We comprehensively and historically perused publications regarding carnivorous SC/SV on PubMed dating from 1964 to 10 January 2024 as well as searched for papers using the search string “[(spermatozoa) AND (cryoprotectant)] AND (carnivores)” that adhered to the PRISMA guidelines (Liberati et al., 2009), as demonstrated in Figure 2. These studies were carefully screened, and eligible works were selected based on the following inclusion criteria: (1) studies published in English, (2) studies aimed at reporting carnivorous SC/SV, and (3) studies reporting CPAs and extenders. We excluded papers focused on other animals or the cryopreservation of different organs as well as review articles; however, the references of such works were scrutinized to find relevant works.

The physicochemical attributes of CPAs used in carnivorous SC were curated from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Then, the chemical properties, molecular structures, canonical simplified molecular input line entry system (SMILS), XlogP3, logS, cellular locations, molecular weights in g/mol, melting points in °C, topological polar surface areas (TPSAs; Ų), and toxicity classes of common small-molecule CPAs were computed.

A knowledge-based approach based on an additive atom/group model starting from the known logP value of a similar reference compound (Cheng et al., 2007) was employed to compute XlogP3. Here, the logP value is a constant defined as log10 (partition coefficient) with the partition coefficient P = [organic]/[aqueous], where [ ] indicate the concentrations of the solutes in the organic and aqueous partitions. A negative value of logP indicates that the compound has a higher affinity for the aqueous phase (more hydrophilic); when logP = 0, the compound is equally partitioned between the lipid and aqueous phases; a positive value of logP denotes a higher concentration of the lipid phase (i.e., the compound is more lipophilic). For instance, logP = 1 indicates a 10:1 partitioning of the organic:aqueous phases. In contrast to XlogP3, logS is related to the water solubility (mol/L) of a chemical and is expressed as a common solubility unit in log10 value. The molecular polar surface area (PSA) refers to the surface area corresponding to the polar atoms and is a descriptor of the passive molecular transport through the membranes for predicting the transport properties of chemicals (Ertl et al., 2000).

SwissADME (http://www.swissadme.ch/) (Daina et al., 2017) was used to compute the biological activities. The bioavailability radar offers a speedy assessment of the medication resemblance of a particle (Daina et al., 2017). Six physicochemical properties (lipophilicity, size, polarity, solubility, flexibility, and saturation) were computed for each of the CPAs, and these descriptors were established with physicochemical ranges along each axis. The pink areas outline the best possible regions for each of the ranges. Briefly, the fitness parameters include lipophilicity (LIPO) as logP (XLOGP3) values ranging from −0.7 to 5.0; SIZE as molecular weights in the range of 150–500 g/mol; polarity (POLAR) as TPSA values ranging from 20 to 130 Ų; insolubility (INSOLU) as solubility with logS (ESOL) values ranging from −6 to 0; insaturation (INSATU) as saturation fractions of carbons with sp3 hybridization (fraction Csp3) ranging from 0.25 to 1; and flexibility (FLEX) as the number of rotatable bonds ranging from 0 to 9 (Daina et al., 2017).

The boiled-egg construction depicts a snapshot of the human intestinal absorption (HIA) and blood–brain barrier (BBB) permeability of a substance (Daina and Zoete, 2016), where the blue dots represent the P-glycoprotein (PGP) substrates (PGP+) and red dots indicate the PGP non-substrates (PGP−). The outer gray area represents compounds with lower gastrointestinal absorption and limited BBB penetration; the white area represents the physicochemical space of the molecule with the highest probability of passive HIA; the yellow area represents the physicochemical space of the molecule with the highest likelihood of BBB penetration.

For the most part, a chemical reaction has some basic equilibrium constant parameters; however, these are not effortlessly obtained from experiments. Herein, we show that the computed thermodynamic properties of CPAs used for carnivorous SC are dependent on Chemeo, which includes high-quality compound properties (www.chemeo.com), and the Joback method that produces an anticipated value of 813.3 K. The actuation entropy (S°), enactment enthalpy (ΔfH°), and Gibbs free energy obtained from the thermodynamic properties contrast between the explicit reactant and change states compared to individual response stages subbed into the progress state theory (Benkert et al., 2007), as shown in Equations 1, 2 to gage the rate steady (K) and preexponential factor (A). The thermodynamic equilibrium constant of each response was obtained from the ratio of the forward to retrogressive rate steady values for the resulting correlation. K = k B T / h p / RT 1 − n exp S ° / R exp − Δ fH ° / RT (1) A = k B T / h p / RT 1 − n exp n exp S ° / R (2) where k_B is the Boltzmann constant, h is the Planck constant, and n is the molecularity.

We reported the initial values of all chemical properties of CPAs, from which their proton affinities were computed. The basicity was determined from the presence or absence of the reaction shown in Equation 3 and its inverse reaction, where A is an amino acid, B is a reference base, and AH⁺ and BH⁺ are the protonated versions of the amino acids and bases, respectively (Gorman et al., 1992; PAaJd, 2006). A + BH + →   AH + + B (3)

In addition, the optimized structures of the chosen CPA mixtures were predicted using density functional theory (DFT), GaussView (Frisch HBS et al., 2009), and Gaussian 09 program (Frisch et al., 2009) with 6-311++G (d, p) premise sets. The potential limitations of the chosen tools were addressed in the study, and the parameters chosen were essential for the transparency and robustness of the study based on careful consideration of their strengths. For example, SwissADME is known for its ability to predict the absorption, distribution, metabolism, and excretion (ADME) properties, which are crucial for assessing the suitability of a cryoprotectant, and PubChem is a widely used and authoritative resource for chemical information.

The features of computational toxicology including absorption, distribution, metabolism, excretion, and toxicity (ADMET) were predicted using a web ADMET server (http://biosig.unimelb.edu.au/pkcsm/prediction) (Pires et al., 2015). This allowed determination and expression of the ADMET boundaries through investigation of human influences on medication (pharmacokinetic properties), neutral nature, and the unwavering quality of the restorative media on small-molecule CPAs. Using the SciDaviz package (Benkert et al., 2007), the correlation distribution properties were plotted in graphs in the vertical direction and structured based on expansion with different CPAs.

A set of endpoints including water solubility (log mol/L), colorectal adenocarcinoma cells (Caco2) permeability (log(Papp) in 10^–6 cm/s), HIA (% absorbed), skin permeability (logKp), PGP substrate, PGP I inhibitor, and PGP II inhibitor was computed to predict the various absorption components of the ADMET assay (https://biosig.lab.uq.edu.au/pkcsm/theory). For instance, using the in vitro absorption data of 674 compounds, pkCSM predicts the apparent cellular permeability coefficient of the compound of interest. For the pkCSM predictive model, high Caco2 permeability translates to values greater than 0.9.

Another set of endpoints including steady-state volume distribution (VDss; human; log L/kg), unbound fraction (human; Fu), BBB permeability (log BB), central nervous system (CNS) permeability (logPs), and total clearance (log mL/min/kg) was computed to predict the various distribution and excretion components of the ADMET assay (https://biosig.lab.uq.edu.au/pkcsm/theory). A set of substrates and inhibitors of the cytochrome (CYP) class including the CYP2D6 and CYP3A4 substrates as well as inhibitors of CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4 were computationally assessed to predict the metabolism part of the ADMET assay (https://biosig.lab.uq.edu.au/pkcsm/theory). An array of toxicological endpoints, including AMES toxicity maximum tolerated dose (MTD; human; log mg/kg/day), Ether-à-go-go-related gene (hERG) I inhibitor, hERG II inhibitor, rat oral acute toxicity (LD50 mol/kg), rat oral chronic toxicity (LOAEL; (log mg/kg-bodyweight/day), hepatotoxicity, skin sensitization, Tetrahymena pyriformis toxicity (log µg/L), minnow toxicity (log mM), and toxicity class, were then computed to predict the various toxicity components of the ADMET assay (https://biosig.lab.uq.edu.au/pkcsm/theory).

Furthermore, a set of toxicological endpoints was employed to show the ecofriendliness of the CPA of interest. For instance, the AMES test is a globally accepted bacterial assay for evaluating the potential genotoxicity, mutagenicity, and even carcinogenicity of chemicals (Zeiger, 2019). The relevancy of the AMES test for SC/SV requires further investigation; however, CPA remnants can alter the reproductive microbiota of packs or pets that have received AI, especially if repeated services have been attempted. This relevancy can be examined by measuring the ability of the CPA to induce reverse mutations at selected loci in several bacterial strains. The cardiotoxic hERG binding of a CPA may seem irrelevant to its evaluation. The human hERG codes for the alpha subunit of the potassium ion channel known as K_v11.1, which is dubbed so because of its role in coordinating the electrical activity of the heart. The flexible nature of hERG to bind to various ligands was used in its selection as a target in the ADMET assay of all compounds to avoid cardiotoxicity (Garrido et al., 2020). Skin sensitization is a predictor of potential adverse effects such as dermatitis for chemicals applied to the skin. The MTD is commonly estimated as the highest dose that can be administered for a specific duration that will not compromise the survival of the animal through causes other than carcinogenicity (Shayne, 2024). Itis noted that MTD is sometimes also referred to as the minimum toxic dose. The acute median lethal dose (LD50) is the dose of a chemical that will kill 50% of the test organisms within 24 h of exposure. Acute toxicity studies are usually conducted for various routes of administration in rodents to provide the baseline values for novel chemicals of interest (Gadaleta et al., 2019). The computation model for LD50 was built on over 10,000 compounds tested in rats and predicted in units of mol/L. In addition, chronic studies such as rat oral chronic toxicity aim to identify the minimum dose of a substance that results in the least observable adverse effect level (LOAEL) and maximum dose for the no observable adverse effect level (NOAEL). This predictor was built using the LOAEL library comprising 567 compounds.

Tetrahymena pyriformis toxicity (log µg/L) is a globally accepted screening test for the detection of environmental toxicants (Maurya and Pandey, 2020). Here, the laboratory-adopted protozoan Tetrahymena pyriformis is the most commonly used ciliated model for primary toxicological research, and the dose that inhibits 50% of its growth is the toxic endpoint, which has been identified as values > −0.5 log µg/L. At a higher organism level, minnow toxicity is used to predict the lethal concentration (LC50) of any chemical that would kill 50% of flathead minnows; here, LC50 values below 0.5 mM (log LC50 <-0.3) indicate high acute toxicity. Finally, based on the hepatic injuries induced by 531 compounds, pkCSM computationally predicts the hepatotoxicity of the compound of interest.

The Toxtree v.3.1.0-1851-1525442531402 software platform (http://toxtree.sourceforge.net/) was employed to predict the toxicity class based on the Cramer rule (Cramer et al., 1976; Munro et al., 1996; Patlewicz et al., 2008). Here, the toxicological hazard of oral administration is estimated from the molecular structure, and three classes are defined as follows. Low oral toxicity or Class I substances have simple chemical structures and efficient modes of metabolism; here, the no observable effect level (NOEL) at the fifth percentile is 3.0 mg/kg-bodyweight/day and human exposure threshold is 1.8 mg/person/day. The NOEL is the highest dose or exposure level of a chemical that produces no noticeable (observable) toxic effects. Intermediate oral toxicity or Class II substances possess structures that are less innocuous than Class I substances but do not contain structural features suggestive of toxicity; here, the NOEL at the fifth percentile is 0.91 mg/kg-bodyweight/day and human exposure threshold is 0.54 mg/person/day. High oral toxicity or Class III substances have chemical structures that permit no strong initial presumption of safety or may even show significant toxicity or have reactive functional groups; here, the NOEL at the fifth percentile is 0.15 mg/kg-bodyweight/day and human exposure threshold is 0.09 mg/person/day. The hazardous substances data bank (HSDB) at https://www.nlm.nih.gov/toxnet/index.html is a reliable index for deciphering the biosafety of extenders and CPAs used in the SC/SV of carnivores. The toxin and toxin target database (T3DB; http://www.t3db.ca/toxins/) was employed to compute the other toxicological features.

The data were described in terms of mean ± standard error of the mean (SEM) for all features of the selected CPAs in this study. Furthermore, all descriptive statistics of the numeric features were analyzed using IBM SPSS Statistics ver. 20 (https://www.ibm.com/spss).

First, 185 articles were screened, and no duplicate works were found. Subsequently, the publications were vetted by title revision, and three review articles were excluded. Thus, 182 articles were selected based on the titles, including those concerned with in vivo experiments, research linked to female animals, and toxic substances, along with publications involving species such as humans, mice, and bulls (Figure 2). Among the remaining articles, 83 full-text articles were assessed and discussed by the coauthors to find the best and most effective CPA candidates for dogs (Table 1). Finally, 62 full-text appropriate papers on SC/SV as well as their physicochemical and some toxicological properties of small molecules were discussed (Table 1).

In summary, Table 1 shows an array of extenders employed in the SC/SV of carnivores. Although various extenders have been commercially distributed as chemically defined media with unknown or lesser-known chemical compositions (e.g., INRA-96 extender^®), many of the commercial blends contain materials ranging from amino acids to biological fluids such as skim milk. SC is an essential biotechnology in canine reproduction, and several studies have attempted to introduce novel CPAs and/or cryopreservation methods to achieve better results, which have been systematically reviewed herein (Table 1). Readers are referred to the references mentioned in Table 1 for detailed comparisons of the SC and SV methods along with the CPAs and their major roles in decreasing cryoinjuries.

The cellular locations of small-molecule CPAs are predicted and presented in Table 2. Briefly, none of the CPAs were localized in the cell membrane, while ethylene glycol (EG), citric acid (CA), glutathione (GSH), dimethyl formamide (DMF), and dimethyl sulfoxide (DMSO) were localized in the cytoplasm. Moreover, EG, glycerol (Gly), fructose (Fru), glucose (Glc), CA, glutamine (Gln), GSH, DMF, methyl formamide (MF), and DMSO were localized in the extracellular compartments. Some CPAs were also localized in the mitochondria (e.g., Gly, CA, Gln, and GSH), lysosomes, Golgi apparatus (e.g., Glc), and endoplasmic reticulum (e.g., GSH and Glc). All physiochemical properties of the CPAs may influence their transfer through membranes. The physicochemical characteristics of some selected small-molecule CPAs used in carnivorous SC/SV were curated from databases to obtain cues regarding their behaviors and toxicities.

The TPSAs of both EG and Gly were less than 100 Ų (Table 2); therefore, they can cross the cell membrane more easily than CPAs that have TPSAs larger than 100 Ų. The TPSAs of the CPAs used for carnivores ranged from 20.3 (DMF) to 160 Ų (GSH), and we cannot categorize the CPAs based on their TPSAs because they do not represent a rule of thumb (Table 2). The hydroxyl groups of EG or Gly make them polar substances that penetrate the cell membrane because of their low molecular weights. Two novel acetamide derivatives (methylacetamide and dimethylacetamide) are also used as CPAs for carnivorous SC/SV (Table 2).

The physicochemical attributes of CPAs used for carnivorous SC/SV are presented in Table 2, and the descriptive statistics of their quantitative features are shown in Supplementary Table S1. Based on the XlogP descriptor, all CPAs except soybean lecithin (Sln) show negative values, meaning that they are hydrophilic; however, they are considered permeating CPAs. The degree of hydrophilicity ranges from −4.5 for GSH to −0.6 for DMSO (Table 2). Among the CPAs, Sln showed an XlogP value equal to 12.9, which reflects the supralipophilic nature of this non-permeating CPA.

Based on the PubChem information, the CPAs were distributed in various cellular locations. All CPAs are distributable in the extracellular compartment as their logS values support this finding (Table 2). The organelle distribution of CPAs can alter all functional aspects of spermatozoa during the thawing and/or freezing phases of SC/SV (Table 2). The logS values of small-molecule CPAs ranged from −15.5 for Sln as the most hydrophobic to 1.33 for DMA as the most hydrophilic agent (Table 2). The molecular weights of the CPAs vary from 59.07 for MF to 758.1 for Sln, while the melting points range from −60.4°C for DMF to 236°C for Sln (Table 2).

All penetrating CPAs show optimal ranges for the physicochemical properties, except GSH, which deviates in terms of flexibility and polarity (Figure 3). Sln is a non-penetrating CPA that deviates in terms of size, insolubility, lipophilicity, and flexibility (Figure 3). None of the CPAs except DMSO exhibit predicted high BBB penetration, as demonstrated by their location inside the yellow ellipse (yolk) outlining BBB absorption (Figure 3). Glc is located in the outer gray area, indicating that it is a compound with lower HIA and limited BBB penetration. Many CPAs are located at the boundary between the gray and white sections. Gln is situated in the white area and represents a molecule with the highest probability of passive HIA. GSH is out of range of the boiled-egg scheme. None of the CPAs are PGP⁺, and they cannot interfere with the transfer of other drugs.

Based on the computational thermodynamic properties, Fru, Glc, and CA do not show proton affinities, whereas EG and Gly are protonated. Figure 4 predicts the optimized geometries of common small-molecule CPAs used for carnivorous SC/SV. At 813° K, the enthalpy of Gly is 3.5% higher than those of DMF and DMSO, whereas that of EG is 1.46% lower in the gaseous phase. Although most exothermic (ΔfH° < 0) cases are spontaneous, certain endothermic (ΔfH° > 0) cases can be included in this as well. Entropy is a thermodynamic parameter that describes the number of possible arrangements for a system in a particular state. The molecular structures demonstrate the increases in randomness of the particles upon melting of the solid, particularly as the liquid vaporizes (entropy S°_gas >> S°_liquid > S°_solid). The entropy of EG S°_gas (311.8 J/mol k) is higher than that of S°_liquid (166.9 J/mol k). The Gibbs free energy (ΔfG° = ΔfH° - T S°) is calculated by combining the effects of enthalpy (ΔfH°) and entropy (S°) on a process. The ΔfG° (813 K) values for all CPAs are spontaneously formed at this temperature because the Gibbs free energies of the CPAs are less than zero. The Gibbs free energy of CA (−931.5 kJ/mol) is higher than those of all other compounds.

Table 3 shows that EG has the lowest proton affinity (PAff) among the CPAs. The PAff of Gln (937.8 kJ/mol) is greater than that of DMSO (832.1 kJ/mol); however, when the alkyl chains are lengthened, the PAff cross (at the butyl acetate and methyl pentanoate pair) becomes slightly smaller for acetates with long chains. Because the protons are accompanied by the absorption of HO⁻ groups, the Fru, Glc, and CA systems have low affinities, with the molecules collecting 20–25 times more on the inside than outside.

Based on the data shown in Table 4, all CPAs reported for carnivorous SC/SV are healthy for the brain except GSH, DMF, MF, and DMSO, which show tendencies to be PGP (transmembrane efflux pump) substrates. None of the CPAs were PGP I and PGP II inhibitors. None of the selected CPAs could penetrate skin, and their logKp values were negative, indicating that they cannot cross skin tissue upon accidental skin exposure. Because the logKp values of all CPAs do not exceed −2.5, they do not have excellent skin permeability. The intestinal absorption percentages of CA and GSH are 0, whereas those of DMSO and Sln are 100% through the intestinal tract (Table 4). In particular, EG, DMSO, DMF, MF, Gly, and Sln show Caco2 permeabilities in this order, whereas Fru, GSH, Gln, Glc, and CA are not Caco2-permeable (Table 4). EG, DMSO, DMF, MF, Fru, and Gly are water-soluble compounds, whereas Sln, Glc, CA, Gln, and GSH show poor water solubilities (Table 4).

It is noted that the distributions of the CPAs in various compartments may alter the rheological properties of both SP and spermatozoa during SC/SV. The low VDss values of all CPAs except Glc indicate high water solubility or high plasma protein binding because the CPAs remain predominantly in the plasma, whereas high VDss values indicate significant concentration in the tissues, for example, due to tissue binding or high lipid solubility (Table 5). The relevancy of the computational data for BBB permeability (logBB), CNS permeability (logPs), and total clearance (log mL/min/kg) of the CPAs are obscured at this step; however, similar definitions like permeability of the cell membrane of the spermatozoa may be pursued computationally in the future (Table 5).

All penetrating CPAs are not substrates or inhibitors of the main CYPs, whereas Sln is a CYP3A4 substrate (Table 6). None of the CPAs are renal OCT2 substrates, and the highest total clearance was found for Sln (Table 6). Computationally, the small-molecule CPAs reported for SC/SV of carnivores can be categorized into three toxicity classes. Here, EG, Gly, Glc, and CA are low-toxicity (Class I); Gln is an intermediate-toxicity (Class II); and Sln, Fru, GSH, DMF, MF, and DMSO are high-toxicity (Class III) compounds (Table 7).

The toxicity profiles of the CPAs are presented in Table 7; however, these data are not useful for daily clinical SC/SV and artificial insemination (AI). Among the CPAs discussed here, only GSH is positive in the AMES test (Table 7). Glc and Gln show hERG I inhibitory effects, whereas the remaining CPAs show no hERG I or hERG II inhibitory effects (Table 7). The rat oral acute toxicity (LD50 mol/kg) doses of EG and Gly are the lowest among the CPAs (Table 7), whereas the NOAEL of rat oral chronic toxicity of Sln is lowest among CPAs (Table 7). Glc and Gln show hepatotoxicity among the CPAs, whereas none of the CPAs show skin sensitization (Table 7). Sln, Fru, EG, CA, GSH, DMF, and MF show T. pyriformis toxicity values greater than −5 log µg/L and are therefore considered toxic (Table 7). Thus, Gly and DMSO are not toxic to T. pyriformis in the computational sense. Among all the CPAs, Sln is the only compound found to be toxic to minnows (Table 7).