*Review* **NADomics: Measuring NAD+ and Related Metabolites Using Liquid Chromatography Mass Spectrometry**

**Nady Braidy 1,2,\*, Maria D. Villalva <sup>1</sup> and Ross Grant 3,4**


**Abstract:** Nicotinamide adenine dinucleotide (NAD+) and its metabolome (NADome) play important roles in preserving cellular homeostasis. Altered levels of the NADome may represent a likely indicator of poor metabolic function. Accurate measurement of the NADome is crucial for biochemical research and developing interventions for ageing and neurodegenerative diseases. In this mini review, traditional methods used to quantify various metabolites in the NADome are discussed. Owing to the auto-oxidation properties of most pyridine nucleotides and their differential chemical stability in various biological matrices, accurate assessment of the concentrations of the NADome is an analytical challenge. Recent liquid chromatography mass spectrometry (LC-MS) techniques which overcome some of these technical challenges for quantitative assessment of the NADome in the blood, CSF, and urine are described. Specialised HPLC-UV, NMR, capillary zone electrophoresis, or colorimetric enzymatic assays are inexpensive and readily available in most laboratories but lack the required specificity and sensitivity for quantification of human biological samples. LC-MS represents an alternative means of quantifying the concentrations of the NADome in clinically relevant biological specimens after careful consideration of analyte extraction procedures, selection of internal standards, analyte stability, and LC assays. LC-MS represents a rapid, robust, simple, and reliable assay for the measurement of the NADome between control and test samples, and for identifying biological correlations between the NADome and various biochemical processes and testing the efficacy of strategies aimed at raising NAD+ levels during physiological ageing and disease states.

**Keywords:** NAD+; nicotinamide; ageing; plasma; biomarker

#### **1. Introducing NADomics as a Tool for Quantification of the NADome in Biological Samples**

NADomics is the high-throughput study of nicotinamide adenine dinucleotide (NAD+) and its related metabolites. NAD+ is an essential coenzyme that is present in all organisms [1]. NAD<sup>+</sup> serves as a major coenzyme for enzymatic reduction–oxidation reactions and ATP production. More recently, NAD+ has also been shown to be a crucial co-substrate for numerous enzymes (i.e., sirtuins, NAD<sup>+</sup> glycohydrolase (CD38), poly(adenosine diphosphate–ribose) polymerases (PARPs)) [2–4]. The term NADomics is an analogy to metabolomics, the study of the metabolome. The word NADome is a portmanteau of NAD<sup>+</sup> and its related metabolome. The NADome is the entire set of NAD+ metabolites that are anabolised or catabolised by an organism or system (Figure 1). The emerging field of NADomics has enabled the identification and quantification of ever-increasing numbers of NAD-related metabolites.

**Citation:** Braidy, N.; Villalva, M.D.; Grant, R. NADomics: Measuring NAD<sup>+</sup> and Related Metabolites Using Liquid Chromatography Mass Spectrometry. *Life* **2021**, *11*, 512. https://doi.org/10.3390/life11060512

Academic Editors: Chiara Villa and Jong-Hyuk Yoon

Received: 11 April 2021 Accepted: 24 May 2021 Published: 31 May 2021


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** NAD+ metabolism in eukaryotic cells. NAD+ anabolism from tryptophan occurs by the de novo kynurenine pathway (KP). NAD+ precursors via the salvage pathway include nicotinamide (NAM), nicotinic acid (NA), nicotinamide riboside (NR), and nicotinic acid riboside (NAR). The enzyme nicotinamide phosphoribosyltransferase (NAMPT) converts NAM to nicotinamide mononucleotide (NMN). Nicotinamide mononucleotide adenylyltransferase (NMNAT1-3) converts NMN to NAD+. NAM can be methylated to N-methyl-nicotinamide (MeNAM) by the action of nicotinamide N-methyltransferase (NNMT). NADH represents the reduced form of NAD+. NADP+ is the phosphorylated form of NAD+. NADP+ can be reduced to NADPH by NAD kinases (NADK1,2). PARPs, Sirtuins, and CD38 NAD+ glycohydrolases are known as NAD+ consumers, leading to the generation of NAM. Nicotinic acid phosphoribosyltransferase (NAPRT) converts NA to nicotinic acid mononucleotide (NAMN), which is then converted to NAD+ by NMNAT1-3. NAR needs to be converted to NAMN to yield NAD+ synthesis via nicotinamide riboside kinases (NRK1,2). NRK1,2 also convert NR to NMN. NAR can form NA via purine nucleoside phosphorylase (PNP). PNP are also capable of converting NR to NAM.

NAD+ and its reduced form NADH can be phosphorylated to NADP+ and NADPH, which serve as major coenzymes in over 400 oxidoreductase enzymes [5]. NAD+ serves as a hydrogen acceptor allowing the transfer of electrons for oxidation–reduction (i.e., redox) reactions leading to ATP production in the mitochondria [1]. NAD+ glycohydrolases (CD38, CD157) are involved in the production of calcium-mobilising messengers, ADPribose (ADPR) and its cyclic form (cADPR) [6]. PARP-mediated ADP-ribosylation uses the ADPR moiety of NAD+ to repair DNA, leading to the breakdown NAD<sup>+</sup> to nicotinamide (NAM) and an ADP-ribosyl product [7]. Sirtuins are a family of class III NAD+-dependent histone deacetylases that exhibit protein lysine deacetylase, and partial ADPR transferase activities [8]. Deacetylation occurs when the modified lysine side chain is coupled to the cleavage of the glycosidic bonds in NAD+, leading to the generation of the deacetylated lysine, acetylated ADPR, and NAM as by-products [9]. These processes are dependent on NAD+ availability, and NAM is an endogenous inhibitor of CD38, PARP, and sirtuins [5].

Continuous replenishment of cellular NAD+ levels is important for normal cellular survival [10]. The de novo NAD+ biosynthesis pathway in most cells is dependent on the amino acid tryptophan via the kynurenine pathway. When the availability of dietary tryptophan is limited, the efficiency of the conversion of tryptophan to NAD<sup>+</sup> decreases below the well-established conversion ratio of 60:1 [11,12]. Nicotinic acid (NA), NAM, NAM riboside (NR), or NA riboside (NAR) can also be used to synthesise NAD+ via the NAD<sup>+</sup> salvage pathway [5]. NAM can be methylated to N-methyl-nicotinamide (MeNAM) by the action of NAM N-methyltransferase [13]. The enzymes nicotinamide phosphoribosyltransferase (NAMPT) and nicotinic acid phosphoribosyltransferase (NAPRT) convert NAM and NA to nicotinamide mononucleotide (NMN) and nicotinic acid mononucleotide (NAMN), respectively [5]. Additionally, the phosphorylation of nicotinamide riboside (NR) and nicotinic acid riboside (NAR) via nicotinamide riboside kinase (NRK) also leads

to the production of NMN and NAMN, which can be converted to nicotinic acid adenine dinucleotide (NAAD) and NAD<sup>+</sup> by nicotinamide mononucleotide adenyltransferase (NMNAT) [14]. NAAD can be amidated to NAD+ by NAD synthetase (NADS) [5].

The NADome is important in physiological processes such as energy production, transcriptional regulation, DNA repair, protein modification, and secondary messenger signalling [5]. Therefore, the application of NADomics in the clinic provides an essential indicator of nutritional status, redox function, and incidence and progression of age-related diseases [15]. A decline in cellular NAD+ levels has been associated with mitochondrial impairments, immune dysfunction, and reduced histone deacetylase activity, which can interfere with several transcription factors and affect gene expression levels [5]. Reduced NAD<sup>+</sup> levels also have a dramatic impact on the activity of PARPs, thus impairing DNA repairing. We and others have demonstrated that intracellular NAD+ levels decline in conditions of metabolic stress in muscle, brain, heart [2], lung [2], liver [2], kidney [2], skin [16], and plasma [17,18] in humans and rats. NAD<sup>+</sup> levels are reduced in tissues and cells exposed to oxidative stress and DNA damage, the overfed liver, the failing heart, the damaged peripheral neuron, and the injured brain [5], and correlate with disease severity of multiple sclerosis [19]. However, it remains unclear whether a depressed NADome is a function of age, although ageing is a major risk factor for the accumulation of metabolic stress.

Promotion of cellular NAD<sup>+</sup> anabolism has been shown to restore NAD<sup>+</sup> levels and reverse some phenotypes of ageing by enhancing cellular repair and stress resistance. Recent studies have shown that administration of the NAD<sup>+</sup> precursors, nicotinamide riboside (NR) and nicotinamide mononucleotide (NMN), can attenuate pathology in several murine models of age-related disorders [20]. Additionally, oral administration of NR has been reported to increase whole blood NAD<sup>+</sup> levels in humans [21]. A 6-week randomised, double-blind, placebo-controlled crossover clinical trial reported that NR supplementation was well-tolerated and effectively promoted NAD<sup>+</sup> metabolism in healthy middle-aged and elderly adults [22]. However, another randomised, placebo-controlled, double-blinded trial showed that NR supplementation had no effect on insulin sensitivity and lipid mobilising effects, and no adverse events due to NR were reported [23]. We also demonstrated that intravenous (I.V) injection of NAD<sup>+</sup> increased plasma and urine NAD<sup>+</sup> levels and exhibited differential effects on the NADome in healthy middle-aged humans. Thus, while it can be argued that there are several strategies to increase NAD<sup>+</sup> levels, understanding the endogenous intracellular and extracellular levels of the NADome are of emerging interest.

#### **2. Clinical Relevance for Detection of the NADome in Biological Fluids**

The NADome is an essential mediator of metabolic pathways that have been associated with ageing and age-related disorders. Several published methods have identified the NADome in whole blood and various cells or tissues, however, very few studies have examined changes in the NADome extracellularly [17,18,24]. A previous study reported that NAD<sup>+</sup> is predominantly an intracellular nucleotide, and only phosphorylated metabolites could be detected using LC-MS/MS [21]. However, emerging evidence suggests that exogenous NAD<sup>+</sup> may cross the plasma membrane and replenish intracellular NAD+ levels in mammalian cells [25,26]. The discrepancy is likely to be related to the differences in the LC-MS/MS methodology and sample extraction. For example, Trammell et al. [27] used a double extraction method and two different chromatographic runs for reduced and oxidised forms of adenine and pyridine nucleotides, extending the analysis time per sample.

Owing to the clinical significance of increasing NAD<sup>+</sup> levels, and the role of NAD<sup>+</sup> in several age-related degenerative diseases, analytical methods for quantification of the NADome can be applied to clinically relevant biological samples, including whole human blood, urine, and CSF. This is because: (1) NAD<sup>+</sup> is released from cells at low amounts, (2) NAD<sup>+</sup> catabolism is an immediate process that leads to production of biologically active products, including NAM and consequently, MeNAM, (3) NAD+ can act directly

on cell surface receptors such as connexin 43 channels and purinergic P2 receptors and therefore must be present extracellularly (Figure 2) [28], (4) MeNAM is the main metabolite of pyridine nucleotide catabolism and is cleared by urinary excretion [29], and (5) there is a lack of NAD-consuming enzymes in the urine compared to plasma.

**Figure 2.** Schematic representation of the role of NAD+ in purinergic signalling. Extracellular NAD+ and ATP are released from damaged cells. ATP binds to the ATP-sensitive P2X7 receptor of monocytic cells. Activation of inflammasomes and caspase-1 induces cleavage of pro-IL-1β and release of bioactive IL-1β. NAD<sup>+</sup> binds to P2Y receptors and activates iPLA2β, leading to the production and release of bioactive mediators which serve as nicotinic agonists.

> NAD<sup>+</sup> and related metabolites present in the blood will be filtered at the glomerulous and will be present in at least the renal filtrate of the proximal tubule. Recent studies have shown that CD38-mediated cADPR production in renal arteries and the distal tubule is necessary for Ca2+-mediated regulation of renal function [30]. Thus, the presence of CD38, as a NAD-dependent ectoenzyme in the distal renal tubule, and which its activity is critical for normal renal function and dependent on the availability of NAD<sup>+</sup> as a substrate, strongly supports the view that intact NAD<sup>+</sup> is still present in the renal filtrate after most of the tubular reabsorption has occurred.

> A unique element of the central nervous system (CNS) is its high oxygen consumption and energy requirements relative to size (i.e., 2% of body weight and uses 20% of oxygen). This high energy demand is vital to sustaining the complex metabolic activities of the CNS and is achieved through accelerated mitochondrial activity using readily available NAD+ in its redox couple. Importantly, elevated mitochondrial activity also results in significant superoxide production and nuclear damage [31], necessitating DNA repair through PARP activity, a NAD<sup>+</sup> catabolising process. In addition to its role in intracellular metabolism, extracellular NAD<sup>+</sup> also exerts direct synaptic effects, reducing synaptic excitotoxicity [32], and may also act as a neurotransmitter [33]. In order to act at the synapse, extracellular NAD+ is clearly present in the CNS, as recently reported [33].

NAD<sup>+</sup> is also a central player in the maintenance and control of biological rhythms coordinated through the 20,000 pacemaker neurons in the suprachiasmatic nucleus in the CNS, where NAMPT/NAD+ drives the circadian clock feedback cycle through SIRT1 and CLOCK:BMAL1 [34]. Numerous NADPH-diaphorase (i.e., nitric oxide-producing) neurites are present on the free surface of the ependymal layer in direct contact with the cerebrospinal fluid (CSF) [35], thus suggesting that NADPH must also be present in the CSF to engage with this enzyme. Additionally, nicotinamide N-methyltransferase is present in the CSF and actively converts CSF NAM into its N-methylated metabolite [36], further highlighting both the presence and potential importance of NAM for the CNS.

#### **3. Analysis of the NADome Using Traditional Techniques**

Several methodologies have been previously used to quantify the NADome. These include specialised HPLC-UV, NMR [37], capillary zone electrophoresis, or colorimetric enzymatic assays [38]. While commonly available biochemical assays are readily available and relatively inexpensive, they provide indirect measurements and require tedious enzymatic manipulation [39,40], limiting their use in the clinic. Moreover, these assays are only available for NAD+, NADH, and ATP, and thus are unable to provide an accurate reflection of changes in other metabolites in the NADome, such as NADP+ and NADPH. For instance, the intracellular levels of NAD<sup>+</sup> have been shown to vary between 1 μM to 1 mM [16]. In addition, the ratio of NADPH:NADP+ was reported to be ~100 using an enzymatic assay which measures the substrate concentration of malic and isocitrate dehydrogenase enzymes [41]. However, direct measurement of these metabolites suggests that the ratio of NADPH:NADP+ may be significantly lower, i.e., 3.3 and 0.04 in the rat liver and heart [42]. Although multiple values are missing in the current literature and biological variability may indeed play a role, it is more likely that analytical variation represents a major contributory factor to the reported differences.

NMR-based experimental approaches have been developed to quantify the NADome in human cells. 1H NMR spectroscopy has been previously used to quantify the NADome in human platelets and erythrocytes [43]. While the relative concentration can be quantified directly from the NMR spectrum for each metabolite, irrespective of the complexity of the sample, this approach may require up to 2 h of data acquisition to achieve a signal-to-noise ratio of at least 5 for a sample containing approximately 2 μM of metabolite [43]. This can limit the application of NMR for the quantification of the NADome in some biological specimens, where some metabolites may be present in very low nanomolar levels [15].

Liquid chromatography (LC) techniques using absorbance for detection provide some advantages for quantifying multiple metabolites in the NADome relatively fast, i.e., within 10 to 60 min. However, HPLC-UV methods are limited due to low sensitivity and the presence of co-eluting contaminants [16]. For example, in a complex biological sample, a single peak may represent the metabolite of interest and other related metabolites that share identical retention times. Some studies have used mass spectrometry to non-quantitatively confirm the nature of the metabolite(s) present in any given fraction [16]. However, this strategy is time-consuming and costly. Since many metabolites in the NADome can be converted into other metabolites either by oxidation/reduction or enzymatic reactions, accurate quantification of NAD+, NMN, NAM, and other metabolites without examination of the entire NADome may be misleading.

#### **4. Quantification of the NADome Using Liquid Chromatography-Mass Spectrometry**

LC coupled to tandem mass spectrometry (LC-MS/MS) has been recently developed for the quantitation of the NADome in biological specimens (Figure 3) [16]. In line with HPLC assays, LC allows the separation of individual metabolites and must be optimised in a similar manner to HPLC methods. The data collected from LC-MS is two-dimensional, i.e., retention time and mass-charge ratio, thus increasing specificity and sensitivity and allowing for the separation of closely related metabolites such as NAD<sup>+</sup> and NADH. Additionally, most LC-MS assays have a limit of quantification in the femtomole range [16].

**Figure 3.** NADomics workflow. The NADomics workflow involves profiling the NADome with statistically significant variations in biological samples, e.g., blood, urine, and CSF. The specific NAD+ metabolite ID including chemical structure and concentrations can be elucidated using LC-MS/MS. Analysis is the final step to elucidate associations between the identified metabolite and its role in physiology and disease.

Hydrophilic interaction liquid chromatography (HILIC) is an emerging separation mode of LC that suits well for the quantification of the NADome. The variant uses polar columns with a stationary phase, whereby polar analytes are eluted from the column by increasing water content of the mobile phase (typically acetonitrile with low amounts of water). HILIC also allows for hydrogen donor interactions between neutral polar species, and weak electrostatic mechanisms under high organic solvent conditions [44]. The retention of the analytes, peak shape, and chromatographic tailing is regulated by the pH of the mobile phase and ion strength attributed to ionic additives such as ammonium acetate and ammonium formate. HILIC allows for a high flow rate due to very low column backpressure by the high organic mobile phase [44].

HILIC separation has been previously used for the quantification of AMP, GMP, UMP, CMP, and IMP in infant formula [45,46]. It has also been adapted for quantitative analysis of cAMP, ATP, and other nucleosides, and mono-, di-, and tri-phosphate nucleotides, thus allowing for the simultaneous analysis of a large number of metabolites in a single run [45,46]. A volatile additive in mobile phase enables smooth hyphenation with mass spectrometry detection and has recently been optimised for the detection of at least 17 different metabolites in the NADome in astrocytes and oocytes [47].

#### **5. Challenges of NADomics in Biological Specimens**

Accurate quantification of the NADome is crucial for evaluating the cellular redox status. In this section, we discuss potential challenges and solutions that have affected previous methods.

#### *5.1. Extraction*

Extraction of the NADome is a major source of analytical variation. Immediate extraction of the NADome in biological samples is ideal. Extraction methods which fail to

inactivate enzyme activities following cell lysis can limit the accurate quantitation of the NADome. For example, the levels of NAD+ in a biological sample can be degraded to 1% of the anticipated value, while the levels of NAM can increase more than 10-fold. The most common method of extraction for most NAD+ metabolites is boiled buffered ethanol [16]. We previously demonstrated that ice-cold 80% methanol was suitable for the extraction and maintaining molecular integrity of the NADome in murine oocytes [47].

A recent study compared the quenching and extraction efficiency of 7 different solvents on the NADome in mammalian cells and mouse tissue [42]. The solvents included a cold aqueous buffer with/without detergent, hot aqueous buffer, cold organic mixtures such as 80% methanol, buffered 75% acetonitrile, and acidic 40:40:20 acetonitrile:methanol:water with 0.02 or 0.1 M formic acid. The study found that extraction with acidic 40:40:20 acetonitrile:methanol:water was most efficient at maintaining the NADome for measurement using LC-MS. However, inclusion of detergent may also be useful, albeit to a lesser extent [42].

Human plasma, serum, urine, and cerebrospinal fluid (CSF) require only the removal of proteins, which can be performed using pre-heated buffered ethanol solution (ethanol:HEPES 1 mM, pH 7.1) [24], ice-cold methanol [17], methanol:acetonitrile, or centrifugal filtration apparatus. Additionally, a recent method quantified the NADome in human cell cultures, erythrocytes, CSF, and primate skeletal muscle without drying steps (using steam drying or speed vac), thus increasing NADome stability [24].

#### *5.2. Internal Standards*

Ionisation suppression is a major problem that should be considered when measuring the NADome. This phenomenon refers to the ability of some sample components to influence the ionisation and detectability of certain analytes [16]. Therefore, the peak height or peak area may not be a true reflection of the peak size in the original complex mixture. Hence, internal standards are necessary to minimise ionisation suppression errors [16]. Optimisation of internal standards is also important to minimise inaccuracies in the quantification of the NADome due to interconversion of some metabolites, i.e., non-enzymatic degradation limits the accurate measurement of the NAD+:NADH and NADP+:NADPH ratios. Spiking with internal standards can monitor interconversion of these metabolites [42].

Previous studies that quantified the NADome using LC-MS/MS assays used internal standards derived from yeast cultured in 13C-glucose-supplemented (13C 15N)-labelled medium [48] or in 13C-glucose with excellent correlation results [16]. However, yeast cell culture facilities may not always be available in-house. Isotopic labels for some NAD<sup>+</sup> metabolite isotopic labels are not commercially available. In the absence of the exact isotypic label, a closely related molecule (structural analog) is recommended [49]. Evans [50] quantified 18 metabolites of the NADome with excellent correlation coefficients using external standards.

We previously used the following internal standards for the quantification of selected NAD<sup>+</sup> metabolites in human and murine cells: 2H4-NAM (MeNAM, NAM, NA, NAMN, NADPH, NAAD, ADPR, cADPR), 13C5-Adenosine (adenosine), 13C5-Cyclic AMP (cAMP, NAD+, NMN), 13C1015N5-ATP (NADH, ATP), 13C5-AMP (AMP), and 15N5-ADP (NADP+, ADP). Most of these internal standards displayed correlation coefficients (r2 ≥ 0.98) [47].

#### *5.3. Analyte Stability*

Instability of pyridine nucleotides is likely to be the biggest challenge when quantifying the NADome in a variety of biological samples. For instance, while reduced forms (i.e., NADH and NADPH) are more stable in alkaline solution, the oxidised forms (NAD+ and NADP+) are more stable in acidic solutions, and this is likely due to acid-catalysed autoxidation of NADH and NADPH [27]. Since time, pH, and temperature are likely to have a major effect on the ability to accurately quantify the NADome, quick quenching of metabolism is essential.

Demarest et al. recently assessed the benchtop stability of the NADome in human red blood cells (RBCs), the epithelial cell line HEK-293T, and primate skeletal muscle [24]. The study demonstrated rapid degradation of NADH and NADH within 30 min in the cellular matrices, whereas NAD<sup>+</sup> and ADPR were only stable for 10 min, and only 20% remained after 30 min. On the other hand, NAM and NMN levels increased after 1 h. This may be due to increased catabolism and degradation of NAD<sup>+</sup> and other metabolites [24]. Interestingly, NMN levels decreased in RBCs and skeletal muscle, while NADP<sup>+</sup> remained stable for 1 h in both RBCs and HEK-293T cells. In the CSF, only NAD<sup>+</sup> and NMN could be detected in the linear range. NAD+ demonstrated greater stability in the CSF compared to the cellular matrices and this was attributed to a reduced NADome in the CSF and/or limited availability of NAD<sup>+</sup> 'consumers', e.g., PARPs and Sirtuins [24].

Therefore, optimising sample collection, storage, and availability of suitable testing protocols is essential to retain and accurately report changes to the NADome in intervention studies. In addition, quenching of samples should be completed without delay to minimise degradation of metabolites by active enzymes and preserve the endogenous NADome.

#### *5.4. Liquid Chromatography*

Perhaps the most well-described LC-MS assay for quantification of the NADome is based on hydrophilic interaction liquid chromatography. One method to quantify the NADome uses two different mobile phases on two porous graphitic carbon reversed phases (Hypercarb, Thermo) in alkaline (NMN, NAMN, ADP, ATP, NAD+, NADH, NAAD, NADP+) and acid separation (NA, NAM, and NR) [16]. We previously demonstrated that an amino column using a dual HILIC-RP gradient with heated electrospray (HESI) tandem mass spectrometry detection in mixed polarity multiple reactions monitoring (MRM) mode could be simultaneously used for the quantification of the NADome in a single chromatographic run in biological specimens [47]. Recently, another study using an Accucore HILIC column identified some metabolites in multiple transitions [24]. For instance, NAD<sup>+</sup> was observed in the NAAD transition. Additionally, NAM was observed in the NA transition, but these metabolites could not be resolved. Picolinic acid, a metabolite in the de novo NAD+ synthesis pathway and an isomer of NA, also co-eluted with NA [24].

Our refined method is an application of hydrophilic interaction chromatography (HILIC)—a major chromatographic system used in metabolic profiling [50]. The retention mechanism is based on partitioning and water is used as the eluent. An amino-modified HILIC Phenomenex Luna NH2 column has been previously shown to demonstrate good retention and chromatographic resolution of water-soluble metabolites, including the NADome, with good peak shape, compared to cyano and/or silica columns, and none of the selected metabolites were observed in multiple MRM transitions [47,50].

#### **6. Concluding Remarks**

It is well-established that concentrations of the NADome represent a useful marker for elucidating the current status of cells, and may likely be an important biomarker in several metabolic and age-related disorders. Thus, accurate quantification of the NADome may be beneficial for researchers in understanding the pathobiology of metabolic disorders and effects of drug candidates. LC-MS/MS represents a rapid, robust, simple, and reliable assay for the measurement of the NADome in clinically relevant biological tissue. LC-MS/MS requires minimal sample processing. Using the amino phase chromatographic separation and commercially available internal standards eliminates cost, and requirement for yeast cultures for labelled metabolites, which are likely to represent major obstacles for measurement of the NADome to be applied in clinical diagnosis. NADomics can be used to provide renewed insights on physiological and pathological processes and may assist in identifying and evaluating potential therapeutic strategies.

**Author Contributions:** Conceptualisation, N.B and R.G. Methodology, N.B. and R.G. Writing draft preparation, N.B. and R.G. Writing review and editing, N.B. and R.G. Visualisation, M.D.V. Supervision, N.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Review* **Sudden Infant Death Syndrome: Beyond Risk Factors**

#### **Serafina Perrone 1,\*, Chiara Lembo 2, Sabrina Moretti 1, Giovanni Prezioso 1, Giuseppe Buonocore 2, Giorgia Toscani 1, Francesca Marinelli 1, Francesco Nonnis-Marzano <sup>3</sup> and Susanna Esposito <sup>1</sup>**


**Abstract:** Sudden infant death syndrome (SIDS) is defined as "the sudden death of an infant under 1 year of age which remains unexplained after thorough investigation including a complete autopsy, death scene investigation, and detailed clinical and pathological review". A significant decrease of SIDS deaths occurred in the last decades in most countries after the beginning of national campaigns, mainly as a consequence of the implementation of risk reduction action mostly concentrating on the improvement of sleep conditions. Nevertheless, infant mortality from SIDS still remains unacceptably high. There is an urgent need to get insight into previously unexplored aspects of the brain system with a special focus on high-risk groups. SIDS pathogenesis is associated with a multifactorial condition that comprehends genetic, environmental and sociocultural factors. Effective prevention of SIDS requires multiple interventions from different fields. Developing brain susceptibility, intrinsic vulnerability and early identification of infants with high risk of SIDS represents a challenge. Progress in SIDS research appears to be fundamental to the ultimate aim of eradicating SIDS deaths. A complex model that combines different risk factor data from biomarkers and omic analysis may represent a tool to identify a SIDS risk profile in newborn settings. If high risk is detected, the infant may be referred for further investigations and follow ups. This review aims to illustrate the most recent discoveries from different fields, analyzing the neuroanatomical, genetic, metabolic, proteomic, environmental and sociocultural aspects related to SIDS.

**Keywords:** SIDS; newborn infant; genetic polymorphism; neurotransmitter

#### **1. Definition and Epidemiology**

Sudden infant death syndrome (SIDS) is defined as "the sudden death of an infant under 1 year of age which remains unexplained after thorough investigation including a complete autopsy, death scene investigation, and detailed clinical and pathological review" [1,2]. SIDS is characterized by an unexpected death during the sleeping period and it typically occurs in the first 12 months of age in a previously healthy infant. Most events take place in the child's home; the out-of-home deaths are most frequent at a relative's home or a child-care setting, especially if the child sleeps in prone position and in a stroller and/or car seat [3]. SIDS is a subcategory of Sudden Unexpected Infant Deaths (SUID) and represents nearly half of these cases; SUID includes SIDS of unknown cause and also events of strangulation in bed or accidental suffocation [4]. The peak of incidence of SIDS is around two to four months of age and 90% of cases occur before six months of age; prevalence of SIDS is higher in boys than girls, at a 3:2 ratio [5].

In the late 1980s prone sleep has been documented as a major risk factor for SIDS, leading in the 1990s to "back-to-sleep" campaigns which have had a great impact on reduction of SIDS rates [6]. SIDS infant mortality has decreased well over 50% for most

**Citation:** Perrone, S.; Lembo, C.; Moretti, S.; Prezioso, G.; Buonocore, G.; Toscani, G.; Marinelli, F.; Nonnis-Marzano, F.; Esposito, S. Sudden Infant Death Syndrome: Beyond Risk Factors. *Life* **2021**, *11*, 184. https://doi.org/10.3390/ life11030184


Academic Editor: Jong Hyuk Yoon

Received: 24 December 2020 Accepted: 22 February 2021 Published: 26 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

countries, especially in the first few years after the beginning of national campaigns [7]. Alongside the SIDS prevention recommendations, perinatal care has experienced numerous other improvements, so it's difficult to attribute the SIDS incidence drop to only the supine sleep practice [6]. Furthermore, the increasing rates of other causes of death such as strangulation in bed and accidental suffocation may represent another explanation of the decline in SIDS rates [8]. Over the first decade of this century, infant mortality from SIDS has experienced a continuous drop in some countries such as Australia, Canada, England and Wales, Germany, Japan and the Netherlands, while it has remained stationary in others, notably the USA and New Zealand [9]. Despite continuous public health efforts concentrating on the improvement of sleep conditions with a special focus on the high-risk groups [10,11], incidence of SIDS still continues to be high. In fact, SIDS still represents a prominent cause of infant death, occurring at a rate of 27/100,000 live births in the United Kingdom and 38/100,000 the United States [12,13]. In Italy incidence is about 1 out of 1000 live births [14]. Therefore, continuous research on the cause and prevention of SIDS is needed.

#### **2. Pathogenesis and Risk Factors**

The most recent evidence suggests that SIDS pathogenesis is a multifactorial condition that comprehends genetic, environmental and sociocultural factors [8]. The Triple-Risk Model, first described in 1994, affirms that SIDS occurs in infants with latent biological vulnerability (brainstem abnormality or genetic pattern), who is exposed to a trigger event or extrinsic risk factor (prone sleeping, airway obstruction) during a critical phase of development [15]. The combination of intrinsic and extrinsic factors which overlap during a period of respiratory, autonomic and cardiac development, usually occurring between two to four months of age, leads to a life-threatening event during a period of sleep. Failure of protective mechanisms during these episodes finally concludes with unexpected death. On the contrary, SIDS is less likely to occur with the removal of one of these factors [16,17].

Multiple extrinsic risk factors for SIDS in the sleep environment of the infants have been examined. Prone sleeping appears to be the most significant risk factor for SIDS. In fact, it is likely to be associated with re-breathing expired gases, suffocation, overheating and decreased arousal [18]. Infants in side lying position are also at high risk to roll into prone position while sleeping [19]. Other factors related to the sleeping environment involve soft bedding, sleeping with blankets, pillows, soft objects, bumper pads, head or face covered during sleep, bedsharing (especially co-sleeping on a couch or sofa) and room or infant overheating [20]. Maternal smoking during pregnancy is associated with a fivefold increase in SIDS events, and postnatal smoking exposure further increases the risk [21–23]. In fact, it's been demonstrated that smoking during pregnancy contributes to the risk due to the disruption of arousal patterns in the sleeping baby along with impairment of autonomic system and cardiovascular response: uterine exposure reduces lung compliance and volume, alters arousal mechanisms and decreases heart rate variability in response to stress, all factors that can negatively affect a baby's ability to respond appropriately to the environment [24,25]. A recent study showed a linear correlation between number of cigarettes smoked daily during pregnancy and the risk of a SUID event. Furthermore, quitting smoking over the three trimesters is strongly associated with great reduction of SUID risk but also diminution of cigarettes smoked daily contributes to a small decrease in risk [26]. Additional risk factors include maternal alcohol use, young maternal age (under 20 years) and poor prenatal care [27]. The combined exposure of alcohol and tobacco beyond the first trimester of pregnancy appears to have a synergistic effect on the risk for SIDS events [28]. Co-sleeping associated with recent use of alcohol or drugs by the parents also increases the risk significantly [29]. The strong association between smoking, alcohol consumption and drugs utilization may also explain, in part, the interaction described between co-sleeping and smoking [29–32]. Bottle-feeding is associated with increased risk, while the use of a pacifier and breastfeeding appear to be protective factors [27,33,34]. Some studies suggest that a significant part of SIDS cases may be closely related to sub-clinical

infection processes [35,36]. In a study conducted by Goldwater et al. in 2020, significantly heavier thymus and brain were found in SIDS victims compared to non-SIDS controls. This finding is related to immune responses in the brain and thymus associated with possible subclinical infections [37].

Intrinsic risk factors are male gender, population subgroups such as non-Hispanic black infants, American Indian or Alaska Native infants [8,38] and prematurity. Preterm birth or low birth weight increases the risk of SIDS events three to four times, suggesting that altered intrauterine environment may contribute to the pathogenesis [39]. Additionally, smoking during pregnancy increases the risk of premature delivery [40]. Furthermore, research suggests that disorders of homeostasis, neuroregulation and cardiorespiratory function associated with brain and brainstem anomalies play an important role in SIDS [16]. In particular, serotonin brainstem abnormalities have been identified in up to 70% of infants who have died of SIDS [16]. Since the serotonin system is associated with several homeostatic functions, these anomalies may possibly lead to a network dysfunction that affects arousal and cardiorespiratory functions [41,42]. In addition to brain vulnerabilities, research has focused on identifying genetic variants related to defects involving autonomic and metabolic functions, neurotransmission and cardiac repolarization, suggested to contribute to SIDS infant's "underlying vulnerability" [2,43,44]. Thus, certain infants may have a genetic predisposition to SIDS or an underlying abnormality in the brainstem, which becomes manifest when the infant experiences environmental challenges (hypoxia, asphyxia, hypercarbia, overheating) during sleep and differentiates a SIDS infant from a healthy infant. In fact, infants without the underlying conditions present an efficient protective brainstem response to homeostatic challenges which promptly manage to avoid SIDS occurrence [2]. The combination of multiple extrinsic and intrinsic factors leads to asphyxia. Vulnerable infants are not able to respond with arousal that prevents re-breathing or apnea. Consequently, asphyxia leads to bradycardia and insufficient gasping breathing, which eventually terminate with death [45].

Most of these life-threatening events occur during the sleep period. In fact, sleep is associated with a reduction of blood pressure, heart rate, respiratory rate and muscle tone, especially in the upper airways. During sleep phases protective reflexes to hypoxia and hypercapnia are also depressed. Blood pressure, cerebral oxygenation and cerebral vascular homeostasis are decreased in the prone position [46,47]. In fact, term infants between two to four months of age show a depressed baroreflex response and decreased arousal in the prone position [48,49]. In premature babies these characteristics in prone position are mostly marked [50,51]. These considerations highlight, indeed, the increased risk of SIDS occurring during sleeping, mainly in the prone position during specific infant development windows.

#### **3. The Brainstem Hypothesis in SIDS**

Neural research in SIDS has concentrated efforts to define the existence and eventual location of a pathological lesion in the brain. No major anatomic signs of neural pathology have been revealed by standard autopsy.

Impairment of the brainstem has been related to sudden death. The brainstem is the primary anatomic site of homeostatic control and sleep/waking regulation in the brain [52]. The brainstem hypothesis in SIDS suggests that developmental abnormalities in specific brainstem regions lead to a failure of protective mechanisms against exogenous stressors associated with asphyxia, hypoxia, hypercapnia, or thermal imbalance during sleep. Defense functions, as central chemosensitivity to carbon dioxide (CO2) and peripheral and central chemosensitivity to oxygen (O2), activate an autoresuscitation mechanism with arousal accompanied by head lifting to avoid asphyxia. Defects in this system include impaired arousal, ineffective respiratory pattern, episodes of obstructive apnea during sleep and autonomic dysfunction [53,54].

Additional evidence of the possible brainstem role in SIDS was reported by Naeye et al. who described astrogliosis in the medullary reticular formation in 50% of SIDS cases. This scarring lesion was interpreted as the result of chronic alveolar hypoventilation and hypoxemia [55]. Further studies confirmed the presence of reactive gliosis in different brainstem regions such as inferior olivary nuclei of the brainstem [56,57]. Gliosis is thought to be a sign of brainstem function impairment, as a result of hypoxia and chronic underventilation [58]. Finding a gliosis marker in the brainstem in SIDS cases supported the validity of brainstem hypothesis [59].

Furthermore, an immature developmental pattern in the SIDS brainstem has also been reported, with brainstem neurons presenting an augmented dendritic spine number [60]. These findings have been recently supported by proteomic investigations that showed abnormalities related to neuronal/glial/axonal growth, metabolism and apoptosis in some brainstem areas such as raphe, hypoglossal nuclei and medullary pyramids. These considerations suggest that brainstem immaturity, as well as gliosis, may be involved in the abnormal central respiratory and arousal control [61].

Multiple neurotransmitter networks anomalies are thought to be responsible for the underlying vulnerability in SIDS infants. These defects involve different brainstem neurochemicals such as catecholamines, neuropeptides, acetylcholinergic metabolites, amines, aminoacids (primarily glutamate), growth factors and some cytokines [62]. A defective binding of peptide neurotransmitter substance P to its receptor neurokinin-1 has been described in nuclei involved in cardiorespiratory and autonomic control. In fact, a defect of medullary substance P network with cerebellar sites may result in failure to activate respiratory and motor responses. In particular, low levels of substance P have been found in the olivary nuclei that control head and neck movements. As a consequence, failure of the protective mechanism such as head lifting or tilting to escape hypoxia may occur in SIDS infants [63]. Another study highlighted the interactions of GABA neurons in the medulla oblongata with the medullary serotonergic system in the regulation of homeostasis [64]. Data suggest that deficits of GABAergic and serotoninergic systems cooperate to generate medullary dysfunction in SIDS [65]. SIDS was also associated with low levels of tryptophan hydroxylase enzyme, which is involved in serotonin synthesis, resulting in decreased serotonin production [66].

However, researches have been mostly focused on the role of the brainstem serotoninergic system. The brainstem serotonin network in the rostral medulla is involved in the activation of the protective respiratory and autonomic reactions in response to exogenous stressors during sleep. According to the serotonin brainstem hypothesis, a defect in the serotonin system leads to failure of the autoresuscitation and arousal reaction that ultimately causes SIDS. A study conducted by Kinney et al. identified a serotonergic impairment in the medullary reticular formation of the brainstem as a "core" lesion in SIDS [65]. The affected serotonin brainstem network is supposed to involve serotonin neurons interconnected among arcuate nucleus, paragigantocellularis lateralis, gigantocellularis, intermediate reticular zone and caudal raphe [65]. These regions have been identified using a quantitative autoradiography which has demonstrated a decreased serotonergic receptor binding in these nuclei in SIDS cases compared to controls [67]. The ascending serotoninergic arousal system is also implicated. In fact, the rostral reticular formation dysfunction is postulated to be transmitted to median raphe and dorsal raphe of the ascending serotonergic arousal system that results in failure of the metabolic challenge response, leading to death [65]. In particular, an alteration of Pet-1 expressing neurons located in the serotonergic raphe system has been related to the impaired recovery system. This suggests a role of Pet-1 neuron activity in the neonatal survival mechanism responding to hypoxia [68]. Furthermore, recent findings by Haynes et al. have described increased serum serotonin levels in SIDS cases compared to controls. Therefore, a high serum serotonin may be utilized as a biomarker in SIDS autopsies to distinguish deaths caused by serotonin-related anomalies [69].

As potential causes of serotonin brainstem disruption in SIDS, the role of maternal and pregnancy factors has been discussed. In fact, analysis of the placentas of newborns who subsequently died of SIDS suggests that an infant's vulnerability originates in the gestational period.

In fact, some maternal factors associated with fetal hypoxia, such as placenta vascular hypoperfusion, maternal anemia and cigarette smoking generate a suboptimal intrauterine environment. These maternal factors are hypothesized to be responsible, in part, for impaired brain development, particularly in the central serotonin system, as the basis of vulnerability in sudden postnatal death [70–72]. Furthermore, some gene polymorphisms of the promoter region of a serotonin transporter protein ("L" allele and "LL" genotype variants) are responsible for increased activity of the serotonin transporter protein, resulting in reduced concentration of serotonin. These genetic variants have been frequently found in SIDS victims [16,62,73,74]. Interestingly, the same variants were also found in additional predisposing conditions [75]. However, some studies reported no association between LL genotype or L allele and SIDS in Caucasians [76,77].

Yet, the role of this polymorphism in other ethnicities as a risk factor in SIDS still needs to be clarified [78]. Additional studies should be carried out to assess the role of population genetics influencing serotonin transporter protein alleles and genotypes distribution in different ethnicities [74]. Different ethnic groups sharing the same social conditions show different SIDS rates [79,80]. Recently, a nearly fivefold variation in high risk of SIDS has been found between ethnic groups in England and Wales [81]. Authors speculated that cultural differences play an important role in infant care rather than genetic factors. Further research into infant-care practices in low-risk ethnic groups might enable more effective prevention of SIDS in the general population. While there is some evidence of how genetic differences might influence susceptibility to certain factors (e.g., inflammatory responses) [82], the knowledge about the functional impact of these differences is still lacking. As a matter of fact, there is no evidence that genetic influences outside of social and environmental factors pose a risk for SIDS.

Brainstem dysfunction is not limited to only one neurochemical network, such as the serotoninergic system. In fact, recent studies conducted by Hunt at al. have reported abnormal expression in nine different proteins within some brainstem nuclei, mainly the raphe nuclei and pyramids, that may relate to developmental and neurological cytoarchitecture abnormalities in SIDS cases [61]. A study by Lavezzi et al. recently described low tyrosine hydroxylase expression in SIDS cases associated with a delayed development of Substantia Nigra pars compacta (SNpc). Moreover, nicotine absorption in the uterus was related to decreased neuron density in SN [83]. SNpc represents the major dopamine brain center with an important role in regulation of the sleep-arousal cycle [84]. Therefore, the deficit of dopaminergic neurons, alongside with smoking exposure, may explain SIDS occurrence in the awakening phase in a significant number of cases [83].

#### **4. Metabolic Predisposition**

Inborn errors of metabolism (IEM) have been described as possible causes of SIDS [85–87]. Metabolomics analysis enables characterization of metabolites produced by cells, tissues and microorganisms, their quantification and interpretation [88]. A study conducted by Graham et al. analyzed data of Nuclear Magnetic Resonance and Mass Spectrometry from the medulla oblongata of infants who died from SIDS. This analysis revealed that fatty acid metabolism was the principal metabolic pathway altered in the brain of infants with SIDS occurrence, thus fatty acids oxidation disorders may represent one of many causes of SIDS. Furthermore, this analysis identified one metabolite (octadecenoyl-L-carnitine) that could be potentially used as a diagnostic tool for early screening to detect infants at greatest risk of SIDS occurrence. Further studies should determine if the same diagnostic biomarkers identified in this study can also be found in blood samples [89].

A recent study conducted a postmortem analysis of short chain fatty acids (SCFAs) values in babies who experienced SIDS or death for causes not related to SIDS. It's described that a SFCAs quantitative profile, involving isobutyric, butyric, hexanoic, valeric, and acetic acids, allows identification of the risk of SIDS [90].

#### **5. Proteomics**

Proteomic analysis has also been conducted with the aim to identify different expression of proteins in SIDS [61]. A recent study applied proteomic techniques to characterize changes in the proteome related to hypoxia, inflammation and apoptosis of SIDS compared to age-matched controls, analyzing heart, medulla tissues and blood samples. Results showed differentially regulated proteins, especially APOA1, GAPDH, S100B, zyxin and complement component C4A in SIDS cases as compared to the controls. All of them appeared up-regulated in SIDS except C4A, which was down-regulated. These findings suggest the role of these proteins as potentially diagnostic biomarkers for SIDS [91]. Using proteomics as a discovery tool, a study by Broadbelt et al. described a significant reduction levels of isoforms of the 14-3-3 protein family, associated with anomalies in TPH2 and serotonin levels and serotonin receptor (5-HT1A) binding in SIDS cases [92]. In fact, findings suggested that the deficit of 14-3-3, necessary for TPH2 modulation [93–95], may lead to TPH2 deficiency and consequently to medullary serotonin system impairment, resulting in SIDS [92].

#### **6. Genetic Predisposition**

Genetic studies have mostly focused on the ion channels of heart mutations. Polymorphism in sodium and potassium channel genes have been reported, including the sodium channel gene SCN5A, which is associated with prolonged QT intervals and may also be responsible for altered autonomic system development [96]. In fact, initial findings have showed that 2% of SIDS cases carried gain-of-function mutations in the sodium channel encoded by SCN5A gene [97]. Studies also described some inherited cardiac diseases such as Long and short QT syndrome, Brugada syndrome, catecholaminergic polymorphic ventricular Tachycardia and hypertrophic cardiomyopathy as monogenic causes in some SIDS cases [44,98–102]. It's been suggested that long QT syndrome (LQTS) could explain 10% of all current SIDS cases [98,103].

With the development of next-generation sequencing (NGS), the analysis of the whole exome is expanding the identification of mutations and also potential pathogenic sequence variations responsible for underlying vulnerability in SIDS deaths. Many researches have described pathogenic mutations in genes associated with cardiac channelopathies responsible for SIDS occurrence in up to 30% of cases as monogenic cause of death [44,101,102]. In a study of variant analysis conducted by Tester et al., only 4.3% of the European SIDS cases possessed pathogenic variations in one of the 90 genetic heart disease genes analyzed [104]. Another recent study by Köffer et al. assessed the percentage of rare and ultrarare variants (respectively minor allele frequency ≤0.2% and ≤0.005%) in genetic heart disease genes. In only 6% of these cases, gene ultrarare variants were considered potentially pathogenic. Therefore, still rare variant interpretations associated with inherited cardiac diseases need to be standardized [105]. A recent study by Liebrechts-Akkerman et al. published in 2020 conducted an analysis using a targeted massively parallel sequencing exon screening, differently from other targeted genotyping SIDS studies. This analysis allowed detecting either novel or previously known exonic DNA variants in selected arrhythmia genes. These findings provided further evidence for cardiac arrhythmias as partial genetic explanation of SIDS. Thus, the authors stress the importance of standardized DNA testing for LQTS and other cardiac arrhythmia genes as an essential element in the ordinary SIDS diagnostic protocol. Furthermore, it's proposed as an additional preventive measure to perform a standard ECG testing in newborns during the first two weeks of life [106]. Another recent study of 2020 by Simma et al. proposed performing standardized neonatal ECG screening in the first days of life with the aim of detecting neonates with a significant transient form of prolonged QT intervals and helping the diagnosis of congenital LQTS. This study suggests applying genetic testing as a second step in case of abnormal ECG results. Furthermore, in this study pathological cases were treated with a beta blocking agent for the first years of life [107].

Implementation of such a strategy would place considerable strain on the health care system. It might also result in increased financial costs and could influence parental behavior. Many questions remain opened: how many lives will ECG save? What do we know about 'normal' and 'abnormal' results that may influence what we do next? Will parents with a 'normal' result be less likely to follow safer sleep advice? Will parents with an 'abnormal' result become very anxious? In absence of clear answers, this topic needs to be carefully thought by planning much more research.

Other studies examined the potential role of noncardiac genes in the pathogenesis of SIDS. According to these studies, 61 genes were identified as potential "SIDS-susceptibility" genes, in the variant analysis typically in the promoter region. Yet, only around 55% of these genes have been involved on a monogenic basis in SIDS [108–112]. Of note, it has been described as a significant overrepresentation of functionally disruptive variants in the SCN4A gene. This gene encodes skeletal muscle voltage-gated sodium channel (Nav1.4), involved in the skeletal respiratory muscle contraction which have a key role in SIDS pathogenesis [113]. A recent study also reported SCN1A variants from exome sequencing in two infants who died of SIDS [114]. A study conducted by Gray et al. performed a SIDS-susceptibility variant analysis of these 61 previously published noncardiac SIDSsusceptibility genes. According to the findings of this research, there is very limited evidence that these specific genes are implicated in SIDS susceptibility in a monogenic basis with autosomal dominant and recessive inheritance. Therefore, it still needs to be investigated whether infant vulnerability to sudden death may be supported by a more complex polygenic inheritance model [115].

The involvement of additional genes regulating the metabolism of neurotransmitters (mainly serotonin and dopamine) has already been described in the previous chapter "The brainstem hypothesis in SIDS" [41,73,74,77].

Finally, genetic predisposition has also been related to increased vulnerability to smoke exposure. In fact, a recent study analyzed the contribution of GSTs (enzymes of glutathione-S-transferase supergene family), involved in detoxification of xenobiotics [116]. A significant correlation has been described between the GSTM1 deletion characterizing the gene polymorphism, resulting in a lack of metabolic activity, and SIDS exposed to smoking. These findings highlight the role of smoking exposure as an important SIDS risk factor connected with a biological predisposition [117].

#### **7. Recommendations on Safe Infant Sleeping Environment**

In November 2016, the American Academy of Pediatrics task force published updated recommendations to reduce risk of SIDS and sleep-related infant deaths. These recommendations are addressed to infants up to one year of age. The strength of the guidelines is based on case-controlled studies as randomized trials cannot be performed for SIDS [16].

The sleeping position is the "strongest modifiable risk factor for SIDS" [16,32]. Infants should sleep in the supine position until one year of age or until the infant is able to roll from back to supine position, unassisted. Of note, supine position has not been related to increased risk of choking or aspiration, even in infants with gastroesophageal reflux disease [118]. Supine position is also recommended for premature infants either in NICU and home settings [40].

A firm flat surface covered only by a thin, fitted sheet should be used. Soft items such as toys, crib bumpers, positioners and pillows should be avoided, as well as loose sheets and blankets, due to risk of suffocation or airway obstruction [16]. Car seats, infants swings or strollers should not be used for routine sleep. The infant who fell asleep in such equipment should be repositioned to on an appropriate surface as soon as possible, due to the risk of head flexion resulting in obstruction of the upper airways [119]. The sleep surface must be located in a hazard-free location, without dangling cords or electric wires [16].

Overheating of infants by overwrapping, head coverings and excessive clothing must be avoided [120]. Room sharing is encouraged at least during the first six months of age, but bed sharing is prohibited until one year of age. In fact, sleeping in the caregiver's bed, couch or chairs put the infant at risk of overheating, sleeping on a soft surface and being rolled over by adults. Conversely, room sharing has been shown to decrease SIDS risk by 50% [27].

Breastfeeding appeared to be one of the strongest protective measures against SIDS. Breastfeeding benefit is stronger when breastfeeding is exclusive, reducing the risk of SIDS by approximately 50% if conducted over the first month of life [121]. Furthermore, it was shown that any breastfeeding, exclusive or with formula supplementation, was able significantly protect against SIDS. In addition, the protective effect increases proportionally to the duration of breastfeeding [121].

The pacifier use for naps or bedtime is also considered a protective factor against SIDS. The exact protective mechanism is still unclear; it is assumed that the use of the pacifier may increase autonomic control and cardiovascular stability which help to maintain a patency of the airways. No adverse effect was found of breastfeeding with the use of the pacifier [33,34]. However, it is recommended to introduce pacifiers to infants only after breastfeeding is well established. The pacifier should not be attached to any strings or cords as these might present a risk of strangulation [33,34,122].

The routine use of home apnea monitors is not recommended in infants, including preterm and infants at risk of SIDS. In fact, cardiorespiratory monitors have not been demonstrated to decrease incidence of SIDS [123]. Furthermore, the usage of these tools may distract from adoption of other effective measures or give false alarms that could lead to overdiagnosis with consequential unnecessary analysis and caregiver anxiety [124].

Prenatal care should be endorsed from early pregnancy. Smoking, alcohol consumption and illicit drugs should be avoided by women during pregnancy, after delivery and during breastfeeding, as these factors significantly increase the risk of SIDS occurrence [16,22,23].

Regular immunization in accordance with Centers for Disease Control and Prevention schedule should be followed as it has been demonstrated to be protective and not associated with SIDS risk [16].

Prone position or "tummy time" is recommended only if supervised, when infants are awake and alert, as its benefits motor development and helps minimize positional plagiocephaly [125]. There is no evidence that suggests that swaddling used as a strategy to promote sleep and calm the infant reduces the risk of SIDS. When swaddling practice is performed, infants should be positioned on their back. In fact, recent studies described an increased risk of SIDS in case the swaddled infants are placed in or rolls to the prone position [126].

Finally, education by health care professionals and implementation of safe sleep practices should be based on the model of SIDS risk-reduction recommendations. A recent study described a functional use of smartphone technology for prevention, by assessing infant sleep safety practices among at-risk communities. It's been observed that photographs processed by coders provide a cost and timesaving assessment that may support safe sleep interventions in clinical and community settings [127].

Ongoing research into the etiology of SIDS and other sleep-related infants' deaths is encouraged to help achieve the ultimate goal of completely eradicating SIDS deaths [16].

#### **8. Conclusions**

SIDS occurrence is associated with multifactorial conditions. While extrinsic factors have been largely recognized and significantly reduced through recommendations on safe sleep worldwide, understanding the underlying intrinsic vulnerability to SIDS still represents a challenge.

Biomarkers such as proteins, metabolites and neurotransmitters have been proposed for early identification of cases at risk. To eliminate SIDS cases related to cardiac channelopaties, it has been also proposed that an ECG is performed on all neonates within two

weeks of life with subsequent genetic analysis if some alterations are detected, but the implementation of such a strategy requires further research.

New "omic" technologies provide a large amount of data that can be analyzed independently and combined, allowing detection of multiple system alterations.

A complex model that combines different risk factors data from biomarkers and omic analysis may represent a tool to identify a SIDS risk profile in newborn settings. If high risk is detected, the infant may be soon referred for further investigations and follow up.

Whole exome sequencing of newborns in NICUs is another promising area to reveal susceptibility to SIDS. [128]. This new technology opens the possibility of extending the concept of precision medicine to an early stage of life.

**Author Contributions:** Conceptualization, S.P., F.N.-M. and S.E.; writing—original draft preparation, C.L., S.M., G.P., F.M., F.N.-M.; writing—review and editing, C.L., S.P., G.B., and G.T.; visualization and supervision, S.P., G.B. and S.E. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by Department of Medicine and Surgery, University of Parma, Project "Realizzazione della Rete regionale sulla prevenzione e lotta alla "Sudden Infant Death Sindrome" (SIDS)".

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article.

**Acknowledgments:** We thank Daniela They, President of Colibri Association and Allegra Bonomi, President of SEMI per SIDS Association for their support.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**


#### **References**


## *Review* **Zinc Metalloproteins in Epigenetics and Their Crosstalk**

**Abdurrahman Pharmacy Yusuf 1, Murtala Bello Abubakar 1,2,\*, Ibrahim Malami 1,3, Kasimu Ghandi Ibrahim 1,2, Bilyaminu Abubakar 1,4, Muhammad Bashir Bello 1,5, Naeem Qusty 6, Sara T. Elazab 7, Mustapha Umar Imam 1,8, Athanasios Alexiou 9,10,\* and Gaber El-Saber Batiha 11,\***


**Abstract:** More than half a century ago, zinc was established as an essential micronutrient for normal human physiology. In silico data suggest that about 10% of the human proteome potentially binds zinc. Many proteins with zinc-binding domains (ZBDs) are involved in epigenetic modifications such as DNA methylation and histone modifications, which regulate transcription in physiological and pathological conditions. Zinc metalloproteins in epigenetics are mainly zinc metalloenzymes and zinc finger proteins (ZFPs), which are classified into writers, erasers, readers, editors, and feeders. Altogether, these classes of proteins engage in crosstalk that fundamentally maintains the epigenome's modus operandi. Changes in the expression or function of these proteins induced by zinc deficiency or loss of function mutations in their ZBDs may lead to aberrant epigenetic reprogramming, which may worsen the risk of non-communicable chronic diseases. This review attempts to address zinc's role and its proteins in natural epigenetic programming and artificial reprogramming and briefly discusses how the ZBDs in these proteins interact with the chromatin.

**Keywords:** epigenetics; epigenome; zinc finger domain; zinc finger motif; zinc finger proteins; zinc metalloproteins

#### **1. Introduction**

Zinc is an omnipresent micronutrient essential for healthy prenatal and postnatal developments in humans and the growth and development of plants, animals, and microorganisms [1,2]. In silico data suggest that about 10% of the human proteome potentially binds zinc [3]. Zinc is present in all body tissues and fluids as a component of over

**Citation:** Yusuf, A.P.; Abubakar, M.B.; Malami, I.; Ibrahim, K.G.; Abubakar, B.; Bello, M.B.; Qusty, N.; Elazab, S.T.; Imam, M.U.; Alexiou, A.; et al. Zinc Metalloproteins in Epigenetics and Their Crosstalk. *Life* **2021**, *11*, 186. https://doi.org/ 10.3390/life11030186

Academic Editor: Jong Yoon


Received: 5 February 2021 Accepted: 23 February 2021 Published: 26 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

2000 proteins (including epigenetically active enzymes) [4–7]. A group of writers, erasers, and readers of epigenetic marks are zinc-dependent [8]. Some of these are enzymes such as histone deacetylases (HDACs) having the zinc itself incorporated in their active sites; thus, it directly partakes in the catalytic process [9]. Others are proteins containing zinc within a zinc-binding domain (ZBD). These domains are important in substrate recognition, self-regulation, integrity, crosstalk, and sometimes catalysis [10,11]. With intracellular zinc within the normal range, these proteins work together in a coordinated manner to shape the plastic epigenome. In contrast, fluctuations in zinc levels or its deficiency and loss of function mutations in the ZBDs of these proteins affect their expression and function and may lead to epigenetic perturbations. [12,13]. These aberrant epigenetic changes may increase the risk of non-communicable chronic diseases such as cancer, diabetes, and cardiovascular diseases with possible multigenerational or transgenerational consequences. Intracellular zinc homeostasis is under the tight regulation of two families of zinc transporters and the zinc-binding proteins metallothionines (MTs) [14,15]. In addition to their zinc homeostasis roles, zinc transporters are also emerging as important modulators of the epigenome [4].

The significance of zinc, its deficiency, its transporters, and some of its enzymes in epigenetics has been partly reviewed [4,8,16]. However, these reviews are not all-inclusive and did not cover other ZBD-containing proteins involved in epigenome programming and editing. Here, we attempt to provide a comprehensive overview of zinc metalloproteins' roles in natural and synthetic epigenetics. We also try to explore how the ZBDs in these proteins interact with the chromatin. Of note, in the context of this article, the term "zinc metalloproteins" encompasses zinc finger proteins (ZFPs), zinc metalloenzymes, and enzymes requiring zinc activation for catalysis.

#### **2. Molecular Bases of Epigenetic Modifications**

Since the late 1980s, epigenetics has been one of the critical areas of research in molecular biology. From the late twentieth century to date, the meaning of the term "Epigenetics" has been evolving, and its exact definition remains elusive [17]. However, this review focuses on the most common definition (the molecular mechanism of transcription regulation) derived from the root words "epi+genetics", which means in addition to or "on the top" of genetics. Based on this concept, it has been defined as the mitotically and meiotically heritable layer of chemical code beyond the deoxyribonucleic acid (DNA) sequence, which regulates the genome, leading to various transcriptional outcomes in different cell types [17–19]. The four basic mechanisms involved in epigenetic regulations include DNA methylation, histone post-translational modifications, remodeling of the chromatin architecture, and gene silencing associated with noncoding RNAs such as microRNA [20–22]. Discussing the detailed molecular mechanisms of these epigenetic events is indeed beyond the scope of this review. Therefore, here, we focus only on zinc's role and that of its proteins in DNA methylation, histone modifications, chromatin remodeling, epigenome editing, and their crosstalk.

#### *2.1. DNA Methylation*

DNA methylation is the physical addition of a methyl (-CH3) group to a DNA sequence. The most common form of this modification in eukaryotes is adding a methyl group to the 5th carbon of cytosine in a DNA sequence to form 5 -methylcytosine (5 -mC) [23–25]. This modification predominantly occurs at cytosine-phospho-guanine dinucleotide-rich regions called CpG islands, which are mainly present around the promoter regions of genes and have been implicated in transcription regulation [26]. Moreover, DNA methylation occurs harmoniously and proportionately on both strands of the methylated DNA [27]. In most cases, cytosine methylation promotes gene silencing principally via the gene promoter's CpG islands' hypermethylation. Meanwhile, the promoter regions of transcriptionally active genes remain hypomethylated [26]. Thus, the methylation status of a gene could affect its expression level.

In mammals, two sets of enzymes regulate the DNA methylation status of the genome. They include the zinc-dependent DNA methyltransferases (DNMTs), which facilitate 5 C methylation and its passive demethylation, and the iron-dependent members of the teneleven translocases (TETs), which catalyze the active demethylation of DNA coupled to base excision repair [23,28,29]. So far, DNMT1, DNMT3a, and DNMT3b are the three known DNMTs that directly add methyl groups from S-adenosyl methionine (SAM) to the human genome [30]. DNMT1 catalyzes the maintenance or restoration of DNA methylation patterns of a hemimethylated double-stranded DNA in somatic cells following replication; therefore, it is called maintenance DNMT (Figure 1) [31]. On the other hand, DNMT3a and DNMT3b establish new methylation tasks on a newly synthesized (unmethylated) double-stranded DNA in the germline and the embryo; they are named de novo DNMTs (Figure 1) [30]. These three enzymes are not mutually exclusive catalytically, as Ren et al. (2018) [31] have reported more evidence suggesting the involvement of each enzyme in de novo and maintenance DNA methylation.

**Figure 1.** Comparison between de novo DNA methylation and maintenance DNA methylation. The figure compares de novo DNA methylation and maintenance DNA methylation. The two forms of DNA methylation catalyzed by DNMTs. (**A**) = unmethylated DNA; (**B**) = hemimethylated DNA; (**C)** and (**D**) = fully methylated DNA; DNMT3a/DNMT3b = DNA methyltransferases 3a/3b, both catalyze de novo DNA methylation (methylation of a newly formed double-stranded DNA during gametogenesis and embryogenesis). DNMT1 = DNA methyltransferase 1: It catalyzes replicationdependent maintenance of DNA methylation existing on hemimethylated DNA. Red dots stand for methyl groups. Source: Icons were obtained and customized in BioRender (biorender.com (accessed on 23 February 2021)).

#### *2.2. Histone Post-Translational Modifications and Chromatin Remodeling*

Histones are highly conserved basic proteins with a net positive charge. They have variable structures and amino acid composition that form the chromatin core. In eukaryotes, the chromatin core comprises a spherical octamer of four pairs of histone variants, namely H2A, H2B, H3, and H4, on which the DNA binds. Additionally, a fifth variant (H1) connects the DNA-bound octamers at regular intervals [32]. Each of the histone variants has an extended tail of amino acids at its N-terminus, which supports epigenetic modifications such as methylation, acetylation, phosphorylation, ubiquitination, ribosylation,

citrullination, and SUMOylation [33–36]. Chromatin organization is achieved due to the tight winding of DNA around the histone octamers to form condensed structures called nucleosomes. Tightly packed nucleosomes form a higher-order structural organization called chromatin [22,37,38]. Based on the degree of packaging, chromatin can be heterochromatin or euchromatin. Heterochromatin is tightly packed and inaccessible to transcription factors, while euchromatin is loose and accessible [26].

Mechanistically, histone modifications regulate transcription in two ways: one is by facilitating the transition between heterochromatin and euchromatin (chromatin remodeling), and the other is by serving as scaffolds or binding sites for reader proteins, which in turn recruit other proteins that write or erase epigenetic marks [37,39,40]. It is conceivable that modifications that strengthen DNA–histone interactions to form tight chromatin usually lead to gene silencing, while those that disrupt this interaction promote gene expression [38]. Interestingly, histone modifications work in a coordinated manner. Different histone marks engage in unique crosstalk or interaction that maintains the chromatin's highly dynamic state. This interaction will be discussed further under this review's subsequent headings, describing zinc's role in the various forms of histone modifications.

#### **3. Molecular Bases of Epigenome Regulation Associated with Zinc Metalloproteins**

Some of the epigenetically active enzymes identified so far are zinc metalloenzymes. Zinc is essential for the catalysis and autoregulation of enzymes such as DNMTs, histone methyltransferases/methylases (HMTs), histone demethylases (HDMs), histone acetyltransferases/acetylases (HATs), histone deacetylases (HDACs), histone E3-ubiquitin ligases (EUBLs), and histone deubiquitinating module (DUBm) complexes [4,41]. Interestingly, zinc is not only required for the integrity, catalysis, and self-regulation of these enzymes but also in recognition of their substrates and their recruitment to binding sites by other zinc-binding proteins [31,42–44]. Furthermore, the methionine synthase and betaine–homocysteine methyltransferase essential in DNA and histone methylation are zinc-dependent [45]. Additionally, zinc finger proteins (ZFPs) are also involved in epigenetic regulations, especially in epigenome editing [46]. ZFPs are the largest group of transcription factors with diverse structures that commonly have a zinc finger domain (ZFD) housing structural arrays of amino acids coordinated by one or more zinc atoms called zinc finger motifs (ZFMs) [10,47].

#### *3.1. Role of Zinc in DNA Methylation*

Each of the three DNMTs is composed of complex multi-functional domains categorized into a C-terminal catalytic domain and an N-terminal regulatory domain [31]. DNMT1 contains approximately 1620 amino acid residues [48]. About 78 of the total amino acids (about 4.83%), which comprises residues 621–698, form a ZBD called the CXXC domain (C for Cysteine and X for any amino acid) [49]. The domain is part of the N-terminal regulatory region, which interacts with the unmethylated CpG islands of a hemimethylated DNA to facilitate the self-inhibition of DNMT1 through an autoinhibitory linker. In this way, the enzyme is prevented from de novo DNA methylation [31,50,51]. Furthermore, the interaction between the ZBD and the catalytic domain of DNMT1 is essential for the enzyme's allosteric activation [4]. Thus, the ZBD forms part of the N-terminal regulatory domain that controls the enzyme's catalytic activity. Additionally, the ability of DNMT1 to recognize hemimethylated DNA is dependent on its interaction with a ZFP called the ubiquitin-like protein 1, containing plant homeodomain (PHD) and a really interesting new gene (RING) finger domains (UHRF1). This protein senses DNMT1 and reads methylation patterns on a hemimethylated DNA with the aid of its PHD and SRA (SET and RING associated) domains [43]. Thus, it recognizes the methylation marks on the DNA and directs the enzyme to those marked regions [44,52]. Similarly, the N-terminal regulatory domains of both DNMT3a and DNMT3b harbor a cysteine-rich complex multi-subunit domain called the "alpha thalassemia mental retardation x-linked DNA methyltransferase3 DNA methyltransferase3L" related domain abbreviated as *ATRX-DNMT3-DNMT3L* or simply ADD

domain. The domain encloses two ZFDs, namely a Guanine-Alanine-Thymine-Alaninelike (GATA-like) and a PHD together with a C-terminal alpha-helix [42,53]. ADD domain facilitates the autoinhibition and allosteric activation of both enzymes [31]. In addition to DNMT3a and b, DNMT3-like protein (DNMT3L) is another member of the DNMT3 family identified in humans [54]. This protein also contains the ADD domain but lacks the catalytic domain and helps in the allosteric regulation of both DNMT3a and DNMT3b [27,42]. In 2001, Bourc'his and co-researchers reported that some offspring of a homozygous knockout (DNMT3L-/-) model of female mice died before midgestation [55]. According to these researchers, the dead fetuses had hypomethylated maternally imprinted genes. This finding suggests the role of DNM3L in the de novo methylation of maternally imprinted regions of the DNA. A year later, Chedin et al. (2002) [54] co-expressed DNMT3L with DNMT3a in a human cell line. Their intervention enhanced the de novo methylation activity of DNMT3a at the targeted DNA sequences irrespective of which sequence is involved, but with little or no effect on DNMT3b. However, studies reviewed by Suetake et al. (2004) [56] have reported similar results for DNMT3b. These observations entail that DNMT3L has a stimulatory role on both DNMT3a and DNMT3b. Mechanistically, the stimulatory role of DNMT3L on DNMT3a/b may not involve their recruitment to the targeted regions. Instead, it may depend on the allosteric interaction between the C-terminal half of DNMT3L, and these enzymes' catalytic domains [56]. Moreover, DNMT3L could bind to DNMT3a/b but not the DNA itself [56], further emphasizing the lack of the protein's catalytic domain. Furthermore, a more recent study indicated that DNMT3L exerts its regulatory function through its PHD-like ZFD (the ADD domain) [57]. In a nutshell, these findings imply that the ZFD (ADD) in the DNMT3 family dictates the enzymes' catalytic domains.

Emerging evidence suggests a strong correlation between cellular zinc levels and the expression and activities of DNMTs in different experimental models. For instance, studies have shown a significant increase in the protein expression levels and activities of DNMT1 and DNMT3A in zinc-deficient human esophageal cancer (EC) cell lines compared to similar cell lines without zinc deficiency [58,59]. Consequently, zinc deficiency improved these cells' radiosensitivity by enhancing the hypermethylation of the microRNA 193b gene promoter via the upregulation of DNMTs. MiR-193b induces radioresistance in EC cells by arresting their cell cycle through the downregulation of Cyclin D1 mRNA [58]. One possible mechanism by which zinc deficiency could upregulate DNMT1 activity is by distorting the N-terminal CXXC domain's integrity due to lack of zinc, which may result in loss of its autoinhibitory function.

Moreover, in another study on hypozincemia-induced cognitive dysfunction in rats, zinc deficiency led to the upregulation of DNMT1 transcription in their hippocampus and subsequent hypermethylation of the brain-derived neurotrophic factor (BDNF) gene and its downregulation [60]. BDNF is highly expressed in the mammalian brain's hippocampus and is one of the critical regulators of learning and memory [61,62]. In contrast, no significant change in DNMT3B gene expression was reported in the study, while DNMT3A expression was downregulated [60]. Furthermore, a more recent study on the effect of zinc supplementation on offspring's cognitive function in rats reported a significant downregulation of DNMT1 and BDNF as well as the upregulation of DNMT3A protein levels in the hippocampus of the F1 neonates of zinc-deficient dams; however, all three proteins were downregulated in the lactating offspring of these rats [63]. According to the researchers, there was no significant hypermethylation of the BDNF gene in the offspring throughout the experiment. Its downregulation at the developmental stage correlates with the neonatal upregulation of DNMT3A, which may be due to the deformation of its regulatory ZBD (the ADD domain).

Additionally, this observation further illustrates the enzyme's de novo methyltransferase activity described earlier. However, there were no significant changes in the protein expression levels of DNMTs and BDNF following post-weaning zinc supplementation, although DNMT3A and BDNF showed an increasing trend [63]. This finding implies that zinc supplementation in offspring could restore the expression of DNMTs and DNA methylation changes in some genes, which an early life zinc deficiency might have perturbed in parents. Thus, it is conceivable from these observations that zinc deficiency affects the DNA methylation statuses of genes through different pathways involving DNMTs in vivo. It is also noteworthy that hypozincemia-induced perturbations in the DNA methylation status depend on the cell's physiological or pathological state.

#### Role of Zinc in Folate-Mediated One-Carbon Metabolism

Zinc indirectly affects DNA and histone methylation through one-carbon metabolism, which is a term used to describe the relationship between folate and methionine metabolic pathways [64]. One carbon unit (such as methyl and formyl groups) is transferred to folic acid from amino acids and then redistributed to other molecules through SAM to facilitate methylation reactions, including DNA and histone methylations [65].

The methyl groups added to DNA and histones usually come from amino acids, such as serine and betaine/trimethylglycine (obtained from choline). Serine donates a 1C unit to form glycine and 5, 10-methylenetetrahydrofolate. Then, the latter is reduced to 5-methyltetrahydrofolate by methylenetetrahydrofolate reductase (MHTR). Then, the activated folate donates its methyl group to homocysteine to form methionine, and thus, THF is regenerated for another cycle. This reaction is catalyzed by 5-methyltetrahydrofolatehomocysteine methyltransferase (MTR), which is also known as methionine synthase [65,66]. MTR is dependent on zinc and biotin [16]. In the active site of MTR, the zinc ion is coordinated by three cysteine residues, namely C217, C272, and C273, in a trigonal bipyramidal structure [67], which is critical for the catalytic activity of this enzyme. Following synthesis, methionine reacts with ATP to generate SAM, which facilitates DNA, RNA, and histone methylation, among other methylation reactions, and homocysteine is regenerated for another cycle. A study has shown that MTR gene expression was significantly downregulated in the liver and kidney of zinc-deficient rats compared to control. These changes could not be reversed by zinc supplementation [68]. The study also reported elevated levels of homocysteine (an immediate substrate of MTR) in the serum of zinc-deficient rats, further indicating the enzyme's downregulation. One possible consequence of this finding is the depletion of SAM (a cofactor of DNMTs and HMTs), thus affecting DNA and histone methylation (Figure 2).

Alternatively, methionine is generated from homocysteine and betaine in another reaction catalyzed by betaine–homocysteine methyltransferase (BHMT) [64,65,69]. This enzyme also requires zinc for its activity [16] (Figure 2). In the active site of BHMT, the zinc atom is coordinated by the thiol groups of three cysteine residues, namely C217, C299, and C300 tetrahedrally, in addition to a fourth, which is coordinated by the OH of tyrosine (Y77). This coordination is also critical for the enzyme's catalytic activity [70]. In a study involving mice initially fed with a high-fat diet and later treated with either a zinc diet (containing 30 g zinc) or a no-added zinc diet, there was a significant downregulation of the BHMT gene and protein expression, lower methionine levels, and reduced homocysteine clearance in the liver of zinc-deficient mice; the mRNA and protein levels of Specificity protein 1/Sp1 (a ZF transcription factor that regulates the transsulfuration pathway of methionine metabolism) were also suppressed in the same group of mice [71]. Notably, these observations reaffirm the vital role of zinc in methionine and SAM synthesis as well as homocysteine clearance, which is mechanistically due to its ability to regulate the expression and function of MTR, BHMT, Sp1, and perhaps other unidentified zinc metalloproteins involved in the various pathways linking homocysteine to SAM. Thus, the findings explain zinc's role in maintaining DNA and histone methylation facilitated by SAM, which is a product of one carbon metabolism and a cofactor of DNMTs and HMTs.

**Figure 2.** Role of zinc in DNA and histone methylation through the folate-mediated one-carbon metabolism. This figure summarizes the pathways linking one-carbon metabolism with DNA and histone methylation. Zinc-dependent enzymes and key intermediates are highlighted in different colors. BHMT (**orange**) = Betaine homocysteine methyltransferase: A zinc-dependent enzyme, which catalyzes the alternate reaction of methionine synthesis; DMG = Dimethylglycine; DNMTS (**dark green**) = DNA methyltransferases: Catalyze DNA methylation and have zinc-binding domains that regulate their activities; HMTs (**dark green**) = histone methyltransferases: Catalyze histone, lysine, and arginine methylation; MTR (**pink**) = Methionine synthase: It is zinc-dependent and catalyzes the direct synthesis of methionine (the direct precursor of SAM) from homocysteine and N5-methyl THF; SAH = S-Adenosylhomocysteine; SAM (**light green**) = S-Adenosylmethionine: The activated methyl donor for methylation reactions; THF = Tetrahydrofolate: The active form of folate that supplies methyl groups to homocysteine to form methionine. Zn2+ (**red**) = Zinc ion. Source: Image was created in Corel Draw X3 Graphics Suite (Pixmantec Rawshooter essentials 2005. Version 1.1.3).

#### *3.2. Role of Zinc in Histone Methylation*

Histone methylation is the addition of one, two, or three methyl groups on the ε-nitrogen of lysine or the guanidino nitrogen of arginine residues of mainly H3 and H4 tails, which is catalyzed by HMTs [72,73]. Histone arginine methylation is catalyzed by protein arginine methyltransferases (PRMTs). In contrast, histone lysine methylation is catalyzed by lysine-specific histone methyltransferases containing or not containing the evolutionarily conserved SET domain (suppressor of variegation, enhancer of zeste, and trithorax) [72,74]. SET is a catalytic domain initially discovered in the expression product of some genes responsible for heterochromatin formation and the white eye phenotype in Drosophila melanogaster [75]. In some SET domain-containing HMTs, the domain harbors either post-SET only or pre-SET and post-SET cysteine-rich ZFDs. These domains interact with the SET domain to maintain its integrity and regulate its catalytic activity [73,74]. For instance, the two yeast HMTs, cryptic loci regulator 4 and histone-lysine N-methyltransferase dim-5 (Clr4 and Dim-5) are SET domain-containing HMTs, and both have pre-SET and post-SET ZFDs flanking their SET domain [76,77]. Clr4 is a reader and a writer of histone 3 lysine 9 (H3K9) methylation mark and facilitates chromatin condensation in Schizosaccharomyces pombe [76]. On the other hand, Dim-5 methylates H3 at K9 and facilitates DNA methylation in Neurospora crassa [78]. The pre-SET ZFD of both enzymes exerts a structural function and encloses three zinc ions in trigonal coordination with nine cysteine residues. Each zinc ion bonds to four cysteine residues tetrahedrally such that three are individually attached while it shares the fourth one with the neighboring zinc atom [76,79]. The post-SET domain in both proteins appears to be a cysteine-rich ZFD enclosing a zinc ion in tetrahedral coordination. It seems to be involved in SAM binding, although its exact function remains elusive [74,79]. Histone lysine methyltransferases are very diverse in eukaryotes, including humans. About seven families and a few orphan

members have been identified in humans. SUV39H1 and SUV39H2 are examples of human homologs of the yeast Clr4 harboring the pre-SET and post-SET ZFMs [79,80]. SET domain-containing HMTs catalyzes the methylation of H3, mainly at K9 and K4 [45,76]. Interestingly, histone lysine methylation does not disrupt the lysine residues positive charge and has fewer effects on DNA–histone interaction. However, unlike DNA methylation that often leads to gene silencing, histone lysine methylation activates or represses genes either by providing a binding site for methyl reader proteins, which in turn recruit the transcription machinery to the targeted genes or by remodeling the chromatin architecture, thus affecting the ability of transcriptional complexes to access DNA [23,52,81]. Generally, the methylation of H3 at K4, K36, and K79 leads to gene activation, whereas methylation at K9 and K27 of H3, as well as K20 of H4, altogether leads to gene repression [45].

#### *3.3. Role of Zinc in Histone Demethylation*

Histone demethylation is the reverse of histone methylation catalyzed by HDMs. It involves the removal of methyl groups from the lysine and arginine residues of histone tails. Histone demethylases are of two types: lysine demethylases (KDMs) and peptidyl-arginine demethylases (PADs) [82]. KDMs are the class of histone demethylases with ZBDs. Based on their catalytic mechanisms, they are subdivided into two groups of six families: a flavin adenine dinucleotide-dependent amine oxidases (AOF/KDM1) and an iron (ii) and alphaketoglutarate-dependent Jumonji-containing (JmjC) dioxygenases (KDMs 2-6) [4,83]. Two members of the KDM1 family, lysine-specific demethylases 1 and 2 (LSD1 and LSD 2), have been identified [80]. LSD2, also known as KDM1B or AOF1, has approximately 822 amino acids, some of which form an N-terminal ZFD with C4H2C2-type and CW-type classes of ZFMs, and specifically demethylates H3K4me1 and H3K4me2 marks [84]. The N-terminal ZFD (residues 50–264) facilitates substrate specificity and maintains the enzyme's active conformation [85]. Mutations in the genes transcribing the ZF components of this domain induce conformational changes in the amine oxidase domain, with subsequent loss of the demethylase activity [86]. This observation suggests its role in the catalytic mechanism of the enzyme. The second family of lysine demethylases encompasses more than 30 proteins, all of which contain the JmjC domain [82]. A typical example of a KDM in this family is KDM2B, also known as *JHDM1B/FBXL10/NDY1*, which demethylates H3K4me3 and H3K36me2. In addition to the JmjC domain, the enzyme contains two other zinc finger domains: the CXXC and the PHD domains, together with an F-box domain. The JmjC domain erases the H3K36me2 mark, the CXXC is for unmethylated DNA binding and recruitment of transcription factors, and the PHD serves as a histone modificatison reader domain [83]. A study has reported the enrichment of the H3K4me3 mark in the thymus of hematopoietic cell-specific knockout mice model of the CXXC domain of KDM2B [87]. This observation supports the earlier mentioned role of this domain in the demethylation of the H3K4me3 mark.

#### *3.4. Role of Zinc in Histone Acetylation*

Histone acetylation is the transfer of a negatively charged two-carbon unit, the acetyl group (CH3CO−), from acetyl-CoA to specific N-terminal lysine residues of histone tails [88]. Being negatively charged, the acetyl group neutralizes the positive charge on the acetylated lysine residues and hence reduces the overall positive charge on the histone protein. Consequently, this leads to the disruption of DNA–histone interactions (electrostatic interactions between negatively charged DNA and the positively charged histones), making the DNA more accessible to the transcription machinery. Therefore, histone acetylation favors euchromatin formation and activates gene expression [4,88–90]. Histone acetylation is catalyzed by HATs, some of which are zinc-dependent. Several families of HATs have been identified in eukaryotes, including humans [91]. Examples of ZBD-containing HATs are the members of the MYST family, which is an acronym derived from the names of two human genes and two yeast genes: human monocytic leukemia ZFP (MOZ), yeast bf2, also known as ySas3 or KAT6 (histone lysine acetyltransferase 6), yeast Sas2 (KAT8), and human Tip60 (KAT5). The defining feature of this class of HATs is the N-terminal MYST domain, which has an intrinsic HAT activity [92]. Some MYST family members contain zinc within a C2HC-type ZFM and a PHD-linked ZFD, which helps in the identification of and interaction with substrates [91,93]. Five members of the MYST family have been identified in humans: two important ones are MOZ and its paralog MORF (MOZ Related Factor), which are also known as KAT6A and KAT6B (previously MYST3 and MYST4), respectively; both catalyze the acetylation of H3 at K9 and K14 [94]. Similarly, the N-terminals of both enzymes enclose two tandem PHD ZFs that enable H3 recognition and binding and regulate the HAT domain [93,94].

#### *3.5. Role of Zinc in Histone Deacetylation*

Histone deacetylation is the hydrolytic removal of an acetyl group from N-acetyl lysine residues on histone tails, which are catalyzed by HDACs. A total of eighteen HDACs hydrolyze the amide bond of N-acetyl lysine residues of histone tails in mammals, including humans [95]. These enzymes are of four categories or classes, which require either zinc or NAD+ for catalysis. Class I (HDACs 1-3, 8), class II (IIa: HDACs 4, 5, 7, 9 and IIb: HDACs 6, 10) and class IV (HDAC 11) are zinc-dependent, while class III (sirtuins 1-7) depend on NAD+ [96–99]. The zinc ion in the active sites of zinc-dependent HDACs coordinates an aspartate-histidine (D-H) dyad and a tyrosine (Y) residue (except in class IIa HDACs, in which the Y residue has been replaced by another H). The coordination maintains the appropriate positioning of the functional groups required for the catalytic process [9,97,100]. In this catalytic mechanism, the zinc ion launches an initial nucleophilic attack on the water molecule (an essential step in the catalytic mechanism common to numerous zinc metalloenzymes, including carboxypeptidase A and carbonic anhydrase). Then, the zinc ion induces a polarity on the carbonyl oxygen of N-acetyl lysine on the histone tail and thus facilitates a broad base nucleophilic attack (on the carbonyl carbon) by the zinc-bound water molecule [9,100,101]. In addition to the central zinc atom typical to all the eleven zinc-dependent HDACs, class IIa HDACs also enclose another zinc atom within a unique and highly conserved (among members of this class IIa) CCHC-type ZFM found adjacent to the entrance of their active sites [102]. Although this motif's exact function is not clear, it may be critical for maintaining these enzymes' catalytic activity and stability and may serve as a potential allosteric site to develop their inhibitors. HDAC inhibitors have been employed in the management of diabetes mellitus (DM) [103–105] and cancer [100,106,107]. Interestingly, some of these drugs' inhibitory mechanism is based on their ability to chelate the zinc ion in their active sites [9].

#### *3.6. Role of Zinc in Histone Ubiquitination*

Ubiquitination is the covalent attachment of one or a chain of ubiquitin molecules. It is one of the diverse post-translational medications of proteins, including histones [108]. Histone ubiquitination (mostly monoubiquitination) occurs predominantly at specific lysine residues of H2A and H2B. It is involved in transcription regulation and DNA repair [42]. This modification regulates transcription by three primary mechanisms: (1) direct remodeling of the chromatin architecture to increase or decrease DNA accessibility to transcriptional complexes, (2) recruitment of proteins that facilitate chromatin remodeling, and (3) as a prerequisite for other histone modifications such as methylation and acetylation [109]. Generally, protein ubiquitination occurs in three stages: activation, conjugation, and ligation catalyzed by ubiquitin-activating enzymes (E1), ubiquitin-conjugating enzymes (E2), and ubiquitin ligases (E3) [110].

The zinc-dependent class of E3 ubiquitin ligases (EUBLs) are the RING domaincontaining proteins that catalyze the monoubiquitination of H2A at K13, K15, K119, K127, and K129 as well as H2B at K34, K120, and K125, respectively [41]. The RING domain harbors two zinc atoms in tetrahedral coordination with a histidine ring and seven cysteine side chains. This arrangement creates a C3HC4-type ZFM with 40–60 amino acid residues arranged in the sequence, Cys-X2-Cys-X [9-39]-Cys-X[1-3]-His-X[2-3]-Cys-X2-Cys-X[4-48]- Cys-X2-Cys, where "X" represents any amino acid [111]. Histone ubiquitination marks are associated with transcription silencing, activation, and DNA repair or participate in crosstalk that regulates other histone marks, such as histone methylation, which is essential in maintaining the dynamic nature of the chromatin [41,112].

In humans, H2A monoubiquitination is catalyzed by three enzymes or enzyme complexes: (1) RING finger 168 (RNF168), which adds a unit of ubiquitin on H2A at K13 or K15 and promotes the non-homologous end-joining pathway in the DNA damage repair response; (2) members of the polycomb repressor complex 1 (RING1A, RING1B and RNF51), which ubiquitinate H2A at K119 and lead to transcriptional repression; (3) the BRCA1/BARD1 complex (BReast CAncer type-1 susceptibility protein/BRCA1-Associated RING Domain protein 1) ubiquitinates H2A at K127 K129 and promotes the homologous recombination pathway in the DNA damage repair response [112]. On the other hand, several RING finger families of enzymes and enzyme complexes that catalyze H2B monoubiquitination have been identified in humans. The RNF20–RNF40 complex, which mainly catalyzes H2BK120ub1, appears to be the predominant writer of H2Bub1 marks [41,112]. Another important class is the MOF–MSL (MOZ-related Factor–Male Specific Lethal homolog) complex comprising MOF, MSL1, MSL2, and MSL3. The MSL1/2 component catalyzes the monoubiquitination of H2B at K34, while the MOF component (a member of the MYST family of HATs discussed earlier) catalyzes H4K16 acetylation [113]. Of note, H2Bub1 at K120 and K34 has been linked to transcriptional upregulation. In contrast, H2Aub1 at K34 and K120 and H4K31ub1 are collectively involved in crosstalk that facilitates H3K4 and H3K79 methylation [41].

#### *3.7. Role of Zinc in Histone Deubiquitination*

Histone deubiquitination is the removal of the ubiquitin moieties attached to lysine residues of histone tails by hydrolysis. H2A/B deubiquitination is catalyzed by the *Spt-Ada-Gcn5-Acetyltransferase* (SAGA) coactivator complex, which is a multi-subunit complex with both histone acetylase and deubiquitinase activity that is structurally and functionally conserved from yeast to humans [114]. In both yeast and humans, the SAGA complex is organized into a functional unit called the deubiquitinating module (DUBm), comprising of four functional subunits: three of them form an N-terminal regulatory region, and the fourth one creates the catalytic domain [115]. The four components in the yeast DUBm are the ubiquitin-specific protease 8 (Ubp8), which is the catalytic domain, and the three N-terminal subunits, namely the SAGA-associated factor 11 (Sgf11), the transcription and mRNA export factor (Sus1), and the SAGA-associated factor 73 (Sgf73). These units altogether serve as a scaffold that maintains the catalytic domain's active form [115]. Of note, Sgf11 and Sgf73 are ZFDs required for the DUBm function and stability, respectively [116]. In humans, the ubiquitin-specific protease 22 (USP22) is the ZFD-containing component of the human SAGA complex that deubiquitinates both H2A and H2B in vitro [114]. Moreover, similar to the yeast complex, this module also contains hATXN7 (human ataxin 7), hATXN7L3 (human ataxin 7 like 3), and hENY2 (human enhancer of yellow 2 transcription factor homolog, which are the ZBD-containing protein orthologs of the yeast Sgf73, Sgf11, and Sus1, respectively. The units interact to allosterically regulate its activity [114].

#### *3.8. Zinc Finger Proteins That Read Epigenetic Modifications*

Apart from enzymes, a plethora of proteins containing ZBDs (collectively referred to as transcription factors) is also involved in epigenetics. These proteins can read/recognize and bind to methylation patterns on the CpG islands of DNA or the various modifications on the N-terminal lysine/arginine residues of histone tails and then recruit other proteins that alter chromatin conformation, leading to varying transcriptional consequences [10,11,117].

#### 3.8.1. DNA Methylation Readers and Their Binding Mechanisms

Based on the commonalities in their methyl binding domains (MBDs) and the binding mechanism, methyl binding proteins (MBPs) are of two types: readers of double (fully) methylated DNA and readers of hemimethylated DNA. The first group (readers of doublemethylated DNA) comprises the best-characterized families of MBPs [10]. The most typical feature uniting these diverse families of transcription factors is the presence of two or more Cys2His2 (C2H2) ZFMs. Each ZFM has approximately 30 amino acid residues arranged in a simple structure: a central zinc atom coordinating two histidine rings protruding from the α-helix and two cysteine side chains extending from the β-sheets (Figure 3). Then, these motifs form binding domains harboring a set of two, three, or more zinc fingers arranged in tandem repeats. Examples include the Broad-complex, Tramtrack, and Bric-a-brac, which are also known as POxvirus and Zinc finger (BTB/POZ) domains containing families such as Kaiso (ZBTB33), ZBTB4, and ZBTB38 [10,117,118].

**Figure 3.** Labeled cartoon representation of a Cys2His2 zinc finger motif. A C2H2 zinc finger is an essential structural component of fully methylated DNA reader proteins with methyl binding domains. The red dot represents the zinc atom; the spiral structure denotes the alpha-helix; the two antiparallel arrows are the beta-sheets. The two five-membered rings extending from the alpha helix are the imidazole rings of histidine; the two curves extending from the beta-sheets and pointing toward the red dot are the thiols of the two cysteine residues. Source: Icon was obtained and customized in BioRender (biorender.com (accessed on 23 February 2021)).

More members of ZF MBPs reviewed by Hodges et al. (2020) [10] include zinc finger protein 57 (ZFP57), a member of the Krüpple-associated box (KRAB) domain family; Krüppel-like factor 4 (KLF4), a member of the Krüppel-like factor family, Wilm's Tumor Gene 1 (WT1); Early Growth Response Protein 1 (EGR1); and CCCTC-binding factor (CTCF), which is also known as 11-ZFP.

These proteins' binding mechanisms are complex and transcription factor-dependent due to variable structures and amino acid composition. However, they generally present a unique interaction model. Here, the arginine residue that precedes the first histidine ring in the C2H2 ZFM forms a hydrogen bond with the 3 -G ring and a Van der Waals interaction with the 5 -mC of the CpG dinucleotide. This interaction creates a 5-methylCytosine/Thymine-Arginine-Guanine triad typical of fully methylated DNA readers [119]. Additionally, a conserved glutamate residue (among the readers of double methylated DNA) also interacts with the 5 -mC via an OHN-type and an OHC-type hydrogen bonds [10]. This mechanism has been extensively reviewed recently [117,120]. Moreover, a conserved lysine residue in ZBTB38 and other ZFPs (whose ZFM sequences were aligned) substitutes the arginine in methylated DNA binding [117]. Thus, this finding demonstrates another possible mechanism for CpG methylation recognition by ZFPs.

On the other hand, the hemimethylated DNA readers are mainly the SRA domaincontaining families of MBPs [121]. UHRF1 and UHRF2 are the members of this family

identified in humans. In addition to the SRA domain, both contain a ubiquitin-like (UBL), a tandem Tudor, a PHD, and a RING domain. Studies have shown that the ability of UHRF1 to recruit DNMT1 to the replication foci depends on the interaction between its SRA domain and hemimethylated DNA [31,43,44]. UHRF2 is also known to read the 5-hydroxymethyl cytosine (5 hmC) mark on DNA through its SRA domain and recruit DNMT1 to replicate foci [122]. 5 hmC is an oxidation product of 5 mC generated by the members of the TET family of proteins in the mammalian DNA demethylation pathway; it is found at the proximity of transcription factor binding sites and is also involved in the regulation of transcription [123,124].

#### 3.8.2. Histone Modifications Readers and Their Binding Mechanisms

Readers of histone modifications are numerous, very diverse, and histone mark specific. The best characterized ZF readers include bromodomain, chromodomain, and PHD readers. They recognize modified and unmodified lysine and arginine residues [125]. The chromodomain and bromodomain proteins recognize methylated lysine and acetylated lysine, respectively. In contrast, the PHD readers acknowledge both modified and unmodified lysine and arginine residues. Thus, PHD proteins are considered versatile readers of histone modifications [40].

The predominant readers of lysine methylation are the chromodomain containing members of chromatin modifiers (collectively referred to as the Royal family), which recognize methylated lysine mainly at H3 and participate in chromatin remodeling. A critical member of this family is the heterochromatin protein 1 (HP1). It binds to H3 methylations primarily at K9 and, to some extend K27, and it facilitates heterochromatin condensation [126]. Another important example of a methylated lysine reader is the Bromodomain PHD finger Transcription Factor (BPTF). This protein is a PHD containing a bromodomain reader considered the largest subunit of the ATP-dependent chromatin remodeling complex called the nucleosome remodeling factor (NURF) in humans. This protein has a preference for H3K4me2/me3 (the marks associated with the transcription start site of active genes) and thus remodels the chromatin state of those genes to enable their expression [127].

Due to their diversity, the binding mechanisms of methyl lysine readers differ from one reader to another. However, most of these proteins form an aromatic pocket with the side chains of tryptophan and tyrosine residues, which help them recognize and bind to various histone methylation marks. On the other hand, readers of unmodified lysine such as the PHD do not form an aromatic pocket; instead, they form hydrogen bonds between the ε-nitrogen of the lysine residue and the hydroxyl hydrogen of either aspartate or a glutamate residue in the PHD domain [125]. Table 1 summarizes the significant ZBDs discussed in this review; their functions and examples of human proteins containing these domains are given.

**Table 1.** Summary of zinc-binding domains and their functions.


#### *3.9. Role of Zinc in Epigenome Editing*

Due to their ability to recognize epigenetic marks, zinc finger proteins (especially the Cys2His2 class of ZFs) are also involved in synthetic epigenetics. It is a popular concept known as epigenome editing [128]. This synthetic form of epigenetic modification entails writing or erasing epigenetic marks on the DNA or histones via the recruitment of the natural catalytic activity of the chromatin-modifying enzymes [46]. Two basic techniques have been employed to achieve this purpose. One of them involves using artificially customized epigenetic mark readers (the Cys2His2 ZFs) to recruit these catalytic domains to the predetermined targeted regions on the DNA and histones [20]. The ZFs are coupled to the chromatin-modifying enzymes, and the resultant machinery is employed to add or erase epigenetic marks.

The abbreviations used in the table are defined hereunder, and the respective letters in each acronym are capitalized. ADD—Alpha thalassemia mental retardation x-linked DNA methyltransferase3 DNA methyltransferase3L; BTB/POZ—Broad-complex, Tramtrack, and Bric-a-brac, also known as POxvirus and Zinc finger; BPTF—Bromodomain PHD finger Transcription Factor; CXXC—C for Cysteine and X for any amino acid; DNMT1—DNA methyltransferase1; DNMT3a/3b/3L—DNA methyltransferase3a/3b/3L; HAT—Histone acetyltransferase; H3K36me2—Histone 3 lysine 36 dimethyl; H3K4—Histone 3 lysine4; H3K9/K14—Histone 3 lysine 9/lysine14; JmjC—Jumonji-containing; KDM2B—lysine demethylase 2B; KRAB—KRüpple-Associated Box; MORF—MOZ-Related Factor; MOZ human MOnocytic leukemia Zinc finger protein; MYST—MOZ, Ybf2 (Sas3), Sas2, and Tip60; PHD—Plant HomeoDomain; RING—Really Interesting New Gene; RNF168—Ring Finger168; PRC1—Polycomb Repressor Complex1; RNF20–RNF40—Ring Finger20–Ring Finger40; SET—Suppressor of variegation 3-9, Enhancer-of-zeste and Trithorax; SET1A— Suppressor of variegation 3-9, Enhancer-of-zeste and Trithorax 1A; SET1B—Suppressor of variegation 3-9, Enhancer-of-zeste and Trithorax 1B; SRA—SET and RING finger Associated domain; UHRF1—Ubiquitin-like, containing PHD and RING finger domains, 1; UHRF2—Ubiquitin-like, containing PHD and RING finger domains, 2; ZFP57—Zinc finger protein 57.

#### Role of Zinc in DNA Methylation Editing

DNMTs coupled to zinc fingers are used to introduce or delete DNA methylations at specific predetermined targeted regions. This technique has been used to alter the DNA methylation statuses of genes, affecting gene expression. For example, a customized array of seven ZFs coupled to the catalytic domain of DNMT3a was employed to induce DNA methylation on the p16 promoter of immunodeficient mice. Consequently, this led to increased metastasis and cancer cells' proliferation due to the p16 gene suppression [129]. P16 is a tumor-suppressor protein that was found to be associated with improved prognosis and early survival of patients with oropharyngeal cancer in China [130]. In ZF arrays, each unit of a ZF recognizes three nucleotide bases. Hence, the customized collection of zinc fingers used in chromatin recognition can bind to about nine to 18 DNA bases as each array has three to six ZFMs (Figure 4).

On the other hand, oxidative (active) DNA demethylation at targeted promoters of specific genes has been induced by the ten-eleven-translocases (TET) family members. This task was achieved by coupling these enzymes with specially designed ZFs that target those genes' promoters. For example, in a study conducted to induce oxidative DNA demethylation on human intracellular adhesion molecule 1 (ICAM-1) gene promoter, a hexameric ZF array was coupled to the TET-2 enzyme and used to target the promoter [131]. This intervention led to demethylation of the ICAM promoter and subsequent reactivation of the expression of the gene.

**Figure 4.** Cartoon representation of the zinc finger (ZF) array. A customized sequential array of six Cys2His2 ZFs: each of the six units containing a red dot surrounded by a coiled strand and two antiparallel arrows represent a Cys2His2 ZF. The red dot is the zinc ion coordinated by the side chains of two cysteines and two histidine residues. ZF arrays are employed in synthetic epigenetics to edit (add or delete) epigenetic marks at predetermined DNA or histone regions. They are usually coupled with writers or erasers to achieve this purpose. Source: Icons were obtained and customized in BioRender (biorender.com (accessed on 23 February 2021)).

#### **4. Summary of Zinc Metalloproteins in Epigenetics and Their Crosstalk**

Zinc metalloproteins involved in epigenetics are summarily classified into five main classes. Class I are the writers or markers: they include zinc metalloenzymes that establish epigenetic marks such as DNA methylation, histone methylation acetylation, and ubiquitination. Examples are DNMTs, HMTs, HATs, and EUBLs. Class II is nicknamed erasers due to their ability to erase or remove epigenetic marks such as histone methylation, acetylation, and ubiquitination. Examples include enzymes such as KDMs, HDACs, and DUBm complexes. Class III is made up of a large family of ZFPs that read epigenetic marks such as DNA methylation and histone modifications and subsequently generate signaling pathways by recruiting an array of proteins involved in the signal transduction process, which ultimately lead to the activation or silencing of genes [39,40,112]. This class is termed readers, and their examples include Kaiso, UHRF1, UHRF2, ZFP57, HP1, BPTF, etc. Class IV is made up of a customized combination of readers coupled with writers or erasers. These assemblies are employed in epigenome editing and are referred to as editors.

A classic example is the hexameric Cys2His2 ZF array associated with the TET-2 enzyme and used for the ICAM gene promoter [128]. Class V are zinc-dependent enzymes nicknamed feeders due to their ability to catalyze critical reactions in the folate-mediated one-carbon metabolism, which is a term used to describe the group of metabolic pathways involved in the supply of one-carbon units such as activated methyl group (in the form of SAM) from folate to DNA, histones, and other biomolecules. In other words, they feed the methylation pathways with active methyl groups for various methylation reactions, including DNA and histone methylations (Figure 5).

Interestingly, readers are involved in crosstalk with writers and erasers, and this interplay regulates the human epigenome, leading to various transcriptional outcomes in both health and disease conditions. For instance, a PHD protein such as UHRF1 can read modified or unmodified histones and then recruit a writer such as DNMT1 to the replication foci. The latter establish epigenetic marks on hemimethylated DNA.

**Figure 5.** Summary of zinc metalloproteins and their roles in epigenome regulation. Figure 5 summarizes the various classes of zinc metalloproteins that come together to regulate the epigenome. Writers, erasers, readers, editors, and feeders represent the names of the types. These classes of proteins work together in a coordinated manner to regulate the epigenome. Examples of enzymes/proteins in each category are given in abbreviations. BHMT = Betaine homocysteine methyltransferase: a zinc metalloenzyme that catalyzes the formation of methionine from homocysteine and betaine, an essential reaction in DNA and histone methylation; chromo = Chromodomain containing proteins such as heterochromatin protein 1 (HP1), an H3K9 methyl reader that facilitates heterochromatin condensation; C2H2 ZFs = 2-Cysteine-2-Histidine-type of zinc fingers, critical readers of fully methylated DNA; DNMTS = DNA methyltransferases, enzymes that catalyze de novo and maintenance DNA methylation at the CpG islands of fully methylated or hemimethylated DNA; HATs = Histone acetyltransferases, enzymes that catalyze histone lysine acetylation; HDACs = Histone deacetylases, enzymes that catalyze removal acetyl groups from acetyl-lysine residues of histone tails; HMTs = Histone methyltransferases, enzymes that catalyze the methylation of histones on lysine residues; KDMs = Histone lysine demethylases, enzymes that remove methyl groups from the methylated lysine residues of histones; MTR = Methionine synthase, an enzyme that catalyzes the synthesis of methionine from homocysteine and tetrahydrofolate, an essential reaction for DNA and histone acetylation; PHD = Plant homeodomain containing zinc finger proteins, they serve as versatile readers of histone modifications; SRA = SET and RING finger Associated domain-containing proteins: they read hemimethylated DNA. USP22 = Ubiquitin-specific protease 22: the catalytic component of the human deubiquitinating module complex that catalyzes histone deubiquitination. ZF MBPs = Zinc finger methyl binding proteins: readers of fully methylated DNA. Source: Image was created in BioRender (biorender.com (accessed on 23 February 2021)).

#### **5. Conclusions and Future Perspectives**

This paper reviewed zinc's epigenetics role, which depends on zinc metalloproteins' involvement in epigenome programming. We also discussed here that these proteins contain ZBDs critical for substrate recognition, self-regulation, and catalysis. Most importantly, we demonstrated that an interplay involving the ZBDs of these proteins maintains the highly plastic epigenome in a dynamic state. Therefore, we conclude that zinc is an essential trace metal in epigenetics. However, despite the plethora of ZBD-containing proteins identified and still being discovered, only a few of them have been employed to manipulate the epigenome. Non-communicable chronic diseases such as cancers, diabetes mellitus, and cardiovascular diseases have been associated with aberrant epigenetic changes in the genes related to these diseases; targeting these genes with customized ZBDs could make a remarkable difference in minimizing their burden. For instance, the hypermethylation of

oncogenes or demethylation of tumor suppressor genes by epigenetic editors could help control various cancers. Furthermore, studies on transgenerational epigenetic effects at different doses of parental zinc exposure on offspring could unveil how zinc deficiency affects future generations' health. Thus, the epigenetic burden of diseases could be minimized.

**Author Contributions:** Conceptualization, A.P.Y., M.B.A., M.U.I.; writing—original draft preparation, A.P.Y.; writing—review and editing, M.B.A., I.M., K.G.I., B.A., M.B.B., N.Q., S.T.E., M.U.I., A.A., G.E.-S.B.; project administration, M.U.I.; funding acquisition, A.A. and M.U.I. All authors contribute to the writing—review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by a grant from the Center for Advanced Medical Research and Training (CAMRET), Usmanu Danfodio University Sokoto, Sokoto, Nigeria. Abdurrahman Pharmacy, Yusuf is a recipient of CAMRET research funded scholarship.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

