of coexpression between the seven TFs, we analyzed each TFs expression in two OSN classes (Figure 2E). In Or92a OSNs, all seven TFs were expressed including E93, the TF that was not required for Or92a expression. Similarly, Or47b neurons expressed E93, the only TF required for expression, and acj6, Fer1, sim, and xbp1. These data show that the seven TFs are expressed in broad and overlapping populations of mature sensory neurons, which do not correlate with sensilla groups or OSN classes. The lack of anatomic correlation of the expression patterns suggests that these TFs are part of a distinct regulatory network separate from the general process of antenna and neuron specification.
The Identified OR Regulators Are Required in Adult Flies
The onset of OR expression takes place during the second half of pupal development, after OSN axon guidance, and is one of the final steps of sensory neuron differentiation (Figure 3A). To rule out a role of the seven TFs in early OSN specification and differentiation, which could affect OR gene expression more indirectly, the pan neuronal markers, Elav and Neuroglian, were analyzed. The overall number of OSNs and axonal projections from the antenna to the brain was not affected in any of the RNAi knock downs, indicating no gross changes in sensory neuron development (Figure S2).
(A) Schematic of the TARGET experiments. Flies were shifted at late pupal stage from 18°C to 29°C (red line), or from 29°C to 18°C (green line); the RNAi was induced specifically at 29°C. (B–D) Or92a and Or47b in situ hybridizations (red) counterstained with DAPI (blue). (B) With the suppression of RNAi at 18°C, the OR was expressed in all genotypes (red staining). (C) The TF knock down at the end of pupal development (shift from 18°C to 29°C) fully suppresses OR expression. (D) Developmental TF knock down (shift from 29°C to 18°C) does not affect OR expression except for zf30c-IR.
Next, to determine the temporal window of TF function in OR expression we used the TARGET system . Here, the IR-mediated gene knock-down can be regulated via a temperature-sensitive Gal4 repressor (GAL80ts) (Figure 3A). At the restrictive temperature (29°C), GAL80ts is inactivated, permitting Gal4 to express the TF-IR in all OSNs (Figure 3A). Flies maintained continuously at 18°C (no TF-IR expression) expressed Or92a and Or47b at the correct antennal location (Figure 3B). In contrast, when the TF-IR flies were shifted after the onset of OR expression to 29°C, Or92a or Or47b expression was lost (Figure 3C).
In a reverse approach, knock down of the TFs during pupal development and a reversal of the wild-type TF expression in early adult stages allowed us to distinguish between earlier developmental roles and a later function in OR gene regulation (Figure 3D). Developmental suppression of acj6, E93, Fer1, sim, onecut, and xbp1 did not affect adult OR expression (Figure 3D), whereas knock down of Zf30c during pupal development reduced OR expression. These data support a view of sensory neuron development where these seven TFs possess a specific OR regulatory function and with Zf30c having an additional earlier role in OSN class specification.
A TF Regulatory Matrix for Drosophila OR Expression
To determine whether this small set of TFs can regulate the full collection of OR genes, we extended our RNAi analysis to the majority of the sensory neurons classes in the Drosophila olfactory system. The resulting OR expression phenotypes were assembled into a regulatory matrix (Figure 4A; for statistics see Table S2). The matrix exposed several general regulatory features. First, all 32 ORs required at least one of the seven TFs for correct expression, demonstrating a prominent role in OR gene regulation for the TFs (Figure 4A and 4C). Second, in line with the wide TF expression patterns across the antenna (Figure 2B), the TFs were required for OR expression in OSN classes indiscriminate of sensilla group or antenna location (Figures 4B and S3), supporting that the TFs disconnect the OR expression from the early antenna patterning and development. Third, the identified seven TFs were required for expression of partly overlapping sets of OR genes (Figure 4A), suggesting a combinatorial mode of OR gene regulation. Last, unique TF combinations were associated with 17 of the 32 ORs expression and each combination ranged from one to six TFs and only two additional TFs would be sufficient to resolve the remaining redundancies. Taken together these data show that the identified small set of TFs in different combinations are required for OR expression in each OSN class.
(A) The regulatory matrix represents in situ hybridizations for 32 ORs/TF-IR, indicated as wild-type levels (gray dots) and lost (black dots) OR expression. Trichoid ORs marked in orange. (B) Bar diagram, representing number of ORs that required each TF for expression. Trichoid ORs marked as orange insets in each bar. (C) Number of ORs regulated by 0–7 TFs depicted as bar graphs. (See Table S2 for statistics and see Figure S3 for domain and sensilla arranged matrix).
The Seven TFs Bind to Different Combinations of Motifs Upstream OR Genes
To address, whether any of the identified TFs bind directly to the regions upstream of each OR, we exploited the well-established vertebrate DNA binding motifs of Acj6, Onecut, and Xbp1. It has been shown that Drosophila Acj6 and Onecut share binding properties with their vertebrate orthologs (Figure S4) ,,. Most vertebrate Xbp1 DNA motifs contain a 6-bp core sequence C/TCACGT ,. In mobility shift assays, recombinant Drosophila Xbp1 bound this core sequence (Figure S4A), demonstrating shared binding properties between the Xbp1 orthologs.
The Acj6-, Onecut-, and Xbp1-DNA binding motifs were used to search 1 kb upstream of 32 OR genes and identified various combinations of the TF binding motifs upstream of each OR (Figure 5A; for location of each motif see Table S3). Most OR promoter regions contained at least one binding site for the TFs required for expression (Figure 5A). The fact that some OR promoter regions lacked predicted binding sites for the required TF suggest either that the Drosophila TF and the vertebrate ortholog have slightly different DNA binding requirements or that the TF in these cases indirectly regulate the OR gene. In vitro binding assays for four of the OR genes showed that all motifs were recognized by the matching TFs (Figure S4C). These data together with the strong correlation between motif and OR gene activation suggest that each OR promoter is bound and regulated by different combinations of these TFs.
(A) Regulatory matrix for Acj6, Onecut, and Xbp1 across 32 ORs comparing RNAi phenotypes (grey and black dots) and predicted TF binding motifs upstream each OR (purple dots). (B) Or19a promoter construct driving CD8::GFP (black) in the correct OSN class, which is lost when either the Acj6 or Onecut motif has been mutated (C) Motif density plot across the first 1,000 bps upstream of each OR gene for Acj6, Onecut, and Xbp1 motifs. (D) Table summarizing the expression produced by the short OR promoter constructs. (E) A 5.7-kb fragment upstream of Or23a produces an endogenous expression pattern, whereas a short 0.21-kb fragment produces ectopic expression in a large number of OSNs across the antenna.
To address whether the motifs were necessary in vivo for OR expression, we focused on the shortest promoter region sufficient for OR expression, Or19a (Figure 5D). The sufficient promoter region contains both an Acj6 and Onecut motif, both TFs required for Or19a expression (Figure 5A). When either of the two motifs was mutated, the expression of the Or19a construct was abolished (Figure 5B). These results demonstrated that the motifs were necessary for promoter function and that the TFs directly regulate OR expression.
Long Range Repression Modulates OR Expression
For Acj6 and Onecut a peak of binding motifs was observed directly upstream of the OR genes (Figure 5C), (for individual predictions see Table S3), which corresponded to a region of high sequence conservation found upstream of most OR genes (unpublished data) . Transgenic constructs containing these conserved regions produced antenna OSN expression for ten tested OR promoters (Figure 5D), suggesting that a short region directly upstream of each OR gene was sufficient for expression. However, half of the short promoter constructs produced misexpression (Figure 5D and 5E); the lack of OSN class specificity implies that distal regulatory regions are required to repress OR expression in some OSN classes. The similarities in behavior for the various OR promoter constructs suggest a common OR promoter organization with a proximal region that produces expression and a distal repressive region that restricts the OR expression to one single OSN class.
The Location of the DNA Binding Motif Determines TF Function
The bioinformatic analysis uncovered DNA motifs in OR promoters that did not require the matching TF for expression (Figure 5A). When the upstream locations of these “nonessential motifs” were plotted, a peak was found downstream of the TATA boxes (Figure 6A; see Table S3 for location of each motif). Conversely, all motifs upstream of ORs that required the matching TF were located upstream of each TATA box (Figure 6B), suggesting that motif location might reflect different TF functions. For example, Or98a, which did not require xbp1 for expression, had an Xbp1 motif downstream of the TATA box (Figure 6D). Moreover, in xbp1-IR flies, Or98a showed ectopic expression in OSNs that normally express Or7a and pairs with Or56a (Figure 6C). The repression of Or98a and the activation of Or7a expression in the same OSN class show that Xbp1 has a dual function in the specification of OR gene expression. One simple explanation might be that Xbp1, when bound far upstream, activates expression of Or7a and, when bound next to the TATA box, hinders transcriptional initiation of Or98a. To address this possibility, the Or98a Xbp1 motif was mutated, which produced misexpression across the central antenna (Figure 6D). These data suggest that the differential activity of Xbp1 can be defined by the location of the binding motif in the regulatory regions of the two OR genes.
(A) Motif density plot, showing motifs found upstream of OR genes that did not require the matching TF (see Table S3 for statistics). (B) Bar graph depicting the total number of motifs located upstream or downstream the TATA box for ORs that either require the TF (“essential”) or not (“nonessential”) for expression. (C) Double in situ labeling of Or98a and Or56a in wild type (Wt) and xbp1-IR antennae revealed ectopic Or98a expression next to Or56a. The RNAi phenotypes are summarized as a matrix (grey, wild-type expression; red, ectopic; and black, loss of expression). (D) One Xbp1 motif (purple) was found next to the TATA box (green) of Or98a. The Or98a promoter construct produced expression in a single domain (light blue oval, black expression). Whereas, the same Or98a promoter construct with a mutated Xbp1 motif (red) produced a distal expansion of the expression.
The Identified TFs Both Activate and Repress OR Expression
To investigate if any of the other six TFs also repress OR gene expression, the knock-downs were reexamined more closely and striking de-repression was observed for two more TFs (Figure 4C). Strong ectopic Or43b expression was found in E93-IR distal antennae (Figure 7A). Double-labeling experiments showed that OSNs with ectopic Or43b expression formed a pair with Or23a OSNs and thereby replacing Or83c in E93 knock-downs (Figure 7A), which suggested that E93 repressed Or43b in these OSNs and was required for Or83c expression. These results indicate a dual regulatory function similar to Xbp1 in which the location of the unknown E93 motif might possibly produce Or83c expression and Or43b repression. The second example of ectopic expression was identified in acj6-IR antennae with Or67a being de-repressed and coexpressed with Or67b (Figure 7B). Both Or67a and Or67b have upstream Acj6 binding motifs (Figure 5A) and the TFs required for Or67b expression were some of the TFs also required for Or67a expression (Figure 7B). Hence, the dual Acj6 function required to separate Or67a and Or67b expression might be determined in a combinatorial fashion possibly by attraction of different cofactors to each promoter.
(A) Double in situ labeling of Or23a and Or43b in wild type (Wt) and E93-IR antennae, the Or43ba expression phenotypes are further depicted schematically and summarized as a matrix (grey, wild-type expression; red, ectopic; and black, loss of expression). (B) Double in situ hybridization labeling of Or67a and Or67b expression in wild type (Wt) and acj6-IR antenna. The resultant phenotypes are further summarized as a schematic and a matrix summary. Note the new pair of Or43b and Or23a when E93 is knocked down (A), and OR coexpression generated in acj6 knock-downs (B). (C) Model depicting how activation and repression of OR expression can specify an OSN class. Activation of OR gene expression (left box); different combinations of a limited set of TFs bind a proximal upstream region and produce OR expression in a broad antenna region. Repression of OR gene expression (right box), distal located repressors together with the dual function of the TFs determined by binding site location or possibly cofactor use, restrict OR expression. The combined sum of OR gene activation and repression produce expression to one single OSN class.
We performed a multilevel systematic analysis of sensory class specification in the Drosophila olfactory system and identified seven TFs to be critical regulators of odorant gene expression. Different combinations of these TFs are required for precise neuron-specific onset of OR gene expressions as well as maintenance in mature OSNs. The systematic analysis further reveals that the identified TFs bind to different DNA motifs through which they can act as both activators and repressors of OR gene expression (Figure 7C).
The Seven Identified TFs Are OR Selector Genes
In 1975, Antonio Garcia-Bellido presented the concept of selector genes, TFs that can determine a particular cell fate. Several levels of selector genes has been found, which control gene programs that individually specify organ, tissue, and cell type . Recently, studies in Caenorhabditis elegans have revealed that one factor and its motif can be enough to assign expression to one neuronal class –. These observations have led to the formulation of the terminal selector gene hypothesis , which put forward that only a small set of TFs are continuously required to express the genes that signify each neuron class, like ORs. However to date few such cases have been identified.
Our systematic analysis presents several observations that suggest the identified TFs to be terminal selector genes for OR expression. First, continuous expression of all seven TFs are required for OR expression in the mature OSNs (Figure 3C). Second, the seven TFs are expressed in the mature OSNs, in various patterns across the antenna (Figure 2B). Third, all 32 tested ORs require different combinations of the TFs for expression (Figure 4A). Last, motifs for the TFs are found upstream of the ORs genes they regulate (Figure 5A) and the motifs are necessary for OR promoter function (Figure 5B). Consequently, it appears that the large number of OR expression patterns are achieved by combinatorial use of a few TFs that function as OR selector genes.
The OR selector genes belong to different protein families (Figure 2A), which indicate that evolution has favored recruitment of TFs with very different DNA binding properties, rather than expansion of one family that shares the basic DNA binding motif, which could secure the fidelity of the combinatorial pattern. On the other hand, the high motif specificity for each factor suggests that loss or gain of motifs for one OR selector might generate a new OR expression pattern and a totally new OSN class. Thus, single OSN class expression and high evolvability, two hallmarks of olfactory system evolution, might in part be due to the combinatorial function of the OR selector genes.
Single OR Expression, a Large Regulatory Cost
How many OR selector genes are required to uniquely express one OR in each OSN class? We identified seven OR selector genes, but given the limitations of RNAi, it is likely that there are a total of at least ten critical TFs to specify all OSN classes. Even this probably low estimate generates a rather high number of TFs considering that Drosophila antennae have 34 OSN classes that express ORs . Theoretically the number of TFs needed for a binary combinatorial code to generate 34 unique outcomes is six (26 = 64). Seven TFs can in theory separate 27 = 128 combinations, and ten TFs designate more than 1,000 combinations, suggesting a large number of unused combinations. This surplus of combinations may be due to the inherent randomness of evolution and the impossibility of creating a streamlined code by chance. Another possibility for this large number is the need for a high degree of fidelity, with little or no ectopic OR expression tolerable for proper functioning of the olfactory system. Extrapolation of our observations to the regulatory requirements of the mammalian olfactory system indicates that at least 200–300 TFs would be required to provide a regulatory system that controls >1,000 mammalian ORs, a daunting number. Therefore, it is reasonable to suspect that the stochastic OR selection mechanism found in vertebrates was added during evolution to accommodate the heavy increase in regulatory costs resulting from an expanded number of OR genes.
Combinatorial Activation and Repression Control OR Expression
To date very few TFs have been found to be restricted to small neuronal populations in neuroepithelia or in the developing brain in general . This situation has motivated the suggestion that combinatorial TF regulation defines broad expression patterns of molecules such as neurotransmitters, but is insufficient to generate the large number of neuron classes in, for example, the olfactory system . Similarly, all seven selector genes in this study are expressed across the antenna but still are required for the expression of some few ORs (Figures 2B and 4A). How can widely expressed TFs then produce restricted expression patterns? We have formulated two explanations. First, our promoter analysis suggests that the OSN class specificity is in part due to repression. Most ORs have a proximal regulatory region next to the gene that is sufficient for expression in OSNs but requires repression from more distal regions for the spatial restriction to each OSN class (Figure 5D). In this model, the expression of the TFs that produce OR expression does not need to be particularly specific as long as they are counteracted by repressive factors. Second, the identified TFs can both activate and repress OR expression dependent on the location of the binding site or by the available cofactors (Figures 6C, 6D, 7A, and 7B). Dual use of the TFs might increase their regulatory power and as a likely consequence the number of TFs required for OR expression to be reduced. We therefore suggest that specification of large numbers of neuron classes in the olfactory system and likely in the nervous system, require two layers of combinatorial coding, one layer of terminal selector genes that produce expression and a layer of repressors that restrict the expression to each class.
Materials and Methods
Virgin flies containing Pebbled-GAL4, UAS-Dicer2, and the OR promoter fusions were mated with males obtained from the VDRC library. The crosses were set up at 25°C, and after 3 d the parental flies were removed and the vials shifted to 27°C. 2–3 d after eclosure, the GFP levels corresponding to OR expression were ranked 0–5, where 5 corresponded to the wild-type level. For all assays, for five females per line crosses were scored blind to the genotype and all lines with phenotypes scored below 2 were retested. A line was considered to have established phenotype if three consecutive crosses included flies that scored below 2. To further validate the established phenotypes, RNAi lines from the VDRC, NIG, and the Transgenic RNAi Project (TRiP) were used, all the different lines gave the same phenotype as in the screen (Table S1). In order to avoid animals with low RNAi efficiency and reduce the risk of false negatives in the regulatory matrix, OR expression phenotypes were only scored from antennae with total loss of Or92a or Or47b GFP.
Mutant Analysis and MARCM
To confirm acj6 function in OR gene regulation, viable offspring from the acj66 mutant crossed to the Or92a promoter fusion were analyzed. For the other mutants, genetic mosaics were generated using the MARCM system , which was visualized with an Or92a promoter fusion with Gal4 driving the expression of UAS-SytGFP . For large clones in the antenna, an ey-FLP insertion on the X chromosome was used , dependent on gene location mosaics were generated in animals of the following genotypes: ey-FLP; FRT40/42 TF mutant/FRT40/42,TubGal80; Or92a-Gal4, UAS-SytGFP, or ey-FLP; Or92a-Gal4, UAS-GFP; FRT80/82 TF mutant/FRT80/82 TubGal80.
Immunostaining and In Situ Hybridization
Immunostaining and in situ hybridization were performed according to previously described methods ; for practical in situ details see . The OR probes were previously used in the OR expression characterization . TF in situ probe templates included sequence from the first coding exon and 1 kb downstream or to the end of the gene and were from genomic DNA and cloned into pBSK. Or49a, Or65a,b,c, and Or69a,b were at the detection limit and excluded from the regulatory matrix analysis.
The primary antibodies used were Rat anti-Elav (7E8A10, DSHB, 1∶500), mouse anti-Acj6 (DSHB, 1∶100), mouse anti-Neuroglian (BP104, DSHB, 1∶50), and Rabbit anti-GFP (TP-401, Torrey Pines, 1∶2,000).
1 kb upstream the translational start site of each OR was scanned with the motifs for HNF6 and BRN3 using weight matrices and programs provided by Genomatix (HNF6.01, BRN3.01, BRN3.02; http://www.genomatix.de/)  and Biobase (HNF6_Q6; http://www.gene-regulation.com/) . The Genomatix Matinspector and the Biobase match program optimized matrix thresholds were applied. Putative Xbp1 binding sites were identified on the basis of a pattern search with the consensus motif C/TCACGT .
Electrophoretic Mobility Shift Assay
The various TF DNA binding domains were cloned into the pGEX-2T vector and bacterial recombinant glutathione S-transferase fusion proteins were purified by glutathione Sepharose 4B beads (Amersham). For the binding assay, single-stranded DNA oligonucleotides were end-labeled with T4 polynucleotide kinase (Roche) and G-32-P ATP (PerkinElmer) with T4 polynucleotide kinase according to the manufacturers' instructions, annealed with the complementary strand, and purified on a microspin column (Roche).
Binding reactions were performed at room temperature for 20 min. The binding reaction included 3 µl recombinant glutathione S-transferase fusion proteins, 3 fmol labeled probe, 10 mM HEPES (pH 7.9), 70 mM KCl, 1 mM DTT, 1 mM EDTA, 2.5 mM MgCl2, 4% glycerol, and 1 µg poly (dI/dC) (VWR). Cold competition was performed by adding DNA oligonucleotides in molar excess 15 min before addition of labeled probe. The samples were separated on a 6% acrylamide TBE gel at 60 V for 90 min. Gels were dried and visualized by the FLA-5100 Multi Gauge system (FujiFilm).
OR promoter fusion lines have previously been described . Pebbled-GAL4 and acj66mutant flies were kind gifts from L. Luo. simH9 was kindly provided by C. Klaembt. UAS-Dicer2 and the TF-IRs for the screen were provided by the VDRC. Additional TF-IR lines were obtained from NIG and TRiP. sim-lacZ flies were obtained from the Szeged Drosophila Stock Centre (Szeged, Hungary), and xbp1k13803, zf30ck02506, zf30c-lacZ, tubP-Gal80ts were obtained from the Drosophila Stock Center (Bloomington, Indiana). For the promoter studies all DNA constructs were injected into w1118 flies, and six to 12 lines were analyzed.
TF knockdown correspond to loss of TF mRNA. In situ labeling of each TF (red) and DAPI (blue) performed on TF-IR antenna, note the tight correlation of loss of xbp1 (red) and Or92a-CD8::GFP expression (green).
Olfactory sensory neuron layers and morphology are unperturbed in the TF knock downs. Antenna from TF-IR flies stained for neuronal markers in red (Neuroglian and Elav) and counterstained with DAPI.
None of the seven TFs were required for OR expression to one antenna domain or sensilla group. Regulatory matrix arranged after the five antenna domains (blue to red) and each sensilla group. Each domain is exemplified by one OR promoter fusion in green, counterstained with ELAV in red. Note that at least three of the seven TFs are required for expression in each sensilla group (basiconic, trichoid, and coeloconic).
Predicted DNA motifs are bound by the identified TFs in vitro. (A) Electrophoretic mobility shift assay (EMSA) performed with radiolabeled probe containing the vertebrate Xbp1 core sequence with (+) or without (−) recombinant Xbp1. Increasing amounts (100-, 200-, 300-, 900-fold excess) of nonlabeled probe were used as cold competitors; 900-fold excess of a probe carrying a mutated motif is shown in the lane labeled “m.” (B) Acj6, onecut, and Xbp1 motifs upstream of four ORs. (C) Expanded EMSA validation of the predicted Acj6, Xbp1, and onecut motifs from the four OR promoters. Radiolabeled motif probe (P) and competition with cold (C) motif probe at 900-fold excess are shown.
All tested IRs and mutants for each TF gave rise to identical phenotypes. Statistics related to Figure 1. OR expression phenotypes for two or more TF-IRs and available mutants for each gene, noted as number of animals with loss of OR expression/number of analyzed animals. Wt, wild type, denotes no loss of expression.
TF-IRs phenotypes for the 32 ORs in the regulatory matrix. Statistics related to Figure 3. OR expression detected by in situ hybridizations on TF-IRs antennas and rated per animal from; 0 (loss) to 5 (control levels) and denoted as phenotype level/animals. OR expression rated above 2 was considered to be wild-type variance.
The RNAi screen and the initial work were carried out while MA was a postdoctoral researcher in Barry Dickson's Lab, IMP, Vienna, Austria. We are grateful to A. Berghard, C. Ribeiro, B. Dickson, and the Alenius lab for critical comments and discussion; M. Sigvardsson for technical advice; and O. Forsberg and K. Bartalska for excellent technical assistance. We thank L. Luo and C. Klaembt for flies.
The author(s) have made the following declarations about their contributions: Conceived and designed the experiments: Screen: MA; expression analysis: MA; mutant analysis: TH; temporal TF requirement: MA; regulatory matrix: MA; bioinformatics: MA; TF analysis: MA; OR promoter analysis: MA; dual TF function: MA. Performed the experiments: Screen: MA; expression analysis: MA; mutant analysis: AB TH; temporal TF requirement: SJ AB TH; regulatory matrix: SJ LA MA; bioinformatics: AS MA; TF analysis: SJ MA; OR promoter analysis: SJ MA; dual TF function: LA MA. Analyzed the data: Screen: MA; expression analysis: LA MA; mutant analysis: AB TH; temporal TF requirement: SJ MA; regulatory matrix: SJ LA MA bioinformatics: AS MA; TF analysis: SJ MA; OR promoter analysis: SJ MA; dual TF function: LA MA. Contributed reagents/materials/analysis tools: Screen: MA; expression analysis: MA; mutant analysis: TH; temporal TF requirement: MA; regulatory matrix: MA; bioinformatics: MA; TF analysis: MA; OR promoter analysis: MA; dual TF function: MA. Wrote the paper: TH MA.
- 1. Zylka M. J, Dong X, Southwell A. L, Anderson D. J (2003) Atypical expansion in mice of the sensory neuron-specific Mrg G protein-coupled receptor family. Proc Natl Acad Sci U S A 100: 10043–10048.M. J. ZylkaX. DongA. L. SouthwellD. J. Anderson2003Atypical expansion in mice of the sensory neuron-specific Mrg G protein-coupled receptor family.Proc Natl Acad Sci U S A1001004310048
- 2. Buck L, Axel R (1991) A novel multigene family may encode odorant receptors: a molecular basis for odor recognition. Cell 65: 175–187.L. BuckR. Axel1991A novel multigene family may encode odorant receptors: a molecular basis for odor recognition.Cell65175187
- 3. Chandrashekar J, Hoon M. A, Ryba N. J, Zuker C. S (2006) The receptors and cells for mammalian taste. Nature 444: 288–294.J. ChandrashekarM. A. HoonN. J. RybaC. S. Zuker2006The receptors and cells for mammalian taste.Nature444288294
- 4. Yau K. W, Hardie R. C (2009) Phototransduction motifs and variations. Cell 139: 246–264.K. W. YauR. C. Hardie2009Phototransduction motifs and variations.Cell139246264
- 5. Johnston R. J Jr, Desplan C (2008) Stochastic neuronal cell fate choices. Curr Opin Neurobiol 18: 20–27.R. J. Johnston JrC. Desplan2008Stochastic neuronal cell fate choices.Curr Opin Neurobiol182027
- 6. Serizawa S, Miyamichi K, Nakatani H, Suzuki M, Saito M, et al. (2003) Negative feedback regulation ensures the one receptor-one olfactory neuron rule in mouse. Science 302: 2088–2094.S. SerizawaK. MiyamichiH. NakataniM. SuzukiM. Saito2003Negative feedback regulation ensures the one receptor-one olfactory neuron rule in mouse.Science30220882094
- 7. Lewcock J. W, Reed R. R (2004) A feedback mechanism regulates monoallelic odorant receptor expression. Proc Natl Acad Sci U S A 101: 1069–1074.J. W. LewcockR. R. Reed2004A feedback mechanism regulates monoallelic odorant receptor expression.Proc Natl Acad Sci U S A10110691074
- 8. Kolterud A, Alenius M, Carlsson L, Bohm S (2004) The Lim homeobox gene Lhx2 is required for olfactory sensory neuron identity. Development 131: 5319–5326.A. KolterudM. AleniusL. CarlssonS. Bohm2004The Lim homeobox gene Lhx2 is required for olfactory sensory neuron identity.Development13153195326
- 9. Hirota J, Mombaerts P (2004) The LIM-homeodomain protein Lhx2 is required for complete development of mouse olfactory sensory neurons. Proc Natl Acad Sci U S A 101: 8751–8755.J. HirotaP. Mombaerts2004The LIM-homeodomain protein Lhx2 is required for complete development of mouse olfactory sensory neurons.Proc Natl Acad Sci U S A10187518755
- 10. McIntyre J. C, Bose S. C, Stromberg A. J, McClintock T. S (2008) Emx2 stimulates odorant receptor gene expression. Chem Senses 33: 825–837.J. C. McIntyreS. C. BoseA. J. StrombergT. S. McClintock2008Emx2 stimulates odorant receptor gene expression.Chem Senses33825837
- 11. Vosshall L. B, Amrein H, Morozov P. S, Rzhetsky A, Axel R (1999) A spatial map of olfactory receptor expression in the Drosophila antenna. Cell 96: 725–736.L. B. VosshallH. AmreinP. S. MorozovA. RzhetskyR. Axel1999A spatial map of olfactory receptor expression in the Drosophila antenna.Cell96725736
- 12. Hallem E. A, Ho M. G, Carlson J. R (2004) The molecular basis of odor coding in the Drosophila antenna. Cell 117: 965–979.E. A. HallemM. G. HoJ. R. Carlson2004The molecular basis of odor coding in the Drosophila antenna.Cell117965979
- 13. Couto A, Alenius M, Dickson B. J (2005) Molecular, anatomical, and functional organization of the Drosophila olfactory system. Curr Biol 15: 1535–1547.A. CoutoM. AleniusB. J. Dickson2005Molecular, anatomical, and functional organization of the Drosophila olfactory system.Curr Biol1515351547
- 14. Endo K, Aoki T, Yoda Y, Kimura K, Hama C (2007) Notch signal organizes the Drosophila olfactory circuitry by diversifying the sensory neuronal lineages. Nat Neurosci 10: 153–160.K. EndoT. AokiY. YodaK. KimuraC. Hama2007Notch signal organizes the Drosophila olfactory circuitry by diversifying the sensory neuronal lineages.Nat Neurosci10153160
- 15. Komiyama T, Carlson J. R, Luo L (2004) Olfactory receptor neuron axon targeting: intrinsic transcriptional control and hierarchical interactions. Nat Neurosci 7: 819–825.T. KomiyamaJ. R. CarlsonL. Luo2004Olfactory receptor neuron axon targeting: intrinsic transcriptional control and hierarchical interactions.Nat Neurosci7819825
- 16. Bai L, Goldman A. L, Carlson J. R (2009) Positive and negative regulation of odor receptor gene choice in Drosophila by acj6. J Neurosci 29: 12940–12947.L. BaiA. L. GoldmanJ. R. Carlson2009Positive and negative regulation of odor receptor gene choice in Drosophila by acj6.J Neurosci291294012947
- 17. Dietzl G, Chen D, Schnorrer F, Su K. C, Barinova Y, et al. (2007) A genome-wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature 448: 151–156.G. DietzlD. ChenF. SchnorrerK. C. SuY. Barinova2007A genome-wide transgenic RNAi library for conditional gene inactivation in Drosophila.Nature448151156
- 18. Adryan B, Teichmann S. A (2006) FlyTF: a systematic review of site-specific transcription factors in the fruit fly Drosophila melanogaster. Bioinformatics 22: 1532–1533.B. AdryanS. A. Teichmann2006FlyTF: a systematic review of site-specific transcription factors in the fruit fly Drosophila melanogaster.Bioinformatics2215321533
- 19. Sweeney L. B, Couto A, Chou Y. H, Berdnik D, Dickson B. J, et al. (2007) Temporal target restriction of olfactory receptor neurons by Semaphorin-1a/PlexinA-mediated axon-axon interactions. Neuron 53: 185–200.L. B. SweeneyA. CoutoY. H. ChouD. BerdnikB. J. Dickson2007Temporal target restriction of olfactory receptor neurons by Semaphorin-1a/PlexinA-mediated axon-axon interactions.Neuron53185200
- 20. Shanbhag S. R, Muller B, Steinbrecht R. A (2000) Atlas of olfactory organs of Drosophila melanogaster 2. Internal organization and cellular architecture of olfactory sensilla. Arthropod Struct Dev 29: 211–229.S. R. ShanbhagB. MullerR. A. Steinbrecht2000Atlas of olfactory organs of Drosophila melanogaster 2. Internal organization and cellular architecture of olfactory sensilla.Arthropod Struct Dev29211229
- 21. McGuire S. E, Mao Z, Davis R. L (2004) Spatiotemporal gene expression targeting with the TARGET and gene-switch systems in Drosophila. Sci STKE 2004: pl6.S. E. McGuireZ. MaoR. L. Davis2004Spatiotemporal gene expression targeting with the TARGET and gene-switch systems in Drosophila.Sci STKE2004pl6
- 22. Turner E. E (1996) Similar DNA recognition properties of alternatively spliced Drosophila POU factors. Proc Natl Acad Sci U S A 93: 15097–15101.E. E. Turner1996Similar DNA recognition properties of alternatively spliced Drosophila POU factors.Proc Natl Acad Sci U S A931509715101
- 23. Nguyen D. N, Rohrbaugh M, Lai Z (2000) The Drosophila homolog of Onecut homeodomain proteins is a neural-specific transcriptional activator with a potential role in regulating neural differentiation. Mech Dev 97: 57–72.D. N. NguyenM. RohrbaughZ. Lai2000The Drosophila homolog of Onecut homeodomain proteins is a neural-specific transcriptional activator with a potential role in regulating neural differentiation.Mech Dev975772
- 24. Acosta-Alvear D, Zhou Y, Blais A, Tsikitis M, Lents N. H, et al. (2007) XBP1 controls diverse cell type- and condition-specific transcriptional regulatory networks. Mol Cell 27: 53–66.D. Acosta-AlvearY. ZhouA. BlaisM. TsikitisN. H. Lents2007XBP1 controls diverse cell type- and condition-specific transcriptional regulatory networks.Mol Cell275366
- 25. Kanemoto S, Kondo S, Ogata M, Murakami T, Urano F, et al. (2005) XBP1 activates the transcription of its target genes via an ACGT core sequence under ER stress. Biochem Biophys Res Commun 331: 1146–1153.S. KanemotoS. KondoM. OgataT. MurakamiF. Urano2005XBP1 activates the transcription of its target genes via an ACGT core sequence under ER stress.Biochem Biophys Res Commun33111461153
- 26. Ray A, van Naters W. G, Shiraiwa T, Carlson J. R (2007) Mechanisms of odor receptor gene choice in Drosophila. Neuron 53: 353–369.A. RayW. G. van NatersT. ShiraiwaJ. R. Carlson2007Mechanisms of odor receptor gene choice in Drosophila.Neuron53353369
- 27. Mann R. S, Carroll S. B (2002) Molecular mechanisms of selector gene function and evolution. Curr Opin Genet Dev 12: 592–600.R. S. MannS. B. Carroll2002Molecular mechanisms of selector gene function and evolution.Curr Opin Genet Dev12592600
- 28. Etchberger J. F, Lorch A, Sleumer M. C, Zapf R, Jones S. J, et al. (2007) The molecular signature and cis-regulatory architecture of a C. elegans gustatory neuron. Genes Dev 21: 1653–1674.J. F. EtchbergerA. LorchM. C. SleumerR. ZapfS. J. Jones2007The molecular signature and cis-regulatory architecture of a C. elegans gustatory neuron.Genes Dev2116531674
- 29. Wenick A. S, Hobert O (2004) Genomic cis-regulatory architecture and trans-acting regulators of a single interneuron-specific gene battery in C. elegans. Dev Cell 6: 757–770.A. S. WenickO. Hobert2004Genomic cis-regulatory architecture and trans-acting regulators of a single interneuron-specific gene battery in C. elegans.Dev Cell6757770
- 30. Way J. C, Chalfie M (1988) mec-3, a homeobox-containing gene that specifies differentiation of the touch receptor neurons in C. elegans. Cell 54: 5–16.J. C. WayM. Chalfie1988mec-3, a homeobox-containing gene that specifies differentiation of the touch receptor neurons in C. elegans.Cell54516
- 31. Hobert O (2008) Regulatory logic of neuronal diversity: terminal selector genes and selector motifs. Proc Natl Acad Sci U S A 105: 20067–20071.O. Hobert2008Regulatory logic of neuronal diversity: terminal selector genes and selector motifs.Proc Natl Acad Sci U S A1052006720071
- 32. Gray P. A, Fu H, Luo P, Zhao Q, Yu J, et al. (2004) Mouse brain organization revealed through direct genome-scale TF expression analysis. Science 306: 2255–2257.P. A. GrayH. FuP. LuoQ. ZhaoJ. Yu2004Mouse brain organization revealed through direct genome-scale TF expression analysis.Science30622552257
- 33. Ma Q (2006) Transcriptional regulation of neuronal phenotype in mammals. J Physiol 575: 379–387.Q. Ma2006Transcriptional regulation of neuronal phenotype in mammals.J Physiol575379387
- 34. Lee T, Luo L (1999) Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis. Neuron 22: 451–461.T. LeeL. Luo1999Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis.Neuron22451461
- 35. Lattemann M, Zierau A, Schulte C, Seidl S, Kuhlmann B, et al. (2007) Semaphorin-1a controls receptor neuron-specific axonal convergence in the primary olfactory center of Drosophila. Neuron 53: 169–184.M. LattemannA. ZierauC. SchulteS. SeidlB. Kuhlmann2007Semaphorin-1a controls receptor neuron-specific axonal convergence in the primary olfactory center of Drosophila.Neuron53169184
- 36. Newsome T. P, Asling B, Dickson B. J (2000) Analysis of Drosophila photoreceptor axon guidance in eye-specific mosaics. Development 127: 851–860.T. P. NewsomeB. AslingB. J. Dickson2000Analysis of Drosophila photoreceptor axon guidance in eye-specific mosaics.Development127851860
- 37. Obernosterer G, Martinez J, Alenius M (2007) Locked nucleic acid-based in situ detection of microRNAs in mouse tissue sections. Nat Protoc 2: 1508–1514.G. ObernostererJ. MartinezM. Alenius2007Locked nucleic acid-based in situ detection of microRNAs in mouse tissue sections.Nat Protoc215081514
- 38. Cartharius K, Frech K, Grote K, Klocke B, Haltmeier M, et al. (2005) MatInspector and beyond: promoter analysis based on transcription factor binding sites. Bioinformatics 21: 2933–2942.K. CarthariusK. FrechK. GroteB. KlockeM. Haltmeier2005MatInspector and beyond: promoter analysis based on transcription factor binding sites.Bioinformatics2129332942
- 39. Matys V, Kel-Margoulis O. V, Fricke E, Liebich I, Land S, et al. (2006) TRANSFAC and its module TRANSCompel: transcriptional gene regulation in eukaryotes. Nucleic Acids Res 34: D108–110.V. MatysO. V. Kel-MargoulisE. FrickeI. LiebichS. Land2006TRANSFAC and its module TRANSCompel: transcriptional gene regulation in eukaryotes.Nucleic Acids Res34D108110