Biology in balance: human diploid genome integrity, gene dosage, and genomic medicine

Abstract

The path to completion of the functional annotation of the haploid human genome reference build, exploration of the clan genomics hypothesis, understanding human gene and genome functional biology, and gene genome and organismal evolution, is in reach.

…we enjoy the freedom to think and are limited only by our imaginations, but we also have more technologies and tools with which to explore life and its very essence. These toys that we have to play with can now generate massive data flow opening up to experimental exploration fundamental questions that have occupied the minds of generations of scientists and philosophers. As geneticists, we also enjoy a privileged position in society. Much of the medical profession and humanity itself are looking toward us as geneticists to provide the next ‘big insight’ into the ailments that afflict us all. – J.R.L. 1 May 2011

[2011 Commencement Watson School of Biological Sciences Cold Spring Harbor Laboratories, Long Island, NY, USA]

Viewing recent accomplishments in science and medical genetics history through the lens of the ‘retrospectroscope’ [1], it is interesting to muse over how we arrived at where we are in genomic medicine (see Glossary) and how disruptive technologies, such as genomics, the internet, digital information theory, and artificial intelligence, have fomented biological discovery and implemented worldwide and humankind advances in medical sciences. The path(s) to discovery was never easy and often punctuated by incremental steps in knowledge acquisition. What is perhaps clear is the emergence, and re-emergence, of two defining principles of the genetics and genomics ‘golden era’: biological balance and complementarity. As members of the species Homo sapiens currently inhabiting this planet Earth, we are a diverse global community, all immersed in a genetics and genomics renaissance to push the conceptual boundaries of biology and medicine (see Outstanding questions).

Outstanding questions.

What are the mutational mechanisms for genomic change?

What are the ‘rules’ governing human genomic instability?

Can we predict genes/genomic intervals more susceptible to SV mutagenesis: a genomic instability index?

What is the relevance of gene dosage to disease ascertainment? Disease biology? Molecular diagnosis? Therapeutics (Rx)?

How do DNA secondary structure mutagenesis and indel mutations contribute to rare disease traits?

The path to gene dosage

As I sit here at home enjoying the sunrise and the emergence from total darkness, interrupted only by the flickering of candlelight after a complete hurricane-induced power failure during the second year of a global pandemic, my thoughts gain focus. We are on the way to the largest coronavirus disease 2019 (COVID-19) disease peak (i.e., omicron) yet seen in the city of Houston, Texas. But I optimistically think about the future and contemplate the genetics and genomics golden era. A timeline of events provides a framework for these two intellectual disciplines and, at times, parallel paths to enlightenment [2,3], as delineated in this 100-year illustrative perspective (Figure 1). Certainly the ‘golden era’ did not happen overnight, nor through any singular epiphany or ‘Ah haaaaaaah!’ eureka moment, but rather hard thinking and hard work prevailed during a century of science.

Figure 1. Gene to genome mutation: a 100-year perspective on clinical genomics.

Above are ‘inflection points’ marked by experimental observations. In the middle of the figure is the timeline and years of notable events (purple) or structural variant (SV) mutagenesis mechanisms (orange) as given. Below are organismal phenotypes or genome phenotype/changes observed. Note genomic duplication and triplication in fruit flies (vertical green broken line rectangle) and genomic duplication in human (vertical green rectangle). Gene dosage (purple), defining genomic disorders (purple), and clinical chromosome microarray analysis (CMA; purple). Gene dosage as a concept began to emerge from studies in the fruit fly Drosophila melanogaster. The role of gene dosage in disease (Dz) began with the identification of the CMT1A duplication by studies of Charcot-Marie-Tooth (CMT) neuropathy, the finding of CMT in chromosome 17 direct duplications, dup(17)(p11.2), and genomic disorders. Abbreviations: AAMR, Alu–Alu-mediated rearrangement; AD, autosomal dominant; CNV, copy number variant; ES, exome sequencing; FoSTeS, fork stalling template switching; MMBIR, microhomology-mediated break induced replication; NAHR, nonallelic homologous recombination; PTLS, Potocki-Lupski syndrome; Sx, syndromes. See [2,3,13,24,25,51,54,116–119].

One is humbled by those who came before us, yet excited and impatient, by an empowerment to explore and unravel the fundamental principles of biology at the very roots of evolution and life as we know, observe, and experience it. Darwin observed the biology and recognized the evolution of different life forms; the Grants, Rosemary and Peter, as detailed in The Beak of the Finch [4], measured evolution in real time. Clan genomics applies evolutionary theory and rare variant alleles in genomic medicine. From the focus of the lens of historical perspective, Mendel took a top-down approach to genetics and inheritance, whilst Watson and Crick took a bottom-up approach to the same; they met in an intellectual sense at the gene. The term ‘gene’ was ascribed to the conceptual ‘factors’ of Gregor Mendel working on garden peas in the backyard of St. Thomas Abbey, his monastery in Brno. McKusick observed, characterized, and catalogued traits in medicine and disease states; a Rosetta Stone we have as a tool for practising genomic medicine. His life’s work is available on the World Wide Web: https://www.omim.org, a body of investigations on which to build and learn for all humanity, the rare disease trait for each gene embedded within a clinical synopsis.

What does seem to be clear is that there are two fundamental principles of life sciences: biological balance and complementarity; these principles underlie all of life science, evolution, evolutionary medicine, and genomic medicine [5]. Complementarity in the sense that a ‘normal distribution’ has extremes at both ends; mirror traits exist, a left hand has a right; each DNA strand of the DNA double helix has a W-C base to pair; a 16S rRNA binds an mRNA at a cognate ribosome binding site; an SNRP/snoRNA recognizes an exon junction; an enzyme active site binds a chemical substrate; enantiomers can bind and affect receptor differently, an antibody and T cell receptor recognize an antigen; and a ligand fits to its cognate receptor, all by complementarity. The emergence of the concepts of gene dosage, both haploinsufficiency and triplosensitivity traits, genomic disorders [6–8], clan genomics [9–11], and clinical genomics [5, 12, 13] can all be tied to this 100-year timeline (Figure 1).

Charcot-Marie-Tooth disease studies illuminate genome mutation

It is perhaps amusing, at least to this observer [13–15], how Charcot-Marie-Tooth (CMT) distal symmetric polyneuropathy (DSP) was thought to both (i) represent a single ‘disease entity’ and (ii) play such a prominent role in unraveling genome mutational processes (Figure 2A). CMT studies provided empirical data to support the concepts of structural variant (SV) mutagenesis mechanisms, genomic instability, genome mutational phenomena, such as complex genomic rearrangements (CGRs) [16]: the CGR including triplication, inversion, chromothripsis, chromoanasynthesis, and multiple de novo copy number variations (MdnCNV, a perizygotic mutagenesis phenomenon)] [10], gene mutation burden [17], gene and genome evolution [18], genomic disorders [19], clinical genomics [13], and gene dosage (Figure 2B,C). Gene dosage is perhaps the epitome of organismal ‘biology in balance’: the organism developmentally, physiologically, and biochemically in biological homeostasis.

Figure 2. Charcot-Marie-Tooth (CMT) neuropathy, distal symmetric polyneuropathy (DSP), gene dosage, and genetic math.

(A) Seemingly one disease, CMT DSP, yet multiple genetic/genomic possibilities. The decades from 1990 to 2019 were notable for a burst in molecular mechanistic understanding of human gene and genome changes. (B) Clinical descriptions of phenotypes and genomic alterations viewed through a retrospectroscope. Blakeslee’s and Bridges’ experimental observations (dark red) with clinical descriptions below (black). Timeline for human duplication (red) and triplication (blue) delineation and as robustly tied to rare disease traits. (C) Below gene dosage (orange) copy number at a locus or gene. Diploid genome copy number math and gene dosage. Below gene dosage (orange) copy number at a locus or gene. Note homozygous duplication copy number variant (CNV) and heterozygous triplication CNV have the same 4n gene dosage but have different consequences for transmission genetics and allelic states [109]. Abbreviations: DS, Down syndrome; ES, exome sequencing; FoSTeS, fork stalling template switching; MdnCNV, multiple de novo CNV; MMBIR, microhomology-mediated break-induced replication; NAHR, nonallelic homologous recombination; SV, structural variant; WGS, whole-genome sequencing.

Entire books have been written on the topic of gene dosage as a concept and with a focus on Down syndrome (DS) [20]. But it would take genomics, understanding the ‘molecular biology of the gene’ [21], and the human genome, to provide a tangible appreciation and experimental approach to the concept of gene dosage. The many facets of gene dosage and its ubiquitous nature can be manifested in all of biology, as well as evolution of genes, diploid genomes, the primate speciation of H. sapiens [22,23], and in autosomal dominant (AD), autosomal recessive (AR), and X-linked (XL) disease traits. Like the universality of the genetic code, gene dosage is a universal concept of life itself.

Gene dosage: haploinsufficiency and triplosensitivity traits

Gene dosage can masquerade as a chromosomal syndrome or genomic disorder [10], or an adult onset neurodegenerative disease, such as Parkinson or Alzheimer disease [10], due to genomic rearrangements of one chromosome of the homologous pair, a cancer due to gene amplification or chromothripsis [10]. Gene dosage can also be observed as a triplosensitivity or haploinsufficiency trait, or even underlie a mirror trait [2]. One can observe gene dosage effects and consequences with gene triplication versus duplication, or homozygous versus heterozygous duplication (Figure S1 in the supplemental information online).

Albert Blakeslee [24] and Calvin Bridges [25], experimenting at Cold Spring Harbor Laboratories (Figures 1 and 2) on Long Island, had their own take on gene dosage working on nuclear inheritance in the plant Datura and the Bar eye phenotype in fruit flies. Upon review of Calvin Bridges’ discussion from his one-page Science article of 1936, I was intrigued by this statement: ‘The respective shares attributable in the total effect to the genic balance change [gene dosage/copy number?] and position effect change [gene dosage/expression?] seems to be at present a matter of taste’ (bracketed text inserted by J.R.L.).

Gene dosage has been interpreted as potentially being due to gene regulation or transcriptional changes that can be a manifestation of a position effect, topologically associated domain (TAD) disruption, and regulatory mutations affecting promoters, repressors, or enhancers and even epigenetic phenomena and Lyonization of XL genes. Gene dosage may underlie such genetic phenomena as dosage compensation, imprinting, position effect variegation, chromosomal position effects, penetrance, and even perhaps zeste-mediated transvection [26]. Or gene dosage can simply be gene copy number changes due to duplication or triplication of a gene, or gene amplification. From these perspectives (Figures 1 and 2), gene dosage is important and relevant to Mendelian disease traits, birth defects, genomic disorders, and cancer genomes [10].

Gene dosage is known to be important to many disease states. The clinical association between trisomy 21 (T21) DS and early onset dementia has been long known. Early studies that put forward experimental evidence for gene dosage of the amyloid precursor protein (APP), a gene that maps to human chromosome 21, in support of the amyloid hypothesis in dementia [27] were seemingly met with disbelief [28,29]. Twenty years later, when APP duplication was identified in AD Alzheimer disease within the French population, the evidence for gene dosage and the experimental support for the amyloid hypothesis was clarified [30] and perhaps irrefutably.

The approach to dementia and neurodegeneration as a gene dosage disorder has been fruitful in the alpha synucleinopathies (gene dosage copy number with SCNA triplication, Online Mendelian Inheritance in Man: https://omim.org MIM: 163890, and duplication in association with Parkinson disease) [31–34] and tauopathies (gene dosage expression) [35]. From this perspective, triplosensitivity and haploinsufficiency disease traits are both manifestations of the gene dosage model [10]. The mirror trait electrophysiological correlates of mouse fluid consumption behavior (i.e., licking) are a mirror trait correlate of neurobehavioral manifestations of a gene dosage effect [36]. Mouse neurobehavioral differences (NDs) and neurobehavioral assays can show properties of mirror traits [37]. Those autism spectrum disorder (ASD)-like mouse behaviors of ASD, a trait observed with the Potocki-Lupski syndrome (PTLS; MIM: 610883) (Figure 1) duplication [38], can be mitigated by environmental enrichment [39].

One does have to wonder how much of neurodegeneration and neurodevelopmental disorders (NDDs) [40] are actually due to gene dosage effects by either copy number changes, transcriptional ‘imbalance’, or ‘protein synthesis/degradation’ perturbations. The latter is exemplified by alpha synucleinopathies and polyglutamine repeat diseases. Even pathology of the ‘sickling cellular phenotype’ of sickle cell anemia (MIM: 603903), as well as therapeutic intervention for sickle cell and beta-thalassemia by BCL11A CRISPR/Cas9-guided molecular therapies [41,42], are gene dosage effects. The oligomerization of prion protein diseases, induced by protein conformational changes, can be thought of as disruption of the homeostasis of protein synthesis and degradation equilibrium. Thus, manipulations of gene dosage effects are essentially homeostatic disturbances of the central dogma of molecular biology (DNA > RNA > protein), as are disease phenotypes resulting from the inability to degrade unusual variant protein polymers (e.g., polyglutamine expansion proteins) or conformers (sickling hemoglobin), or regulate tissue mRNA levels by either mRNA synthesis or degradation by PTC-containing mRNA via the nonsense-mediated decay (NMD) pathway.

The preponderance of genetic and genomic data implicates such a gene dosage model/gene action model in macular degeneration observed in ABCA4 Stargardt (STGD1; MIM: 248200) macular dystrophy, an AR rare disease trait, and the role of heterozygous LoF alleles in susceptibility to age-related macular degeneration [43–45]. The disease neurobiology of the resulting photoreceptor neurodegeneration is due to defective transport by a retinal-specific ABC transporter ABCA4 (also known as ABCR and Rim protein, a retinal photoreceptor-specific ABC transporter [46]), of visual cycle products in the retinal photoreceptors (i.e., rods and cones). Cis-acting modifiers in the ABCA4 locus contribute to the penetrance of the major disease-causing variant allele in STGD1 [47]. This allele has been the subject of controversy, given its high minor allele frequency in some populations, resulting in a questioning of its potential pathogenicity. This interpretive questioning seems to occur particularly when, as a biallelic variant in the homozygous state, heterozygous versus homozygous allele states represent a gene dosage effect of a particular variant allele.

Gene dosage of APP in T21 and Robertsonian translocation DS, the latter due to T21 from either rob (14;21) or rob (21;21), and the different implications from nondisjunction T21 versus translocation DS, with respect to recurrence risk counseling and transmission genetics [48], are perhaps not readily apparent. It relates to the chromosome and gene copy number, that is, karyotype (N = 45, XX or 45, XY) in the parents versus 46 chromosomes in translocation DS. The clinical association of congenital heart disease (CHD) with T21, causing DS, was known for decades and even narrowed to a DS critical region and genomic interval, but it would take experiments in fruit fly to support a gene dosage model and evidence for a multilocus pathogenic variation (MPV) or two-locus gene dosage hypothesis. The orthologous human genes for each map within the DS critical genomic interval [49].

Evidence from Yuan-Harel-Lupski (YUHAL; MIM: 616652) syndrome also implicates a two-gene dosage model. In this YUHAL instance, both dosage-sensitive genes responsible for the triplosensitive traits (i.e., CMT1A and PTLS), peripheral myelin protein-22 (PMP22) and RAI1, physically map to the same genomic duplication copy number variant (CNV) [10] (i.e., the latter duplication mutation is a segmental aneusomy rather than a whole chromosome trisomy, as in T21-associated nondisjunction DS and CHD).

Two-locus models, wherein one gene modifies the penetrance or clinical expression of a human disease-causing variant allele [in one illustrative example, Huntington disease (HD)-associated polyglutamine expansion], have been investigated experimentally in Drosophila melanogaster. It took genetic experiments in a D. melanogaster model of HD to understand the dominant-negative (antimorphic) nature of the polyQ allele and how engrailed as a modifier implicates HD as a semidominant disease trait [50] rather than a ‘pure’ AD trait in the Gregor Mendel sense.

Bridges and the ‘birth’ of the gene dosage hypothesis

For me, the review of Bridges work and the birth of the gene dosage hypothesis were stimulated by the elucidation of the CMT1A duplication [51] (CMT1A; MIM: 118220) (Figures 1–3). The Bridges studies, and documentation of CMT1A-associated DSP as a semidominant trait [52,53], implicated four potential explanatory models for duplication-mediated disease: (i) gene dosage, (ii) gene interruption, (iii) point mutation in the duplicated gene, and (iv) position effects (Figure 3). I was very excited when we began systematically to experimentally exclude the gene interruption and position effect models [54]. There was perhaps some amusing irony in publishing a paper in the journal Human Mutation, showing no PMP22, single-nucleotide variant (SNV) mutation (i.e., no point mutation of PMP22 in seven kindred segregating the CMT1A duplication for multiple generations). These studies essentially ruled out the ‘point mutation’ in the duplicated gene, causing CMT as a model [55]. I was also, in retrospect, happy that we did not find the de novo PMP22 point mutation associated with CMT1A [56] before we found the CMT1A duplication, as we would have had to screen hundreds of CMT patients and families to find it.

Figure 3. The gene dosage hypothesis and structural variant mutagenesis.

Four models affecting gene action for a hypothetical ‘Charcot-Marie-Tooth (CMT) disease gene’ formulated upon delineation of the CMT1A duplication. I recall being at the European Society of Human Genetics and the session chair allowing me two or three slides. Thus, I took out my Kodachrome and used a Sharpie to draw an ‘X’ through three models and then showed what we were about to publish with DNA markers flanking the CMT1A locus and motor nerve conduction velocity (NCV) studies on segmental aneusomy showing reduced NCV pathognomonic for CMT1A. These experimental studies resulted in the gene dosage hypothesis published in the first issue of Nature Genetics. This paper was published in 1992 (see Figure 1 for 100-year perspective) and marked the beginnings of testing the gene dosage hypothesis. During the ensuing 30 years, many additional experimental approaches and studies in humans, mice, and rats [82,83] supported the gene dosage hypothesis in disease biology and therapeutic intervention.

Human chromosome 17p12 CNV (Figures 2, 4, and 5), primarily the CMT1A duplication and HNPP deletion alleles (Figure S1 in the supplemental information online), are the most frequent variant alleles found in CMT patients, families, and neuropathy disease clinical populations [57]. Whilst CNV alleles at the PMP22 locus (i.e., CMT1A duplication [51], HNPP deletion [58], and triplication of the locus) (Figure 4, Figure S1 in the supplemental information online) have illuminated gene dosage and its role in DSP, SNV studies and allelic series have crystallized gain-of-function (GoF) versus loss-of-function (LoF) mutation effects (Figure 5) for CNV and the role of GoF/LoF alleles in both AD and AR DSP disease traits.

Figure 4. Genomic triplications result from different mutational mechanisms.

Two different mutational mechanisms, nonallelic homologous recombination (NAHR) and fork stalling template switching FoSTeS/microhomology-mediated break-induced recombination (MMBIR), result in type I and type II structures. Type I occurs by NAHR and results in triplication on a duplication chromosome inherited from a parent. Note, type II requires two template switches (TS), can occur de novo, and can lead to DUP-TRP-DUP or the Carvalho structure: DUP-TRP/INV-DUP [110,111,120,121]. Also note the overall pattern, as illustrated by broken red direct tandem repeat lines, of a tandem duplication on one chromosome homolog of the homologous pair or one allele. See [62,111,118,122].

Figure 5. Crowdsourcing world science in human genetics and genomics: thousands of families, hundreds of laboratories, many countries.

Allelic series studies in Charcot-Marie-Tooth (CMT) and distal symmetric polyneuropathy (DSP) reveal PMP22 and MPZ as dosage-sensitive gene loci with gain-of-function (GoF) and loss-of-function (LoF) effects. The illustration represents 30 years of peripheral nerve neurobiology and neurogenetics from researchers too numerous to count (i.e., TNTC and >100 laboratories) and properly cited publications, from laboratories around the globe! Suffice to say, multiple human chromosomal, genomic, and gene mutational processes were delineated by ascertaining a CMT patient/subject by virtue of a gene dosage effect. (A) The eponym CMT refers, in deference, to the elucidators of the disease. Note the astute clinical observation of familial occurrence of the trait by these clinician investigators even before either nuclear inheritance or Mendel’s laws were elucidated. (B) PMP22 gene dosage and DSP are most often the result of the recurrent nonallelic homologous recombination (NAHR)-derived duplication or deletion copy number variant (CNV); the CMT1A duplication (MIM#: 118220) and HNPP deletion (MIM#: 162500). Nevertheless, rare variant and ultra-rare variant CNV encompassing PMP22 do exist. Intriguingly, duplication involving the upstream regulatory region and not including coding sequences can drive the gene dosage effect potentially by involving an enhancer or promoter [63]. (C,D) Gene allelic series and GoF versus LoF alleles. Note PMP22 studies define the CNV gains as GoF alleles, whilst single-nucleotide variant (SNV) studies reveal LoF alleles by nonsense or frameshift allele or HNPP deletion of PMP22. MPZ, encoding peripheral myelin protein zero, allelic series shows the different clinical entities (all DSPs) that can be observed. Note changes to Cys cause more severe disease; as does homozygosity of hypomorphs. The different DSPs are color-coded for mutant alleles identified: CHN (green), CMT1 (yellow), CMT2 (red), DSN (pink), RLS (orange). The majority of RLS patients are found to have the CMT1A duplication, whilst the descendant of the original family described was found, by DNA studies of autopsy material, to have a P0 pathogenic MPZ variant allele [112]. Note: several CMT2 alleles convey an ‘axonal phenotype’, perhaps hinting at the intimate connection between glia (Schwann cells) and neurons (axons). Also of interest, both homozygous duplication and homozygous PMP22 and MPZ ‘recessive trait alleles’ result in DSN; of note, one such PMP22 recessive [i.e., biallelic, trait allele (T118M)] results in severe axonopathy [86,91,113]. Abbreviations: CHN, congenital hypomyelinating neuropathy; CMT1, type 1 ‘demyelinating’ Charcot-Marie-Tooth disease; CMT2, axonal CMT; DSN, Dejerine-Sottas neuropathy; HMSN, hereditary motor and sensory neuropathy; HNPP, hereditary neuropathy with liability to pressure palsies; RLS, Roussy-Levy syndrome.

CMT and mutational burden

Of note, several mutational models (Figure 2A) were proposed based on the experimentally ascertained genomic variation from personal whole-genome sequencing (WGS) [59] and exome sequencing (ES) analysis [17,60] of CMT and DSP patients, including studies on my own genome. These include duplication/deletion CNV, reciprocal duplication/deletion events by nonallelic homologous recombination (NAHR) [61], the role of new mutation in inherited neurological disease, triplication CNV and CMT/DSP [either type I triplication by NAHR from a CMT1A duplication parent or type II triplication from de novo mutation by fork stalling template switching (FoSTeS)/microhomology mediated break induced replication (MMBIR) [62]] (Figure 4), exonic dropout through deletion CNV [63] and exonic duplication CNV [63,64], WGS for a rare disease trait gene [59], mutation burden [17] from ES, and hypermutation CNV and the MdnCNV phenomena of perizygotic mutagenesis [65,66].

The complexity of genomic architecture at the CMT1A locus, and disease etiology of CMT by gene dosage, allows for so many genomic mutational ways to get to CMT and a DSP phenotype (Figures 2 and 5; Figure S1 in the supplemental information online). Also, many different genes and combinations of possible alleles at a locus can cause the CMT1 DSP as either a dominant (AD) rare disease trait, an XL disease trait, or the biallelic pathogenic variation of an AR disease trait (Figure 5) [67].

Some of the work of two great geneticists, William Allan in 1939 [67] and J.B.S. Haldane in 1941 [68], were derived from human genetic studies of CMT families. In some ways, when one reads the paper by William Allan, it appears that he recognized a role for gene dosage by suggesting that the AR biallelic CMT trait (both gene copies mutated) was usually more severe and of earlier onset; he perhaps understood, and elucidated, genetic heterogeneity noting both AD and XL CMT (also referred to as peroneal atrophy by Howard Tooth in Cambridge, UK; Figures 2B and 5) trait families [67]. Moreover, William Allan may have understood locus heterogeneity of human disease traits, as evidenced by his differentiating the three inheritance patterns (AD, AR, and XL) for CMT disease trait family segregation [67].

Genomic disorders, gene dosage, and mirror traits

Mouse studies provided a rich source of experimental data allowing insights into pioneering organisms like H. sapiens and their disease trait biology. Studying different human and mouse alleles at a locus further informed the genetics and genomics of disease. These studies suggest human allelic diversity, both allelic heterogeneity and allelic affinity [69] in humans, could further inform disease biology and neurobiology. The latter, allelic affinity, delineated a syndrome, Peripheral demyelinating neuropathy (CMT), Central dysmyelination (PMD), Waardenburg syndrome, and Hirschsprung disease (PCWH; MIM: 609136), due to variation in SOX10, a gene mapping to Ch22q13.1, and defined a new disease trait. PCWH includes demyelinating disease of the peripheral and central nervous system (CNS) [69] and glial support cell phenotypes, all seemingly neural crest derivatives of cells of the developing nervous system: oligodendrocytes, Schwann cells, and enteric nervous system cells involved in Hirschsprung.

Interestingly, for PCWH, the last exon PTC that escapes from NMD appeared to show a ‘polarity gradient’ [69]. Genetic polarity can be observed as effects on gene function that generally can be ascribed to the fact that genetic information consists of a linear sequence of nucleotides; the mechanism by which that information is transcribed into mRNA, and the mRNA is translated into the final chain of amino acids, in a sequential manner [70]. Curiously, such ‘polarity’ in the SOX10 last exon mutation and PCWH is not unlike that observed in polycistronic operons in bacterial multigene polycistronic mRNA transcriptional units; polar mutations, such as PTC or insertion elements/transposon mutations, that occur in ‘upstream genes’ [71] of operons like lac (and its z, y, and a genes), cause polarity.

Chromosome engineering enabled creation of a deletion and a duplication mutant chromosome 11 in a mouse, a genomic interval maintaining conserved synteny with human Ch17p11.2, allowing models for both PTLS (MIM: 610883) (Figure 1) and Smith-Magenis syndrome (SMS; MIM: 182290) genomic disorders and amongst the first genomic disorders modeled in the mouse. Duplication [Dp(11)] and deficiency [Df(11)] strains that were heterozygous for either the PTLS duplication CNV or SMS deletion (deficiency, Df) were constructed to model these genomic disorders [72] (Figure 6A). A null allele was constructed in the dosage-sensitive Rai1. When paired with the Dp(11) chromosome, it restored Rai1 dosage to the normal N = 2 genomic balance complement and rescued the gene dosage phenotype. Such a mouse strain allelic combination at a locus, leaving dozens of genes mapping within the segmental aneusomy in the ‘trisomic N = 3’ state, provides strong support for a ‘driver gene’ gene dosage effect. These data showed that weight curves were restored to wild-type (wt) levels and craniofacial dysmorphology corrected, consistent with rescue of the gene dosage effect (Figure 6B,C) [73].

Figure 6. Mouse genetics and genomics, chromosome engineering, gene dosage, and mirror traits.

(A) Construction of chromosome engineered mouse duplication and deletion alleles, germline transmission, and propagation of Dp(11) and Df(11) strains; founder animals ‘for Smith-Magenis syndrome (SMS) and Potocki-Lupski syndrome (PTLS)’ were called Ann and Jim, respectively, by Dr Katherina Walz. (B) PTLS and SMS animals. (C) Restoration of Rai1 to N = 2 copy number in PTLS mice corrects gene dosage and defines driver gene effect. (D) Mirror traits of PTLS and SMS animal models due to gene dosage. (E) SMS models for obesity and metabolic syndrome and gene × environment (G × E) gene dosage effect. Note Dp(11) allele causes lean phenotype and protects from diet-induced metabolic syndrome. Abbreviations: HF, high-fat; RC, regular chow; WT, wild-type.

Remarkably, growth curve studies revealed a mirror trait for Dp(11) (underweight) versus Df(11) (overweight) animals (Figure 6D). Even neurobehavioral traits could be rescued by restoration of genomic balance [37]. The PTLS mouse was lean, thin, and long (mouse)/tall (human), matching the human phenotypic trait (Figure 6Di,Dii), whilst the SMS mice were overweight and even, like SMS patients, developed metabolic syndrome [74]. However, one of the most characteristic features of SMS is the craniofacial gestalt, or dysmorphology, allowing a recognizable pattern of human malformation [75]. Three-dimensional imaging and computational facial recognition analyses can readily discern the SMS phenotype [76]. The engineered mouse models recapitulate the craniofacial dysmorphology with a more fully penetrant phenotype in the chromosome deletion Df (11) animal versus the Rai1 knockout (KO) allele [77,78]; these data speak to penetrance and gene dosage. Could transvection be involved?

These mouse chromosome engineering data show that not all null alleles have the same effect in whole animals; Rai1 KO are less penetrant than chromosome deletions that delete both Rai1 and surrounding regulatory genomic intervals, Df(11). Differently sized chromosome deletion CNVs have measurable differences in penetrance [77]. These latter genomic deletion studies implicate communication between the Rai1 loci on homologous chromosomes, a genetic phenomenon known as transvection. Such observations suggest that, like the compound inheritance gene dosage model (CIGD) for disease, subtle changes in gene expression and dosage can affect penetrance, as has been found in human and mouse CAKUT and operable congenital scoliosis [10,79,80].

Even more exciting findings from these genetic studies involving engineered mouse chromosomes were results showing that gene × environment (G × E) effects could be observed and studied (Figure 6E). Wt animals fed regular chow established the normal weight and growth parameters. The PTLS Dp(11) mice did not gain weight when fed high-fat (HF) chow, whereas their wt littermate controls became morbidly obese on an HF diet and developed metabolic syndrome [74]. The duplication CNV in the Dp(11) strain was fed HF chow to test G × E effects. The duplication allele conveyed a trait of protection from weight gain and obesity, regardless of genetic background (Figure 6E).

Moreover, neurobehavioral traits could be investigated for the Dp(11) and Df(11) animals. Data were accumulating showing ASD and schizophrenia (SCZ) were mirror traits. In the year 2000, the PTLS duplication CNV was associated with ASD [54]. ASD is also associated with 16p11.2 deletion CNV, 1q21.1 duplication CNV, and macrocephaly, whilst SCZ is associated with 16p11.2 duplication CNV, 1q21 deletion CNV, and microcephaly. Macrocephaly and microcephaly are mirror traits [9]. Occipitofrontal head circumference is a quantitative trait and a domain of unknown function (DUF) that is important to brain size expansion during primate speciation maps to 1q21.1; deletion of the DUF1220 domain, the copy number of which is implicated in human brain size pathology and evolution, results in the phenotypic extreme of microcephaly [22,23,81].

For the Dp(11) animals, one could also rescue the neurobehavioral assays that suggested an ASD-like trait in the PTLS mouse model, analogous to that observed in patients, by rearing Dp(11) mice in an enriched environment; environmental manipulation could affect disease trait expression [39]. These latter studies provided independent experimental proof-of-concept and compelling data that infant stimulation can assist children with developmental delay or intellectual disability and other NDs.

Gene dysfunction, dosage, and mitigating disease by molecular therapy (Rx)

Confirmation of the gene dosage hypothesis (Figures 3–5; Figure S1 in the supplemental information online), and its underlying premises for molecular therapeutic intervention, have been demonstrated experimentally in the rat animal model for the CMT DSP disease associated with the CMT1A/PMP22 duplication [82,83] (Figure 5). Whereas the gene dosage mechanism has generalized the concept to other CNV-mediated disease, point mutation and allelic series, particularly of the PMP22 and myelin protein zero genes, PMP22 and MPZ, and the mouse Pmp22 Tr and Tr^J alleles (Figure 5C,D), have provided insight into allele gene dysfunction for LoF (i.e., LoF: null, hypomorph, antimorph) and GoF (i.e., GoF neomorph, hypermorph) mutations. Studies of such mutant alleles for PMP22 (Figure 5C) provided evidence that CMT1A duplication disease is a triplosensitive trait, and HNPP deletion-associated disease causes a haploinsufficiency susceptibility trait [84], resulting in a nerve pathology of tomaculous neuropathy [85], a sausage-like swelling of the myelin, seemingly due to separation of myelin layers (Figure 7A,B).

Figure 7. Myelin biophysics, glial–nerve cell neurobiology, cell biology of neuron function, and peripheral nervous system therapy.

Misfolded proteins, cellular apoptosis, and curcumin-based Rx for distal symmetric polyneuropathy (DSP). (A) P₀ protein, MPZ, structural studies by Shapiro et al. reveal a tetramer structure that interdigitates between tetramers on opposing membrane interfaces at the myelin intraperiod line: a colloidal ‘glue’, more like molecular ‘velcro’, holds together the myelin layers, as illustrated by the Lemke News & Views [114]. Quantitative apoptosis assay by annexinV/popidium iodide (PI) flow cytometry sorting documents a cellular phenotype of an increased cellular apoptosis rate for misfolding mutations that are associated with more severe DSP. (B) Curcumin molecular therapy mitigates and reverses the increased apoptosis rate for MPZ misfolding mutations. (C) Curcumin reverses the increase rates for PMP22 point mutation. Note, PMP22 wild-type control showing subtle apoptosis rate increases in comparison with empty vector control, consistent with a gene dosage affect and the adult onset of disease for the CMT1A DSP, autosomal dominant disease trait. (D) The PMP22 point mutation animal model. Pmp22 Tr^J clinical phenotype is mitigated and restored to biological homeostasis with dietary supplementation with curcumin in a dose-dependent manner, that is abrogated by removal of curcumin. Could simple dietary supplementation assist patients with misfolding mutants? Whole-animal model organism studies demonstrate oral curcumin (i) penetrates peripheral nerve, (ii) reduces Schwann cell apoptosis, and (iii) mitigates clinical consequences of Tr^J mutation in a (iv) dose-dependent specific manner, that is (v) reversed upon withdrawal of molecular therapy. See [91,95,96,114,123].

From this perspective, the HNPP deletion [86] acts through a G × E effect (G × E: PMP22 gene dosage and lifetime accumulated nerve injury/damage) that can result in multifocal neuropathy [87] and carpal tunnel syndrome [88]. It will remain intriguing to explore all mutation types (SNV, indel, CNV) and allelic types (the null, hypomorph, antimorph, neomorph, and hypermorphs of Muller) [89] and their effects on rare disease traits for each human gene.

The incredible observation by Saher and colleagues [90] on myelin membrane biological homeostasis, and correction of a mouse gene dosage perturbation and myelinopathy disease by simple dietary cholesterol supplementation, warrants further exploration of myelin biology. These investigators showed correction of the CNS demyelinating disorder Pelizaeus-Merzbacher disease (PMD; MIM: 312080), an XL disease trait most often caused by a duplication and a PLP1 gene dosage effect, through simple membrane biology manipulations via diet [90]. In this PMD animal model case and cholesterol supplementation study, the dietary manipulation (adding cholesterol to the diet) provided benefit to the animal. Perhaps other myelin disorders can also be mitigated or corrected through dietary changes. Could feeding fatty hamburgers and stopping statins help alleviate some neurological and neuromuscular signs and symptoms in selected individuals?

MPZ studies have provided further insights into disease pathobiology and pathogenetic mechanisms (Figure 5D). Of note, at two different codon/protein amino acid positions, Ser63 and Arg98, a change to Cys causes a more severe disease pathology than other amino acid changes involving the same codon; changes to Cys would be predicted to perturb protein tertiary structure and potentially disrupt proper protein folding (Figure 5D). Animal model studies reveal disease pathology at the cellular phenotype level results from induction of the unfolded protein response (UPR) [91,92]. Gene dosage studies in MPZ show gene amplification can cause severe DSP ([93] and observations with a 5× amplification of MPZ [94]). Moreover, while some SNV point mutations cause no disease in parental carriers, homozygous probands have a severe AR DSP trait, consistent with a gene dosage effect (Figure 5D [64]). Such studies clearly show MPZ is a dosage-sensitive gene.

Cellular phenotype studies reveal MPZ and PM22 point mutations associated with more severe DSP disease cause a larger percentage of induced apoptosis in cell populations by quantitative annexin V/propidium iodide flow cytometric assays in comparison with wt protein and empty vector controls (Figure 7) [95,96]. They also show such variant alleles cause mutant proteins to reduce transit through the endoplasmic reticulum (ER) and induce UPR. The increase in cellular apoptosis rate can be mitigated and reversed by curcumin supplementation, seemingly acting through chaperone induction to relieve ‘cellular constipation’ in the ER of glial Schwann cells. Whole-animal studies of the PMP22 point mutation CMT DSP disease model, Trembler-J (Tr^J), with curcumin provide a measurable organismal clinical response in a dose-dependent manner; clinical response is abrogated by removal of curcumin molecular therapy (Figure 7F). These observations (Figure 7) encourage the use of a general strategy for molecular therapy of misfolding mutants, and clinical trials for curcumin supplementation, in misfolding mutations involving other gene loci.

Looking towards the future: staring into my crystal ball

Patient databases such as DECIPHERⁱ [97] and ClinVarⁱⁱ will continue to catalogue human variation and remain rich sources of human allelic variation; DECIPHER integrates both CNV and SNV with human genome resources. The understanding of evolution and copy number [9,10,22,23] of (i) low copy repeats in the human genome, (ii) genes and gene function, (iii) paralogous genes and gene families, (iv) pseudogenes, and (v) even protein domains will continue to thrive with the applications of human genetics and genomics in medicine. As will our understanding of chromosome, genome, gene, and protein domain mutations and gene paralog function, genome mutation, mutation burden, MPV, and CIGD models. Moreover, further understanding of human gene and genome architecture may have implications for gene conversions, recombination/repair mechanisms such as NAHR (DSBR) and FoSTeS/MMBIR (oeDNA repair; SSBR or DSBR?), and CRISPR/Cas9 therapeutic interventions of human disease traits [98].

Furthermore, human genome architecture and SV mutagenesis new mutation studies (Figures S2 and S3 in the supplemental information online) may contribute, as CNV deletion alleles, to AR rare disease trait gene identification, understanding dosage-sensitive genomic disorder ‘driver genes’, and molecular diagnoses in clinical genomics [99]. These new mutation studies could also provide insights and influence: prenatal diagnostics, allele frequency distributions in populations, and human developmental genomics (Figure S3 in the supplemental information online). Understanding mutagenesis mechanisms, human genome mutation rates, and the roles of new mutation versus inherited variant alleles will continue to improve our understanding of the genetic influences in rare disease traits and somatic mutagenesis events that might result in a cancer. Such studies may also reveal both developmental selection and organismal fitness from an evolutionary perspective [99].

Reverse genetic and genomics approaches to defining disease and disease biology, including dissecting paralogous gene function in organismal traits, may become the ‘new norm’ for disease description; moreover, molecular diagnosis by personal genome clinical genomics (ES, WGS) may become a first-line clinical test, even before establishing a differential clinical diagnosis. The gene will play a central role in science and medicine; it is the ‘currency’ of genetics. Allelic series will help to define rare disease traits and the genes and gene variation-causing disease traits, as exemplified by and documented for [100] White-Sutton syndrome (WHSUS; MIM:6163640), something that disease cohort studies could not do. Allelic series may also (i) provide insights into genetic heterogeneity; (ii) delineate signaling pathways; (iii) elucidate paralog functions; and (iv) in combination with quantitative phenotyping and computational analyses of OMIM disease trait clinical synopses, illuminate the biology of disease and blended phenotypes [101–103,124]. Hypomorphic regulatory mutation alleles will need integration with clinical genomic studies. Could quantitative trait loci (QTL) and gene dosage expression be the way in? Perhaps Calvin Bridges was asking the right question all along!

Future genome-wide association studies (GWAS) will have to adjust for allele counts [104], to deal better with AR versus AD disease traits, and perhaps phase new mutation variants into structural haplotypes. Future genome AD disease trait studies, genomics in disease cohorts, genomics in ‘neurotypical’ populations, and genomics in clinically selected populations, clinical genomics in all populations of the world, and molecular therapeutic intervention studies might all benefit from better understanding of gene dosage effects. Gene dosage, at both the gene copy number and cellular transcriptional change levels, as well as by (i) mosaicism, (ii) Lyonization, (iii) perhaps transvection effects, and (iv) manifestation of penetrance for an XL disease trait (H. Hijazi et al., unpublished), can all effect, and be affected by, gene action and interaction. Neurodevelopmental delay and ND [105,106], such as ASD and SCZ, will continue to be better categorized through genomic medicine, and therapeutic strategies for intervention developed; genetics is the way into the neurobiology of these diseases.

Studies of admixed populations and populations with elevated coefficients of consanguinity will also illuminate the genetics and genomics of disease and the roles of new mutations, MPV, and inherited alleles in disease. Family recruitment and genomic studies by research ES showed evidence for a 79.8% solved rate in NDD in the admixed Turkish population [40]. These NDD studies also (i) extend the ‘one disease multiple contributory genes genomic hypothesis’ [103] to distributive runs of homozygosity/absence of heterozygosity (ROH/AOH) in admixed populations and (ii) extend the clan genomics hypothesis to the idea of ‘de novo copy number neutral SV mutagenesis’ (i.e., haplotype reconstruction) and MPV to AR + AR disease trait blended phenotypes in admixed populations.

Biallelic variation, AR disease trait inheritance, and combinations of alleles at a locus underlie the CIGD model of Wu and Zhang, resulting from a rare variant coding SNV and relatively common ‘small mutation effect size’ noncoding variant alleles [107]. Gene dosage may underlie a substantial fraction of birth defects and the developmental genomics of birth defects of the spine [107], kidney (CAKUT and unilateral renal agenesis) [79], lung [108], and limbs (R. Duan et al., unpublished). Approaches that better recognize gene dosage and gene action may align with the clan genomics hypothesis [9,10] and perhaps increase further the utility of population variation, as catalogued in gnomAD and other public databases (Figure S3 in the supplemental information online and [115]).

The Puerto Rican founder allele and, specifically, the COL27A1 founder in a geographic isolation bottleneck and Steel syndrome on the island of Puerto Rico, was informative for trait penetrance. The genetic and genomic data from the USA and in populations from countries mapping to multiple continents lend further global support to the clan genomics hypothesis. Exploring the clan genomics hypothesis worldwide may provide insights into human biology, including insights applicable to clinical genomics and disease (Figure 8) [9–11]. Moreover, gene dosage effects potentially can be mitigated by molecular therapies [82,91,95,96], including nucleic acid pharmacotherapeutics like those utilized to deliver COVID-19/20/21/22? vaccines to the world.

Figure 8. Searching for rare disease trait alleles around the globe; family-based genomics.

Clan genomics and the practice of genome medicine worldwide. Studies of a skeletal dysplasia, Steel syndrome (STLS; MIM#: 615155), identified the COL27A Founder allele in Houston, TX, amongst a family (two individuals with trait) of Puerto Rican descent and homozygotes due to identity-by-state (IBS). In other regions of the globe, homozygous alleles occur by identity-by-descent (IBD). Notably, evidence exists for carrier state alleles (i.e., heterozygous variant alleles) and susceptibility to orthopedic complaints and potentially short stature [11]. Abbreviations: BHCMG, Baylor-Hopkins Center for Mendelian Genomics; EMR, electronic medical record; ES, exome sequencing.

Highlights.

Both the human genetic and genomics sciences will play an increasing role in genomic medicine practice.

Disease represents perturbations from biological homeostasis.

Biological balance, complementarity, and gene dosage are fundamental principles of biology, human organismal developmental biology, and rare disease traits.

Molecular diagnostic and therapeutic intervention demand better understanding of a locus gene dosage effects.

Glossary

AAMR

Alu-Alu-mediated rearrangement, a specific form of microhomologous/microhomoeologous recombination between repetitive DNA sequence elements in the human genome

Aneusomy

a specific chromosome deviates from the euploid/diploid state

Chromoanasynthesis

massive genomic rearrangement of a chromosome and an apparent rebirth or resynthesis of a damaged chromosome with a single chromosome-focused large number of new structural variation, including copy number gains and inversions, and can include complex genomic rearrangements (CGRs) such as DUP-TRP-DUP

Chromothripsis

single chromosome shattering event thought to be repaired potentially by non-homologous end joining (NHEJ). Typically, a copy number neutral event or with multiple genomic interval inversions and some deletions all present on one chromosome

CIGD

the genetic essence of the compound inheritance gene dosage model of Nan Wu (PUMC, Beijing, China) and Feng Zhang (Fudan University, Shanghai, China) is the pairing of a small-effect allele (e.g., hypomorph) with a null. This pairing of hypomorph (can be a noncoding variant such as a GWAS signal or eQTL variant, or coding variant with higher population frequency than a typical LoF allele) with a large-effect allele (e. g., null) in the zygote that can effectively result in gene expression below that observed with haploinsufficiency and affect gene action in terms of the penetrance of the disease trait. The genetic ‘key’ is that homozygosity for such a small-effect allele has no trait penetrance; thus, it does not affect Hardy-Weinberg equilibrium and will not be selected against in the organismal population. The combination of alleles is the genetic essence of the concept, a concept that in some ways can be thought of as a biallelic AR trait in the Gregor Mendel sense

Clan genomics

posits that genome variation and human disease traits result from rare variants (SNV + CNV) with large effect that have arisen recently in the family/clan/population history. Therefore, new mutations in you and (such as copy number neutral SV with identity-by-descent) in your recent ancestors, and novel combinations of variant alleles from your parents, account for many medically actionable variant loci

Complex genomic rearrangement (CGR)

sometimes referred to as Gross Chromosomal Rearrangements (GCR) in yeast wherein pulsed-field gel electrophoresis (PFGE) may have enabled resolution of structure. In humans, CCR (Complex Chromosomal Rearrangements) refers to a cytogenetically visible chromosomal rearrangement in which one can observe evidence for more than 2 different chromosomes involved

Copy number variant (CNV)

deviation, of >100 W-C bp, from the diploid genome content at a specific map position

Developmental genomics

multigene/locus phenomena that may come about because of specific cell population or tissue interactions of developmental processes, such as signal transduction gradients, tissue inductive interactions, organogenesis, and stochastic fluctuations in gene dosage expression during cell migration, the latter perhaps particularly relevant to cells migrating to the distal aspects of the body axis planes (e.g., distal-proximal developmental plane of limb development)

DSBR

double-strand break repair, a homologous recombination (HR) pathway/mechanism

DSP

distal symmetric polyneuropathy

FoSTeS

fork stalling template switching, a DNA replicative recombination repair mechanism that biochemically processes one-ended, double-stranded, DNA (oeDNA)

Genome mutation

mutation of the genome that can involve more than one gene or locus

Genomic disorders

conditions that result from genomic rearrangements, rather than W-C DNA sequence bp changes, and in which a specific local genomic architecture (direct or inverted low copy repeats or repetitive elements) can incite genomic instability and perturb genome integrity and chromosome structure

Genomic medicine

the use of personal genome variant data: to enable a differential diagnosis, make a molecular diagnosis(es), identify new mutations, assist patient and family clinical management, direct or implement therapeutic gene/variant-based oligonucleotide therapy approaches in prenatal testing, prepregnancy family planning, newborn screening, population health, and to molecularly characterize the molecular pathology of tumor or other surgical specimen (such as for chromothripsis or chromoanasynthesis)

Indel

mutation of genomic sequence characterized by insertion or deletion of <100 W-C base pairs

dnCNV

multiple de novo CNV, a perizygotic organismal mutagenesis phenomenon

MMBIR

microhomology mediated break induced replication; a form of DNA replicative recombination/repair

Multilocus pathogenic variation (MPV)

variants at more than one genetic locus that are associated with pathology

Mutation burden

a load or burden of variant alleles (i.e., mutations) in genes whose proteins act in a functional biological unit or interactome, biochemical pathway, transcriptional unit or network, developmental pathway, functional unit (e.g., neuron/glial support cells), or organ system

Nonallelic homologous recombination (NAHR)

utilizes paralogous sequence substrates for W-C base pairing and leads to genomic duplications, deletions, and inversions

ND

neurobehavioral differences

NDD

neurodevelopmental disorders

oeDNA

one-ended, double-stranded, DNA; results at collapsed forks after replication through a nick or single-strand break; also can result at telomere ends of chromosomes that are not healed by telomerase or in telomerase-negative mutants, such as observed in Caenorhabditis elegans

Polarity

genetic polarity phenomena are those which result from the sequential nature of the functioning of the genetic material. Functioning of the genetic material can be considered at two levels: the sequential replication of the chromosome and the expression (transcription and translation) of the gene

Segmental aneusomy

a specific segment of a single chromosome is deleted or duplicated; if found in a group of patients that share a clinical synopsis, can then be referred to as a microdeletion or microduplication syndrome, many of which are genomic disorders

SNV

single-nucleotide variant of the DNA double helix

SSBR

DNA single-strand break repair

Transvection

a genomic process of chromosome homolog interaction; it is mediated by the zeste gene in Drosophila