Uncovering functional insights into human pathogenic variants in CDK19 using Drosophila models

Abstract

Heterozygous missense variants in CDK19 have been found in patients diagnosed with Developmental and Epileptic Encephalopathy-87 (DEE87) who present with global developmental delay, intellectual disability and other neuromuscular deficiencies. Two missense variants in CDK19, Y32H and T196A, were first proposed to be loss of function based on experiments in Drosophila models of DEE87. Subsequently, it was proposed that Y32H is a gain of function with elevated kinase activity. We present a detailed functional evaluation of these dominant missense variants in several contexts. We use fly models of DEE87 in which endogenous cdk8, the fly ortholog to human CDK8 and CDK19, is depleted through RNA interference (RNAi) while expressing the human genes. Depletion of Drosophila cdk8 causes thicker muscle myofibrils, fused mitochondria, and climbing defects. The expression of wild-type human CDK19 in a fly cdk8 knockdown background rescues these defects, highlighting functional conservation. In our assays, we used a cdk8 depleted background and individually expressed either the variant or wildtype CDK19 to compare the function of the variants relative to wild-type. We demonstrate that Y32H can rescue defects caused by cdk8 depletion, while T196A is unable to due to possible loss of function. Further, we find that supplementation of the fly diet with an antioxidant improves T196A phenotypes. Our Drosophila studies allowed us to assay these variants for further insight into their functional nature and to obtain translational knowledge that may be applied back to human health.

Introduction

Decades of research have established Drosophila as a powerful tool for investigating fundamental biological processes and in understanding human disease mechanisms and functions of rare gene variants implicated in syndromes [1–3]. Many elegant and diverse assays can be used in flies to probe functional readouts of protein-coding gene products. This can be extended to assay patient-derived variants of such proteins that may be linked to a syndrome, which offers insights into potential therapeutic interventions for the patients that harbour them. Here we describe our work to better understand two missense patient variants in the human CDK19 protein.

Cyclin dependent kinase 19 (CDK19) and its paralog, Cyclin dependent kinase 8 (CDK8) are both classified as CDKs that regulate transcription [4, 5]. They can both bind to their partner cyclin, Cyclin C, and other mediator proteins to form conserved kinase modules. These modules can then interact with RNA polymerase II to modulate gene expression in a mutually exclusive way [4–7]. We recently showed that fly Cdk8 can regulate mitochondrial dynamics in fly muscle tissues by promoting activity of the fission regulator, Dynamin-related protein 1 (Drp1) [8]. Furthermore, we showed that depletion of fly cdk8 affects fly lifespan, locomotion, muscle myofibril and mitochondrial morphologies, and leads to elevated muscular reactive oxygen species (ROS) [8]. These effects can be rescued upon expression of wild-type human CDK19, highlighting functional conservation. Elevated expression or altered protein activity of either CDK19 or CDK8 has been linked to diverse ailments in humans, including cancer and developmental syndromes [9–16]. A syndrome that is linked to de novo heterozygous missense variants in CDK19 was termed Developmental Epilepsy and Encephalopathy-87 (DEE87) (OMIM 618916) [11–14]. Patients harbouring a single missense variant in CDK19 present with global developmental delay, intellectual disability and other neuromuscular deficiencies including ataxia and hypotonia [11–14]. In recent years, numerous patient variants in CDK19 have been characterized [11–14] and they generally cluster to two key regions of the protein, namely the adenosine triphosphate (ATP) binding region and the activation segment [17–19]. These are both located in the serine–threonine kinase domain, the functional and key domain of CDK19 [17–19].

Two de novo missense variants in human CDK19, Y32H and T196A, were the first to be characterized [12]. These protein variants are a result of a missense mutation in the CDK19 gene [12]. The variant Y32H contains a histidine in place of a conserved tyrosine at amino acid position 32 [12]. This residue is found to be part of the Gly-rich loop also known as the ATP binding pocket [17–19] and is implicated in the binding of ATP as part of kinase activity. The variant T196A contains an alanine in place of a conserved threonine at amino acid position 196 [12]. This residue is located within the activation segment and implicated in kinase activation, substrate binding and phosphorylation [17–19]. The clinical description of the patients harbouring these variants was accompanied by specific experiments in Drosophila to gain functional insight into the nature of these variants. The genetic strategy employed in this work consisted of using the GAL4-UAS system to express an RNAi construct to deplete fly cdk8 and express a complimentary DNA (cDNA) corresponding to either the wild-type CDK19 or a patient variant [12]. This was done to assess the effects of CDK19 patient variant expression in a cdk8 depleted background compared to the effect of wild-type CDK19 expression in that same background for a readout on variant function [12]. The authors tested numerous parameters including viability, lifespan, and sensitivity to mechanical stimuli [12]. A theme emerged in their work pointing towards a loss of function nature for both variants in that they failed to rescue defects caused by total loss or severe depletion of cdk8 as was seen with wild-type CDK19.

Another group of researchers dissected the nature of several de novo missense variants in CDK19 using a zebrafish model [11]. They employed a combination of genetic and biochemical strategies in their work to gain more insight into variant function [11]. Notably, they assessed kinase activity of two of these variants, Y32H and G28R, in vitro [11]. They assessed auto-phosphorylation and the phospho-status of known targets of CDK19 [11]. They determined that Y32H displayed elevated auto-phosphorylation and target phosphorylation compared to wild-type CDK19, leading them to assert that Y32H may be a gain of function variant [11]. They did not assess T196A in this assay; however, they found that G28R had reduced phosphorylation readouts compared to wild-type CDK19 [11]. From these findings they proposed G28R to be a loss of function variant. Further, they speculated that while affecting different locations on the protein, G28R and T196A converge on the same end result of being loss of function [11]. Further, in zebrafish phenotypic assays they found that both variants caused developmental defects, despite having opposite effects on kinase activity. In summary, they asserted that Y32H is a gain of function variant and that both gain of function and loss of function variants in CDK19 can lead to the same developmental syndrome, DEE87 [11]. This highlighted that functionally distinct variants can still converge on the same developmental syndrome, demonstrating a need for extensive and robust assays to obtain more information on the nature of these patient variants.

Given these contrasting views on the functional nature of CDK19 patient variants, we aimed to reconcile these insights and uncover further functional insights into these variants. In this current study, we assessed effects of these variants on viability, lifespan, muscle function and morphology with respect to wild-type CDK19 in a cdk8 depleted background. We previously described numerous defects due to depletion of Drosophila cdk8 in adult indirect flight muscles (IFMs), which served as valuable readouts for CDK/Cdk function [8]. To carry out these assays, we used a genetic background in which fly cdk8 was depleted using RNAi and the patient variant or wild-type CDK19 cDNA constructs were expressed. Under these experimental conditions, ubiquitous depletion of cdk8 yields reduced lifespan, climbing defects, and muscle and mitochondrial dysmorphology. These can all be phenotypically rescued by expression of wild-type CDK19. In addition, we show that expression of Y32H or T196A in this model results in dominant pathogenicity under more severe conditions. Indicating that both variants can lead to the same defects, similar to what is seen in the patients. However, under less severe conditions we observe the Y32H background shows no pathogenicity while the T196A background maintains its pathogenicity, causing severe lethality. These findings suggested that the experimental conditions at 25°C would allow us to dissect functional readouts of these patient variants in a less pathogenic context. We show that the cdk8 depletion climbing defect and muscle mitochondrial dysmorphology can be phenotypically rescued by expressing wild-type CDK19 and Y32H, but not T196A. This would be supportive of a functional Y32H variant and a loss of function T196A variant. Further, it was previously demonstrated that fly Cdk8 can phosphorylate Drp1 in vivo at Serine-616 (S616) to induce mitochondrial fission [8]. By extension of functional conservation, the same was proposed for CDK19. We find evidence that the phospho-status of Drp1 at S616 in adult muscle tissue may be elevated in Y32H flies and is significantly reduced in T196A flies. This would be supportive of a gain of function Y32H and loss of function T196A respectively. Further, we show that T196A climbing is mildly improved upon supplementation of the antioxidant drug NACA and there is no harmful effect to fly climbing ability with the supplementation of NACA. NACA can also robustly induce mitochondrial fission in all backgrounds indicating an ability to overcome the mitochondrial dysmorphology in the cdk8 depletion background that is normally sustained with T196A expression in that background. It also effectively reduces the muscular ROS in T196A adults to that of control fly muscle. The Y32H adult muscle did not harbour elevated ROS compared to control flies prior to NACA supplementation and demonstrated no effect after NACA addition. Our findings with NACA indicate a subtle uniqueness in backgrounds created by these patient variants under these experimental conditions in Drosophila, which could be subject to further research in future studies.

In summary, we present our Drosophila model to further dissect the nature of two de novo missense variants in CDK19. This model can be applied to study other CDK19 variants linked to DEE87 or to other variants implicated in various human syndromes. We provide evidence in numerous contexts that Y32H can functionally compensate for Cdk8 depletion in a manner similar to wild-type CDK19. Conversely, we demonstrate that T196A is unable to functionally compensate for Cdk8 depletion, indicative of a loss of function nature with respect to wild-type CDK19. Our findings raise the possibility of assessing NACA supplementation with other CDK19 variants in Drosophila models of DEE87. Further investigations of this nature may open the door to a potential variant-specific patient treatment regime in the future. Our studies further highlight the need for diverse functional analyses to dissect variant function in human syndromes.

Results

CDK19 variants Y32H and T196A localize to unique regions in the CDK19 protein possibly in close spatial proximity

The amino acid substitutions in CDK19 associated with DEE87 which we further characterize are found to localize to two distinct regions of the protein, namely the ATP binding domain and the activation segment (Fig. 1A). Numerous de novo missense CDK19 patient variants have been found in these domains [11–14]. Upon predictive protein modelling and annotation, these residues appear to be in close proximity with one another in the three-dimensional protein structure (Fig. 1A) [20, 21]. This would be a reasonable model since the region binding ATP needs to be near the activation segment to successfully phosphorylate substrates. Patients that harbour these variants are heterozygous because they have one wild-type allele and one variant allele of CDK19 (Fig. 1B). We have leveraged the GAL4-UAS system in Drosophila to model DEE87. In these models, an RNAi-based knockdown of endogenous fly cdk8 is combined with expression of either wild-type human CDK19 or a de novo missense variant under either 25°C or 29°C conditions (Fig. 1B). The purpose of this is to modulate the degree of RNAi-based depletion and variant construct expression, as GAL4 effects are often temperature sensitive [22]. A summary illustration (Fig. 1B) highlights the patient genetic environment with respect to CDK19 and fly models that can be used to assess functional readouts of these patient variants. We find that the wild-type and variant CDK19 protein constructs are expressed at comparable levels in adult tissues (Fig. S1A) and that these proteins are abundant in cytoplasmic and nuclear fractions, consistent with previous work [8, 12]. Given that these proteins are expressed at equivalent amounts, we can begin to make functional comparisons based on phenotypic readouts across numerous assessment methods.

Figure 1.

CDK19 variant localization on protein sequence and model. (A) Human CDK19 protein sequence cartoon highlighting the amino acid substitutions, p.Tyr32His and p.Thr196Ala, each resulting from a de novo missense mutation. AlphaFold2 protein model for human CDK19^WT (UniProt ID: Q9BWU1) with an enlarged field of view of the two amino acid locations: p.Y32 and p.T196. (B) Graphical representation of the model used in the paper leveraging the GAL4-UAS system for expression of an RNAi construct against fly cdk8 and cDNA construct for human CDK19^X (x = WT, Y32H, T196A) and the patient genetic environment for context.

Modulating ectopic expression of Y32H and T196A in a cdk8 knockdown background causes variable effects on fly eclosion and lifespan

It was previously shown that cdk8 homozygous mutant flies die at early larval stages of development [23]. This effect can be phenotypically rescued upon ubiquitous expression of either fly cdk8 or human CDK19 using actin-Gal4 [12]. In contrast, the patient variants Y32H and T196A were unable to rescue viability in this same background under those experimental conditions [12]. It was from these findings that both variants were proposed to be loss of function with respect to the wild-type CDK19 [12]. We find that using daughterless-Gal4 (da-Gal4) to perform RNAi-based ubiquitous knockdown of fly cdk8 at 29°C, reduces the percent of pupae that successfully eclose as viable adults compared to control (Fig. 2A). This eclosion defect is significantly rescued upon expression of wild-type CDK19 to a level not significantly different from control (Fig. 2A). Expression of either Y32H or T196A in this fly cdk8 depleted background significantly reduces the percent of pupae that eclose as adults compared to the wild-type CDK19 background (Fig. 2A). Under the 29°C growth conditions, the percent of pupae that eclosed in the Y32H and T196A backgrounds were approximately 3.2% and 0.5% respectively, significantly lower than the mean of 40% in cdk8 depletion alone (Fig. S2A). These findings showcase that both variants are unable to rescue the cdk8 depletion eclosion defect like the wild-type CDK19 can. Instead, both variants induce a dominant pathogenicity yielding a loss in the number of viable adults further than cdk8 depletion on its own. The mechanism of pathogenicity at this stage was not yet clear; however, these findings mirrored those of Chung et al. [12] and warranted adapted conditions before proceeding with further functional assays. The eclosion assay at 29°C was our sole experiment at this temperature.

Figure 2.

Expression of Y32H and T196A in a cdk8 knockdown background causes variable effects to fly eclosion and lifespan. Percent of pupae that successfully eclosed as adults when ubiquitously expressing the variants in a ubiquitous fly cdk8 knockdown background at (A) 29°C and (B) 25°C (total of nine vials per genotype from n = 3 biological replicates). Statistical analyses consisted of an ordinary one-way ANOVA followed by a Dunnett multiple comparisons test for both temperature conditions. Results are mean ± s.e.m. (^***P < 0.001, ^****P < 0.0001, n.s., no significance). (C) Average percent viability of adult male flies for the listed genotypes and (D) approximated day of 50% viability (n > 45 flies per genotype, n = 3 biological replicates). Statistical analyses are a log-rank score for survival trends (χ² = 18.98, P = 8.0e-04).

It has been documented that the GAL4 transcription factor activity is temperature sensitive [22, 24], thus raising flies at 25°C instead of 29°C would reduce the degree of GAL4 activity and subsequently reduce cdk8 depletion and CDK19 construct expression. We demonstrate the eclosion defect in the cdk8 depletion background corresponds to a mean residual cdk8 expression of approximately 20–25% compared to control, and all backgrounds at 29°C demonstrate comparable cdk8 depletion (Fig. S1D and E). We demonstrate that using the 25°C growth conditions and da-Gal4 abrogates the eclosion defect in the cdk8 knockdown background, resulting in a percent eclosion consistent with the white-RNAi control (Fig. 2B). These experimental conditions yield a mean residual cdk8 expression of approximately 30% compared to white-RNAi control flies (Fig. S1B) and all backgrounds showed a comparable residual cdk8 expression with respect to control (Fig. S1C). The residual cdk8 expression we demonstrate at 25°C is in contrast to the residual cdk8 expression of 15%–20% [8] or 5%–10% [12] used in previous studies. Our less severe cdk8 depletion couples with an abrogated lethality in this background.

We found that expression of Y32H under these conditions did not have a pathogenic effect on eclosion as was seen at 29°C (Fig. 2A and B), with no significant difference in the percent eclosion between the cdk8-RNAi and Y32H backgrounds (Fig. S2B). In contrast, expression of T196A maintains a dominant pathogenic effect in this background with approximately 12% of pupae eclosing, significantly fewer than the wild-type CDK19 and cdk8 depletion backgrounds (Fig. 2B, S2B). These findings under less severe cdk8 depletion demonstrate that Y32H expression does not affect eclosion while T196A expression does. This represented an area of divergence for these variants and warranted further study into the functional nature of the variants through other readouts. Furthermore, it illustrates that the variants are having an effect distinct from wild-type CDK19 which links them to their pathogenicity in human patients. With the non-lethal cdk8 depletion background at 25°C, it provides us with a background in which to investigate variant function more fully. Thus, all further work utilized the 25°C experimental conditions to investigate the functional nature of these two variants compared to the wild-type CDK19 when expressed in the cdk8 depletion background. Our initial characterizations mirror analyses performed by Chung et al. [12]. The key difference being that in their assays depletion of cdk8 alone caused 95% lethality with few escapers, while under our assay conditions at 25°C using da-Gal4, cdk8 depletion has no effect on eclosion as a readout for viability. In this context, we could more easily assess functional readouts on these patient variants.

We next assessed adult fly lifespan at 25°C as a readout for overall fly health by determining how long flies of each genotype lived and approximating the time at which 50% of flies had died and 50% still survived. This is commonly termed the day of 50% viability. We noted a severe reduction in the number of viable adults from the T196A background based on our previous findings and hypothesized that this background may experience a shortened lifespan. Under these assay conditions the respective days of 50% viability were 51 days for the white-RNAi control and 29 days for the cdk8-RNAi background (Fig. 2D). This indicates the notable drop in adult fly lifespan upon cdk8 depletion on its own. Expression of wild-type CDK19 in the cdk8 depleted background rescued lifespan with 50% viability climbing to around day 46, while the Y32H and T196A backgrounds reached 50% viability around day 27 and day 12 respectively (Fig. 2D). The T196A background showed a robust drop in viability compared to all other backgrounds while the Y32H background performed like the cdk8 depletion background and the wild-type CDK19 background performed relatively in line with the control (Fig. 2C and D). Upon carrying out a Kaplan–Meier statistical analysis (log-rank score) to obtain a chi-squared (χ²) value for these lifespan data [25], the χ² was 18.98 which equates to a p-value of 8e-04. This indicates a significant difference in the lifespan trends of these backgrounds (Fig. 2C). This also indicates that these two variants are having a unique effect on fly lifespan with respect to each other and the wild-type CDK19.

We demonstrate that Y32H expression, while having no effect on eclosion at 25°C, does affect fly lifespan compared to wild-type CDK19. However, T196A affects lifespan to a greater severity than Y32H and wild-type CDK19. We find here further evidence that these variants are functionally distinct relative to the wild-type and with respect to each other. We note that there are less viable adult flies from the T196A background compared to the other backgrounds. This severe reduction in the number of T196A progeny would imply that the viable flies that successfully eclose are escapers, representing the most weakly affected within this genotype. Of the T196A progenies that failed to develop or eclose, we observed distinct phenotypes including elongated L3 larvae, a deformed L3-pre-pupa in a pupal case, and pharate progenies (Fig. S2C).

Y32H can rescue fly cdk8 depletion climbing defects like wild-type CDK19 while T196A does not

In the lifespan assessment, we noted that the T196A genotype is affected earliest and significantly with respect to the Y32H or wild-type CDK19 backgrounds when comparing the day of 50% viability (Fig. 2C and D). Taking this into consideration, we wanted to choose an adult lifecycle timepoint that would allow us to consistently assess phenotypes without running into viability issues. Therefore, we selected flies at the three-day old adult stage for all further assessments. This is a well-used stage across Drosophila literature and also consistent with staging used to assess adult phenotypes caused by cdk8 depletion [8]. We aimed to follow up on the lifespan and viability readouts for variant function by assessing a more specific readout, neuromuscular health. We recalled that patients harbouring these dominant de novo missense variants in CDK19 present with neuromuscular defects [12] and that these phenotypes are conserved in flies with severely depleted functional cdk8 [8, 12]. We wanted to determine what effect the variants would have on the cdk8 depletion climbing defect compared to the wild-type CDK19. Thus, we aimed to assess adult fly climbing ability as illustrated (Fig. 3A) as a readout on variant function to determine if they can rescue the cdk8 depletion climbing defect as wild-type CDK19 can.

Figure 3.

Y32H rescues cdk8 knockdown climbing defects like wild-type CDK19, while T196A does not. (A) Graphical representation of fly climbing assessment for the listed genotypes. (B) Average percent of male flies that reach or pass the 5 cm target line from five assays across a total of n = 6 biological replicates for the indicated genotypes. The data from each genotype is a total of 30 climbing assays individually repeated six times (n = 6 biological replicates and n = 30 adult males per genotype). Statistical analysis used was an ordinary one-way ANOVA followed by a Dunnett multiple comparisons test. Results are shown as bar plots with mean height reached ± s.e.m. (^****P < 0.0001, n.s., no significance). (C) Violin plots showcasing climbing distribution on a per assay basis representing the height climbed by each fly for five climbing assays for all six independent biological replicates.

We assayed flies from each genotype to determine how many could reach or pass the 5 cm line in a given 15 second assay. We took an average from five independent climbing assays to obtain an average percent of flies that reach the max height as an endpoint assessment. Further, we plotted the per assay height reached by each adult fly across all assays as an overall climbing distribution. These data were obtained from five adult flies per genotype averaged from five replicate climbing assays with a total of n = 6 independent biological replicates. Under these experimental conditions, flies with depleted cdk8 present with significant climbing defects compared to control flies (Fig. S3B and C) and this is significantly rescued upon expression of wild-type CDK19 (Fig. 3B, S3D). Remarkably, expression of Y32H also significantly rescued the cdk8 depletion climbing defect (Fig. S3B and C) to a degree that is not significantly different from the wild-type CDK19 background (Fig. 3B, S3D). This illustrates that expression of wild-type CDK19 or Y32H can rescue the cdk8 depletion climbing defect in a comparable way. This indicates an ability of Y32H to functionally compensate for a reduction in fly Cdk8 under these conditions.

Further, we find that the escapers from the T196A background have a climbing ability that is not significantly different from the cdk8 depleted flies (Fig. S3B and C) and significantly reduced compared to the wild-type CDK19 background (Fig. 3B, S3D). These findings demonstrate that T196A is unable to functionally compensate for depletion of fly Cdk8 in the way that wild-type CDK19 can. We find evidence through this neuromuscular readout that Y32H is likely a functional variant while T196A is likely a reduced function or loss of function variant.

Given we assessed adult males in these climbing assays and the T196A background was severely lethal, our number of flies assessed is limited per biological replicate. To account for this, we increased our biological replicate count to ensure for more coverage. The pooled climbing data for males and females demonstrates the same degree of significance as we show here (Fig. S3C and D). This would account for a total of 60 adults assessed per genotype across five total climbing assays in six independent biological replicates for a total of 300 individual climbing assays performed (Fig. S3C and D).

Y32H and T196A flies demonstrate similar myofibril and distinct mitochondrial phenotypes that correlate to pDrp1^S616 levels as a readout for kinase function

We see that expression of Y32H can phenotypically compensate for reduced endogenous cdk8, similar to wild-type CDK19, while T196A cannot. With these insights in mind, we aimed to assess adult muscle phenotypes for a more specific readout on variant function. We investigated muscle fiber and mitochondrial morphologies in the flies expressing the variants. We wanted to determine the effect of the variants on cdk8 depletion muscle morphology defects compared to wild-type CDK19 in the cdk8 depleted background. Our rationale was that depletion of endogenous fly cdk8 has been linked to muscle fiber thickening and mitochondrial fusion/elongation [8]. These defects can be rescued upon ectopic expression of wild-type CDK19 [8], demonstrating functional conservation. However, no such observations have been made for these de novo missense patient variants of CDK19. We used Drosophila adult IFMs to assess myofibril thickness, sarcomere unit length, and mitochondrial morphology as a readout for variant function with respect to wild-type CDK19 under cdk8 depletion conditions. Adult IFMs are a well-studied and suitable muscle group to assess these characteristics and they are remarkably similar to vertebrate cardiac muscle in terms of structure and mode of power output [26, 27]. IFMs are also known for their use in assessing effects of protein variants on muscle cell function and morphology [26, 27], as we aimed to do here.

Under these conditions, ubiquitous cdk8 depletion yields significantly thicker muscle myofibrils compared to control flies (Fig. S4A). This myofibril thickening can be rescued upon expression of wild-type CDK19 to a myofibril thickness not significantly different from the white-RNAi control (Fig. 4A and B, Fig. S4A). These findings were also demonstrated with muscle-specific depletion of cdk8 [8]. Notably, both Y32H and T196A fail to rescue this phenotype as these backgrounds harbour myofibrils significantly thicker than the wild-type CDK19 background (Fig. 4B). These findings are not significantly different than the cdk8 depletion background on its own (Fig. S4A). This indicates a failure of either variant to rescue the cdk8 depletion myofibril thickening compared to the wild-type CDK19. We next assessed the sarcomere unit length and observed all backgrounds had consistent sarcomere unit lengths compared to the wild-type CDK19 background (Fig. 4C). We use the myofibril thickness and sarcomere unit length as proxies for muscle dimensions. A change in myofibril thickness but not sarcomere unit length indicates a net change to fiber thickness. These findings demonstrate that both variants are unable to rescue the cdk8 depletion muscle myofibril thickening as the wild-type CDK19 can, indicating a functional divergence in both variants with respect to the wild-type. It also demonstrates that under these conditions both variant backgrounds have the same resulting phenotype with regards to muscle myofibrils.

Figure 4.

Y32H and T196A flies show similar muscle myofibril thickness and variable mitochondrial phenotypes that correlate to pDrp1^S616 levels. (A) Rhodamine-phalloidin staining (F-actin) to visualize filamentous actin and indirect immunofluorescence of ATP5α (mitochondria) to visualize mitochondrial morphology in the indicated genotypes. Scale bar: 5 μm. (B) Quantification of adult fly indirect flight muscle myofibril width. (C) Quantification of adult fly indirect flight muscle sarcomere unit length. (D) Quantification of adult fly indirect flight muscle mitochondrial length. Quantification of muscle and mitochondria shown as box and whisper plots with solid line at the mean and whiskers for range with 15 muscle regions assessed per genotype. (E) Western blot showing the total level of phospho-Drp1^S616 and Drp1 to assess pDrp1^S616/Drp1 ratio in the listed genotypes from adult thorax tissue. Densitometry done in Fiji (ImageJ) to show (F) total Drp1 protein level and (G) phospho-status of Drp1 at position S616 both normalized to alpha-tubulin loading control. Results are presented as bar plots mean intensity ± s.e.m. all muscle tissue is from 3-day old adult males, raised at 25°C, with n = 5 unique muscle tissue per genotype repeated in n = 3 biological replicate gels. Statistical analyses are ordinary one-way ANOVA followed by a Dunnett multiple comparisons test. (^*P < 0.05, ^****P < 0.0001, n.s., no significance).

Alongside the myofibrils we assessed muscle mitochondria. We used mitochondrial length as a proxy for morphology as has been done previously when looking at effects of modulating cdk8 levels on adult IFM mitochondria [8]. The benefit to this is the mitochondria are constrained between myofibrils, making it easy to assess their morphology. Mitochondria that are longer would be considered fused/elongated while those that are shorter would be considered punctate. Under these experimental conditions, the cdk8 depletion background harbours significantly longer mitochondria than the white-RNAi control (Fig. S4B) representative of a more fused/elongated phenotype. This mitochondrial dysmorphology can be phenotypically and significantly rescued by expression of wild-type CDK19 (Fig. 4A, D). We find that expression of Y32H can also significantly reduce mitochondrial length compared to the cdk8 depletion background (Fig. S4B). The mitochondrial lengths in the Y32H and wild-type CDK19 backgrounds are not significantly different from one another (Fig. 4D). This indicates an ability of Y32H to rescue the cdk8 depletion mitochondrial dysmorphology to a similar degree as wild-type CDK19. Conversely, expression of T196A fails to rescue the mitochondrial dysmorphology as the mitochondrial lengths in the T196A background were not significantly different from the cdk8 depletion background (Fig. S4B). Also, the T196A mitochondria are significantly longer compared to the wild-type CDK19 background (Fig. 4D). These findings demonstrate that Y32H can functionally compensate for Cdk8 depletion in a comparable way to wild-type CDK19, while T196A is unable to do so.

While assessing mitochondrial morphology of the variant expressing backgrounds, we also observed that the Y32H background contained more distinct mitochondria from all fixed assessment regions in the muscle (Fig. S4C). This demonstrates that while harbouring a punctate morphology in adult IFMs as we just demonstrated, the amount of mitochondria assessed from this background is greater than all other backgrounds.

The kinase activity of Cdk8 is required for regulation of mitochondrial morphology through phosphorylation of Drp1 at the serine-616 residue (S616) [8]. Given the functional conservation between fly Cdk8 and human CDK19 [8, 12], we propose this function is conserved. To more precisely determine the functional nature of these patient variants in CDK19, we aimed to assess the phospho-status of Drp1 at S616 in adult IFM tissue lysates. For this assay, the adult thorax was chosen for its high muscle to non-muscle tissue ratio. We find that total levels of Drp1 protein relative to alpha-tubulin loading control are consistent across all backgrounds (Fig. 4E and F, Fig. S4D). Under these experimental conditions, cdk8 depletion alone does not lead to reduced phospho-Drp1 (pDrp1) as the pDrp1:Drp1 ratio is not significantly different from control (Fig. S4E) or the wild-type CDK19 background (Fig. 4G). The pDrp1:Drp1 ratio in the Y32H background trends higher, though not significantly. In contrast, the pDrp1:Drp1 ratio is significantly lower in the T196A background with respect to the cdk8 knockdown (Fig. S4E) and wild-type CDK19 (Fig. 4G) backgrounds. Under these conditions we demonstrate less severe cdk8 depletion is insufficient to reduce pDrp1 levels and expression of wild-type CDK19 in that background is insufficient to elevate them. Further, that expression of Y32H may elevate the pDrp1 level while T196A significantly reduces it. These findings demonstrate once again that these patient variants are functionally distinct from each other and in comparison to the wild-type CDK19. Applying the established relationship between CDK19/Cdk8, the phospho-status of Drp1, and mitochondrial morphology [8] to this result, these data provide in vivo evidence of a functional, possibly gain of function Y32H variant. There is further elucidation required to confirm this, a subject of future studies. In contrast, we find evidence for a loss of function T196A variant that may be acting in a dominant negative manner on endogenous fly Cdk8 to yield a significantly reduced pDrp1 level. This would be one possible explanation for the T196A findings. In any case, we reaffirm our assertion that Y32H is a functional variant and T196A is a loss of function variant.

T196A flies have elevated muscular ROS

We previously demonstrated that cdk8 depletion correlates with elevated ROS in neuronal and muscle tissue [8]. Given our functional insights and investigations thus far, we hypothesise that Y32H is a functional variant and T196A is a loss of function variant. We wanted to determine the effects of expressing these variants in a cdk8 depleted background on the level of ROS in adult muscle. Our rationale for this assessment was that a functional Y32H would compensate for depleted Cdk8 in fly tissues and likely result in reduced ROS, similar to wild-type CDK19. In contrast, a loss of function T196A would be unable to compensate for reduced Cdk8 and we would likely observe a sustained or elevated level of ROS. To do this, we can use dihydroethidium (DHE), a spectroprobe which reacts with superoxide radicals to produce a fluorescent signal that can be detected and quantified [28].

We observe under our experimental conditions, ubiquitous cdk8 depletion does not lead to elevated muscular ROS compared to the white-RNAi control (Fig. S5A). This is in contrast to elevated muscular ROS upon muscle specific cdk8 depletion shown previously [8]. We observe a tighter distribution in the DHE intensity in the Y32H background compared to cdk8 depletion (Fig. S5A) and wild-type CDK19 (Fig. 5A and B), though not significantly. We do find a significant increase in the mean DHE fluorescent intensity for the escaper T196A flies compared to the cdk8 depletion (Fig. S5A) and wild-type CDK19 (Fig. 5A and B) backgrounds. This finding demonstrates that under these conditions neither wild-type CDK19 nor Y32H affect muscular ROS in a ubiquitously depleted cdk8 background, while T196A elevates the ROS in this background. The mechanism by which ROS is elevated in the T196A background is not clear. Nevertheless, we demonstrate a subtle yet unique divergence in the phenotype created by these two patient variants.

Figure 5.

T196A adults display elevated muscular ROS that can be suppressed through NACA supplementation. (A) DHE staining to visualize ROS in adult indirect flight muscle (IFM) for the listed genotypes raised on standard food media (a-e) and NACA supplemented food media (a’-e’). Scale bar: 100 μm. (B) Quantification of mean DHE intensity from 15 fixed boxes from a total of nine adult male fly IFM from standard food and NACA food progenies. (C) Comparison of all standard food progenies to the T196A NACA food progenies. Quantifications shown as box and whisper plots with solid line at the mean and whiskers for range with n = 9 muscle regions assessed per genotype. All muscle tissue is from 3-day old adult males, raised at 25°C, with n = 9 muscle across three biological replicates. Statistical analyses shown are ordinary one-way ANOVA followed by a Dunnett multiple comparisons test. (^**P < 0.01, ^***P < 0.001, n.s., no significance). The letters a-e and a’-e’ represent the five genotypes raised without NACA and with NACA respectively.

Supplementation of the antioxidant NACA can reduce elevated ROS in T196A flies

Given our finding of elevated muscular ROS in T196A escaper adult muscle, we wanted to assess if this elevated ROS could be sequestered by supplementation of an antioxidant. Our rationale for this assessment is that antioxidant compounds have previously been used to suppress elevated ROS in flies expressing patient variants of other genes [29]. One such antioxidant compound is called N-acetylcysteine amide (NACA), which is a derivatized form of the more than half century old FDA approved drug, N-acetylcysteine (NAC) [30]. NACA has a more suitable half-life, improved tissue permeability, and can be converted into L-cysteine to eventually be integrated into reduced glutathione (GSH) production among other possible routes [30–33]. GSH is the chief antioxidant in biological organisms across genera [30–33], and we aimed to leverage this NACA/GSH relationship in our assay.

We set out to determine if NACA supplementation could suppress the elevated ROS in T196A adult muscle. To do this, we supplemented NACA into the fly medium at a concentration of 80 μg/ml and raised all flies on NACA supplemented food for the entirety of their lifecycle. This concentration of NACA was used based on a previously determined effective concentration in fly studies [29]. We observe that while there was no significant effect to other backgrounds, the escaper T196A adult muscle had significantly lower DHE intensity compared to their standard food medium raised counterparts (Fig. 5A and B). This is indicative of reduced muscular ROS in the T196A background. Notably, the ROS in these escaper T196A flies raised on NACA is not significantly different from the other backgrounds pre-NACA supplementation (Fig. 5C). Thus, we show that expression of T196A is contributing to elevated muscular ROS that NACA can suppress.

Supplementation of NACA indiscriminately induces mitochondrial fission in adult muscle tissue and couples with climbing improvement in T196A flies

It has been well established that mitochondria are the majority contributors to cellular ROS, up to almost 90% [34]. Further, it was previously demonstrated that more severe muscle specific depletion of cdk8 yielded mitochondrial dysmorphology and elevated muscular ROS [8]. Given this, and that NACA efficiently suppressed elevated muscular ROS in the T196A background, we aimed to assess mitochondrial morphology from adult muscle to determine what effect NACA has on mitochondria. Essentially, we wanted to assay if NACA could effectively overcome the mitochondrial dysmorphology upon cdk8 depletion that T196A expression fails to rescue. Our rationale for this assessment is based on preliminary evidence that compared white-RNAi and cdk8-RNAi IFMs from flies raised with or without NACA supplementation. This preliminary finding demonstrated that NACA could potentially induce robust mitochondrial fission in muscle, successfully overcoming the mitochondrial dysmorphology phenotype resulting from cdk8 depletion (Fig. S6A and B). To assess this, we raised these backgrounds on NACA supplemented food and assessed adult muscle myofibril and mitochondrial phenotypes as we have done previously.

Under these experimental conditions, there is no significant difference in the adult IFM myofibril width from adults raised with or without NACA supplementation (Fig. 6A and B). We do; however, find a significant reduction in mean mitochondrial length across all backgrounds upon NACA supplementation (Fig. 6A, C). This is indicative of a robust shift to a punctate mitochondrial morphology in adult muscle tissue. Notably, these findings confirm our preliminary findings with cdk8 depletion and NACA while offering new insight that NACA supplementation can overcome the sustained mitochondrial dysmorphology in the T196A background as well. This indicates that NACA can indiscriminately induce mitochondrial fission in all backgrounds regardless of the mitochondrial morphology and counteract the T196A failure to rescue the cdk8 depletion mitochondrial morphology.

Figure 6.

Dietary supplementation of the antioxidant drug, NACA, induces robust mitochondrial fission in adult muscle coupled with climbing improvement in T196A flies. (A) Rhodamine-phalloidin (F-actin) to visualize filamentous actin and indirect immunofluorescence of ATP5α (mitochondria) to visualize mitochondrial morphology in the indicated genotypes raised on NACA supplemented food. Scale bar: 5 μm. (B) Quantification of adult fly indirect flight muscle myofibril width comparing flies raised with or without NACA supplementation. (C) Quantification of adult fly indirect flight muscle mitochondrial length comparing flies with or without NACA supplementation. All muscle tissue is from 3-day old adults, raised at 25°C, with n = 5 muscle per genotype repeated in three biological replicates for a total of 15 unique muscle regions assessed. Data presented as box and whisper plots with solid line at the mean and whiskers for range. (D) Average percent of male flies that reach or pass the 5 cm target line from five assays across a total of n = 6 biological replicates for the indicated genotypes raised with or without NACA supplementation. The data from each genotype are a total of 30 climbing assays individually repeated six times (n = 6 biological replicates and n = 30 adult males per genotype). Statistical analysis used was an ordinary one-way ANOVA followed by a Dunnett multiple comparisons test. Results are shown as bar plots with mean height reached ± s.e.m. (E) Violin plots showcasing climbing distribution on a per assay basis representing the height climbed by each fly for five climbing assays for all six independent biological replicates raised with or without NACA supplementation. Statistical analyses consisted of one-way ANOVA tests followed by a Dunnett multiple comparisons test. (^****P < 0.0001, n.s., no significance). The letters a-e and a’-e’ represent the five genotypes raised without NACA and with NACA respectively.

Following these findings of NACA on the mitochondrial phenotypes in adult muscle, we wanted to determine overall neuromuscular health of these flies after NACA supplementation. We used the climbing assessment to compare flies that were raised with or without NACA supplementation. Our rationale for this assessment is that the poorest performing adult flies in the climbing assessments harbour elongated/fused mitochondria. We have now demonstrated that this can be rescued upon NACA supplementation, and so we aimed to reconcile this with a neuromuscular assay. Using the adult IFMs as a readout for muscular health, we propose that muscle defects brought on by cdk8 depletion that T196A fails to rescue may be contributing to a locomotion defect. The climbing abilities of these flies were assessed and the white-RNAi, wild-type CDK19, and Y32H backgrounds performed comparably with or without NACA supplementation (Fig. 6D and E). These backgrounds harboured no climbing deficiency to begin with meaning at the very least NACA supplementation at this dose does not have a harmful effect on fly locomotion. The cdk8-RNAi and T196A backgrounds show a trend higher in locomotive ability when assessing endpoint height reached, though not significant (Fig. 6D). On a per assay basis; however, the cdk8-RNAi and T196A backgrounds both show overall groupwise improvement in climbing ability (Fig. 6E). These findings demonstrate that NACA can indiscriminately induce mitochondrial fission in adult muscle tissue. Further, while having no detrimental effect to Y32H flies, NACA supplementation can yield modest improvement in T196A escaper fly phenotypes. This indicates another branching point for these variants in terms of their nature with respect to each other and with respect to the wild-type CDK19.

Discussion

With the identification of many novel missense variants, some of unknown significance or contrasting functional effects, it is important to have robust and diverse assays to study the impacts of these missense variants. Simple tests of pathogenicity do not always reveal underlying activities, depending on experimental conditions. So, while pathogenicity can be established by certain assays, a more in-depth investigation is often needed to identify the functional nature of such variants to fine tune potential therapeutic measures to avoid ineffective treatments.

Upon initial discovery and characterization of the dominant patient variants in CDK19, CDK19^Y32H (Y32H) and CDK19^T196A (T196A), a set of Drosophila experiments were carried out to establish pathogenicity and functionality [12]. These findings demonstrated that Y32H and T196A were unable to rescue lethality caused by cdk8 deficiency or aberrant phenotypes caused by severe cdk8 depletion in the way that wild-type CDK19 could [12]. The prevailing hypothesis based on these results was that both of these de novo patient variants are loss of function and in human patients may be acting in a dominant negative manner [12]. Following this work, another group independently assessed Y32H and another variant, G28R, using zebrafish as a model system. Based on in vitro kinase assays, they proposed that Y32H may be a gain of function variant (hypermorph) and speculated that while G28R and T196A affect distinct regions of CDK19, they may both result in a loss of function [11]. Both research groups utilized distinct genetic and/or biochemical strategies under different conditions and model systems, creating a broad yet contrasting image on the functional impacts of these variants. In our work, we aimed to reconcile the current understanding on the nature of these patient variants by adapting the Drosophila model system previously used. We did so by reducing the severity of the cdk8 depletion conditions used previously to assess viable adult progeny for various readouts to further characterize the effects of these variants. Understanding the effect of a missense variant on the biochemical nature of the protein can inform us of possible deleterious effects at the cellular level, and reveal insights into their clinical features. Of note, our studies confirm that both variants can exert a dominant and pathogenic effect and that if indeed Y32H is a gain of function and T196A is a loss of function, then both can result in similar phenotypes. A summary of the effect the variants and wild-type CDK19 have on the cdk8 depletion phenotypes can be found in Table 1. This encapsulates the ability of the variant to rescue the cdk8 depletion defect as the wild-type CDK19 can, fail to rescue the defect, or produce no effect when compared to the cdk8 depletion background.

Table 1.

Summary of the assessed functional readouts for wild-type CDK19, Y32H, and T196A compared to the phenotypes present upon Cdk8 depletion.

Genotype	Eclosion at 29°C	Eclosion at 25°C	Lifespan	Climbing Ability	Myofibril Morphology	Mitochondrial Morphology
da > Cdk8-RNAi	Reduced	No effect	Reduced	Reduced	Thick	Fused/elongated
da > CDK19WT; Cdk8-RNAi	Rescued	No effect	Rescued	Rescued	Rescued (Thin)	Rescued (Punctate)
da > CDK19Y32H; Cdk8-RNAi	Further reduced	No effect	Failed to rescue	Rescued	Failed to rescue	Rescued (Punctate)
da > CDK19T196A; Cdk8-RNAi	Further reduced	Reduced	Further reduced	Failed to rescue	Failed to rescue	Failed to rescue

Summary of the working understanding of CDK19 patient variant functionality in the context of its ability to rescue a defect caused by cdk8 depletion. The first row corresponds to the cdk8-RNAi background and the effect of cdk8 depletion on each corresponding assessment or assayed parameter compared to a white-RNAi control background. For the three backgrounds containing a CDK19 construct alongside cdk8 depletion there is the resulting effect of CDK19 expression on the cdk8 depletion phenotype. The effect of the wild-type CDK19 or a patient variant on the cdk8 depletion phenotype was scored as no effect, failed to rescue, or rescued.

It has been established that patients harbouring one missense allele of CDK19 are diagnosed with DEE87 syndrome [11–14]. Further, many of the phenotypes including developmental delay, epilepsy, and effects to broader brain development are shared despite harbouring variants that affect different regions of CDK19 [11–14]. The work of Zarate et al. [11], pushed our understanding on the nature of these variants and demonstrated for the first time that gain of function and loss of function variants in CDK19 can both act in a dominantly pathogenic manner. While we demonstrate a similar finding in our eclosion assessment at 29°C, it is only through reducing the severity of the experimental conditions do we ascertain functional nuances in these variants. This underscores the reason behind the numerous functional assays we carried out at 25°C. It allowed us to obtain viable adult progenies that could provide insight into distinct functional natures of Y32H and T196A based on the ability of these variants to rescue effects of Cdk8 depletion as the wild-type CDK19 can.

The eclosion assessment at 29°C using the da-Gal4 driver was done to orient us to the experimental conditions we created and determine our ability to broadly reproduce the findings of Chung et al. [12]. We understood that we could leverage the temperature sensitive activity of GAL4 to modulate the degree of cdk8 depletion and the CDK19 transgenic expression to reduce the severity of the experimental conditions. The 29°C temperature conditions are often used in fly research and strikes the balance between maximal GAL4 activity and Drosophila health span [22, 35]. It has been demonstrated previously that maintaining adult flies at 29°C affects healthy lifespan after around day 20 [36]. Since we are simply looking at successful eclosion as a readout for viability and not adult lifespan at 29°C, these effects are minimized. Nevertheless, we can compare the viability of our white-RNAi control at 29°C and 25°C to ensure the temperature-related effects are minimized. We demonstrate a robust and comparable degree of adults eclosed between the two temperatures of this control background, allowing us to proceed at 25°C ensuring temperature-related effects are minimized.

Proceeding to our further assays, we find under the cdk8 depletion conditions at 25°C that Y32H can compensate for reduced fly cdk8 across numerous phenotypic measures like wild-type CDK19 can, while T196A cannot. We demonstrate that Y32H expression is not pathogenic to the fly host, but T196A expression is, and there is a significant reduction in the number of viable T196A progeny. These limited viable adult progenies are escapers of this genotype, likely harbouring some variability that allows them to undergo development and eclose as adults. We reinforce that both variants when expressed under more severe conditions can exert the same dominant pathogenicity; however, upon adjusting the severity of those conditions we can tease apart functional nuance of these two variants. The expression of T196A has a greater effect on adult fly lifespan compared to Y32H and further the Y32H variant demonstrates the ability to compensate for depleted fly cdk8 in the same way wild-type CDK19 can in the adult fly locomotion assays. The theme emerging from these findings is variants with distinct functional natures can still be implicated in the same developmental syndrome. Also, we suggest that understanding patient variants more generally will require more analyses than scoring lethality or survivorship to dissect their functional nature.

We used adult IFM tissue as a readout to assess if the patient variants could rescue muscle-related defects caused by cdk8 depletion in the same way wild-type CDK19 can, as has previously been established [8]. We show that both variant backgrounds sustain the muscle myofibril thickening induced upon cdk8 depletion, while wild-type CDK19 rescues this thickening. Indicating that both variants are unable to rescue this phenotype in the same way the wild-type can. The nuance in the nature of these variants returns when we examine muscle mitochondrial morphology. The depletion of cdk8 leads to muscle mitochondrial dysmorphology, specifically, a more fused/elongated morphology in adult muscle. The expression of wild-type CDK19 rescues this effect and Y32H does as well. The mitochondrial lengths, as a proxy for morphology, are not significantly different in the white-RNAi, wild-type CDK19, and Y32H backgrounds. The T196A background demonstrates sustained mitochondrial fusion/elongation and is not significantly different from cdk8 depletion alone. Indicating the inability of T196A to compensate for reduced Cdk8 in adult fly muscle tissue as wild-type CDK19 and Y32H can. This provided further insight that these variants are likely functionally distinct and implicated Y32H as a functional variant and T196A as a loss of function variant.

A relationship between Cdk8/CDK19, Drp1 phospho-status, and mitochondrial morphology has previously been established implicating Cdk8 in the phosphorylation of Drp1 at S616 [8]. Through conserved functional nature, this same activity was proposed for CDK19. We demonstrate that under our conditions there is evidence for elevated pDrp1 in the Y32H background and significantly reduced pDrp1 in the T196A background compared to the cdk8 depletion and wild-type CDK19 backgrounds. For the Y32H findings, at the least, we find evidence of a functional variant and speculate about a gain of function (hypermorphic) nature. Further testing would be needed to determine the significance of this finding, a subject of future studies. Conversely, the T196A findings illustrate a loss of function nature at the least but offer some room for speculation as well. If a kinase is loss of function and is expressed, as we find under our conditions, then it may be able to bind targets and prevent productive interactions of the endogenous wild-type with those targets. This would be the nature of a dominant negative allele and if T196A is affecting the ability of fly Cdk8 to phosphorylate Drp1, then we might observe a reduction in the pDrp1 readout. The broader theme once again is these variants are functionally distinct with respect to each other and the wild-type CDK19.

Another distinct phenotype between these two variant expressing fly backgrounds is elevated muscular ROS in T196A adults. We demonstrate under our experimental conditions that ubiquitous cdk8 depletion alone is not leading to elevated muscular ROS, but expression of T196A in this background is. We supplemented an antioxidant drug, NACA, into the fly food media to determine if this elevated ROS could be effectively suppressed. In fact, NACA suppresses the elevated muscular ROS in T196A adults such that the readout for ROS is not significantly different from white-RNAi control flies. Alternatively, Y32H expression did not affect muscular ROS compared to cdk8 depletion on its own. To current knowledge, patient tissue samples have not been known to have undergone testing for levels of ROS. Our work provides insight into a novel direction in the characterization of patients diagnosed with DEE87 and might allow for researchers to tease apart subtle but present differences in human patient variants in CDK19.

Further, the supplementation of NACA can indiscriminately induce mitochondrial fission in adult IFM tissue from all backgrounds we examined. There is a general correlation between the punctate mitochondrial morphology and overall fly health and locomotion. Our findings with NACA and climbing illustrate a mild assay to assay climbing improvement in T196A adults and demonstrate that NACA does not affect the Y32H background at this dosage. We note that while the differences in climbing ability are mild in T196A with or without NACA supplementation, it is a direction that could be pursued further in future studies possibly assaying varying concentrations of NACA as well. Our Drosophila model allows us to reduce the severity of the cdk8 depletion to create a viable, yet sensitized, background to assay the effects of these variants by assessing various phenotypic readouts. It also allows for the ability to tease apart unique phenotypes as a result of the distinct variant nature we provide evidence for. This Drosophila model may also, in the future, be adapted to test effects of other patient variants in CDK19, its paralog CDK8, or further on to other syndrome associated patient variants where applicable.

In summary, our work reveals further evidence for a functional Y32H and loss of function T196A. We summarize these findings to encapsulate the working understanding on the nature of these two patient variants in Table 1. We present one particular finding that suggests Y32H could be a gain of function, pending further work in future studies. Our work also reinforces that both gain of function and loss of function missense variants can lead to same syndrome in human patients. It is possible that both gain of function and loss of function variants can interfere with development, normal cellular processes, and activity of the wild-type allele to produce such effects in human patients. This is what we see where the global developmental delay, intellectual disability, and epilepsy phenotypes are conserved in nearly 100% of all patients that have been characterized thus far [11–14]. Further, there is great potential in assessing the nature of these and other patient variants using Drosophila models and that further testing could unlock new possibilities in therapeutic approaches. One such goal could be to prescribe therapies to human patients on a variant-specific basis. We demonstrate that Y32H and T196A under our experimental conditions can create subtle but unique backgrounds, by way of their functional natures, that may be targeted. These methods may be applied to research into personalized treatment regimens for DEE87 patients harbouring variants in CDK19 or other genes linked to various syndromes that affect human health.

Materials and methods

Protein sequence alignment and modelling

To produce a predictive protein model for CDK19, AlphaFold2.0 Protein Structure database was used [20, 21]. The PDB file corresponding to the predictive model of CDK19 was input to PyMOL: The PyMOL Molecular Graphics System, Version 1.20 Schrödinger, LLC, and the model was annotated at position 32 and position 196 corresponding to Y and T respectively on the wild-type CDK19 protein.

Drosophila culture and collection

Drosophila melanogaster flies were raised and maintained on a standard cornmeal-molasses medium in accordance with the recipe outlined by the Bloomington Drosophila Stock Center. Standard food media and 80 μg/ml NACA supplemented media were used as indicated. Fly stock lines were maintained at room temperature, and crosses were carried out at 25°C and 29°C as indicated. The following fly strains were used: ;;da-Gal4 (derived from BL #55850), ;;UAS-Cdk8-RNAi (BL #35324), ;;UAS-white-RNAi (BL #33762). The stocks with a BL number are sourced from Bloomington Drosophila Stock Center (Bloomington, IN, USA). The following stock lines:

;UAS-CDK19^WT; UAS-cdk8-RNAi

;UAS-CDK19^Y32H; UAS-cdk8-RNAi

;UAS-CDK19^T196A; UAS-cdk8-RNAi

were generated in a previous study [12] and provided by Dr Hyunglok Chung (Houston Methodist, Texas, USA). Adult flies were anesthetized using CO_{2 (g)} and collected in polystyrene vials for use when and where appropriate.

Eclosion assessment

Progenies raised at 25°C and 29°C were counted at the pupal stage and allowed to eclose to determine the number of viable adults per genotype. The cross for each genotype was carried through three technical replicate vials allowing a two-day egg lay per vial and repeated for a total of three biological replicates. The total number of pupae and adults from nine vials were determined and quantified to express as percent of pupae that eclose as adults, as a readout for viability.

Lifespan assessment

Sex separated adult flies were maintained at 25°C to assess lifespan. Adult male lifespan data are shown in Fig. 2. These flies were transferred to fresh food every third day and survivorship was scored daily from Monday to Friday, quantified, and expressed as average percent viable per day. Each data point represents the average percent viability from three biological replicates representing n > 45 adult flies. All genotypes were scored until death. The day of 50% viability was also approximated from the graph as indicated by a horizontal black line. A log-rank score was carried out on these data to assess statistical significance of survival trends and to retrieve a chi-squared (χ²) value corresponding to a p-value for significance.

Climbing assessment

Sex separated 3-day old adult flies raised at 25°C were subjected to negative geotaxis assays. Adult male climbing data are shown in Fig. 3. To assess climbing, five flies of each genotype were placed in empty polystyrene vials with another empty vial taped to the top to form a closed cylinder. The bottom vial was marked with dashes at 1 cm increments ranging from 0 cm to 5 cm. The vials containing the flies were gently and simultaneously tapped down and a timer was set for 15 seconds to score the height reached by the flies at the end of the time course. This constituted one assay. A total of five assays were done and the average percent of flies that reached or passed the 5 cm target line from five assays was plotted per genotype. This was then repeated for a total of n = 6 independent biological replicates to comprise a total of 30 assays per genotype per sex. In total, including both sexes, 60 adult flies were assessed per genotype. The individual height reached by each fly on a per assay basis was also plotted for overall genotype climbing distribution. These assays were recorded with a Pro 12 MP Digital Camera.

RNA extraction and qRT-PCR for gene expression quantification

Drosophila tissues were cleaned in ice-cold phosphate-buffered saline (PBS), vortexed to homogenize, and put through a Qiagen RNA extraction protocol using RNeasy Mini Kits (Qiagen 74 134). RNA was converted to cDNA using the OneScript Plus cDNA Synthesis Kit (ABM G236) for subsequent transcript level quantification. qRT-PCR was performed using PowerTrack SYBR Master Mix (Applied Biosystems—ThermoFisher—Lot # 29999–452) on a StepOne Real-time PCR System (Applied Biosystems). Primers used:

rp49 F AGCATACAGGCCCAAGATCG

rp49 R TGTTGTCGATACCCTTGGGC

18S F CTGAGAAACGGCTACCACATC

18S R ACCAGACTTGCCCTCCAAT

cdk8 F CATCCGGGTGTTTCTGTCG

cdk8 R CAGCCCGATGGAACTTAATGAT

To quantify expression level of the target genes of interest, the 2^-ΔΔCt method was used. The calculations were done using rp49 as the internal reference gene to quantify relative expression level. Expression data were input to GraphPad (Prism 10) to produce bar graphs ± s.e.m.

Immunofluorescence staining of adult fly muscle

Adult fly thoraces were separated in ice-cold PBS and fixed in 4% paraformaldehyde (PFA) for 15 minutes at room temperature with rocking. Samples were washed twice for 10 minutes in PBST (0.1% Triton X-100). Tissues were blocked in 5% bovine serum albumin (BSA) in PBST for 1 h at room temperature and then incubated with primary antibody in block solution overnight at 4°C with rocking. The following primary antibody was used: mouse anti-ATP5α (1:500, Abcam ab14748). Following this, tissues were washed with PBST and incubated with the appropriate secondary antibody: FITC conjugated anti-mouse (1:500, Jackson ImmunoResearch Laboratories) in combination with Rhodamine-Phalloidin prepared in methanol (1:500, ThermoFisher R415) for two hours at room temperature with rocking. Samples were mounted on 1 mm spaced glass bridge slides in VectaShield Antifade Mounting Medium (Vector Laboratories, BioLynx) and imaged using a Zeiss LSM880 Airyscan Confocal microscope and processed using Fiji (ImageJ).

Western blotting of adult thorax lysates

Adult thoraces were lysed mechanically in 1x Cell Lysis Buffer (Cell Signaling Technologies, #9803), supplemented with 1x Protease Inhibitors (Roche), and 1 mM phenylmethylsulfonyl fluoride. Protein lysates were prepared in 1x sodium dodecyl sulfate (SDS) sample buffer and resolved using polyacrylamide gel electrophoresis (PAGE). Samples were then transferred to nitrocellulose membranes. and blocked with 5% BSA made up in 1x TBST (Tween-20). The following primary antibodies were used: rabbit anti-Drp1 (1:5000; Cell Signaling D6C7), rabbit anti-pDrp1S616 (1:5000; Cell Signaling 4494), and mouse anti-alpha-tubulin (1:5000; DSHB AA4.3). HRP conjugated secondary antibodies (Jackson ImmunoResearch) were used for chemiluminescent imaging using Clarity Western ECL substrate reagent (Bio-Rad, #170–5061). Images were acquired by Amersham Imager 600 (GE) and densitometry was done in Fiji (ImageJ) to quantify phospho-status of Drp1 with respect to total Drp1, normalized to alpha-tubulin loading control.

Quantification of muscle and mitochondrial phenotypes

Adult fly indirect flight muscle images were input to ImageJ and a box of a fixed size (12.5 μm²) was placed on the image to encompass a region of muscle myofibrils and mitochondria. A total of five unique thorax samples were imaged for each genotype repeated for a total of three independent biological replicates, representing a total of 15 unique muscle regions assessed. Myofibril widths, sarcomere unit lengths, and mitochondrial lengths were determined manually using Fiji (ImageJ). Data collected from adult muscle were pooled for box and whisper plot generation using GraphPad (Prism 10).

ROS staining and quantification

Briefly, adult thoraces were isolated in ice-cold PBS and incubated in 30 mM DHE (Cayman Chemical—#12013) for 30 minutes in the dark at room temperature with gentle shaking. Following DHE incubation, tissues were incubated in 8% PFA and washed twice for 5 minutes before mounting on 1 mm spaced glass bridge slides in VectaShield Antifade Mounting Medium (Vector Laboratories, BioLynx). Images were taken within two days using a Zeiss LSM880 Airyscan Confocal microscope. Fluorescent intensity was measured using Fiji (ImageJ). Data collected from adult muscle were pooled for box and whisper plot generation using GraphPad (Prism 10).

Statistical analyses

We used GraphPad Prism for statistical analysis and generation of figures. Statistical analyses were done with the default settings of the software (* indicates P < 0.05, ** indicates P < 0.01, *** indicates P < 0.001, **** indicates P < 0.0001). Violin plots, bar graphs, and box and whisker plots were made using GraphPad Prism 10. All statistical analyses carried out were one-way ANOVA followed by a Dunnett multiple comparisons test unless otherwise indicated.

Web resources

Clustal Omega Alignment Software: https://www.ebi.ac.uk/jdispatcher/msa/clustalo.

Conflict of interest statement: Authors declare no competing financial interests.