Molecular and genetic characterization of sex-linked orange coat color in the domestic cat

Genetic mapping of Sex-linked Orange

Assuming Sex-linked orange to be a derived trait with a single, shared origin, we constructed haplotypes across the linkage interval for hemizygous orange males to delineate a 1.28 Mb interval in which all orange cats shared a common haplotype (Figures 1B, S1A, S1B). Within this interval, there were 51 variants - SNVs, indels, or structural variants - present on orange haplotypes but absent from nonorange haplotypes and the felCat9 reference genome assembly¹⁹. However, 48 of those 51 variants are evident in whole genome sequences (WGS) from pedigreed cats of seven breeds –Abyssinian, Bengal, Chartreux, Egyptian Mau, Ocicat, Havana Brown, and Toyger—that do not permit orange or calico coat color in the breed standard (https://cfa.org/breeds/). The three remaining variants include an SNV within a fixed SINE element (chrX:110,431,605), a nucleotide repeat expansion allele (chrX:109,896,099), and a 5,074 bp deletion (chrX:110,432,079-110,437,152).

The 1.28 Mb genetically defined interval is contiguous with no gaps or ambiguities¹⁹. Within this interval, the SINE-based SNV and the nucleotide repeat expansion are in non-conserved regions that lack signatures of gene regulatory function. On the other hand, the deletion lies between alternative transcriptional start sites for Rho GTPase Activating Protein 36 (Arhgap36) (Figure 1B), spans an ultra-conserved mammalian noncoding element²⁰, and harbors transcription factor binding sites associated with elevated Arhgap36 expression in neuroblastomas^21,22 (Figure 1C). The deletion is perfectly concordant with orange coat color in 188 additional cats (145 orange, 6 calico/tortoiseshell, and 37 nonorange cats, Table 1). Taken together, these observations provide strong genetic and genomic evidence that the 5 kb deletion causes Sex-linked orange.

Altered gene expression in orange cats

There are 17 annotated genes within the 1.28 Mb genetic interval and, in skin biopsies from orange and nonorange cats, expression by qRT-PCR was detectable for seven of those genes. Six genes showed no difference in expression levels; however, Arhgap36 expression was elevated 13-fold in orange relative to nonorange skin (Figure 2A, n=4 orange and 4 nonorange cats, Student’s t-test P-value=0.013). We also measured the expression of pigmentation genes that are transcriptional targets of Mc1r signaling. Levels of Dct, Mitf, Pmel, and Tyrp1 RNA are significantly reduced in orange cat skin (Figure 2B), similar to mouse models of impaired Mc1r signaling^23,24. As described below, functional studies from our group and others^25,26 provide a mechanistic link between Arhgap36 overexpression and pheomelanogenesis and, in what follows, we refer to orange and nonorange individuals as Arhgap36^del (Arhgap36^del/Y or Arhgap36^del/del) and Arhgap36⁺ (Arhgap36⁺/Y or Arhgap36^+/+), respectively.

• Download figure
• Open in new tab

Figure 2: Altered gene expression in orange cats.

(A) RNA levels of genes in the 1.28-Mb Sex-linked orange candidate interval, measured by qRT-PCR from skin biopsies of orange and nonorange adult cats (n=4 per group, *P=0.013, Bonferroni-adjusted Student’s t-test). (B) RNA levels of pigment cell-specific genes, measured by qRT-PCR from skin biopsies of orange and nonorange male, neonatal cats (n=3 per group, * P<0.10 and ** P<0.01, Student’s t-test). (C) Localization of Arhgap36 RNA (pink and red, in brightfield and fluorescence images, respectively) and eumelanin (black) in histologic sections of hair bulbs from fetal skin biopsies. Fluorescence images permit visualization of Arhgap36 RNA without potential interference from eumelanin. (D) Top panels – UMAP projections of single-cell RNAseq from fetal epidermal skin, colored by shared nearest neighbor clustering and annotated according to cell lineage. Middle and bottom panels – UMAP feature plots of Arhgap36 and Dct expression, respectively.

Arhgap36 is normally expressed in neuroendocrine organs, including the adrenal gland, hypothalamus, and pituitary gland²⁷. We examined Arhgap36 expression in these and several other tissues and observed no difference between Arhgap36^del and Arhgap36⁺ cats (Table S1). However, using in situ hybridization to examine mRNA in neonatal skin, Arhgap36 expression was not detectable in dermis, epidermis, or hair follicles from Arhgap36⁺cats, but exhibited robust and specific expression in hair follicle melanocytes from Arhgap36^del cats (Figure 2C).

To further explore the relationship between gene expression and the Arhgap36 deletion, we carried out single-cell RNA sequencing (scRNA-seq) studies of fetal skin preparations. Pigment cells are a small fraction of all skin cells but, at 44 to 48 days of gestation, developing hair follicles have yet to erupt through the skin²⁸, which allows melanoblasts to be enriched from epidermal sheets using antibodies against Kit. In scRNA-seq libraries prepared from Arhgap36^del/Y and Arhgap36⁺/Y littermates, co-clustering of both samples identified spatially coincident fibroblasts and keratinocytes populations, but spatially distinct melanoblast populations, in UMAP projections (Figure 2D). As described below, hundreds of genes are differentially expressed between Arhgap36^del/Y and Arhgap36⁺/Y melanoblast populations, including Arhgap36 itself (expressed at high levels in Arhgap36^del/Y cells, middle panel, Figure 2D) and targets of Mc1r signaling such as Dct (lower panel, Figure 2D). Notably, Arhgap36 expression was detectable only in melanoblasts and only from the Arhgap36^del/Y sample.

A total of 457 genes show differential expression between Arhgap36^del/Y and Arhgap36⁺/Y melanoblasts (adjusted P-value < 0.05, Wilcoxon rank sum test, and > 2-fold expression difference, Figures 2D, S2A), nearly an order of magnitude more than comparisons of equivalently sized keratinocyte populations. To refine and validate the set of differentially expressed genes between Arhgap36^del and Arhgap36⁺ cells, we examined a melanoblast-enriched scRNA-seq library from an Arhgap36^+/del female, in which X chromosome inactivation causes clonally inherited mosaicism in an otherwise isogenic background.

Clustering revealed two large (and two smaller) melanoblast populations defined by 473 differentially expressed genes (adjusted P-value < 0.05, Wilcoxon rank sum test, and > 2-fold expression change, Figures 2D, S2B). Because Arhgap36 is expressed predominantly in one of the two populations (P-value =1.93×10⁻¹⁴², Figure 2D), we infer that the clustering is driven by whether the Arhgap36⁺ or Arhgap36^del bearing X chromosome was inactivated. Considering the intersection between the two experiments— comparison of male littermates, and comparison of melanoblast populations from a mosaic female—286 genes are differentially expressed, with 166 and 120 genes showing higher and lower expression, respectively, in Arhgap36^del melanoblasts (Figure S2C, Table S2).

Two distinct biologic functions are apparent from our scRNA-seq results. First, Arghap36^del cells exhibit a significantly reduced expression of pigment cell-specific genes required for eumelanin (but not pheomelanin) synthesis, including Pmel, Oca2, Slc24a5, Tyrp1, and Dct (Figure S3A). These observations provide a straightforward explanation for the similar red/yellow hair color phenotypes of orange cats and loss-of-function Mc1r mutants in cats^9,10 and other mammals¹². Second, Arhgap36^del cells exhibit a significant reduction in expression of Pkib as well as 6 genes that encode cAMP-degrading phosphodiesterases, including Pde3a, Pde3b, Pde4b, Pde4d, Pde9a, and Pde10a (Figure S3B, Tables S2 and S3). Pkib encodes an inhibitor of the PKA catalytic subunit (PKA_C) and its downregulation would be expected to increase PKA_C activity. Likewise, downregulation of phosphodiesterases would be expected to increase intracellular cAMP levels, also increasing PKA_C activity. In summary, decreased expression of eumelanogenic genes that are normally activated by cAMP together with decreased expression of genes that normally reduce PKA_C activity suggests that the fundamental defect in Arhgap^del cells affects signal transduction downstream of cAMP and PKA_C.

In the littermate and the calico datasets described above (Figure 2D), Arhgap36 expression was both melanoblast- and Arhgap36^del-specific. We also generated a scRNA-seq library from whole skin of an Arhgap36^del/Y embryo at mid-gestation. We identified multiple different cell types besides melanoblasts including myoblasts, vascular and lymphatic epithelium, neural crest cells, dendritic cells, and dermal fibroblasts, but melanoblasts were the only cell type in which Arhgap36 expression was detected (Figure S4). These results together with the absence of any transcript counts from Arhgap36⁺ melanoblasts shows that ectopic expression of Arhgap36 in pigment cells (but not neural crest cell types) is a molecular signature for the Sex-linked orange mutation.

Arhgap36 function in pigment type switching

Shortly after we discovered the Arhgap36 deletion and associated increase in gene expression in orange cats, Eccles et al. reported that Arhgap36 functioned not as a Rho GTPase inhibitor but as an inhibitor of PKA signaling²⁵ by directly binding the PKA_C and targeting it for lysosomal degradation. Transcriptional activation of Mc1r-regulated genes depends on nuclear phosphorylation of the Cyclic AMP-response element binding protein (CREB) transcription factor by PKA_C^29–31; thus, an Arhgap36-mediated depletion of PKA_C protein would provide a functional model connecting melanocyte-specific ectopic expression of Arhgap36 to its effects on coat color.

Human ARHGAP36 encodes five alternative isoforms that differ in their transcriptional initiation sites and amino-terminal sequence that helps determine intracellular localization. Three of the human isoforms (2, 4, 5) localize predominantly to the plasma membrane, which is required for inhibition of PKA_C ^21,25,32. In bulk RNA sequencing of fetal epidermis from orange cats, we identified two alternative transcripts initiating from the same promoter that encode orthologs of human isoforms 4 and 5 (Figure 3A).

• Download figure
• Open in new tab

Figure 3: Arhgap36 expression leads to depletion of PKA_C.

(A) Diagram of cat transcripts corresponding to different human Arhgap36 isoforms and their predicted protein lengths^{27, 21,25,32}. Human isoform 2 does not have a homologous cat transcript; there are two transcripts that can give rise to isoform 5, one of which begins with an untranslated first exon that lies upstream of the Arhgap36 deletion. (B) Expression of mCh-ArhGAP36-iso4 or mCh-ArhGAP36-iso5 results in depletion of the catalytic subunit of PKA (PKA_C). Representative confocal images of MNT1 cells expressing mCh (left column), mCh-ArhGAP36-iso4 (middle column) or mCh-ArhGAP36-iso5 (right column) and immunostained with antibodies against PKA_C. Cells expressing mCh had similar levels of endogenous PKA_C immunofluorescence compared to neighboring non-transfected MNT1 cells (left column), while expression of mCh-ArhGAP36-iso4 (middle column) or mCh-ArhGAP36-iso5 (right column) led to decrease PKA_C immunofluorescence signal. Calibration bar: 10 μm. (C) Quantification of the average PKA_C, PKA_RIIα, PKA_RIIβ immunostaining intensity of mCh-ArhGAP36 or mCh-expressing cells calculated as the average intensity of each antibody for all cells expressing mCh compared to cells that did not express mCh in a single image. In contrast to decreased PKA_C expression in cells expressing both isoforms of mCh-ArhGAP36, no significant differences were observed in the expression levels of PKA_RIIα and PKA_RIIβ in MNT1 cells (see representative images in Fig. S6). Each dot represents one image, ≥50 cells/condition from n≥3 independent experiments; students T-test, **** P < 0.0001, ns not significant.

• Download figure
• Open in new tab

Figure 4: Mechanism of Arhgap36-dependent production of red/yellow pigment.

In Arhgap36⁺ melanocytes (left), constitutive Mc1r activity stimulates cAMP production, allowing dissociation and nuclear translocation of PKA_C from the PKA apoenzyme, followed by transcriptional activation of melanogenic genes required for eumelanin production. In Arhgap36^del melanocytes (right), ectopic expression of Arhgap36 binds to and targets PKA_C for degradation, downregulating Tyrp1, Dct, and Pmel, which leads to pheomelanin production.

Previous work on Arhgap36 and PKA_C utilized kidney cell lines (MDCK and HEK293T cells)²⁵ and motor neurons²⁶. To investigate Arhgap36 function in relevant cellular context, we generated and expression constructs encoding the domestic cat Arhgap36 isoforms 4 and 5 tagged with mCherry (mCh) at the N-terminus (mCh-ArhGAP36-iso4 and mCh-ArhGAP36-iso5) and expressed them in human melanoma MNT1 cells. Both isoforms localized at or near the plasma membrane and partially in intracellular compartments (Figure 3B). MC1R activation by α-melanocyte stimulating hormone (α-MSH) induced a larger cAMP accumulation in Arhgap36 expressing MNT1 cells compared to control (mCh) transfected cells (Figure S5). The elevated cAMP levels are consistent with the Arhgap36-associated transcriptional downregulation of cAMP-metabolizing phosphodiesterases in our scRNA-seq data and indicate that Arhgap36 attenuation of MC1R signal transduction occurs downstream of cAMP. When we used immunostaining to compare the expression of endogenous PKA in transfected MNT1 cells, and we found that levels of PKA_C but not PKA_R were significantly decreased (Figure 3B and C, S6A and B). The ability of cat Arhgap36 isoforms to deplete PKA_C, but not perturb cAMP responsiveness in MNT1 cells is consistent with Arhgap36 function in other cellular contexts and supports a functional mechanism wherein ectopic Arhgap36 expression attenuates MC1R signaling by inhibiting PKA, thereby dampening melanogenic genes expression and prompting a switch from eumelanin to pheomelanin synthesis.

Methods Biological samples

DNA samples from cats used to refine the Sex-linked orange genetic interval and to test association with the Arhgap36 intronic deletion include samples form Schmidt-Küntzel et al. (2009) and Kaelin et al. (2012), collected from cats in Frederick, Maryland, the Northern California Bay Area, and Port Alegre, Rio Grande do Sul, Southern Brazil^11,17. An orange female cat from a spay-neuter clinic in Northern California (Sja149) and a nonorange female Abyssinian (Cinnamon) were used for whole-genome resequencing. The latter cat was resequenced independently and used to build the felCat9 assembly¹⁹.

Genomic DNA was extracted from buccal swabs or whole blood using a QIAamp mini kit (Qiagen), and from tissues (testes/ovary) by proteinase K digestion, followed by chloroform extraction and ethanol precipitation.

Fetal skin tissues for single-cell RNA-seq and Arhgap36 in situ hybridization were harvested from incidentally pregnant cats at spay-neuter clinics in Northern California. Additional tissue biopsies for gene expression studies were collected from cats euthanized for reasons unrelated to this study, at Auburn University, College of Veterinary Medicine and animal shelters in Alabama and Northern California.

Tissue preparation for single-cell RNA-seq, in situ hybridization, and qPCR are described in relevant sections below.

All animal work was conducted in accordance with relevant ethical regulations and under a protocol approved by the Stanford Administrative Panel on Laboratory Animal Care (APLAC-9722).

Genotyping

The genetic interval from Schmidt-Küntzel et al.¹⁷ was refined by Sanger genotyping of 65 SNVs discovered in Montigue et al.⁴⁴ (Figure S1A) and in our resequencing data (Figure S1B). All PCR primer sets used for genotyping are provided in Table S4.

The Arhgap36 intronic deletion polymorphism was genotyped with two PCR reactions, using a common forward primer upstream of the deletion, and reverse primers located within or downstream of the deletion. Genotypes were assigned based on the presence or absence of the appropriately sized amplicon on an agarose gel. Sex was determined by PCR of a Y chromosome specific Sry amplicon. Genotyping primer sets are provided in Table S5.

Candidate interval refinement, whole genome resequencing, and variant detection

We analyzed haplotypes in O/Y male cats to identify ancestral recombination breakpoints that could narrow the genetic interval. By genotyping 27 male cats (17 orange and 10 nonorange) at 41 SNV loci, we observed that all orange cats were monomorphic at three markers that were polymorphic in the nonorange cats (Figure S1A), which delineated a 1.4 Mb genetic interval (Figure S1A).

To identify all orange-associated variants within this interval, we generated high coverage WGS from two female cats, one orange (O/O) and one nonorange (o/o). Truseq PCR-free whole genome sequencing libraries (Illumina) were constructed using 1 ug of genomic DNA. Each library was sequenced on four lanes of an Illumina HiSeq2500 sequencer to generate 125 nucleotide paired-end reads, which were mapped to the felCat9 genome assembly with BWA-MEM⁴⁵, after trimming with CutAdapt⁴⁶. Coverage depths for the orange and nonorange libraries was 45.1x and 51.9x, respectively. Single-nucleotide and short insertion/deletion variants were identified with GATK⁴⁷, Freebayes⁴⁸, and Bcftools⁴⁹ variant callers. We used Lumpy⁵⁰ to detect structural variation in the Sex-linked orange genetic interval which were then verified with the Integrative Genome Viewer (IGV, Broad Institute).

From the WGS data, we identified 24 SNVs within the 1.4 Mb interval; additional haplotype analysis of six male orange cats (Figure S1B) narrowed the candidate interval to 1.28 Mb. Within this region, 51 variants—SNVs, indels, and/or structural variants—distinguished the O/O from the o/o data; of these variants, 48 were found in breeds that do not permit orange or calico coat color, leaving three as candidates for the orange causal variant.

Quantitative RT-PCR

Dorsal skin biopsies from eight adult cats (four orange and four non-orange), preserved in RNAlater (ThermoFisher), were homogenized with a Kinematica polytron rotor-stator, PT 10-35 GT in TRIzol (ThremoFisher). Biopsies from skin, liver, kidney, heart, adrenal gland, pituitary gland, and hypothalamus were harvested from six cats (three orange and three black), euthanized 28-35 days after birth. Tissues preserved in Qiazol were homogenized using a handheld rotor-stator. In all cases total RNA was isolated using a Purelink mini columns (ThermoFisher) after on-column DNAseI treatment.

For each sample, 500ng of total RNA was reverse transcribed with Superscript IV (ThermoFisher) using oligo dT and random hexamer primers. Quantitative PCR was performed with a LightCycler 1.2 (Roche) using LightCycler FastStart DNA Master SYBR Green reagents (Roche), using primer sets in Table S5. Gapdh served as a reference gene for relative quantification.

Single-cell RNA sequencing and analysis

Prenatal staging was based on embryonic crown-rump length over the course of the domestic cat’s 63-day gestation, as described in previous work²⁸. Skin dissociations were performed on four cat embryos. One embryo was at a mid-gestation stage (4cm C-R length, 28-32 days post coitus), when hair follicle development in the skin initiates. The remaining three embryos were at a late gestation stage (8-9 cM C-R length, 44-48 days post coitus), when nascent hair is formed but has yet to emerge from the skin, facilitating dermal-epidermal separation. The mid-gestation embryo was an orange male (Arhgap36^del/Y). The late gestation embryos consisted of orange and nonorange male littermates (Arhgap36^del/Y and Arhgap36⁺/Y) and a calico female (Arhgap36^del/+).

Single cell preparations from the mid-gestation embryo were performed following a previously developed workflow, designed to enrich for basal keratinocytes but which also captures melanoblasts. Briefly, embryos were treated with Dispase (Stem Cell Technologies) and skin was removed by scraping with a No.15-blade scalpel. Isolated skin was dissociated with trypsin (Gibco) and red blood cells were removed with a red blood cell lysis solution (Miltenyi Biotec). Dissociated cells were incubated with at anti-CD49f antibody conjugated with phycoerythrin (ThermoFisher Scientific, clone NKI-GoH3 12-0495-82) and the antibody-bound fraction was captured after incubation with anti-phycoerythrin beads and enriched using a column-magnetic bead separation procedure (Miltenyi Biotec).

Single cell preparations from the late gestation embryo followed a modified protocol to enrich for melanoblasts. Dorsal skin was excised with micro scissors and treated overnight in Dispase (Stem Cell Technologies). Epidermal sheets, comprised of keratinocytes and migrating melanoblasts, were isolated by gentle scraping with a No.15-blade scalpel and dissociated with trypsin. Red blood cells were removed with a red blood cell lysis solution (Miltenyi Biotec). To enrich for melanoblasts, ten million dissociated cells were incubated with a phycoerythrin-conjugated, CD117 (c-Kit) Monoclonal Antibody (clone 2B8, eBioscience, ThermoFisher). The antibody bound fraction was then captured using a column-magnetic bead separation procedure (Miltenyi Biotech). In each case, the bound cell fraction was ∼200,000 cells, or ∼2% of the unenriched population.

3’ end, single-cell barcoded libraries were prepared from cell suspensions using 10x Genomics single-cell RNA sequencing platform for targeted capture of 6000 cells, using chemistry v2.0 and v3.0 for mid-gestation and late gestation libraries, respectively. Each library was sequenced on a single Illumina HiSeqX lane to generate ∼450 million reads (Table S6), which were aligned to the domestic cat fca126 genome assembly (GCF_018350175.1 with NCBI Felis catus Annotation Release 105) using the 10x Genomics CellRanger software (v7.1.0). The ‘cellranger count’ pipeline was used to output feature-barcode matrices. The ‘cellranger aggr’ pipeline was used to aggregate data from the orange and nonorange male littermate (Arhgap36^del/Y and Arhgap36⁺/Y) libraries into a combined matrix.

Cell clustering and differential expression analysis were performed using Seurat (v5.0.0)⁵¹. Gene expression matrices were filtered to retain cells with (a) >200 expressed features (genes), (b) between 2,000 and 100,000 UMIs counts (unique transcripts), and (c) < 10% of total expression from mitochondrial genes. Data normalization, scaling, and variance stabilization were performed with SCTransform, and PCA and UMAP embedding were used for dimensionality reduction. A Wilcoxon rank sum test for differential expression was performed between melanoblast populations from male littermates (Fig. S2A) or between the two largest melanoblast populations from the female calico sample (Fig. S2B). Melanoblast gene expression levels were aggregated by sample ID (male littermates) or by cluster ID (female calico) to convey normalized, relative expression values across genes as transcripts per million transcripts (TPM, Table S2).

RNA sequencing for Arhgap36 isoform expression analysis

An unenriched aliquot of 2.5 million dissociated epidermal skin cells, isolated from the same late gestation stage orange male (Arhgap36^del/Y) used for scRNA-seq, were resuspended in 200 ul of TRIzol (ThremoFisher). Total RNA was isolated using a Directzol RNA mini column (Zymo Research) and used to construct a strand-specific RNA sequencing library (New England Biolabs). The library was sequenced to generate 52 million 150 nucleotide paired-end reads on a partial lane of an Illumina Novaseq sequencer. Reads were aligned to the felCat9 assembly using STAR v.2.7.5⁵². Expression of Arhgap36 isoforms (Figure 3A) were determined by counting read pairs that overlap isoform specific exons.

In situ hybridization

Fetal skin from near-term (stage 21-22²⁸) orange and black littermates (N=8 biologic samples, 1 orange and 1 black sample from four litters) was dissected and fixed within 12 hours of the spaying surgery.

Tissue genotypes were confirmed by PCR.

5 mM paraffin embedded sections were processed for RNA in situ detection using RNAScope⁵³ 2.5 HD Detection Reagent – RED according to manufacturer’s instructions (Advanced Cell Diagnostics).

Arhgap36 probes were designed to target 255-1205bp of XM_006944007.5. All photomicrographs are representative of at least ten anagen follicles from each biological replicate and were captured on a Leica DMRXA2 system with DFC5500 camera and LASV4 (v.4.2.0) software.

Arhgap36 expression constructs and transfection

Two Arhgap36 isoforms expressed in orange cat skin (XM_006944008.4 and XM_019824288.3) and corresponding to human isoforms 4 and 5 (Figure 3A) were cloned in the pcDNA4/TO vector (Invitrogen) containing the mCherry coding sequence in AflII/HindIII restriction sites. Arhgap36 isoforms were inserted after the mCherry coding sequence using a GAGA linker. The constructs were generated and sequence-verified by Genscript.

MNT1 cells were grown in DMEM supplemented with 18% FBS, 10% AIM-V, 100 units/ml Penicillin-Streptomycin at 37 °C and 5% CO₂. For the immunostaining experiments cells were plated on acid washed and poly-L-Lysine-coated #1 glass coverslips (Neuvitro, GG-12) at ∼70% density in a 6-well plate and incubated at 37 °C overnight. Cells were transfected using PolyMag Neo Magnetofection Reagent (OZ Biosciences) using 1 μg of plasmid DNA and 1 μl Polymag Neo solution for each well.

Cells were fixed for immunostaining 14-18 h post-transfection.

Immunostaining, Imaging, and Image analysis

MNT1 cells were fixed in 4% Paraformaldehyde (Sigma Aldrich) for 20 min at room temperature and washed with PBS. Cells were incubated with primary antibodies diluted in blocking solution consisting of PBS with 0.2% (wt/vol) saponin (Acros), 0.1% (wt/vol) BSA (Fisher Bioreagents), 0.02% (wt/vol) sodium azide (Sigma Aldrich) for 30 min at RT, then washed with PBS 3 × 5 min, as previously described (Calvo et al., 1999). Primary antibodies used were: mouse monoclonal anti-PKA_C (BD Biosciences, 610981, 1:100), mouse monoclonal anti-PKA_RIIa (BD Biosciences, 612243, 1:100) and mouse monoclonal PKA_RIIb (BD Biosciences, 610626, 1:100). For visualization, cells were incubated for 30 min at RT with anti-mouse secondary antibodies (goat or donkey) conjugated with Alexa Fluor 488 diluted in blocking solution (1:1,000). Cells were washed 3 × 5 min with PBS and mounted with VECTASHIELD antifade mounting medium (Vector Laboratories) onto glass slides. The immunostained cells were imaged using an Olympus FV3000 confocal microscope.

Confocal images were analyzed using ImageJ software. Brightness and contrast were adjusted to visualize MNT1 cells expressing low levels of mCh-Arhgap36 or mCh (used as control). Every transfected cell was selected for comparison with 10 wild type cells in each image. For each image we calculated the average fluorescence intensity of the mCh-Arhgap36 or mCh expressing cells as a fraction of the average intensity of the wildtype (non-transfected cells). All the images from one transfection were averaged. Every transfection was equivalent to one experiment.

cAMP imaging

The FRET-based genetically encoded cAMP indicator EPAC-S^H187 (Epac H187) was a generous gift from the Jalink Laboratory (Netherlands Cancer Institute). Cells were transfected with Epac H187 and ∼24 h after transfection cells were serum-starved in OPTI-MEM (ThermoFisher Scientific) for another ∼24 h.

Coverslips were transferred to an imaging chamber with Ringer’s solution. Sequential fluorescence images were acquired with MetaMorph software on an inverted microscope every 10 s using CFP and FRET filter cubes: λ_ex = 430 nm and CFP and YFP emissions were detected simultaneously using 470±20 nm and 530±25 nm band-pass filters. After acquiring 18 baseline images (3 min), 1 μM NDP-αMSH (Sigma-Aldrich) was added, followed by a mix of 25 μM forskolin (FSK, Sigma-Aldrich) and 100 μM 3-isobutyl-1-methylxanthine (IBMX, Sigma-Aldrich) added to elicit maximal cAMP response, used for normalization. Fluorescence emission intensities were quantified as F=F_CFP/F_YFP. Normalized fluorescence intensities were quantified using F_norm(t) = (F_cell(t)− F_min)/(F_FSK+IBMX − F_min), where F_cell is the fluorescence of an intracellular region of interest, F_FSK+IBMX is the maximal fluorescence with FSK and IBMX, and F_min is the baseline fluorescence before stimulation. Light-induced changes in fluorescence intensity were quantified using MetaMorph and Excel software (Microsoft). NDP-αMSH, FSK and IBMX were solubilized in DMSO (Sigma-Aldrich) at >100x the final concentration, so that the final DMSO concentration in the imaging chamber remained < 1% (v/v) for all experiments.