Chromosomes in the DNA era: Perspectives in diagnostics and research

Walter Flemming, a German anatomist, was the first to analyze structures and processes in the interphase nucleus in a systematic way using a light microscope. He described the process of “mitosis,” and visualized and denominated “chromatin” for the first time in 1879. Therefore, he is recognized as the founder of cytogenetics. In 1956, Tjio and Levan correctly determined that the constitutional human chromosome number is 46. Accordingly, from 1956 onward several well-known clinical syndromes were assigned to have their underlying cause in aneuploidy (e. g., trisomy 21 leading to Down’s syndrome). However, structural aberrations were (almost) not detectable until the introduction of banding cytogenetics in 1971 by Lore Zech. She also contributed to the field of tumor cytogenetics, where the link between abnormal chromosomes and cancer had been established since 1914 owing to a seminal paper by Theodor Boveri presenting his “chromosome theory of tumor development” [8, 9].

The highest incidence of constitutional chromosomal aberrations in humans is found in first-trimester spontaneous abortions. An overall rate of >50% prenatal fatal abnormalities can be seen in these abortions, and the observed spectrum of aberrations is similar in natural and assisted reproduction [10, 11]. Compared with the general population, incidences of chromosomal aberrations are higher in patients with mental retardation, dysmorphic features, pre- and postnatal growth abnormalities, absent/preterm puberty, infertility, and recurrent abortions. Taken together, the incidence of gross chromosomal abnormalities is estimated to be 1 in 135 live-born children, around 40% of them being phenotypically affected [12]. A cytogenetically unresolvable or an apparently balanced de novo aberration found in a phenotypically affected individual is the starting point for further analysis; molecular cytogenetics and/or molecular karyotyping to identify possible causative submicroscopic imbalances, complex rearrangements or disrupted genes are nowadays routinely applied in such cases [13]. However, if banding cytogenetic analysis has been skipped and a chromosomal imbalance (>50 kb) is detected, e. g., by CMA, it is extremely important to verify the molecular karyotyping results at the chromosomal level. This is necessary to distinguish an unbalanced insertion from an unbalanced translocation or a small supernumerary marker chromosome from a simple duplication. Also, subsequent analyses of the parents is necessary to estimate recurrence risks.

Furthermore, identification and characterization of chromosomal aberrations play an important role in tumor cytogenetic diagnostics of leukemia, lymphoma, and solid tumors. Acquired, cytogenetically detectable aberrations are frequently observed in these conditions and can be the most complex results of human genome reshuffling [6, 14].

In general, the prerequisites for cytogenetic approaches are living cells that still divide or may be stimulated to divide again. The huge advantage of this technique is the “whole genomic view,” which enables screening for any kind of numerical and gross structural aberration at a single cell level; the latter is a unique opportunity to also detect low-level mosaicism [15]. Additionally, banding cytogenetic analyses can be performed at relative low equipment and consumables costs compared with higher resolution techniques (Table 1). For acquired somatic clonal chromosomal abnormalities in tumor cytogenetics, the result can help to classify the tumor type, may influence tumor therapy and prognosis, and can be used to monitor the disease course in the patient.

Molecular cytogenetics originally comprised two approaches: primed in situ hybridization (PRINS) and fluorescence in situ hybridization (FISH). Nowadays, PRINS is only rarely used in research as it turned out to be exclusively suitable for repetitive regions. FISH was established between 1986 and 1989 for human chromosomes as a single-, two- or three-color approach. Previously, ISH was only available as a radioactive methodology. In 1992, chromosome-based comparative genomic hybridization (CGH) was developed. CGH was later developed toward molecular karyotyping (i. e., array CGH). Since 1996, various multicolor FISH probe-sets have been established. First, only whole chromosome paints were used. Afterward, all kind of probes such as partial chromosome paints, centromeric probes and locus-specific probes were combined for subtelomeric, (peri-)centromeric and/or FISH-banding probe sets [7]. Since the 2000s, it has become possible to apply FISH to study genomic architecture of 3D-preserved interphase nuclei. This approach is still only used for research purposes, even though a possible impact on clinical consequences was reported (see also part (ii) of this article).

FISH is nowadays routinely applied in tumor cytogenetics, including interphase-directed approaches in solid tumors and leukemia [6]. For constitutional genetics, it has recently been shown that FISH is the only routine diagnostic approach capable of detecting the parental origin of disease-causing submicroscopic inversions relevant for the offspring. Such submicroscopic events may lead to microdeletion or microduplication in the putative progeny of such inversion carriers [16]. Nonetheless, patients with a suspected microdeletion syndrome (including subtelomeric imbalances) are nowadays tested using molecular karyotyping rather than FISH. Although in the past, FISH was used as a “prenatal quicktest” to screen for the most frequent second-trimester aneuploidies, this is nowadays most often examined using molecular genetics tests via microsatellite analysis. However, interphase FISH performed in somatic tissues other than blood or fibroblasts is an important tool for ruling out the possibility of cryptic mosaicism, e. g., in the case of female infertility and a banding cytogenetic karyotype of 45,X[2]/46,XX[27]/47,XXX[1], FISH in buccal mucosa may confirm X‑chromosome mosaicism [7].

Metaphase directed FISH is especially important for the characterization of acquired and constitutional chromosomal rearrangements, not being resolvable by banding cytogenetics alone. Cryptic complex rearrangements, orientation of inserted DNA fragments and submicroscopic deletions or duplications may be detected and characterized by FISH [7]. Another important field addressable by FISH is the characterization of heterochromatic variants, which may also be distinguished from euchromatic balanced rearrangements [5]. Finally, FISH is often used as the “second method” to confirm a molecular karyotyping result (Table 1).

For around 10 years chromosomal microarrays (CMA), based on the original CGH method (see above), have been routinely used to screen for genome-wide imbalances. In addition, single-nucleotide polymorphism (SNP)-based techniques are used to test for loss of heterozygosity (LOH). The resolution depends on the platform used and the diagnostic indication and can be as low as 5–10 kb. CMAs and banding cytogenetics are both genome-wide techniques, but the resolution of CMA is 100–1,000-fold higher. Thus, CMA is also referred as molecular karyotyping, even though no in situ view on the chromosome and no information on ploidy or heterochromatic genomic regions are available [5]. Based on CMA, numerous new recurrent microdeletion and microduplication syndromes were reported, e. g., Weise et al. [4]. At the same time, more and more copy number variations with uncertain clinical significance and combinatorial effects have been identified. As CMA utilizes DNA from a relatively large cell pool, small mosaics within the sample may stay undetected. Furthermore, CMA in routine diagnostic settings is able to identify neither balanced aberrations (translocation, inversion, insertion) that might have an effect on the recurrence risk or have a position effect, nor DNA sequence or epigenetic changes including imbalances below the specific resolution of the array platform. Finally, the exact nature of an imbalance cannot be resolved by CMA alone (Fig. 1) and needs subsequent (molecular) cytogenetic analyses. Moreover, the investigation of the parental or de novo origin of a CMA-detected imbalance can help to estimate recurrence risks in single families [11].

Using CMA, the highest diagnostic yield is available in mentally retarded individuals with a normal karyotype; here, an additional detection rate of around 20% has been reported [17]. Other patients with submicroscopic genomic imbalances include individuals with autism, epilepsy, dysmorphic features, developmental delay, and congenital malformations, or a combination of the aforementioned characteristics. However, up to ~80% of infertile patients with a small supernumerary marker chromosome would be missed in CMA [18].

The main driving forces for developing new techniques in genetic diagnostics are the limitations of the established ones (Table 1). This also holds true for PCR-based approaches developed over the last decades (e. g., multiplex ligation-dependent probe amplification—MLPA), and for the latest achievement: high-throughput next-generation sequencing (NGS) techniques. NGS dramatically reduced sequencing costs and time when entering routine diagnostics to replace Sanger sequencing of single genes. However, the most comprehensive strategy for finding new causative mutations by NGS is to run whole-genome (WGS) or whole-exome (WES) sequencing (ideally in family trios). Targeted gene panels are straightforward for reducing time and costs for the health system in routine diagnostics of defined clinical subgroups (e. g., certain malformations, neurological phenotypes, malignancies, etc.). Consequently, targeted NGS-based tests have been developed that can deal with very small amounts of cell-free DNA and DNA ratio differences, as in non-invasive prenatal testing (NIPT). Currently, NIPT is applied globally as a prenatal screening test focusing on trisomies 13, 18, and 21, and on gonosomes [19]. This led to the secondary phenomenon that the incidence of children born with inborn disease-related copy number variations (CNVs), which were formerly picked up by GTG banding, FISH, and/or CMA, became epidemic [20, 21]. In fact, this has had drastic consequences not only for individual families, insufficiently informed about the drawbacks of the NIPT test, but also for national health systems [19‐21].

Hence, besides having a high diagnostic yield, NGS has limitations (Table 1). A recent retrospective study by Höchstenbach et al. [22] nicely illustrates the spectrum of missed diagnoses when WGS might be used as a “one fits all” test. At least 8.1% of GTG/FISH/CMA-detected abnormalities are missed: 73.3% in the premature ovarian failure due to low-level gonosomal mosaicism group, 25.6% in couples with recurrent miscarriages because of undetected Robertsonian translocations (Fig. 1), and 0.35% in mentally retarded patients. Thus, clinicians, clinical laboratory geneticists, and the patient/family need to be aware of expected pick-up rates and the type of abnormalities that can escape the applied diagnostic methods. Another side effect when new state-of-the-art techniques become routine is the decreasing number of analyses of “old-fashioned” methods that might lead to fading competency in the neglected field [23]. Therefore, traditional methods should not be bypassed just because newer approaches become available. The choice of stepwise diagnostics (e. g., from low to high resolution) may be preferable. An example is the need for subsequent diagnostics in defining the type of trisomy found by CMA, which can either be a free trisomy and most likely sporadic, or arise from a (parental) translocation being connected with an enhanced probability of recurrence in further offspring and a UPD risk for future pregnancies. To elucidate the underlying type of trisomy, only cytogenetics can help (Fig. 1).

In conclusion, reasonable (preferably stepwise) combinations of well-established new high-throughput methods will lead to a maximum diagnostic yield for the patients and their families, while keeping in mind methodological limitations, advantages, and disadvantages (Table 1). Even a simple karyotype may resolve the diagnosis (Fig. 1). On the other hand, WGS may serve as a first-line test to find causative single-gene mutations, LOHs, CNVs, and structural rearrangements, given that this can be accompanied by a third-generation sequencing revolution providing long read sequence information. Nevertheless, karyotyping will still be helpful and necessary, as heterochromatic regions are barely covered by sequencing approaches.

However, all patients, in industrial countries and in developing nations, deserve the best and most straightforward strategies for a quick and comprehensive diagnosis. Therefore, clinically customized and cost-minimizing analyses should be used before starting with novel “one fits all” methods (Fig. 1).

Owing to the application of high-throughput settings and cytogenetic approaches in clinical diagnostics, there is a tremendous output of large amounts of diagnostic data and metadata collected in open databases. This is not only fruitful for better genotype–phenotype correlation and comprehensive genetic counseling, but also provides new and deeper insights into fundamental genome biology and disease mechanisms. One example is the identification of recurrent microdeletion and -duplication syndromes by CMA collected in several databases (e. g., Decipher, ISCA, ECRUCA) and the identification of the underlying pathomechanism of nonhomologous recombination triggered by low copy repeats. Subsequently, predisposing rearrangements such as inversions were discovered in parents of affected individuals and are now the subject of evaluation in routine FISH diagnostics to estimate the recurrence risk for those families [16]. Even combining all available standard approaches in human genetic diagnostics may be insufficient to solve the disease-causing gene. But putting all information together would still be helpful in terms of understanding general principles such as genome, chromosomal, and nuclear organization. For example, the recent characterization of a disease mechanism facilitated by a deletion acting in trans was only accessible by taking interphase architecture into account [24]—see the next part of this paper.

The spatial and temporal fine tuning of nuclear architecture is critical for replication, transcription, DNA repair, and cell cycle progression. Recent technical developments include chromosome conformation capture and chromatin immunoprecipitation with subsequent sequencing methods (summarized in Schmitt et al. [25]), together with super resolution microscopy (summarized in Cattoni et al. [26]). Thereby, these methods provide an alternative view into the highly dynamic and complex organization of the nucleus.

The anatomy of the nucleus came into spotlight on the discovery that chromosomes are not fully decondensed in the interphase nucleus [27] and are randomly placed but occupy a preferred area that is also known as chromosome territory [28]. One key method for accessing the nucleus architecture is sequence-specific multicolor FISH. Another modification for reaching a more in vivo situation was to perform studies in 3D-preserved interphase nuclei. Accordingly, there is strong evidence that defined chromosomal positioning is a prerequisite for the correct functioning of living cells. Thus, the comparison of the chromosomal constitution in healthy and disease-affected human brains [29], of sperm in healthy and infertile individuals [30, 31] or in leukemic and normal bone marrow [32, 33] can enlighten the as yet not understood pathomechanisms of many human diseases. Additionally, the impact of extra chromosomes on the nuclear architecture has been studied using this technique in humans [34] and other species [35], in addition to the general position of chromosomes in species other than humans [36, 37].

Recently, it has been confirmed that the precise order of chromosomes is not only restricted to the interphase and prophase nucleus [38] but also seems to be conserved with respect to the parental origin of homologous chromosomes down to the metaphase stage; a functional relevance cannot be neglected any longer [39].

As outlined above for diagnostic chromosome analysis (in the first part of this article), different techniques for studying the human genome involve specific limitations. This also holds true for studies in 3D nuclear architecture, where no single technique is able to address all the questions to be studied. This is mainly because of remarkable differences in throughput, resolution, and reproducibility of the available approaches. In addition, most technical possibilities are static snapshots without temporal dynamic chromatin information, lacking information on cell context and cell variability within a population. The recently launched 4D nucleome project is aimed at overcoming that kind of limitation [40], but a strong need for standards and guidelines is required to ensure data reproducibility [41].

Although chromosomal aberrations can have a direct effect on the phenotype (by altered gene dosage as outlined in the first part of this article) a frequently discussed and barely accessible mechanism is the so-called “position effect” [42]. It is known that genes are aligned linearly along a chromosome. Whereas the influence of enhancers or inhibitors in front of or near to a gene on its functionality is well-accepted, close proximity to a heterochromatic region or direct interaction of genetic regions located Mbs apart from each other (in cis) or even in trans, has long been debated. Meanwhile, it is well-known that the expression of genes distant from a given breakpoint can be affected because of disturbed higher-order chromatin organization (e. g., topologically associated domains, TADs; reviewed in Spielmann et al. [43]). Apart from the structural organization of the nucleus and folding of chromosomes into local aggregates such as TADs, chromosomes can directly interact with each other. This was first observed as the “intermingling” of chromosome territories and is also referred as “chromosome kissing” or “nonhomologous chromosomal contacts” [44, 45]. Such interchromosomal interactions contribute to the formation of nuclear compartments, e. g., when the short arms of human acrocentric chromosomes build up the nucleolus, and/or form a gene expression regulation network with regulatory elements interacting in trans (summarized in Maass et al. [46]). The close proximity between nonhomologous chromosomes facilitates not only the exchange of chromosomal material, appearing in as many as 1:1,000 for Robertsonian translocations [12] but also drives partner selection for other (partly recurrent) translocations [47]. This phenomenon could be tissue-specific, illustrated by recurrent tumor-specific translocations [33, 48]. The direct influence of a chromosomal aberration on 3D nuclear architecture was recently shown for a HDAC4 deletion located in 2q37, resulting in altered “chromosome kissing” among chromosomes 2, 12, and 17 [24].

The common bases for acquired or constitutional chromosomal rearrangements and for chromosomal changes between species are DNA double-strand breaks. One specific class of cytogenetically visible breaks of decondensed chromatin are fragile sites (FSs) that are considered as regions of chromosomal instability with overlapping signatures for breakpoints repeatedly observed in tumors [49‐51], in constitutional rearrangements [4, 52, 53], and also as evolutionarily conserved breakpoints [54‐58]. In addition, those breakage-prone regions are conserved beyond the mammalian linage [59] and seem to be a general and conserved feature of chromosome biology. Therefore, FSs are exemplary structures for focusing deeper into mechanisms of chromosome, gene, and genome evolution.

Fragile sites can be induced under different culture conditions inhibiting proper DNA replication and resulting in unreplicated stretches of DNA visible as chromosomal breaks or gaps on a cytogenetic view. Up to now, more than 230 different FSs have been described at a genomic resolution of 5–10 Mb [60]. So far, only few have been mapped at the molecular level. This is because these so-called common FSs can only be observed in low frequencies (mostly below 0.1%) and are not linked to a specific DNA sequence but rather reflect regions of enhanced breakage probability with variable sizes (up to several Mb). Besides these “common FSs,” which apparently contribute to regular chromosome structure and biology and are not of clinical relevance, “rare FSs” exist, which break at specific repeats and can segregate within families or are associated with certain syndromes [54]. It has become evident that within common FSs, large stretches of DNA-activated replication origins are lacking [61]. Instead, origin scarcity is connected to the even greater effect of replication hindrance on the timely replication of this region. Under-replicated DNA, therefore, parts will be left behind, somehow escaping check point activation before mitosis onset, as treatment with low doses of a potent replication inhibitor can do [62]. During the metaphase such zones of incomplete replication can appear as relaxed chromosomal parts that are reminiscent of gaps or breaks present within each human individual. A recent study on molecular features of fine-mapped FS [63] revealed that these regions are in general gene-poor but at the same time enriched in disease related and Online Mendelian Inheritance in Man(OMIM)-annotated genes. Additionally, they comprise an increased enrichment of CNVs [64, 65], most likely mediated by the imperfect repair of these breaks. Those CNVs include euchromatic gene-carrying sequences, leading to copy gains and losses, and are therefore a substrate for population variability and evolutionary processes. This could be identified as an enrichment of single gene and pseudogene family members located at different genomic FS regions [63].

In conclusion, FSs seem to be not only reused genomic puzzle edges in karyotype evolution but also provide the infrastructure to spread gene (copies) over the genome as a source of evolutionary adaptation. Collectively, this postulated model of “FS-driven gene and genome evolution” awaits further exciting insights into the trade-off between the risk for genomic diseases and cancer on the one hand and genetic variability and flexibility for evolutionary adaptation on the other.

The “chromosomic view,” including nuclear architecture, inter- and intrachromosomal interactions, and FS as underlying principles of chromosomal evolution (including the formation of new genes) and disease, is the unifying tool for understanding all these different aspects of genetics. This applies to future research directions and to the most urgent integration of three-dimensional nuclear organization into human genetic diagnostics.