Functional studies of genetic variants associated with human diseases in Notch signaling-related genes using Drosophila

Abstract

Rare variants in the many genes related to Notch signaling cause diverse Mendelian diseases that affect myriad organ systems. In addition, genome- and exome-wide association studies have linked common and rare variants in Notch-related genes to common diseases and phenotypic traits. Moreover, somatic mutations in these genes have been observed in many types of cancer, some of which are classified as oncogenic and others as tumor suppressive. While functional characterization of some of these variants has been performed through experimental studies, the number of ‘variants of unknown significance’ identified in patients with diverse conditions keeps increasing as high-throughput sequencing technologies become more commonly used in the clinic. Furthermore, as disease gene discovery efforts identify rare variants in human genes that have yet to be linked to a disease, the demand for functional characterization of variants in these ‘genes of unknown significance’ continues to increase. In this chapter, we describe a workflow to functionally characterize a rare variant in a Notch signaling related gene that was found to be associated with late-onset Alzheimer’s disease. This pipeline involves informatic analysis of the variant of interest using diverse human and model organism databases, followed by in vivo experiments in the fruit fly Drosophila melanogaster. The protocol described here can be used to study variants that affect amino acids that are not conserved between human and fly. By ‘humanizing’ the almondex gene in Drosophila with mutant alleles and heterologous genomic rescue constructs, a missense variant in TM2D3 (TM2 Domain Containing 3) was shown to be functionally damaging. This, and similar approaches, greatly facilitate functional interpretations of genetic variants in the human genome and propel personalized medicine.

1. Introduction:

1.1. Genetic variants associated with human diseases in Notch signaling related genes and the increasing demand for experimental studies to characterize their functional consequences

Notch signaling is required for the development of most, if not all, organ systems, and continuous to be required post-developmentally for tissue homeostasis and physiology (1-3). Experiments in diverse model organisms including fruit flies, nematode worms, and mice have revealed the importance of Notch signaling in various cell and tissue types in vivo (4, 5). Additionally, information gathered by physicians and clinical geneticists has elucidated the role of this evolutionarily conserved signaling pathway in human health (6, 7). Through genetic and genomic studies of rare disease patients, a number of inherited or de novo pathogenic variants in genes that are part of the core Notch signaling pathway have been discovered as causes of various diseases (8, 9). These include NOTCH1, DLL4 and RBPJ in Adams-Oliver syndrome (10), JAG1 and NOTCH2 in Alagille syndrome (11), DLL3, LFNG and HES7 in spondylocostal dysostosis (12), NOTCH3 in CADASIL (Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy) (13), and PSEN1 and PSEN2 in early-onset Alzheimer's disease (EOAD) (14). Many of these clinical studies were supported by experimental data that showed disease associated variants have functional consequences on Notch signaling, which solidified the pathogenicity of the variants identified in patients.

Discovery of new human disease genes is happening at an unprecedented pace due to advancements in high-throughput sequencing technologies including whole-exome sequencing (WES) (15, 16). For example, dominantly inherited and de novo heterozygous variants in DLL1 were identified as a cause of a new syndrome named “Neurodevelopmental disorder with nonspecific brain abnormalities and with or without seizures” in a recent study using this technology (17). For some genes that have already been linked to a known disease, variants with different functional consequences can be shown to cause a completely different disease with a distinct phenotype, a phenomenon that is referred to as ‘phenotypic expansion’ (18). For example, the skeletal disorder Hajdu-Cheney syndrome was found to be caused by late truncation variants in NOTCH2 in 2011 (19, 20), a gene previously identified in 2006 as a cause of Alagille syndrome which is a multisystem disorder primarily affecting the liver (21). In another example, lateral meningocele syndrome and infantile myofibromatosis, which are rare pediatric disorders affecting different organ systems, were found to be caused by rare variants in NOTCH3 in 2015 (22) and 2013 (23), respectively. The same NOTCH3 gene was previously linked to CADASIL which is an adult-onset brain vascular disorder first reported in 1977 (24) and molecularly mapped in 1996 (25). Hence, in addition to new human disease gene discoveries that are likely to continue for many years to come (26), documentation of new phenotypic expansions for known disease genes will equally be important for us to understand how rare genetic variants cause diverse human disorders. Because many genetic variants identified through these discovery studies require experimental validation, there is an increasing demand for collaborative studies that involve scientists who can perform functional assays on the candidate genes of interest (27).

In addition to variants that cause rare diseases, large-scale genome wide-association studies (GWAS) have found associations between certain variants in or near Notch-related genes and common diseases or other human traits. This includes the association of variants in NOTCH4 to schizophrenia (28), DLL1 to type 1 diabetes (29), and ADAM10 and NOTCH3 to late-onset Alzheimer’s disease (LOAD) (30, 31). More recently, an intronic variant in NOTCH4 has been associated with the risk of developing critical illness upon SARS-CoV-2 infection in COVID-19 patients (32), suggesting that this catalog of variant-phenotype association in Notch-related genes will continue to increase as human genomics advances. While most variants identified through GWAS lie in putative regulatory regions of genes, which are often difficult to functionally validate, some studies have focused on identified coding variants that can be experimentally tested to determine their functional consequences (31, 33, 34). As more and more personal WES and whole-genome sequencing (WGS) data become available through large scale epidemiological studies such as the “All of Us” research program in the United States (35), demand for experimental functional studies of variants associated with common diseases or traits will likely increase (36).

While rare inherited variants and germline mutations have been studied in various diseases and phenotypic traits, somatic mutations in Notch signaling related genes have also been studied in various types of cancer including leukemia and solid tumors (37). In some cancers, activation of Notch signaling is oncogenic, while in others it is tumor suppressive (38, 39). For example, mutations that cause ligand-independent activation of NOTCH1 have been found in majority of T cell acute lymphoblastic leukemia (T-ALL) (40), whereas mutations that inactivate the same gene have been found in multiple types of squamous cell carcinoma (41). Therefore, similar to inherited variants and germline mutations, it is important to functionally categorize somatic mutations that are found in cancer samples to determine whether they are truly pathogenic or not (42).

Functional studies of genetic variants linked to Notch signaling related genes have been primarily performed through cell based assays (43-45). Although reporter assays and expression analysis of downstream target genes in well established cell lines such as 293T, HeLa and CHO (Chinese Hamster Ovary) cells are highly sensitive and effective, some functional deficits may be missed because these systems do not fully recapitulate the complex cellular interactions that are present in an in vivo setting. Since Notch signaling can involve complex cellular interactions between the signal sending and receiving cells and is a highly context dependent pathway, it is ideal to study the consequence of disease-linked/associated variants in a whole organism for a more comprehensive analysis. To date, most in vivo functional characterization of disease associated variants in Notch signaling related genes have been carried out in the mouse (46, 47). While this mammalian model organism is an excellent model system that has many similarities with human and can be manipulated through diverse genome editing technologies (48, 49), other popular genetic model organisms such as fruit fly, nematode worm and zebrafish have unique characteristics that become advantages when studying variants that have been associated with various diseases (50, 51).

The fruit fly, Drosophila melanogaster, is a genetic model organism that has been extensively used to perform functional characterization of disease-linked variants over the past two decades (52). Drosophilists have access to numerous genetic technologies that allow them to manipulate the fly in almost any way one can imagine. These include sophisticated genome editing technology using CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) and highly specific site-directed transgenic technology using the ΦC31 integrase (53-55). A vast amount of knowledge on gene expression and function in Drosophila is being actively incorporated into FlyBase (http://flybase.org/) by skilled curators and is accessible to anyone interested in taking advantage of this information (56). Furthermore, tens of thousands of useful genetic reagents are publicly available from stock centers including the Bloomington Drosophila Stock Center (BDSC, https://bdsc.indiana.edu/), the Kyoto Stock Center (https://kyotofly.kit.jp/). the Vienna Drosophila Stock Center (https://stockcenter.vdrc.at), and the National Institute of Genetics of Japan (https://shigen.nig.ac.jp/fly/nigfly/) In addition, many affordable cellular, molecular and antibody resources can be obtained through Drosophila Genomics Resource Center (https://dgrc.bio.indiana.edu/) and Developmental Studies Hybridoma Bank (https://dshb.biology.uiowa.edu/). Because Notch signaling research originated from studies of flies with notched wings, and most components of the core signaling pathway are conserved between Drosophila and human, fly geneticists are in an excellent position to functionally study the effect of variants of unknown significance (VUS) in human genes whose fly ortholog have been implicated in Notch signaling.

In this chapter, we describe a workflow to assess the function of a rare variant identified in a Notch signaling related gene using Drosophila melanogaster. This protocol is specifically intended for readers who have a certain degree of expertise in fly genetics who are contacted by physicians and clinical geneticists that have identified a potential disease causing human variant in a gene that has been linked to Notch signaling in the past. If one is new to fly research, we direct the readers to the following resources to first familiarize themselves with basic principles and knowledge related to Drosophila work (57-59). Scientists, including fly biologists, can register in matchmaking registries including ModelMatcher (https://www.modelmatcher.net/) to connect with clinicians, patients and other scientists who wish to collaborate on a gene of interest. Upon being contacted by a potential clinical collaborator, a scientist should first evaluate the quality of the variant of interest from a human genetics/genomics perspective, collect information about the gene of interest in human, flies and other species, gather previously published regents to study the gene of interest, generate new genetic reagents that are tailored, and perform in vivo assays that will allow functional assessment of the variant of interest. First, we describe the general workflow related to the non-benchwork tasks associated with this process. Next, we will describe a series of protocols related to the functional study of a missense variant in TM2D3 (TM2 Domain Containing 3) (33) as an example of how such a functional study can be designed and executed at the bench using Drosophila.

1.2. Bioinformatic analysis of a genetic variant of interest, selection of an appropriate model system and overall experimental design

Once a rare variant in a gene of interest is identified in an undiagnosed patient as a potential cause of disease through WES or WGS, or a coding variant in a gene is found to be statistically significantly associated with a common disease or a phenotypic trait, physicians and clinical geneticists will often wish to understand the functional consequence of such a variant. While there are many bioinformatic programs designed to predict the functional consequence of genetic variants, there are still limitations to in silico methods (60). For example, while most nonsense and frameshift variants lead to loss-of-function (LOF) of the gene of interest, some truncating variants can be gain-of-function (GOF) alleles (61). Furthermore, functional consequences of missense variants are even more difficult to predict because there are so many possible outcomes (62). A missense variant can be a complete LOF (amorph), partial LOF (hypomorph), gain of endogenous function (hypermorph), gain of negative/toxic function (antimorph, also often referred to as ‘dominant negative’), gain of new function (neomorph) or have no functional consequences (isomorph) (63). Because there are no computational methods that can accurately classify variants according to these functional categories, clinicians often seek support from bench scientists to conduct experimental studies (64). If you are contacted by a clinician who is requesting experimental validation of a variant of interest, we recommend one to first perform literature and database searches to gather information about the gene of interest before hitting the bench.

1.2.1. Gather information about the disease of interest, patient or patient cohort, the method that was used to identify the candidate variant and the rationale that was used to rule out other candidates.

1.2.1.1. For rare disease cases in which WES or WGS was performed to identify the candidate variant, it is often important to know whether the WES/WGS was performed as a singleton (sequence of the patient is available but not for other relatives) or as a trio (sequence of the patient and his/her parents). Since each personal genome has many polymorphisms that are irrelevant to the disease of interest, singleton sequence data possesses more noise compared to trio sequence data (65, 66). Whether the variant of interest has been validated by Sanger sequencing or not can also be important because next-generation sequencing technologies still have considerable error rates (67). Moreover, a scientist should discuss with the clinician how other candidate variants identified in the patient were ruled out during the variant prioritization process to understand the logic behind the variant of interest being selected as the prime candidate. This process is crucial because it minimizes the possibility that one is working on a variant that is not the actual cause of the disease the patient has.

1.2.1.2. For genetic variants that are identified in large-scale association studies through epidemiological studies, it is important for the scientists to know the statistical power of the study, to determine whether the case and controls were properly matched, and to discuss the rationale behind why the clinicians and biostatisticians think the variant of interest is likely to be directly contributing to the disease phenotype rather than being just a marker that is co-segregating with the true culprit (68). Such communication with clinical collaborators during the early phase of the project is important because it provides a better understanding of the larger project and the precise reason that the variant should be experimentally tested.

1.2.2. Assess the likelihood of the variant being pathogenic using public human genomic databases and variant pathogenicity prediction tools.

For candidate variants for undiagnosed rare diseases, it is often useful to know whether damaging variants in the gene of interest are likely to be under selective pressure in the general population. Moreover, it is important to search whether the variant of interest has been previously reported in other individuals with or without disease. In addition, especially for missense variants, it is often useful to know whether multiple in silico pathogenicity prediction programs predict that the variant of interest is likely to have functional consequences or not. One can gather such information using an online tool called MARRVEL (Model organism Aggregated Resources for Rare Variant ExpLoration, http://marrvel.org/). which is a publicly accessible search engine that integrates data from various public databases (69). MARRVEL aggregates information from population genomic databases such as gnomAD (70) and disease cohort databases including ClinVar (71) and DECIPHER (72). In addition, MARRVEL provides pathogenicity prediction scores from nine algorithms including CADD (73) and REVEL (74). To allow users to determine whether the gene of interest has been previously linked to a genetic disorder in humans, MARRVEL displays information from OMIM (75) in a concise manner. For more information about MARRVEL and how to interpret the information provided by this tool, we recommend the readers to reference the following tutorial articles (76, 77).

1.2.3. Identify the orthologous genes in key model organisms and gather biological information from different species.

If the human gene of interest harboring the candidate variant has not been well characterized, it is often important to gather information about its orthologous genes in multiple experimental model organisms to judge whether the variant of interest in the candidate gene is likely to explain the phenotype of the patient. For example, if manipulation of the orthologous gene of interest has been shown to cause similar or related phenotypes in model organisms, the chances of a rare variant in this gene being pathogenic can increase. On the other hand, it is important to keep in mind that the lack of phenotypic similarities between model organisms and patients is usually insufficient to rule out a candidate variant because some genes may have acquired human specific roles. In addition, if the variant identified in the patient causes different types of functional alterations to the encoded protein compared to the genetic manipulations that have been performed in model organisms (e.g. knock-out, knock-down, knock-in, overexpression), there could be some discrepancies between the patient and model organism phenotypes. MARRVEL can also be used to quickly identify the orthologous candidate genes in eight species (mouse, rat, frog, zebrafish, fruit fly, nematode worm, fission yeast, budding yeast) based on an integrated ortholog prediction tool called DIOPT (78). Users can further obtain a summary of Gene Ontology (GO) terms and expression information in each species. MARRVEL further provides hyperlinks to each model organism’s gene pages for in-depth analysis of the previous literature and large scale initiatives (69). Other integrative tools provided by the Monarch Initiative (79), the Alliance of Genome Resources (80) and Gene2Function (81) are also useful when performing cross species searches. Such ‘vertical integration’ of gene function information across multiple species often helps formulate a hypothesis that can be experimentally tested in a model organism with a conserved ortholog.

1.2.4. Design specific experiments to test variant function in the proper organism and experimental settings.

Upon gathering sufficient evidence that the variant of interest is a good candidate worth functionally testing, one should select a model system to pursue this experimentally. This choice is highly dependent on the gene of interest, patient’s phenotypes and scientists’ expertise. For genes that encode proteins with known biochemical or cell biological roles, quantitative functional studies can be designed and conducted in vitro. For genes that encode proteins that are difficult to study in vitro, or those that do not have a known molecular function, experiments in vivo can often provide hints that cannot be obtained through in vitro work. When performing experiments in vivo, previous knowledge regarding phenotypes that are seen upon knock-out, knock-down and overexpression of the gene of interest provide informative entry points to design specific experiments. When multiple organisms can be used, one should consider the pros and cons of each model. For example, while a mouse model may lead to establishment of a true ‘disease model’, it can be time, labor and budget consuming. If the purpose of the initial experiment is to simply test whether a variant of interest has functional consequences, invertebrate model organisms such as Drosophila and C. elegans will likely allow one to assess this in a shorter time frame with a smaller budget. For a more global overview on ways to design variant function studies in Drosophila, we refer the readers to the following tutorial and review articles (55, 82).

2. Materials

2.1. Materials for mutant and transgenic fly generation (for subsections 3.2-3.5)

Reagents:

10 mM ATP (NEB, P0756S)

10X T4 ligation buffer (NEB, B0202S)

Antarctic Phosphatase (NEB, M0289S)

BbsI (NEB, R0539S)

BsaI-HFv2 (NEB, R3733S)

CutSmart buffer (NEB, B7204S)

ddH₂O (double distilled autoclaved water)

DH5α competent bacteria

LB media

Max Efficiency Stbl2 competent bacteria (Thermo Fisher # 10268019)

Miniprep or Midiprep kit (e.g. QIAprep Spin Miniprep Kit or Plasmid Midi Kit, QIAGEN, 28104 or 12143)

NheI (NEB, R0131)

P[acman] clone CH322-146A15 (BACPAC resources, CH322-146A15)

pattB (DGRC, 1420)

pBH vector (83)

PCR Purification Kit (e.g. QIAquick PCR Purification Kit, QIAGEN, 27104)

pCFD3-dU6:3gRNA plasmid (Addgene #49410)

Q5® Site-Directed Mutagenesis Kit (NEB, E0554S)

T4 DNA ligase (NEB, M0202S)

T4 Polynucleotide Kinase (NEB, M0201S)

TM2D3 (NM_078474) Human Tagged ORF Clone (Origene, RC203014)

XhoI (NEB, R0146S)

XbaI (NEB, R0145S)

yellow[wing2+] template vector (84)

Equipment and supplies:

0.2 ml PCR tubes (e.g. Eppendorf)

1.5 ml centrifuge tubes (e.g. Eppendorf)

15 ml centrifuge tubes (e.g. Corning)

Centrifuges (e.g. Centrifuge 5424 and 5810R, Eppendorf)

Microinjection needles (e.g. Femtotips® I capillaries, Eppendorf)

Microinjector (e.g. Femtojet® 4x, Eppendorf)

Micropipettes and pipette tips (e.g. Gilson, Eppendorf)

Stereomicroscope (STEMI2000, Zeiss)

Fly stocks:

y¹ M{vas-int.Dm}ZH-2A w*; PBac{y+-attP-3B}VK00037 (BDSC #24872)

y¹ w* (isoX); nos-Cas9 stock (derived from BDSC #78782)

2.2. Materials for egg hatching assay and neurogenic assay (Section 3.7-3.8)

Reagents:

Active yeast

Dechorionation solution: 50% bleach in H₂O

Devitellinisation solution: 100% methanol/100% n-heptane

Fixation solution: 3.7% formaldehyde in PBS (pH 7.4)/100% n-heptane

Phosphate buffered saline (PBS)

Mounting solution (e.g. VECTASHIELD® Antifade Mounting Medium with DAPI, H-1200, Vector Labs)

Normal donkey serum (for blocking solution)

Primary antibody: rat monoclonal anti-ELAV antibody (use at 1:100 dilution, DSHB, 7E8A10)

Rehydration and Washing solution: 0.05% Triton X-100 in 1x PBS (PBST)

Secondary antibody: Alexa Fluor® 488 AffiniPure Donkey Anti-Rat IgG (use at 1:200 dilution, Jackson ImmunoResearch)

Equipment and Supplies:

1.5 ml microcentrifuge tubes (e.g. Eppendorf)

20 ml liquid scintillation vials (e.g. Millipore Sigma)

24-well cell culture plate (e.g. Corning)

Dissection microscope with a digital camera [e.g. MZ16 microscope (Leica) with MicroFire microscope camera (Optronics)]

Egg laying bottles and grape juice plates

Laboratory rocker (e.g. Bio-Rad)

Laser confocal microscope (e.g. LSM 710, Zeiss)

Mesh filter (~70 μm)

Nail polish

Paint brush

Fly stocks:

BDSC Stock #10: amx¹ lz^g v¹/C(1)DX, y¹ f¹ (see Notes 1)

BDSC Stock #7770: Df(1)Exel9049, w¹¹¹⁸/Binsinscy (see Notes 2)

y w amx^Δ (generated in Section 3.2)

amx[+] (generated in Section 3.5)

TM2D3[+] (generated in Section 3.5)

TM2D3[P155L] (generated in Section 3.5)

Detailed fly genotypes shown in Figure 4:

Figure 4: Functional studies of a late-onset Alzheimer’s disease-associated rare missense variant in TM2D3 based on Notch signaling related phenotypes in vivo in Drosophila.

(A) Results of an egg hatching assay using the classical amx allele over a molecularly defined deficiency obtained from the Bloomington Drosophila Stock Center (BDSC). None of the eggs laid by the amx¹/Df(1)Exel9094 mothers hatch, while a significant portion of this can be rescued by introduction of a wild-type amx genomic rescue construct (+amx). The maximum rescue expected in this experiment is 75% because Df(1)Exel9094 removes a nearby essential gene [Df(1)9094/Y males are lethal regardless of amx function]. Humanized amx genomic rescue transgene (+TM2D3) has about 50% activity of the fly amx genomic rescue transgene and can rescue the egg hatching defect of amx¹/Df(1)Exel9094 progenies. In contrast, the humanized amx genomic rescue construct with the late-onset Alzheimer’s disease (LOAD) associated p.P155L variant (+TM2D3^p.P155L) lacks this activity. (B) Results of an egg hatching assay using the newly generated amx null allele via CRISpR. The results obtained from amx^Δ are similar to the results using amx¹/Df(1)Exel9094, further confirming the results using the classic allele. (C-F) Representative images of the nervous system of embryos that are mutant for amx rescued by different genomic rescue constructs. Neurons are visualized by immunofluorescence staining against Elav, a pan-neuronal nuclear marker. (C) Embryos from amx¹/Df(1)Exel9094 mothers show a strong neurogenic phenotype, which is a classic sign of Notch signaling defect during embryonic neurogenesis. (D) A wild-type amx genomic rescue transgene can fully rescue this defect. (E) A humanized amx genomic transgene expressing the reference human TM2D3 can also rescue this defect in a subset of animals. (F) A humanized amx genomic transgene expressing human TM2D3 with the p.P155L variant cannot rescue the neurogenic defect, indicating this missense variant is deleterious to a Notch signaling related function of this gene in vivo. Similar results have been obtained using the amx^Δ allele (data not shown). Scale bars = 100 μm. Images were adopted from (33).

• Experiments using the new amx^Δ allele

  • amx¹ lz^g v¹
  • amx¹ lz^g v¹; amx[+]
  • amx¹ lz^g v¹; TM2D3[+]
  • amx¹ lz^g v¹; TM2D3[P155L]
  • Df(1)Exel4049, w¹¹¹⁸/Binsincy
• Experiments using the new amx^Δ allele

  1. y w amx^Δ
  2. y w amx^Δ; amx[+]
  3. y w amx^Δ; TM2D3[+]
  4. y w amx^Δ; TM2D3[P155L]

3. Methods

In this section, we describe the functional analysis of a rare missense variant in TM2D3 that was identified as a novel risk allele of LOAD through a large epidemiological study by the CHARGE (Cohorts for Heart and Aging Research in Genomic Epidemiology) consortium. The CHARGE consortium is a collaborative study that integrates multiple large and well-phenotyped longitudinal cohort studies in the US and in Europe (85). In one of their studies, the consortium performed an exome-wide association study using the Exome Chip technology (86) to identify coding single nucleotide polymorphisms (SNPs) associated with increased risk of LOAD (33). Unlike EOAD which is caused by dominantly inherited pathogenic variants in a small number of genes (e.g. PSEN1, PSEN2, APP), LOAD is thought to be caused by multiple genetic and environmental factors that converge onto a common pathogenic pathway (87). Through this work, a rare missense variant in TM2D3 (rs139709573, NM_078474.3:c.464C>T, NP_510883.2:p.Pro155Leu) was identified as a risk allele for LOAD with an odds ratio of 7.5 [95% confidence interval (CI)=3.5–15.9; p=6.6x10⁻⁹] in an Icelandic cohort within the CHARGE consortium (33). In addition to the increased risk, LOAD patients who carry this allele showed an earlier age-at-onset (hazard ratio=5.3; 95% CI 2.7–10.5; p=1.1x10⁻⁶) compared to LOAD patients who do not, indicating that this allele may also be somehow contributing to the acceleration of LOAD pathogenesis.

While epidemiological data suggested that TM2D3 likely plays a role in LOAD, there were several issues that needed to be addressed before drawing a strong conclusion. First, the function of TM2D3 was completely unknown at the time in vertebrate species including human, mouse, frog and zebrafish. Therefore, even if there was a strong associative signal between a rare variant in TM2D3 with LOAD, it was difficult to develop a logical hypothesis to explain the molecular link between the gene and the disorder. Second, when the clinicians informatically assessed the probability of the variant of interest being pathogenic, multiple computational pathogenicity prediction programs including CADD and PolyPhen predicted that the p.P155L variant was likely benign. The p.P155 residue in human TM2D3 is not conserved in most species (F in mouse, N in Xenopus and zebrafish, R in Flies, V in Worm), even though the gene itself is highly conserved in metazoan species. Therefore, computational algorithms concluded that the Proline to Leucine change in this position is likely to be tolerated because this amino acid did not seem to be under selective evolutionary pressure. Finally, the p.P155L variant was identified as being significantly associated with LOAD in the Icelandic cohort within the CHARGE consortium but this signal was not seen in other cohorts. The reason for this was because this allele is about ten times more abundant in Iceland [major allele frequency (MAF): 0.45%] compared to other CHARGE cohorts in European countries and in the US with European ancestry (MAF: 0–0.06%). This enrichment is likely due to a genetic drift that happened during the establishment of the Icelandic population (88). Therefore, to provide independent evidence that the p.P155L variant is associated with LOAD in other cohorts, ten times more patients and control individuals had to be recruited into these other cohorts, which was far beyond the scope of the original study. In summary, the lack of knowledge about the gene function, bioinformatic data predicting the variant likely has no functional consequence, and inability to obtain independent epidemiological and genomic evidence to support that p.P155L in TM2D3 is a strong LOAD associated variant prohibited the clinicians to make strong statements regarding the validity of their interesting findings.

The counterargument that TM2D3 is a strong LOAD candidate gene came from pioneering studies performed in Drosophila between 1930s-2000s. The fruit fly ortholog of TM2D3 is almondex (amx), which was identified in the mid-1930s through an X-linked mutant allele that caused slightly reduced and narrowed eyes, reminiscent of a shape of an almond seed (89). amx mutants also showed female sterility, but the eye morphological phenotype was eventually lost in the amx mutant stocks that had been maintained (90). Hence, amx became the official name of the gene that is responsible for female sterility phenotype in the original amx¹ allele. In a seminal paper in 1983, Lehman et al. showed that amx mutant females are sterile because all embryos laid by homozygous amx mutant mothers shows a strong neurogenic phenotype, similar to embryos defective in genes such as Notch, Delta, neuralized and Enhancer of split (91). Since these genes are core components of the Notch signaling pathway, this finding suggested that amx plays an important role in embryonic Notch signaling as a maternal effect gene (92, 93). However, its molecular function remains unknown. In 2008, through genetic epistasis experiments, Michellod and Randsholt proposed that amx functions at the γ-secretase mediated cleavage step of Notch activation (94). Because γ-secretase is linked to Alzheimer’s disease through PSEN1 and PSEN2, which encode the core catalytic subunit of this protein complex (95, 96), a potential functional link between TM2D3 and LOAD emerged.

To further provide support that TM2D3 and LOAD are indeed functionally linked, it became essential to provide functional evidence that the p.P155L variant in this gene alters the function of the encoded protein in some way. Because there were no cellular or biochemical assays that can report the function of TM2D3, we decided to test this in a heterologous system in Drosophila by first determining whether the function of amx and TM2D3 in Notch signaling is conserved, and further assessing how the p.P155L variant impacts its function. To achieve this, we used the female sterility and embryonic neurogenic phenotype as functional readouts for amx/TM2D3 function in vivo, and either obtained or generated genetic reagents to test our hypothesis. Below, we describe our experiments in detail in hopes that this protocol will stimulate ideas on performing functional studies of other variants in this, or other, genes whose Drosophila orthologs have been linked to Notch signaling.

3.1. Identification and acquisition of pre-existing genetic reagents in Drosophila that are useful for functional studies of the variant of interest

Prior to performing the actual experiments in Drosophila, it is often important to gather information on what kind of genetic reagents (e.g. mutant strains, transgenic strains, reporter strains) have been generated for the gene of interest through previous studies, and which stocks are available from public stock centers.

3.1.1 Go to FlyBase (http://flybase.org/) and visit the gene page of interest. For amx, go to http://flybase.org/reports/FBgn0000077. Navigate the page to obtain information on mutant alleles and transgenes that have been reported in previous studies and corresponding phenotypes associated with each reagent. Although similar information can be obtained from general literature searches utilizing resources such as PubMed (https://pubmed.ncbi.nlm.nih.gov/). FlyBase presents manually curated information in a highly organized manner that greatly facilitates literature mining. For a general tutorial on how to navigate through FlyBase, we recommend the readers to the following articles (56, 97).

3.1.2 Identify the reagents useful for your study, and attempt to obtain them. If a genetic reagent is available from a public stock center, it will usually be listed under the “Stocks and Reagents” box in the FlyBase gene page. For amx, go to http://flybase.org/reports/FBgn0000077#stocks_reagents. If the genetic reagent of interest cannot be found in public stock centers, you can try to reach out to the corresponding author of a recent publication that used that allele/transgenic line to request the necessary reagent. For this study, we identified a classic mutant allele of amx from the BDSC that carries the amx¹ allele (FlyBase ID: FBal0000504) which has been used in most previous studies. Because we realized that this strain carries other mutations in nearby genes (which could be the case for many mutations that have been generated more than several decades ago), we also identified a small deficiency, Df(1)Exel9049 (FlyBase ID: FBab0047358), that removes the coding region of amx as well as several neighboring genes. The two X-chromosomes from the following two strains from BDSC can be combined in trans to generate clean hemizygous amx female flies [amx¹ /Df(1)Exel9049] that are female sterile and show maternal effect neurogenic phenotypes that can be used for functional studies (See subsections 3.7 and 3.8).

BDSC Stock #10: amx¹ lz^g v¹/C(1)DX, y¹ f¹ (see Notes 1)

BDSC Stock #7770: Df(1)Exel9049, w¹¹¹⁸/Binsinscy (see Notes 2)

3.2. Generation of new LOF reagents for the fly gene of interest

While previously reported mutant alleles are often very useful and can help kickstart a project, it can eventually become necessary to generate new mutant lines for your gene of interest. In the case of amx, the publicly available stock from the BDSC discussed above (amx¹) carries a very closely linked secondary mutation (lz^g) that causes a number of phenotypes independent of amx that can interfere with the functional interpretation when homozygous (hence, we can only use the amx¹ chromosome in trans with a deficiency that covers amx but not lz to perform clean experiments). Below, we describe a protocol to generate a new clean null allele of amx using a CRISPR-based method.

3.2.1. Select an effective method to knock-out the gene of interest.

The CRISPR-based yellow wing dominant marker insertion cassette technique offers a straightforward approach to generate molecularly defined amorphic (null) alleles that can be generated using standard equipment in a Drosophila lab (84). While it is possible to generate LOF alleles by CRISPR via designing a single guide RNA (sgRNA) to make a double stranded break within the coding region of a gene of interest and screen for small indel events that causes a frameshift through non-homologous end joining (NHEJ) (98-101), this process can be tedious since it typically requires a labor intensive molecular screening step to identify the desired gene editing event. In addition, a frameshift allele may not always be a null allele because it has the potential to generate a partial LOF, a dominant negative or a GOF allele depending on the gene and the nature of the frameshift. A method based on homology directed repair (HDR) allows full replacement of the gene of interest, allowing the researcher to be certain that the allele they are generating will be a null allele. Moreover, by replacing the gene of interest with a knock-in cassette that carries a dominant visible marker to replace the region of interest (ROI), it becomes possible to trace the desired gene editing event by visual inspection of the fly rather than relying on molecular techniques, greatly simplifying the mutant identification. Furthermore, the same visible marker can be used in downstream applications to follow the mutant allele, simplifying genetic crosses.

The technique described here uses CRISPR/Cas9 to induce double strand breaks flanking the ROI while providing a template to replace the open reading frame of the GOI with a construct that expresses the yellow gene driven by cis-regulatory elements that direct its expression in the fly wing (yellow[wing2+]) (Figure 1) (84). Specifically, we document the utilization of this approach to generate a molecularly defined null allele for amx by replacing the coding region of this gene with the yellow[wing2+] marker. This marker can be easily visualized in a yellow⁻ background, thus allowing robust identification of the flies that underwent gene editing (see Notes 3). Other dominant markers such as yellow[body+] (84), Kozak-GAL4 (102) and 3xP3-DsRed (103) can also be utilized depending on the application and specific needs.

Figure 1: Generation of a null allele of amx using the CRISPR/Cas9-based yellow wing dominant marker insertion strategy.

(A) Schematic diagram of the amx locus on the Drosophila X-chromosome. The coding region of genes are shown as orange boxes and the untranslated regions are shown in grey. CRISPR-mediated homology directed repair (HDR) was used to knock-out the amx gene by replacing it with the yellow[wing2+] cassette. The position of the sgRNA cleavage sites are shown using scissors. The upstream (UHA) and downstream homology arms (DHA) are shown using blue and red boxes, respectively. This donor plasmid also has a Kanamycin resistance (Kan^R) cassette shown using a green box, which will not be integrated into the locus of interest. (B) Schematic diagram of the microinjection and mutant selection process in vivo in flies. A cocktail of sgRNAs and donor plasmid is injected into the posterior pole of fly embryos that express Cas9 in their germline [y¹ w* (iso X); nos-Cas9]. After these embryos reach adulthood, they are crossed to a yellow white (y w) mutant strain and their progenies are screened for positive gene editing events. The flies with successful integration of the yellow[wing2+] cassette will have a darker wing color compared to flies that do not carry this cassette in a yellow mutant background. (C) Photograph of y w mutant male flies with (right) or without (left) the yellow[wing2+] cassette. The fly on the right is a null allele of amx generated through this strategy (y w amx^Δ), which has a darker wing color compared to a y w mutant with an intact amx gene.

3.2.2. Define the ROI and surrounding sequences.

Visualize the ROI using genome browsers (GBrowse or JBrowse) and download the FASTA format sequence of the gene of interest and surrounding regions from FlyBase (https://flybase.org/). Using these tools, one can identify the 5’UTR, coding sequence, 3’UTR of the gene of interest to determine the approximate ROI to delete (Figure 1A). For amx, this information can be obtained through the following links.

GBrowse Link: http://flybase.org/cgi-bin/gbrowse2/dmel?Search=1;name=FBgn0000077

JBrowse Link: http://flybase.org/jbrowse/?data=data/json/dmel&loc=FBgn0000077

FASTA file of the extended gene region (amx ± 2kb upstream and downstream)’: http://flybase.org/download/sequence/FBgn0000077/gene_extented

3.2.3. Select appropriate sgRNA sequences.

To delete the ROI, one must identify two Cas9/sgRNA target sites that flank the ROI. The Cas9 endonuclease is directed to its target site by a combination of a PAM (protospacer adjacent motifs) sequence “NGG (N is any DNA base)” and a stretch of 20 nucleotides preceding the PAM complementary to the sgRNA sequence (104). To identify Cas9 sites that flank the ROI, one can use the CRISPR Optimal Target Finder (http://targetfinder.flycrispr.neuro.brown.edu/) (103) as follows. Alternatively, DRSC/TRiP Functional Genomics Resources (https://www.flyrnai.org/crispr3/web/) also offers a similar platform with a genome viewer that can facilitate the process of sgRNA selection (105). Throughout this protocol, we refer to the genomic region 5’ to the translational start site as “upstream” and the genomic region 3’ to the translational stop site as “downstream”. Because amx is encoded on the antisense strand of the reference genome, the upstream region of amx is in the 3’ direction of the reference genome, and the downstream region of amx is in the 5’ direction (Figure 1A).

3.2.3.1. Copy and paste the amx 5’UTR sequence into CRISPR Optimal Target finder and select find CRISPR targets. This returns a list of potential sgRNA sequences in the 5’ UTR.

3.2.3.2. Evaluate the options and select the most appropriate sgRNAs. Select a target site near the coding sequence with minimal predicted off-targets, preferably with no off-targets on the same chromosome as the gene of interest. For amx, we selected 5’-TCCATTTAAGTTGTGACCATTGG-3’ (PAM sequences are underlined here) for the upstream cut site (sgRNA #1) and 5’-GAAGATCTTGCTATTCCTAATGG-3’ for the downstream cut site (sgRNA #2) as two gRNA sequences that flank the open reading frame of amx (Figure 1A).

3.2.3.3. (Optional) Determine the predicted efficiency of CRISPR-mediated cleavage. Enter the sequence of the gRNAs into the DRSC/TRiP functional Genomics resources CRISPR Efficiency Prediction tool (https://www.flyrnai.org/evaluateCrispr/) (105). This tool displays the predicted efficiency of each potential sgRNA in a genome viewer. In our experience, a sgRNA with a score of 5.5 or higher produces consistent results. If the score is low, select another sgRNA sequence and repeat this process until one identifies a good sgRNA that can be used.

3.2.3.4. Verify the sgRNA target sequences of the Drosophila strain that one will be editing to avoid polymorphisms at the sgRNA and PAM sites. This can be done by Sanger sequencing the region of interest using genomic DNA and appropriate primers, or through whole-genome sequencing using next-generation sequencing technologies (106). If there is a polymorphism, redesign the sgRNA so the sequence matches with the genomic DNA of the strain of interest, or select another sgRNA target site.

3.2.4. Design and subclone the sgRNA expression constructs.

After selecting the two sgRNAs to cut the ROI, one must develop a construct that allows the expression of these sgRNAs in vivo. The following protocol follows the procedure developed by Port et al., (107) which is outlined in detail in http://www.crisprflydesign.org/wp-content/uploads/2014/05/Cloning-with-pCFD3.pdf.

3.2.4.1. Obtain the pCFD3-dU6:3gRNA plasmid (Addgene #49410). This plasmid allows the efficient expression of the sgRNA of interest upon injection into Drosophila based on the U6:3 promoter (107). There are a number of other sgRNA expressing plasmids that can be used for this purpose, which are listed in https://www.crisprflydesign.org/plasmids/

3.2.4.2. Order oligoDNAs to subclone the sgRNA sequences.

For sgRNA #1:

5’-gtcgGAAGATCTTGCTATTCCTAA-3’ (sense strand)

5’-aaacTTAGGAATAGCAAGATCTTC-3’ (antisense strand).

For sgRNA #2:

5’-gtcgTCCATTTAAGTTGTGACCAT-3’ (sense strand)

5’-aaacATGGTCACAACTTAAATGGA-3’ (antisense strand)

Here, “gtcg” and “aaac” sequences are added to the sense and antisense strands to generate an overhang to ligate the oligoDNA into the pCDF3-dU6:3gRNA plasmid cleaved with BbsI.

3.2.4.3 Set up the following reaction to anneal and phosphorylate the sense and antisense oligoDNA in PCR tubes to generate double stranded DNA with 5’-phosphorylation:

1 ul of sense strand oligo (100 uM)

1 ul of antisense strand oligo (100 uM)

1 ul 10X T4 ligation buffer (NEB, B0202S)

0.5 ul T4 Polynucleotide Kinase (NEB, M0201S)

6.5 ul ddH₂O (double distilled autoclaved water)

(10 ul in total)

Place the PCR tubes in a thermocycler and set the following program

37°C for 30 minutes

95°C for 5 minutes

Ramp down to 25°C at the slowest setting

End reaction

3.2.4.4. Digest the pCFD3-dU6:3gRNA plasmid with BbsI (NEB, R0539S) and purify the digested product using a DNA purification kit according to the manufacturer's protocol (e.g. QIAGEN, QIAquick PCR Purification Kit, 28104).

3.2.4.5. Ligate, transform and grow the E coli colony carrying the sgRNA expression constructs. For each annealed oligo pair, set up the following reactions in PCR tubes.

50 ng BbsI-digested pCFD3-dU6:3gRNA plasmid from 3.2.4.4

1 ul of 1:200 dilution of annealed oligos from 3.2.4.3 (sgRNA #1 and sgRNA #2 in separate tubes).

1.5 ul 10X T4 ligation buffer (NEB, B0202S)

1 ul T4 DNA ligase (NEB, M0202S)

ddH₂O to total volume of 15 ul.

(15 ul in total)

Incubate the tubes at room temperature for 10 to 30 minutes or overnight (o/n) at 16°C. Transform the ligated DNA into a competent bacteria (e.g. DH5α) using standard transformation protocols. pCFD3-dU6:3gRNA carries an Ampicillin resistance marker. Pick up single colonies (we recommend selecting at least 4 colonies from each reaction) and grow them on liquid LB media (~3 ml). Miniprep the culture and verify the sequence of the insert using Sanger sequencing using the standard M13 reverse primer (5’-CAGGAAACAGCTATGAC-3’). Select the colony that carries the correct insert and generate glycerol stocks for long-term storage of each sgRNA plasmid. Perform miniprep or midiprep to obtain sufficient amounts of sgRNA plasmids for injection (see 2.6).

3.2.5. Design and assemble the HDR donor plasmid

Using Golden Gate cloning (108), assemble the yellow[wing2+] donor plasmid to replace the ROI. The yellow[wing2+] donor plasmid can be assembled from four DNA fragments: destination vector pBH (available from Drs. Benjamin Housden and Norbert Perrimon) (83), yellow[wing2+] template vector (84), and two homology arms corresponding to the upstream and downstream regions flanking the deleted segment (Figure 2). Golden Gate cloning utilizes type IIS restriction enzymes to make unique pairs of 5’-overhangs between each fragment to its neighboring fragment in the correct order for assembly (109). By alternating digestion and ligation cycles in a thermal cycler multiple fragments can be assembled together. The specific protocol applied to construct the targeting plasmid to generate the amx null allele is described below. For general tutorials on Golden Gate cloning, we refer the readers to the following articles (110-112). Alternatively, one can use other molecular strategies such as Gibson cloning (113) to assemble a donor plasmid for a specific gene of interest, which will not be described here.

Figure 2: Generation of an amx-yellow[wing2+] homology directed repair donor construct using the Golden Gate cloning strategy.

(A) In order to clone the upstream (UHA) and downstream homology arms (DHA) of the homology directed repair (HDR) donor construct, perform PCR using specific primers and genomic fly DNA. In addition to the segments that anneal with the genomic DNA, the primers designed here have features that facilitate the subcloning of these fragments using the Golden Gate strategy. (B) In addition to the two homology arms generated by PCR, this protocol requires two plasmids, one that provides the vector backbone of the final product (pBH vector shown on the left) and another that provides the yellow[wing2+] cassette. (C) Assembly of the amx-yellow[wing2+] HDR plasmid through the Golden Gate reaction. By mixing the UHA and DHA from (A), the two plasmids from (B), a type IIs restriction enzyme BsaI and a DNA ligase, the four segments will be assembled into one plasmid through repetitive digestion and ligation reactions based on the specific overhangs created by the BsaI digestion (shown as overhangs ① to ④).

3.2.5.1. Choose an appropriate type IIs restriction enzyme to be used for Golden Gate cloning. To generate the yellow[wing2+] donor plasmid to replace the coding region of amx, we chose BsaI because the recognition sequence of this enzyme is not present in the homology arms that we designed.

3.2.5.2. Design and generate the upstream and downstream homology arms. Design PCR primers to amplify the upstream and downstream homology arms. For details on how to design primers and amplicons for Golden Gate cloning, see Marillonnet and Werner (110).

3.2.5.2.1 Design the primers to amplify the upstream homology arm (UHA). For the UHA, the end of the annealing portion of the reverse primer is defined by the Cas9 cut site of dsRNA #1. For the UHA of amx, we designed and synthesized the following two oligo DNA to amplify a ~1kb region upstream of the sgRNA #1 cut site.

Forward Primer: 5’-agagagGGTCTCgTATAgagagagtacctgctctttcactcc-3’

Reverse Primer: 5’-ctctctGGTCTCtTTCCcattggccgctttagtcgtaggag-3’

Here, “ctctct” and “agagag” are random nucleotides 5’ to the BsaI binding site that were added to facilitate the digestion reaction, “GGTCTC” is the recognition sequence for BsaI, the following “t” or “g” is a random single base spacer sequence required to align the BsaI cut at the overhang, “TATA” on the forward primer is the overhang to anneal with the pBH vector (overhang ① in Figure 2A), “TTCC” is the overhang to anneal with the yellow[wing2+] cassette (overhang ② in Figure 2A) and the underlined sequences are the annealing portions of the primer.

3.2.5.2.2 Design the primers to amplify the downstream homology arm (DHA), similar to 3.2.5.2.1. For the DHA, the beginning of the annealing portion of the forward primer is defined by the Cas9 cut site of dsRNA #2. For the DHA of amx, we designed and synthesized the following two oligo DNA to amplify a ~1kb region downstream of the sgRNA #2 cut site.

Forward Primer:

5’-agagagGGTCTCgATCCggaatagcaagatcttctcaaaaacgtgtac-3’

Reverse Primer: 5’-agagagGGTCTCgGACCgagtgctccctgctaaaaccatgc-3’

Here, “agagag” are random nucleotides 5’ to the BsaI binding site that were added to facilitate the digestion reaction, “GGTCTC” is the recognition sequence for BsaI, the following “g” is a random single base spacer sequence required for BsaI, “GACC” on the forward primer is the overhang to anneal with the pBH vector (overhang ④ in Figure 2A), “ATCC” is the overhang to anneal with the yellow[wing2+] cassette (overhang ③ in Figure 2A) and the underlined sequence is the annealing portions of the primers for PCR amplification of genomic DNA.

3.2.5.2.3 Amplify the UHA and DHA by PCR based amplification of genomic DNA extracted from the flies that will be used for gene targeting (Figure 2A). To knock-out amx, we utilized the y¹ w* (isoX); nos-Cas9 stock (nos-Cas9 Derived from BDSC #78782). This strain expresses Cas9 in the germline and its X-chromosome has been isogenized and fully sequenced (84).

3.2.5.3 Perform the Golden Gate assembly reaction (Figure 2B-C)

Set up the Golden Gate reaction by preparing the following solution. In a PCR tube, add the following:

20 ng pBH-donor plasmid

Equimolar amounts of upstream homology arm

Equimolar amounts downstream homology arm

Equimolar amounts yellow wing insert plasmid

1 ul CutSmart buffer (NEB, B7204S)

1 ul 10mM ATP

0.5 ul T4 ligase (NEB M0202S)

0.5 ul restriction enzyme BsaI-HFv2 (NEB, R3733S)

ddH₂O to total volume of 10 ul.

(10 ul in total)

Perform the Golden Gate reaction in a thermal cycler using the following program:

Step 1: 37°C for 3 mins

Step 2: 20°C for 2 mins

Step 3: Repeat “Stepl and Step2” 9 times

Step 4: 37°C for 5 mins

Step 5: End program

3.2.5.4 Transform the product into competent bacteria (Max Efficiency Stbl2, Thermo Fisher) and grow these cells on a LB plate appropriate selection media (pBH is Kanamycin resistant). Pick up single colonies (we recommend selecting 4-8 colonies) and grow them on liquid LB media (~3ml). Miniprep the culture and verify the sequence through DNA fingerprinting and/or Sanger sequencing. Select the colony that carries the correct insert and generate glycerol stocks for long-term storage of the HDR donor template plasmid. Perform miniprep or midiprep to obtain a sufficient amount of sgRNA plasmids for injection (see 3.2.6).

3.2.6. Microinject the sgRNA and HDR donor plasmids and establish the genome edited stocks.

3.2.6.1 Make a sgRNA/Donor vector injection solution containing the following.

Homology Directed Repair Donor vector (200-300 ng/ul) from 3.2.5

Upstream sgRNA expression construct (25 ng/ul) from 3.2.4 (sgRNA #1)

Downstream sgRNA expression construct (25 ng/ul) from 3.2.4 (sgRNA #2)

(30 ul in total)

3.2.6.2. Microinject the injection solution into flies that express Cas9 in the germline. To generate the amx null allele, we microinjected the injection solution into the posterior region of y¹ w* (iso X); nos-Cas9 embryos that were within 1 hour after egg laying using standard transgenic techniques (114) (Figure 1B). Injected embryos are designated G0, which are mosaic animals.

3.2.6.3. Raise G0 to adults and set crosses to obtain F1 flies. Collect males and virgin females and cross single males to 3-5 virgin yellow white (y w) mutant females and 2-3 virgin females to 1-2 y w males in individual vials. These crosses produce F1 offspring with potential gene knock-out and yellow wing cassette insertion.

3.2.6.4. Screen the F1 flies for insertion of cassette and establish stable stocks. Raise F1 offspring to adults. Anesthetize F1 adult flies using CO₂ and a diffuser pad. Under a stereomicroscope (0.63 – 5X) at low power, examine wings of each F1 fly for dark pigment resulting from the yellow wing dominant marker construct (Figure 1C). Pick male flies with dark colored wings and individually cross them to appropriate balancer stocks (e.g. FM7c).

3.2.6.5. Molecularly validate the targeting event. Collect 6-8 adult flies, isolate Genomic DNA from each balanced line and perform PCR across each homology arm to the insert followed by Sanger sequencing to confirm the editing event. The PCR primers must start from the genomic DNA flanking the homology arms to ensure correct insertion. We recommend designing primers ~50 base pairs outside of the homology arm ends and the primers within the yellow[wing2] insert. Primers 5’-GCACTGCGCTCAAACTATTAGTACCCC-3’ and 5’-CGGAGCCAGACTTCAGACGGG-’3 can be used to sequence the upstream and downstream homology arms, respectively.

3.3. Generation of a genomic rescue construct for the fly gene of interest.

In addition to mutant flies, it is important to generate flies that express a reference and variant protein of interest to assess the functional consequence of the genetic variant of interest. This can be accomplished using the UAS/GAL4 system (115) by cloning the human or fly cDNA under the control of the upstream activation sequence (UAS) and expressing the reference and variant forms with ubiquitous or tissue/gene specific GAL4 drivers (82). Although many GAL4s, including gene specific T2A-GAL4 (116) drivers, are available to express the proteins of interest in any tissue or developmental time point, the protein of interest expressed through this system tends to be overexpressed compared to endogenous levels. Alternatively, one can engineer the variant of interest in the endogenous gene using scarless genome editing (84, 117) or perform rescue experiments of a LOF allele using a genomic rescue construct that contains the variant of interest (118). However, this strategy does not allow testing of variants that affect non-conserved amino acids. As a hybrid approach, one can first humanize the fly genomic rescue construct and then introduce the variant of interest. Here, we describe the methodology we used to humanize the fly amx genomic rescue construct with human TM2D3, and to further introduce the p.P155L variant linked to AD that affects a residue that is not conserved between Drosophila Amx and human TM2D3 (33) (Figure 3).

Figure 3: Generation of a series of amx genomic rescue constructs to test a disease-associated variant in TM2D3.

(A) Schematic diagram of the amx locus on the Drosophila X-chromosome. The coding region of genes are shown as orange boxes and the untranslated regions (UTR) are shown in grey. Two primers (Oligo 1 and 2) were designed to amplify a ~3.3kb region that contains the amx gene and neighboring sequences. (B) Construction of the amx[+]-attB genomic rescue construct by subcloning the PCR amplicon from (A) into pattB using XhoI and XbaI digestion and ligation. pattB contains a mini-white⁺ cassette (red box) as a dominant visual marker that can be used to track the transgenesis event in a white⁻ mutant background. It also carries an Ampicillin resistance (Amp^R) cassette (green box) for selection in bacteria. (C) Construction of a humanized amx genomic rescue construct (TM2D3[+]-attB) based on amx[+]-attB and two intermediate vectors. Human reference TM2D3 coding sequence was PCR amplified from a cDNA plasmid and integrated into a plasmid in which the amx coding sequence was replaced by a spacer sequence flanking two BsaI recognition sites (Intermediate Vector 2). A construct in which the upstream region of the amx genomic rescue amplicon with a BsaI site was cloned into pattB (Intermediate Vector 1) as a precursor of Intermediate Vector 2. Golden Gate cloning with BsaI and overhangs ①/② allows swapping of the spacer sequence in Intermediate Vector 2 with human TM2D3 coding sequence to generate the humanized genomic rescue construct. (D) Generation of the humanized amx genomic rescue construct carrying the variant of interest (TM2D3[P155L]-attB) through site-directed mutagenesis. The variant of interest is shown using an asterisk.

3.3.1. Design of a genomic rescue fragment for the gene of interest

In order to generate a genomic rescue construct that correctly expresses a specific gene of interest, it is often necessary to capture not only the coding sequence but also its accompanying upstream and downstream untranslated regions (UTRs), the promoter sequence, and the cis-regulatory elements (CRE) necessary for the correct temporal and spatial expression pattern. Ideally, this genomic rescue fragment (GRF) should not include the complete sequence of neighboring genes, so that it is only capable of expressing the specific gene of interest, and can therefore show specificity of any given phenotype it rescues. For most genes, the CRE that drive gene expression are not well mapped. Therefore, the boundaries of the GRF are generally determined by the surrounding genomic structure. The GRF should be limited in size by the constraints of the plasmid that are being used. Here, we used pattB (DGRC, 1420) as the plasmid backbone, which is suited to integrate DNA fragments that are <6kb into specific attP docking sites using the ΦC31 integrase-mediated transgenesis system (119). This construct carries a mini-white⁺ marker (120), which allows users to screen for transgenic flies by examining their eye color in a white⁻ background. For larger GFRs, one can utilize fosmid or BAC (bacterial artificial chromosome) based backbones (121, 122), which will not be discussed here.

3.3.1.1 Examine the sequences flanking the gene of interest by using the GBrowse or JBrowse feature on FlyBase as in 3.2.2.

3.3.1.2 Select a genomic fragment that likely contains all the CRE of the gene of interest. For amx, we selected an approximately 3.3 kb fragment that includes a short segment of the 3’ end of the upstream gene Larp7, the coding sequence and associated UTRs of amx, and a short segment of the downstream gene Dsor1 including most of Dsor1 intron 1 (Figure 3A). Partial sequences in the neighboring genes were included in case they contain distant CRE that regulate the expression of amx, but functional Larp7 and Dsor1 will not be generated from this GRF due to lack of intact coding regions.

3.3.1.3 Design PCR primers flanking the chosen sequence with appropriate binding sites for restriction endonucleases to subclone this fragment into pattB. We selected the following two oligo DNAs and Xho I and Xba I as upstream and downstream restriction enzymes, respectively.

Forward Primer (Oligo 1): 5’-ctctctCTCGAGgttatgttgcctacattttggtgctcac-3’

Reverse Primer (Oligo 2): 5’-cttctTCTAGAgcgtcgcatcgtcagtgaggc-3’

Here, “ctctct” and “cttct” are random nucleotides 5’ to the restriction enzyme binding site that were added to facilitate the digestion reaction, “CTCGAG” and “TCTAGA” are recognition sequences for XhoI and XbaI, respectively. The underlined sequences are the annealing portions of the primers.

3.3.2. Perform a PCR reaction to amplify the GFR.

Different sources of genomic DNA can be used as a template, including genomic DNA isolated from a wild-type fly strain or BAC clones that contain the GFR. We performed PCR from a P[acman] clone CH322-146A15 (BACPAC resources, CH322-146A15), which is a clone that was generated from genomic DNA extracted from the Drosophila strain used to build the reference genome (123, 124).

3.3.3. Subclone the GFR into the pattB plasmid.

Digest the PCR product and pattB plasmid with appropreate restriction endonucleases. Here, we used XhoI and XbaI. After dephosphorylation of the pattB plasmid using a DNA phosphatase (e.g. Antarctic Phosphatase, NEB, M0289S), perform a ligation reaction using a DNA ligase (e.g. T4 DNA ligase, NEB, M0202S), according to the manufacturer's protocol to generate the genomic rescue plasmid.

3.3.4. Validate the GFR-pattB genomic rescue construct using Sanger sequencing.

First perform end sequencing using sequencing primers that anneal to the vector backbone. For pattB, one can use the following primers.

pattB insert sequencing primer 1:

5’-CATTTAATTTAGAAAATGCTTGGATTTCACTG-3’

pattB insert sequencing primer 2:

5’-CCGCACCGCGGCTTCGAGACC-3’

To verify that the entire GFR is correct, design gene specific sequencing primers to cover the whole GFR. Identify a colony with the proper sequence and keep a glycerol stock for long term storage. We refer to this construct as amx-pattB (Figure 3B). Perform miniprep or midiprep to obtain a sufficient amount of this genomic rescue construct for injection (see 3.5).

3.4. Design and construct a humanized genomic rescue construct

Since the amx gene contains only a single exon, the coding region of the fly amx genomic rescue construct can be relatively easily replaced with a human TM2D3 coding sequence to generate a humanization construct. From this construct, human TM2D3 protein will be expressed in the same spatial and temporal pattern of the fly Amx protein, likely at comparable levels (although there may be some differences in translational efficiency and protein stability due to sequence differences). The following protocol describes the precise method used to construct the humanized genomic rescue construct in Jakobsdottir et al (33). Although such cloning can be performed through simple restriction enzyme digestion and ligation reaction if appropriate restriction enzymes cut sites can be found, absence of convenient cut sites forced us to perform this in a multi-step procedure, involving the generation of intermediate plasmids that are suitable for Golden Gate cloning (Figure 3C). It is worth noting that recent advances in multi-fragment assembly have greatly simplified cloning, including the development of commercial kits such as the Gibson Assembly® Master Mix (NEB, E2611S) and NEBuilder® HiFi DNA Assembly Master Mix (NEB, E2621S). These and other techniques can now be used instead of the method we used, but the method described here can be more cost effective since it does not require researchers to purchase customized commercial kits.

3.4.1 Obtain a human cDNA plasmid that contains the full length open reading frame. TM2D3 has two splicing isoforms and the difference between the two is the presence or absence of an alternatively spliced exon. We obtained the longer isoform of TM2D3 (NM_078474) from a commercial source (Origene, RC203014) for this project.

3.4.2 Clone the upstream region of the GRF with a BsaI restriction site at the 3’ end into the pattB vector to make the first intermediate vector. The forward primer for the upstream region is the same as the forward oligo used for the cloning of the amx GRF (Oligo 1, see 3.3.1.3). The reverse primer will provide a reverse facing BsaI site to leave an upstream 5’ overhang to insert the human TM2D3 coding sequence and an XbaI site to insert into the pattB plasmid.

Reverse Primer (Oligo 3):

5’-agagagTCTAGAagagaGGTCTCgtttggtcagatagagcggggaatacgc-3’

Here, “agagag” is a series of random nucleotides 5’ to the restriction enzyme binding site that was added to facilitate the digestion reaction, “TCTAGA” is a recognition sequence for XbaI, “agaga” is a spacer sequence, “GGTCTC” is the recognition sequence for BsaI, the following “g” is a random single base spacer sequence required for BsaI, and the underlined sequence is the annealing portions of the primers.

Perform PCR using pattB-amx plasmid as a template and subclone the amplicon using restriction endonucleases XhoI and XbaI into pattB. Confirm the plasmid using Sanger sequencing and generate a glycerol stock. We refer to this vector as ‘Intermediate Vector 1’

3.4.3 Clone the downstream region of the GRF with a BsaI restriction site at its 5’ end into the Intermediate Vector 1 to make the second intermediate vector. The forward primer will provide a forward facing BsaI site to leave a downstream 5’ overhang to insert the human TM2D3 coding sequence, and is flanked by NheI upstream of the BsaI site that can be used to clone the insert into Intermediate Vector 1. The reverse primer for the downstream region is the same as the reverse primer used for the cloning of the amx GRF (Oligo 2, see 3.3.1).

Forward Primer (Oligo 4):

5’-gcggcGCTAGCGGTCTCttttagttttacaggggttaataggtcataagctaac-3’

Here, “gcggc” is a series of random nucleotides 5’ to the restriction enzyme binding site that was added to facilitate the digestion reaction, “GCTAGC” is a recognition sequence for NheI, “GGTCTC” is the recognition sequence for BsaI, the following “t” is a random single base pair spacer sequence required for BsaI, and the underlined sequence is the annealing portions of the primer.

Perform PCR using pattB-amx plasmid as a template and Oligo 4 and Oligo 2 as forward and reverse primers, respectively, to subclone the amplicon using NheI and XbaI into Intermediate Vector 1 (from 3.4.2). It is important to keep in mind that this insert can ligate in either the forward (correct) or reverse (incorrect) orientation. In the correct orientation, an NheI/XbaI hybrid site internal to the BsaI flanked spacer sequence is formed, providing a way to distinguish the two products. Confirm the correct plasmid using Sanger sequencing and generate a glycerol stock. We refer to this vector as ‘Intermediate Vector 2’ (see Notes 4). The region between the two BsaI sites of this plasmid is a spacer sequence that will be removed and replaced with the human TM2D3 coding sequence in subsequent steps.

3.4.4 Insert the TM2D3 coding sequence into the amx genomic locus using Golden Gate assembly.

The Intermediate Vector 2 from 3.4.3 now has the upstream and downstream components flanking an ends-in BsaI removable cassette that will leave two different 5’ overhangs when digested with the BsaI restriction enzyme. The TM2D3 coding sequence can be swapped with this cassette to make the final humanized rescue construct. Although the pattB vector has two additional BsaI sites within its sequences, Golden Gate assembly can be used to insert the TM2D3 coding sequence while preserving the integrity of the pattB backbone using the following procedure:

3.4.4.1 PCR amplify the TM2D3 coding sequence from a human cDNA plasmid (Origene, RC203014) and flank the amplicon with BsaI restriction sites using the following primers. In addition to sequences that anneal with the TM2D3 open reading frame, additional sequences have been added to generate 5’ overhangs compatible with the upstream and downstream cohesive ends generated when the spacer cassette is removed from Intermediate Vector 2 upon BsaI digestion.

Forward Primer (Oligo 5):

5’-gcggcGCTAGCGGTCTCtCAAAatggcgggaggggtgctccc-3’

Reverse Primer (Oligo 6):

5’-agagagTCTAGAagagaGGTCTCgTAAActaaatgtacaaagagccatctgctgg -3’

Here, “agagag” and “gcggc” are series of random nucleotides 5’ to the restriction enzyme binding site that were added to facilitate the digestion reaction, “GCTAGC” and “TCTAGA” are recognition sequences for NheI and XbaI which were inserted for validation purposes but are not strictly necessary (will be cleaved off upon BsaI digestion), “GGTCTC” is the recognition sequence for BsaI, the following “t” or “g” are random single base spacer sequences required for BsaI, “CAAA” and “TAAA” are overhangs (overhangs ① and ②, respectively, in Figure 3C) for ligating the TM2D3 coding sequence in place of the spacer sequence that is removed upon BsaI digestion, and the underlined sequences are the annealing portions of the primers.

3.4.4.2 Set up a Golden Gate reaction in a PCR tube including the following:

20 ng Intermediate Vector 2 from 3.4.3

Equimolar amounts of the TM2D3 amplicon from 3.4.4.1

1 ul CutSmart buffer (NEB, B7204S)

1 ul 10 mM ATP

0.5 ul T4 ligase (NEB M0202S)

0.5 ul restriction enzyme BsaI-HFv2 (NEB, R3733S)

ddH₂O to total volume of 10ul

(10 ul in total)

3.4.4.3 Run the Golden Gate program in a thermal cycler.

Step 1: 37°C for 3 mins

Step 2: 20°C for 2 mins

Step 3: Repeat Step 1 and 2 ten times.

Step 4:20°C for 15 mins

(This final ligation step is necessary because pattB has additional BsaI sites)

3.4.4.4 Transform the product into chemically competent bacteria (e.g. DH5α or equivalent strain) and grow these cells on a LB plate with appropriate selection media (pattB is Ampicillin resistant) o/n. Pick up single colonies (we recommend selecting 8-10 colonies) and grow them on liquid LB media (~3ml) o/n. Miniprep the culture and verify the construct via DNA fingerprinting and Sanger sequencing. We refer to this plasmid as TM2D3[+]-attB. Select the colony that carries the correct insert and generate glycerol stocks for long-term storage of this humanized genomic rescue plasmid.

3.4.5 Introduce the variant of interest into the humanized genomic rescue construct.

Perform site directed mutagenesis of the humanized genomic rescue construct to introduce the variant of interest (Figure 3D). Here, we describe a protocol using the Q5® Site-Directed Mutagenesis system (NEB) to introduce the p.P155L into the humanized amx genomic rescue construct (TM2D3[P155L]-pattB). Additional methods such as QuikChange II Site-directed mutagenesis system (Agilent) and Phusion Site-Directed Mutagenesis system (ThermoFisher) can also be employed, which will not be discussed here.

3.4.5.1 Set up a mutagenesis reaction using high-fidelity DNA polymerase (Q5 Hot Start High-Fidelity DNA Polymerase included in the Q5® Site-Directed Mutagenesis kit) and appropriate primer pairs in a PCR tube. Perform a mutagenic PCR reaction according to the manufacturer's protocol. To introduce the p.P155L variant (NM_078474.3:c.464C>T), we designed and used the following mutagenesis primers and followed the protocol for Q5® Site-Directed Mutagenesis.

Forward mutagenesis primer:

5’-CCTGTCCTCGGCAGCGCTACCTTGCCAACTGCACGGTGCGGG-3’ (forward)

Reverse mutagenesis primer:

5’-CCCGCACCGTGCAGTTGGCAAGGTAGCGCTGCCGAGGACAGG-3’ (reverse)

Here, the underlined base corresponds to the mutation that will be introduced.

3.4.5.2. Ligate the amplified product and digest the template plasmid using the KLD enzyme mix included in the Q5® Site-Directed Mutagenesis kit based on the manufacturer's protocol.

3.4.5.3. Transform the product into chemically competent bacteria (e.g. DH5α or equivalent strain) and grow these cells on a LB plate with appropriate selection media (pattB is Ampicillin resistant) o/n. Pick up single colonies (we recommend selecting 12 colonies) and grow them on liquid LB media (~3ml) o/n. Miniprep the culture and verify the construct via DNA fingerprinting and Sanger sequencing (see Notes 5). Select the colony that carries the correct insertion and generate glycerol stocks for long-term storage. We refer to this plasmid as TM2D3[P155L]-pattB. Perform miniprep or midiprep to obtain sufficient amounts of this mutagenized humanized genomic rescue plasmid for injection (see 3.5).

3.5. Generation of transgenic flies that carry the original and humanized genomic rescue constructs

3.5.1 Make a plasmid injection solution. Adjust the genomic rescue constructs from 3.3 & 3.4 to 200-300 ng/ul with ddH₂O.

3.5.2 Microinject the injection solution into flies that express ΦC31 integrase in the germline. A number of stocks are available from BDSC (https://bdsc.indiana.edu/stocks/phic31/phic31_int.html). For this project, we selected a line which expresses ΦC31 integrase in the germline using a vasa promoter and carries an attB docking site on the 2nd chromosome (y¹ M{vas-int.Dm}ZH-2A w*; PBac{y+-attP-3B}VK00037, BDSC: 24872). Inject the solution prepared in 3.5.1 into the posterior region of y¹ M{vas-int.Dm}ZH-2A w*; PBac{y+-attP-3B}VK00037 embryos that were within 1 hour after egg laying using standard transgenic techniques (114) (Figure 1B). Injected embryos are designated G0 and are mosaic.

3.5.3 Raise G0 to adults and set crosses to obtain F1 flies. Collect males and virgin females and cross single males to 3-5 virgin yellow⁻ and 2-3 virgin females to 1-2 yellow⁻ males in individual vials. These crosses produce F1 offspring with potential transgene insertion.

3.5.4 Screen the F1 flies for insertion of transgene and establish stable stocks. Raise F1 offspring to adults. Anesthetize F1 adult flies using CO₂ and a diffuser pad. Under a stereomicroscope (0.63 – 5X) at low power, examine wings of each F1 fly for red eye color (Figure 1C). Pick male flies with dark colored wings and individually cross them to appropriate balancer stocks (e.g. CyO). Crossing scheme to establish stable stocks using the ΦC31 transgenesis system can be found in (82).

3.5.5 Molecularly validate the targeting event. Collect 6-8 adult flies, isolate Genomic DNA from each balanced line and perform PCR with appropriate primer pairs to confirm the successful integration of the transgenes (119, 121).

3.6. Establishment of fly stocks to be used for functional assessment of the variant of interest.

To perform rescue experiments, one must combine the genomic rescue transgenes generated in 3.5 with mutant alleles obtained or generated in 3.1 or 3.2. The amx gene is located on the X-chromosome, and the genomic rescue constructs have been inserted onto the 2nd chromosome. Using standard genetic techniques in Drosophila, construct the following fly strains that have an amx mutant allele and a genomic rescue transgene. For tutorials on designing fly crossing schemes, we refer the readers to the “Fly Pushing” book by Greenspan (57).

• amx¹ lz^g v¹; amx[+]

• amx¹ lz^g v¹; TM2D3[+]

• amx¹ lz^g v¹; TM2D3[P155L]

• y w amx^Δ; amx[+]

• y w amx^Δ; TM2D3[+]

• y w amx^Δ; TM2D3[P155L]

3.7. Functional study of the variant of interest based on female fertility

Studies have reported that maternal loss of amx in Drosophila leads to production of embryos that fail to hatch from eggs (89). Therefore, egg hatching rate can be used as a functional measure of amx activity in vivo. Furthermore, if the humanized genomic rescue transgene also rescues this defect, egg hatching rate can be used as a quantitative measure to assess the functional difference between the reference and variant human TM2D3 transgenes. Here, we present a detailed protocol of egg hatching assay and analysis.

3.7.1. Generate or collect the amx mutant flies for the experiment

3.7.1.1. For the experiment using the classic amx¹ allele from 3.1, cross the amx¹ lz^g v¹ males with or without the rescue transgene on the 2nd chromosome ( − ; amx[+]; TM2D3[+]; TM2D3[P155L]) to Df(1)Exel4049, w¹¹¹⁸/Binsinscy virgin females (P0 generation).

3.7.1.2. In the following generation (F1), collect virgin female flies that are amx¹ lz^g v¹ /Df(1)Exel4049, w¹¹¹⁸; (with or without rescue transgene)/+.

3.7.1.3. For the experiment using the new amx^Δ allele from 3.2, collect virgin females from respective stocks (see Notes 6).

3.7.2. Embryo collection, quantification and analysis

3.7.2.1. Cross the amx mutant females to amx mutant males (amx¹ lz^g v¹/Y or y w amx^Δ/Y) and place them in a plastic bottle with a grape juice plate supplemented with active yeast paste. Keep the cross at 25°C for two days.

3.7.2.2. Replace with a new grape plate and allow flies to lay eggs (F2 generation) on the plate with some yeast paste at 25°C for ~6 hours.

3.7.2.3. Place the grape plates with embryos in a humid-controlled incubator for 24 hours at 25°C.

3.7.2.4. Use a fine paint brush washed with water to collect embryos on a mesh filter.

3.7.2.5. Dechlorinate embryos with 50% bleach for 2 minutes. Check under a dissection microscope to confirm that the outer egg shell (chorion) has been successfully removed.

3.7.2.6. Rinse the dechorionated embryos with running water to completely remove the bleach.

3.7.2.7. Move the embryos into a 24-well cell culture plate well that contains ~500 ul of PBS. To facilitate the qualification process, do not overcrowd the embryos (<100 embryos per well) so that the embryos do not overlap.

3.7.2.8. Let the embryos float to the top of the PBS and take a photograph of each well.

3.7.2.9 Incubate the plate at 25°C for 24 hours. During this time period, all normal embryos should have already hatched from their egg case.

3.7.2.10 Take a photograph of each well after the 24-hour incubation and count the number of larvae that have successfully hatched. Embryos that look completely white should be excluded from the calculation because they are most likely to be unfertilized eggs.

3.7.2.11 Calculate the egg hatching rate by dividing the number of larvae by the number of eggs being laid (% = hatched eggs/total eggs x 100). Repeat the process at least three times to obtain biological replicates (Figure 4A-B) (33, 125).

3.8. Functional study of the variant of interest based on the embryonic neurogenic phenotype

In addition to failure of hatching, embryos laid by amx defective mothers also show a robust embryonic neurogenic phenotype, which is a sign of Notch signaling mediated lateral inhibition (126, 127). Therefore, the embryonic neurogenic phenotype can be used as an indirect measure of Notch signaling activity in developing embryos to assess the functional impact of the disease-associated variant in vivo.

3.8.1. Generate or collect the amx mutant flies for the experiment.

Follow the crossing scheme outlined in 3.7.1 to obtain the necessary flies.

3.8.2. Embryo collection and fixation

3.8.2.1. Cross the amx mutant females with or without genomic rescue transgenes to amx mutant males (amx¹ lz^g v¹/Y or y w amx^Δ/Y) and collect/process the embryos as in 3.7.2.1-3.7.2.6.

3.8.2.2. Use a fine brush to transfer the embryos to 3ml of fixation solution in a 20ml scintillation vial.

3.8.2.3. Add 3ml of 100% heptane (1:1 to formaldehyde/PBS) to the scintillation vial. n-heptane and formaldehyde/PBS form two phase layers and dechorionated embryos remain at the interface.

3.8.2.4. Incubate the embryos for 20 minutes at room temperature on a laboratory rocker to fix the embryos while constantly shaking the vials.

3.8.2.5. Remove the lower formaldehyde/PBS layer using a micropipette.

3.8.2.6. Add 3 ml of 100% methanol (methanol will sink below the n-heptane layer) and then shake vigorously by hand for 30 seconds.

3.8.2.7. Place the scintillation vial on the bench until two phase layers of heptane and methanol are formed. Fixed and devitellinated embryos will settle to the bottom in the lower (methanol) layer.

3.8.2.8. Remove the upper n-heptane layer, the debris at the interface and most of the methanol until embryos are only covered in a small amount of methanol.

3.8.2.9. Wash the embryos with fresh 100% methanol to remove any remaining fixative and n-heptane. Transfer the embryos in methanol to 1.5ml centrifuge tubes. Embryos in this condition can be preserved long-term at −20°C until use.

3.8.3. Staining and mounting

3.8.3.1. Rehydrate and permeabilize the embryos from 3.8.4 by incubating them in 0.05% Triton-X in PBS (PBST) three times for 20 minutes each at room temperature.

3.8.3.2. Incubate the embryos with a primary antibody against a neuronal nuclear marker (rat monoclonal anti-ELAV antibody, 1:100, DSHB, 7E8A10) in a blocking solution (5% normal donkey serum in PBST) at 4°C o/n.

3.8.3.3. Wash the embryos three times with PBST for 20 minutes each at room temperature.

3.8.3.4. Incubate the embryos with a secondary antibody (Alexa Fluor® 488 AffiniPure Donkey Anti-Rat IgG, 1:200, Jackson ImmunoResearch) in blocking solution (5% normal donkey serum in PBST) for two hours at room temperature.

3.8.3.5. Wash the embryos three times with PBST for 20 minutes each at room temperature.

3.8.3.6. Remove as much PBST as possible and add sufficient amount of mounting media (VECTASHIELD® Antifade Mounting Medium with DAPI, H-1200, Vector Labs) into the tube to cover sample. Transfer the embryos in the mounting media on to a glass slide with a trimmed pipette tip and cover with a coverslip. Secure the coverslip with nail polish.

3.8.4. Imaging and Analysis

Image the samples with a fluorescence microscope with an appropriate laser and filter. For best results, use a laser confocal microscope and make a z-stack out of multiple optical sections (Figure 4C-F) (33, 125).

4. Notes

1. C(1)DX, y¹ f¹ (FlyBase ID: FBab0000080) that is used to maintain the amx¹ lz^g v¹ chromosome is a compound X-chromosome. All males in this stock are amx¹ lz^g v¹/Y and all females in this stock are C(1)DX, y¹ f¹/Y.

2. Binsinscy (FlyBase ID: FBba0000019) that is used to maintain the Df(1)Exel9049, w¹¹¹⁸ chromosome is a balancer chromosome for the X-chromosome. This chromosome carries a recessive female sterile mutation in addition to several dominant and recessive markers. Although there are homozygous Binsinscy females present in this stock, they can be distinguished from the Df(1)Exel9049, w¹¹¹⁸/Binsinscy flies based on the semi-dominant Bar mutation (B¹) present on Binsincy.

3. Another advantage of using this system is that the yellow[wing2+] knock-in mutant flies can be further used to perform scarless gene editing of the gene of interest. For more information on this application, see Li-Kroeger et al (84).

4. One benefit of generating this Intermediate Vector 2 is that it can be used as a template to make additional constructs that utilize the regulatory region of amx. For example, one can insert components such as GAL4, EGFP or Luciferase instead of human TM2D3 to generate additional reagents to study this Drosophila gene in more detail.

5. To avoid the risk of introducing second site mutations during the mutagenic PCR step, one can subclone the humanized genomic fragment into a new attB plasmid. This will eliminate the possibility that there is a second site mutation on the vector backbone that may interfere with the transgenesis process (e.g. mutation in the attB site or mini-white⁺ marker).

6. One advantage of using a clean null allele is that it reduces the amount of fly work required to perform the egg hatching and embryonic neurogenic assays. Because the amx^Δ chromosome does not carry any other mutations that affect this assay, one can use this chromosome in a homozygous state by collecting virgin females directly from each stock. This is in contrast to the amx¹ chromosome which carries a closely linked lz^g mutation that needs to be neutralized by placing this chromosome over a deficiency that does not affect lz.