Parent-of-origin effects on nuclear chromatin organization and behavior in a Drosophila model for Williams–Beuren Syndrome

Prognosis of neuropsychiatric disorders in progeny requires consideration of individual (1) parent-of-origin effects (POEs) relying on (2) the nerve cell nuclear 3D chromatin architecture and (3) impact of parent-specific miRNAs. Additionally, the shaping of cognitive phenotypes in parents depends on both learning acquisition and forgetting, or memory erasure. These processes are independent and controlled by different signal cascades: the first is cAMPdependent, the second relies on actin remodeling by small GTPase Rac1 – LIMK1 (LIM-kinase 1). Simple experimental model systems such as Drosophila help probe the causes and consequences leading to human neurocognitive pathologies. Recently, we have developed a Drosophila model for Williams–Beuren Syndrome (WBS): a mutant agnts3 of the agnostic locus (X:11AB) harboring the dlimk1 gene. The agnts3 mutation drastically increases the frequency of ectopic contacts (FEC) in specific regions of intercalary heterochromatin, suppresses learning/memory and affects locomotion. As is shown in this study, the polytene X chromosome bands in reciprocal hybrids between agnts3 and the wild type strain Berlin are heterogeneous in modes of FEC regulation depending either on maternal or paternal gene origin. Bioinformatic analysis reveals that FEC between X:11AB and the other X chromosome bands correlates with the occurrence of short (~30 bp) identical DNA fragments partly homologous to Drosophila 372-bp satellite DNA repeat. Although learning acquisition in a conditioned courtship suppression paradigm is similar in hybrids, the middle-term memory formation shows patroclinic inheritance. Seemingly, this depends on changes in miR-974 expression. Several parameters of locomotion demonstrate heterosis. Our data indicate that the agnts3 locus is capable of trans-regulating gene activity via POEs on the chromatin nuclear organization, thereby affecting behavior.


Introduction
Genome plasticity is ensured by the architecture of specific nuc lear loci and nuclear localization of transcriptional ma chinery (MedranoFernández, Barco, 2016;Iourov et al., 2019) where the chromatin organization is the prioritydriven factor (Ito et al., 2014;MedranoFernández, Barco, 2016;Li et al., 2018). The outcome of recent achievements in systems biology is the notion that the plasticity of 3D chromatin architecture of nervous cell nuclei plays the leading role in cognition and neuropsychiatric disorders (MedranoFernández, Barco, 2016;Kim et al., 2018;Iourov et al., 2019). The epigenetic component is still an underestimated source of psychomotor disturbances and neuronal diversity (SavvateevaPopova et al., 2017). Therefore, a new field of human biomedical research named molecular cytogenetics and cytogenomics (Iourov et al., 2008), or chromosomics (Liehr, 2019) has evolved. The main goal of chromosomics is the study of chromosomes, their 3D architecture in the interphase nucleus, the outcomes of chromosomal subregion plasticity and gene interactions for shaping interindividual and intercellular genomic variations in normal behavior and disease.
Recently, this topical problem has turned out to be the pur suit of understanding the concomitant role of active forgetting, since the antithesis to learning acquisition is the forgetting or memory erasure (Davis, Zhong, 2017). Both processes are independent and controlled by different signal cascades: learning acquisition and memory consolidation occurs via cAMP cascade, its components being CREB and C/EBP. Active forgetting relies on actin remodeling cascade responsible for structural alterations of neurons and synapses: small GTPase Rac1 -LIMK1 (the key enzyme of actin re modeling LIMkinase 1) and its phosphorylation substrate сofilin. The absence of Rac1-dependent forgetting causes the autistic spectrum disorders. Expression changes (active or non-active state) of LIMK1 and cofilin lead to different neuro pathologies. The most studied example embracing all the aforementioned facets of manifestations is Microdeletion (Deletion) Williams-Beuren Syndrome, or WBS in 7q11.23. WBS deletion leads to cardiovascular pathology, cognitive deficit in visuospatial construction and hypersociability (KaiserRogers, Rao, 2005;Nikitina et al., 2014c). This is because longterm synaptic plasticity and depression deter mine successfulness of learning and memory, as well as of locomotor behavior, and depend on epigenetic regulation of LIMkinase 1 (LIMK1) gene, one of approximately 28 genes uncovered by WBS deletion. Epigenetic regulation of LIMK1 activity involves DNA methylation, chromatin remodeling and the noncoding RNAmediated process (Smrt, Zhao, 2010).
LIMK1, a member of serine/threonine (Ser/Thr) family kinases regulated by the RhoGTPase pathway, is the key enzyme of actin remodeling cascade. Dendritic spines are actinrich structures, and spine dynamics is driven mainly by actin remodeling, thus sharing several molecular pathways with dendrite growth (Smrt, Zhao, 2010). Increasing sets of evidence suggest that nuclear actin also plays a pivotal role in transcriptional regulation and DNA repair. Interestingly, monomeric actin is a stoichiometric subunit of a variety of chromatin remodeling complexes. A shift between monomeric and polymeric states modifies activity of histone deacetylases (KlagesMundt et al., 2018).
Recently, we have developed a simple and appropriate Dro sophila model for chromosomics (SavvateevaPopova et al., 2017) using the properties of the agnostic locus har boring LIMK1 gene (X:11AB). This region possesses the properties of intercalary heterochromatin. Being a hotspot of chromosome breaks, ectopic contacts, underreplication and recombination, the region attains strain-specific architecture marked by single base changes and small insertion/dele tions. The EMSinduced temperaturesensitive (ts) mutation agn ts3 carries the insertion of transposable element (TE) from Tc1/ mariner superfamily ~460 bp downstream 3′UTR of Drosophila LIMK1 gene (dlimk1), as well as A/Trich 28 bp insertion within intron 1 of dlimk1 capable of pairing with 5′ TIR of the TE.
When maintained at 29 °C, agn ts3 shows a temperature sensitive lethality at all stages of development except for the imaginal stage. At normal temperature, the adult flies show drastic learning acquisition and memory retention defects, as well as locomotor impairments and amyloidlike inclusions (Nikitina et al., 2014b;Kaminskaya et al., 2015). Stress exposure (heat shock for 30 min at 37 °C) suppresses these manifestations (Nikitina et al., 2012(Nikitina et al., , 2014a. Also, agn ts3 mu tation leads to: (1) LIMK1 and p-cofilin increase in the adult brain and salivary glands of 3rd instar larvae at 22-25 °C and a fall down to the level of the wild type strain  (2) high level of tsinduced recombination within agn ts3 region; (3) 3fold increase in frequency of nonallelic ectopic contacts (FEC) within 2L arm of the chromosome 2 Parent-of-origin effects on nuclear chromatin organization and in the 11В X chromosome region (Medvedeva et al., 2010). Additionally, miRNAs expression including the bio markers for human neuropathologies is drastically reduced in agn ts3 relative to the wild type strains (SavvateevaPopova et al., 2017). agn ts3 -specific nuclear organization is shaped in early embryogenesis alongside with formation of chromosomal hetero chromatin regions. Intrinsic agn ts3 FEC is maternally inherited (Medvedeva et al., 2010). Therefore, agn ts3 is a promising model for studies on parent-of-origin effects (POEs) on prog eny considered as significant causative factors of psychiatric disorders (Zayats et al., 2015). For instance, a 1.5 Mb WBS deletion recurrently arises de novo and depends on POEs: ma ternal origin leads to more severe developmental abnormali ties and microcephaly (Pérez Jurado et al., 1996). Moreover, when WBS deletion has a paternal origin, expression levels of a number of genes within the WBS deletion decrease. Among these genes crucial for the brain development is a gene for ge neral transcription factor III (GTF2I). It regulates transcrip tion by binding to DNA and histone deacetylase (HDAC) (Col lette et al., 2009). The main goal of the study is the analysis of POEs role in shaping quantitative traits, namely learning acquisition, memory retention, locomotion, and miRNAs ex pression while using the advantages of the Drosophila model for POEs in progeny from reciprocal crosses between agn ts3 and the wild type strain Berlin (Fig. 1).
To meet the requirements of chromosomics, FECs between the region X:11AB and the other bands of the X chromosome may be estimated as an indicator of chromosomal spatial organization. This approach is justified by the existence of latereplicating genomic territories including the underrepli cated regions of polytene chromosomes. These regions over lap with latereplicating regions of mitotically dividing cells (Belyakin et al., 2005). Suppression of SUUR gene responsible for underreplication of intercalary heterochromatin and, as a consequence, for the ectopic pairing leads to death in early em bryogenesis (Belyaeva et al., 1998). Additionally, to elucidate the contribution of DNA sequence homology as components of epigenetic regulation in the ectopic chromatin pairing we have developed the special software package Homology Seg ment Analysis.

Materials and methods
Drosophila stocks. The fly stocks used belong to Biocollec tion of Pavlov Institute of Physiology of the Russian Aca de my of Sciences: • Berlin, a wild type strain; • Canton S, a wild type strain; • agn ts3 , a temperaturesensitive mutation on Canton S genetic background within agnostic locus (X:11AB) af fecting dlimk1 activity. The reciprocal hybrids between agn ts3 and Berlin were used because Berlin dlimk1 sequence is closer to FlyBase reference sequence (SavvateevaPopova et al., 2017). At the same time, the reciprocal hybrids agn ts3 × Canton S (the genetic back ground for agn ts3 ) and Canton S × agn ts3 demonstrate exactly the same cognitive behavior as reciprocal hybrids with Berlin (Vasiljeva et al., 2019) approving the usage of Berlin. Figure 1 shows the X chromosome and autosomes archi tec ture in Berlin and agn ts3 female and male parents and F1 fe male and male progeny from reciprocal crosses with the accent on the putative hairpin formed in the X chromosome by 28 bp A/T rich insertion within intron 1 of dlimk1 gene and 3′ end of Tc1/mariner element.
Flies were maintained on standard Drosophila yeastraisin medium at +22 ± 0.5 °С under a 12-h light/dark cycle. For the memory and locomotion tests, males were collected upon eclosion without narcotization and kept individually in culture vials till the behavioral experiments on the 5th day.
Estimation of frequency of ectopic contacts (FECs). The acetoorcein squash preparations were prepared from salivary glands of III instar D. melanogaster female larvae. 20 to 30 animals were examined, therefore the number of analyzed chromosomes varied from 300 to 500. The examples of ectopic contacts are presented on Fig. 2. The number of nonhomologous contacts between the region X:11AB and different bands of the X chromosome was calculated and ex pressed as per cent of the total number of the examined nuclei. FECs in parents and F1 reciprocal hybrids were compared using Student's t-test. Identification of genes localized in the X chromosome bands forming contacts with the 11AB re gion was performed using NCBI Genome Data Viewer database (https://www.ncbi.nlm.nih.gov/genome/gdv/) and molecular function of identified genes was derived from FlyBase (https:// flybase.org).
Bioinformatics analysis of DNA segments homology. D. melanogaster genome sequence (release 6) was taken from (Zerbino et al., 2018). Special software package Homo lo gy Segment Analysis searching the matches of short single stranded DNA fragments within the chromosome areas involved in the ectopic pairing in Drosophila has been deve loped. The software written in Python 3 can be freely down loaded from (Zhuravlev, 2019a). Software version from git (commit 41719cddc6283edbd79c5bf2aee237cde48d4b7d) was used. The algorithm of the program is described in brief in (Zhuravlev, 2019b For other tested regions, all parameters were the same except for the regions location. Preparation of miRNAs libraries and bioinformatic analysis. The detailed description of the procedure is given in (SavvateevaPopova et al., 2017). Extract RNA reagent (Evrogen, Russia) was used for total RNA extraction from adult 5 days old males. To obtain the fraction of small RNA, 25 µg of total RNA were separated using 15 % polyacrylamide gel electrophoresis in the presence of Urea (8 M) following excision of small RNA fraction corresponding to 21-29 nts. Illumina TruSeq Small RNA prep kit (Illumina, USA) was used for small RNA libraries preparation. Sequencing was performed on an Illumina HiSeq 2000 platform.
Locomotor activity. Computeraided automatic device for simultaneous registration of 20 animals is described in (Za kharov et al., 2012). The experiment lasted for 1 h. Sponta neous locomotor activity of flies was detected in a plate with eight chambers and transparent cover using highresolution video camera. Software used for locomotor activity analysis is freely available at (Zakharov, 2017).
The following parameters of locomotion were assessed: activity index (%); run frequency (the number of run bouts in 100 seconds); running speed (mm/s). The full record was divided into 1 s quanta, and the mean speed of fly movement in each quantum was calculated. If the result was less than the threshold value (5 mm/s), the fly was considered to be resting during this time quantum; otherwise, it was considered moving. Neighboring quanta with similar movement pattern were merged in intervals of moving and resting. Activity index is determined as a time spent in movement. Running speed is an average fly speed, determined using only intervals of movement. The Kruskal-Wallis analysis of variance with the multiple comparison of mean ranks was used to compare all the experimental groups.
Learning acquisition and middle-term memory formation in Drosophila males. Detailed description of learning/ memory assessments in conditioned courtship suppression paradigm (CCSP) and specially designed software for obser vation and statistical analysis (randomization test) of learning indices based on courtship indices is given in (Kamyshev et al., 1999). CCSP employs the natural stimuli of Drosophila courtship. Both virgin and fertilized females emit an aphro disiac pheromone, attracting a naïve male without courtship experience. However, a fertilized female rejects a male at the courtship stage of attempted copulation via emitting an aversive pheromone. Repetitive rejections during 30 min training provoke a kind of learned helplessness when a male stops courting another female. This courtship suppression might last for one hour when test female is virgin and for eight hours when fertilized. Males with defective memory formation continue to court after such a training as vigorously, as naïve males.
The courtship index (CI, percentage of time spent in court ship) was calculated for each male. The learning index (LI) was computed according to the formula: where CI NA and CI TR are the mean courtship indices for in dependent samples of naïve and trained males, respectively.

Results
Analysis of spatial nuclear organization delimited to FECs formed by the X chromosome region 11AB Spatial nuclear organization was analyzed using microsco pic images of polytene chromosomes in larvae salivary glands. The results of comparative analysis of FECs between the 11AB region and the other X chromosome regions in Parent-of-origin effects on nuclear chromatin organization Parent-of-origin effects on nuclear chromatin organization Berlin, agn ts3 and their hybrids are presented in Table 1. The columns 1-4 show the pattern of FECs between the 11AB re gion and other X chromosome regions pertinent to listed strains. The absence of significant differences of FECs be tween co lumns, i. e. 1 vs 2 and 3 vs 4 indicates matroclinic inheritance, 2 vs 3 indicates hybrid-specific frequencies, 2 vs 4 and 1 vs 3 pinpoints the patroclinic inheritance. The polytene chromosome bands in hybrids demonstrate differences in FECs, a part of them showing either matroclinic properties or properties of the father strain. In certain bands, FECs are similar in reciprocal hybrids (hybrids-specific, i. e. FEC is equal for hybrids, being different from at least one parent) or depend on the direction of a cross, but significantly differ from that of parents. NCBI Genome Data Viewer software helped to reveal the genes located in the X chromosome bands forming the ecto pic contacts. Based on the assumption that shared location of genes determines their functional features (Liu et al., 2019), it was worth elucidating what biological processes are under the influence of epigenetic factors related to allelic parentoforigin. Therefore, the grouping of genes implied their involvement in control of a certain biologic process (Fig. 3). Figure 3, a presents provisional definition of the biological processes controlled by genes within bands forming contacts with 11AB region either in matroclinic or in reciprocal hybrid specific manner. Among the functional groups of genes having increased number in agn ts3 × Berlin relative to the reciprocal hybrid are genes for motor proteins (8, 4.5fold) and sensory receptors (2, 3.5fold). Genes involved in neurodevelopment, oxidativereduction process and proliferation do not present in Berlin × agn ts3 progeny. Heterosis, when the number of genes involved in hybrid-specific contacts is more than 2-fold higher than in both parents, is manifested for oxidativereduction process (3), signal transduction (4), proliferation (9) and non coding RNAs (13).
The biological processes and genes involved in ectopic contacts with 11AB in a mode of father strain are shown in Fig. 3, b. Among them are the genes responsible for chromatin remodeling (5, 5fold increase in agn ts3 × Berlin relative to the reciprocal hybrid), reactive oxygen species metabolic process (3, 3fold), and metabolism (6, 3fold). Sensory perception and signal transduction groups do not present in Berlin × agn ts3 .
Data on biological processes and the number of genes in volved in contacts with 11AB region exclusively in hybrids or with FECs prevailing those of parents (FEC heterosis), are shown on Fig. 3, c. The ratio of gene numbers for agn ts3 × Berlin and Berlin × agn ts3 reveals functional significance of other gene groups. The magnitudes of differences may be ranged as follows: lipid metabolism and neurotransmitter secretion (20, 8fold); metabolism (6, 2.3fold); cell proliferation (9, 6fold; as in previously published evidence on agn ts3 (Tokmacheva, 1995)); chromatine remodeling (5, 3.5fold). Reparation/ recombination group does not present in Berlin × agn ts3 .

Prof iles of the localized fragment frequencies (LFFs)
Using our specially designed software Homology Segment Analysis, we analyzed correlation between ectopic pairing and distribution of small identical fragments within contact ing regions.
The region X:11AB (~790 kb) involved in Drosophila X -X:11AB ectopic pairing was selected as the source of small 30 nt fragments, and the profile of X:11AB localized fragment frequencies (LFFs) was constructed for the X chro mosome (Fig. 4). The region X:14B-15B was selected as a control region having almost equal length (~790 kb).
Both LFF and the FEC (X -X:11AB) distributions signifi cantly differ from normal. Therefore, we calculated Spear man's rank-order correlation coefficients (r S ) for LFF and the strain-specific FEC (X -X:11AB) ( Table 2). A positive correlation between LFF and FEC is observed within the re

479
ГЕНЕТИКА ЖИВОТНЫХ / ANIMAL GENETICS gion of interest X:11AB. The influence of the fragment length on r S value has been tested. For 30 nt fragments, r S is highly significant for all Drosophila strains ( p < 0.001). As to LFF (X:14B-15B), there is no significant correlation with FEC (X -X:11AB) ( p ≥ 0.01) pointing to specificity of analysis of DNA homology within the region of ectopic pairing. For 20 nt fragments, there is a falsepositive correlation between LFF (X:14B-15B) and agn ts3 × Berlin FEC (X:11AB). Pro bably, 20 nt fragments are too short to display DNA homolo gy specifically associated with ectopic pairing. The increase in fragments length from 20 to 50 nt leads to decrease in LFF (X:14B-15B) -FEC (X -X:11AB) unspecific correlation     Parent-of-origin effects on nuclear chromatin organization and simultaneous increase in LFF -FEC (X:11AB) specific correlation. However, there is a lack of LFF (50 nt) -ECF correlation for agn ts3 . Hence, 30 nt fragment length seems to be optimal in search for the identical fragments within the candidate chromosomal regions involved in ectopic pairing.
To find out what DNA sequences impact the ectopic pairing the most, we calculated r S between Berlin FEC and the localization frequencies for the several specific X:11AB 30 nt fragments having the highest number of occurrences (NO) in the X chromosome (Table 3). The maximal correlation is evident in the part of 372 bp middle repetitive DNA sequence (marked with asterisk (Waring, Pollack, 1987)). The majority of fragments with significant positive r S values appear to be the parts of a ~50 bp repeat (marked with #) having significant selfcomplementarity. This sequence also shows an almost complete identity to another part of 372 bp repeat. Such repeats with slight sequence variations occur in both DNA strands of the X chromosome. r S is much lower for (at) 15 repeat and is nearly absent for (gt) 15 , (tc) 15 and (ca) 15 repeats, although their NO can be higher. Noteworthy, all r S values for 372 bp repeat fragments are lower compared to r S for the whole set of the localized fragments (see Table 2), hence all of them seemingly impact ectopic pairing.

miRNAs expression profile
Analysis of miRNAs expression in Berlin, agn ts3 and their re ciprocal hybrids demonstrates significant differences between males of parent strains and hybrids in content of 44 miRNAs. For the reciprocal hybrids, the heat maps of miRNAs expres sion are presented in Fig. 5. Among 44 miRNAs 10 miRNAs belong to the same cluster of testis-specific miRNAs (Mo hammed et al., 2014). However, only miR980 and miR974 are involved in memory processes. Expression of miR980 suppresses Drosophila memory (Guven-Ozkan et al., 2016), while lowered expression of miR974 impairs memory forma tion (Busto et al., 2015).
The heat map represents the RPMnormalized and log 2 trans formed counts of miRNAs reads with zscale normaliza tion of the rows. Thirty percent of lowexpressed miRNAs were removed from further analysis. Only the miRNAs with altered content in a hybrid compared to at least one parent are shown.

Behavioral analysis of parent strains and their reciprocal hybrids
Locomotor behavior. Figure 6 presents the parameters of locomotion. As to activity indices, they are significantly dif ferent in agn ts3 and Berlin, both hybrids differ from agn ts3 and Berlin × agn ts3 differs from Berlin. Running speed and run frequency in agn ts3 and Berlin are similar. Both hybrids dif fer from agn ts3 and Berlin × agn ts3 from Berlin. However, the hybrid running speed exceeds that of parents, demonstrating heterosis. Comparatively to parents, run frequency in hybrids is intermediate. At the same time, any alterations in locomotor parameters in hybrids are similar and unidirectional. Learning acquisition and memory formation. In all strains, CIs of males decrease after training with fertilized females compared to naïve flies (Fig. 7, a). Learning/memory scores in reciprocal hybrids show that memory formation (3 hours after training), but not learning acquisition (0 hours after training), demonstrates patroclinic inheritance (see Fig. 7, b).

Discussion
Studies of genomewide associations between DNA polymor phisms and phenotypic traits have revealed genetic variants predisposing to different mental diseases. Findings pinpoint ing the role of POEs in genetic risk for neuropathology open new possibilities for therapy and preventive medicine (Zayats et al., 2015). In this study, we estimated FECs in reciprocal hybrids considering an impact both of genetic variants of agn ts3 gene and epigenetic factors (POEs) in spatial nuclear architecture, learning/memory formation and spontaneous locomotor activity.
To exploit the advantages given by the model, we delimited the analysis of spatial nuclear organization to FECs formed by the X chromosome region 11AB harboring dlimk1 gene.
Our assumption was that FEСs partly reflect the restricted homology of short DNA sequences in different, seemingly "nonhomologous" regions.
As shown in this study for the short identic fragments within the contacting regions, FECs correlate with LFFs. Although the correlation is rather moderate (~0.35), it is highly specific for 30 nt fragments. Many factors affect ectopic pairing, such as DNA homology, the distance between the interact ing regions and epigenetic factors causing the interstrain FECs differences (Zykova et al., 2018). Our computational algorithm concerns only fragments aligned with the X chro mosome without gaps. This reveals the partial homology of interacting bands. Thus, the interacting chromosomal areas may be significantly larger than 30 or 50 nts. Although dif ferent mechanisms are involved in ectopic pairing, including POEs, our data indicate the significant role of DNA sequence itself. However, as FEC-LFF correlation is mainly observed for specific DNA fragments, their pairing mediated by some proteins or noncoding RNAs cannot be ruled out.
Most of the found 30-50 nt fragments are similar to the D. melanogaster dispersed 372 bp A/Trich noncoding repeat (Waring, Pollack, 1987). This moderately repeated sequence is located in the euchromatin of the X chromosome between the regions 4 and 14A in ~300-400 copies per haploid genome. The 372 bp repeat is a part of 1.688 g/cm 3 class of satellite DNA (1.688X repeats) (Jagannathan et al., 2017). siRNA from the 1.688X repeats is involved in dosage compensation in recognition of the X and autosomal chromatin, thereby delimiting activities of male-specific lethal (MSL) complex to sex chromosomes through upregulation of the X chromo some (Menon et al., 2014).
The data obtained consider the common mechanisms of ectopic contacts formation and dosage compensations. Seem ingly, POEs might influence the spatial chromatin organiza tion, thereby affecting behavioral performances.
This indicates that each Drosophila strain possesses its own pattern of ectopic contacts with the region 11AB. The polytene chromosome bands are heterogeneous in their modes of regulation of ectopic pairing. A part of them is regulated by genes of either maternal, or paternal origin. A separate class is comprised of regions manifesting only hybrid properties. Similar РОЕs were observed for the pattern of methylation and nucleosome distribution within the imprinted loci in humans and plants (Dong et al., 2018;Zink et al., 2018).
In both reciprocal crosses, bands 7A, 9A and 13B display the maternal properties. When mother is agn ts3 , FECs in these bands significantly decrease. In this case maternal control of spatial localization and therefore, of gene expression is ge netically determined. These are genes controlling membrane receptor regulation (PPYR1) and signal transduction (gce), chromatin remodeling (Top1, HDAC6), axon guidance and chemosensory jumping behavior (acj6 ).
In the cross agn ts3 × Berlin, the number of genes with known functions contacting with 11AB with maternal-specific fre quency is 2fold higher than in reciprocal cross. Possibly, this is due to the agn ts3 -specific miRNAs pattern of expression. The role of miRNAs in maternal inheritance and expression in embryogenesis is sparsely studied. As we have shown earlier, the expression level of miR9, miR34 and miR124 differs in agn ts3 from that in Berlin, Canton S and Oregon-R (SavvateevaPopova et al., 2017). miR9 and miR124 are also expressed in early development (0-12 hrs) (Sempere et al., 2003), miR34 is detected in embryos till zygotic reduction (Soni et al., 2013). As known, the switch from maternal to zygotic development program occurs between the second and the third hours of embryonic stage, hence miRNA found in early development have maternal origin (Schier, 2007). These miRNAs targets are Swi/Snflike complex, neuralprogenitor specific npBAF, repressor-element-1-silencing transcription factor (REST belonging to 1 class of histone deacetylases (HDAC1/2) and silent information regulator 1 (SIRT1) -3 class of NAD + dependent histone deacetylases involved in heterochromatin formation, Bourassa, Ratan, 2014). They are involved in neurogenesis, dendrite morphogenesis and axon guidance which depend on global chromatin remodeling. Therefore, it is not surprising that in cross agn ts3 × Berlin ectopic contacts between 11AB and regions containing genes involved in chromosome remodeling are formed with fre quency characteristic for the maternal genome. The products of these genes are: Tip60 -histone acetyltransferase, HDAC6histone deacetylase; mxc -regulator of histone synthesis of Polycomb group; Top1 -DNA topoisomerase. However, only two regions containing genes HDAC6 and Top1 are present in cross Berlin × agn ts3 .
Noteworthy, new knowledge about topologically associat ing domains (TADs) indicates that polytene, diploid, and embryonic TADs condensation along the chromosome axis is just the same everywhere (Eagen et al., 2015). Moreover, comparison of TADs with 3D chromatin organization revealed by the Hi-C method confirms that the interphase nucleus spatial organization into TADs is directly represented by band ing pattern of polytene chromosomes (Kolesnikova, 2018). Therefore, this allows to bridge the ratio of genes forming ectopic contacts in a mode of either maternal, or paternal strain in reciprocal crosses and their physiologic manifestations. The later might result from alterations in the 11AB region architecture. As shown in Fig. 3, a, the agn ts3 like matroclinic mode of inheritance is pertinent to genes responsible for actin and microtubulesbinding proteins with motor function and neurodevelopment. Noteworthy, the state of actin remodeling determining neurologic manifestations is a diagnostic feature of agn ts3 (SavvateevaPopova et al., 2017).
The regions with FECs similar in reciprocal hybrids, but differing from parents, i. e. manifesting hybrid properties, contain a large set of genes responsible for motor functions. Figure 3, c shows genes and biological processes for chro mosomal regions forming ectopic contacts with X:11AB only in hybrids or mainly in hybrids compared to parents. These processes are pertinent to main manifestations of agn ts3 : meiosis, reparation, recombination; transcription factors; me tabolism; proliferation; actinbinding proteins, microtubule associated proteins.
Interestingly, in the cross Berlin × agn ts3 the chromosomal bands 8D, 12E, and 19D demonstrate FECs significantly ex ceeding these of parents. These bands contain genes involved in taste and odor perception and neurodevelopment, in particu lar of the mushroom bodies of the brain. The other examples of father strain manifestations might result from activities of transacting factors, such as miRNAs (Wittkopp et al., 2006). miR974 is involved in memory processes: its lowered expression impairs memory formation (Busto et al., 2015). Decrease in its content in olfactory neurons and the mushroom body V2 neurons promotes 3hour memory. Noteworthy, the content of miR974 is decreased both in agn ts3 and in progeny of Berlin × agn ts3 (impaired 3hour memory) and is similar to wild type in the agn ts3 × Berlin cross (normal memory). Likely, miR974 might act as transacting factor presumed to regulate genes in patroclinic mode. The prevailing role of the paternal genome in memory formation is evident in Canton S and agn ts3 reciprocal hybrids (Vasiljeva et al., 2019).
Taken together, our data indicate that the agnostic locus might belong to the class of quantitative trait loci (QTL) (Qin et al., 2019).

Conclusion
One of the requirements of predictive and personalized medicine is consideration of POEs for prognosis of clinical phenotype of many multifactorial neuropsychiatric disorders. These different and individual manifestations of cognitive abilities and motor functions in patients with the same disease, i. e. behavioral plasticity, results from genome plasticity pro voked by 3D chromatin architecture of the nerve cells nuclei. The evolutionary gene conservation approves the usage of simple low cost, fast and efficient models as Drosophila to probe the causes, consequences and mechanisms of patho logy leading to human disease (Peffer et al., 2015). The Dro sophila agnostic LIMK1 gene is a good candidate for linking the neuronal activity (spine remodeling, neurite outgrowth, trafficking of intracellular components, postsynaptic density functioning) and genetic apparatus (transcription machinery, chromatinremodeling factors). Additionally, quite recent and unexpected findings (Davis, Zhong, 2017) reveal a new target of intellectual disabilities: learning acquisition and memory erasure (forgetting) are governed by different signal cascades, correspondently cAMPdependent and actin remodeling cas cade small GTPase Rac1 -LIMK1 (the key enzyme of actin remodeling LIMkinase 1) and its phosphorylation substrate сofilin. The absence of Rac1-dependent forgetting causes the autistic spectrum disorders. Expression changes (active or non-active state) of LIMK1 and cofilin lead to different neurological disorders. Therefore, in the tradition of Russian genetic school (Lobashev et al., 1973), the agnostic gene might be a functional link between genetic and cytogenetic processes within the nervous system and serve as a model for elucidating both the maternal and paternal modes of transgen erational inheritance.