Evolutionary activation of acidic chitinase in herbivores through the H128R mutation in ruminant livestock

Summary Placental mammals' ancestors were insectivores, suggesting that modern mammals may have inherited the ability to digest insects. Acidic chitinase (Chia) is a crucial enzyme hydrolyzing significant component of insects' exoskeleton in many species. On the other hand, herbivorous animal groups, such as cattle, have extremely low chitinase activity compared to omnivorous species, e.g., mice. The low activity of cattle Chia has been attributed to R128H mutation. The presence of either of these amino acids correlates with the feeding behavior of different bovid species with R and H determining the high and low enzymatic activity, respectively. Evolutionary analysis indicated that selective constraints were relaxed in 67 herbivorous Chia in Cetartiodactyla. Despite searching for another Chia paralog that could compensate for the reduced chitinase activity, no active paralogs were found in this order. Herbivorous animals' Chia underwent genetic alterations and evolved into a molecule with low activity due to the chitin-free diet.


INTRODUCTION
Genomics and fossil records suggest that the ancestors of placental mammals mainly consumed insects. 1 Therefore, extant species may have inherited a gastrointestinal system able to digest chitin abundantly in insects.
To further investigate the crucial area, we constructed coded by mouse exons 3-4 and cattle exons 5-11 (C7) and by mouse exon 3 and cattle exons 4-11 (C8) ( Figures 1C, S4, and S5). Both chimeras displayed significantly lower activity than C3 (containing mouse exon 5) ( Figure 1D), suggesting that exon 5 represents the critical region in the cattle Chia activation. Therefore, we exchanged exon 5 in cattle Chia by the mouse sequence (chimera C9) ( Figures 1C, S4, and S5), causing restoration of about 80% of the mouse enzyme activity, suggesting that exon 5 is involved in the low activity of cattle Chia ( Figure 1D).

H128R activated the chitinolytic activity of cattle Chia
Ten residues are different in exon 5 between both enzymes ( Figure 2C). Therefore, we constructed chimeras C10 and C11 that introduced a part of the mouse exon 5 regions to the cattle enzyme (Figures 2A, S4, and S5). The activity of the chimera (C11) with the cattle sequence in the middle region of exon 5 was significantly reduced ( Figure 2B). These results strongly suggest that residues at positions 117, 121, and 128 are involved in the low activity of cattle Chia ( Figure 2C).
To identify the amino acids involved in this activation, we introduced K117Q, S121T, and H128R to cattle Chia ( Figures 2D, S4, and S6). The mutant Chia protein carrying 3 amino acid mutation (3Mut) showed higher activity than the WT mouse enzyme, indicating that these amino acids were indeed involved in Chia's low activity ( Figure 2E).
To narrow down the amino acids involved in this activation, we introduced single amino acid substitution to cattle Chia (Cattle117Q, Cattle121T, and Cattle128R mutants) ( Figures 2D, S4, and S6). The mutant (MT) proteins carrying K117Q and S121T showed no enzyme activation. However, the H128R mutant achieved a $10-fold activity increase ( Figure 2E). These results indicate that histidine at position 128 is the cause of low activity in cattle Chia.
AlphaFold2, an artificial intelligence program, 39,40 predicted the WT and MT cattle Chia protein structures. By calculating the pKa of the protein, 41 H128 (pKa = 6.87) was located on the a-helix and interacted with E169 on another a-helix ( Figure 2F, left). Chia has a catalytic domain (CatD) consisting of the triose phosphate isomerase (TIM barrel) fold, which involves the groove of the substrate bond by essential tunneling. For mutant enzymes, R128 (pKa = 11.66) does not interact with E169 ( Figure 2F, right). Therefore, this amino acid substitution at position 128 may alter the interaction between the two a-helices of the TIM barrel, affecting the overall structure and catalytic function.

A dietary habit of the bovids determines the amino acid at position 128 in Chia
Bovidae is the broadest family in Ruminantia, comprising 143 species, including several domesticated species (cattle, goat, and sheep). 42 To estimate the timing of the low activity of cattle Chia, we analyzed the completeness of 41 bovid Chia Open Reading Frames (ORFs). The Chia nucleotide sequences of these bovids were obtained from the NCBI Genome database (https://www.ncbi.nlm.nih.gov/genome/; Tables S1 and S2; Data S1, available in the supplemental information at https://doi.org/10.1016/j.isci.2023.107254). A phylogenetic tree was obtained from TimeTree ( Figure 3A).
Ruminant species can be divided into three feeding behaviors: browsing (dicotyledonous plants including leaves, stems, bark, fruits, etc.), grazing (grass), and mixed feeding (both browsing and grazing). 43 Based on the previous study, 44 we classified each of the 34 bovids ( Figure 3A).
Most bovids conserved H128, suggesting that the low activity of the Chia event occurred in a common ancestor of these lineages. However, some species (bush and Harvey's duiker) retained R128 ( Figure 3A, lower). Therefore, referring to this phylogenetic tree, it is possible that Bovidae's Chia activity reduction event occurred multiple times ( Figure 3A).

H128R mutation can activate ruminant livestock Chia
Duiker is found primarily in wooded rather than grassy areas, feeds on leaves, buds, seeds, and fruits, and often consumes small animals and insects. As shown in Figure   The selective constraints were relaxed in ruminants Chia in Cetartiodactyla Cetartiodactyla (Artiodactyla and Cetacea) is one of the most diverse orders of mammals. It includes artiodactyls (cattle, deer, giraffes, pigs, hippos, and camels) and cetaceans (whales and dolphins). Within these species, there are extensive variations in morphology and habitat. Most are herbivores, pig and peccary are omnivores, and whales are carnivores, feeding on chitin-containing organisms such as shrimp, squid, and krill.
To further investigate the molecular evolution of the Chia gene in herbivores, additional 47 species covering all major families of Cetartiodactyla were added to the analysis ( Figure 4A; Tables S1 and S2).
Within the CODEML program in the Phylogenetic Analysis by Maximum Likelihood (PAML) package, 45 we used two pairs of branch models to test whether Chia in herbivorous branches is subject to positive selection. We evaluated the fit of the following branch models to the data: (1) the null one-ratio model (M0), which assumes the same u value for all branches; and (2) a two-ratio model (M2), in which two different u values were estimated for herbivorous (u foreground ) and omnivorous/carnivorous (u background ) branches were allowed to have different u. We then compared the fit of models using likelihood ratio tests (LRTs). Unexpectedly, herbivores had lower u (u foreground = 0.360) than the background u of species with a non-herbivorous diet (u background = 0.430). However, M2 did not show better fitness than M0 (c 2 = 2.8,  (Table 1).
We also used the RELAX program to test whether the strength of selection purification at these sites differed between the phylogenetic branches. 46 RELAX results supported the hypothesis that selection was relaxed in Chia with herbivorous species (k = 0.45, p < 0.001, LR = 130.33, Table 2).

Chia5 has no complementary paralogs in ruminants
Emerling et al. reported that modern mammals possess up to five Chia paralogs (Chia1-Chia5). 33 The Chia molecule analyzed in this study (Figures 1, 2, 3, and 4) and mouse Chia correspond to the Chia5 paralog. Furthermore, the NCBI Gene database indicates that the cattle genome contains two Chia genes corresponding to Chia2 (Gene ID: 786961) and Chia3 (Gene ID: 101903127) ( Figure 5A). However, Chia2 is considered a pseudogene in cattle, and similar to Chia3 that has three stop codons in the ORF. Investigation of the conservation of Chia2 in the Cetartiodactyla genomes revealed that some species of Bovidae, Cervidae, and Suina possess genes that encode for CatD or full-length chitinase consisting of the CatD and chitin-binding domain (CBD) ( Figure 5B; Table S3).
Hence, we investigated whether the Chia2 genes found in certain species, including buffalo, goat, and pig, could compensate for the limited chitinase activity of Chia5. To accomplish this, we expressed these Chia2 and buffalo Chia5 in E. coli, as previously described (Figures 5C, S13, and S14). We found that buffalo Chia5 (histidine at position 128), has weak activity, similar to cattle, goat, and sheep Chia5 ( Figure 5C). On the other hand, and in correspondence to previous reports, omnivorous pig Chia5 showed high activity (Figure 5C). Furthermore, Chia2 exhibited only faint activity in all analyzed species ( Figure 5C).
We performed a comparative analysis of the amino acid sequences of mouse Chia (Chia5) with Chia2 of Cetartiodactyla and Tupaia chinensis (tree shrew). This species feeds on insects and has five Chia paralogs. The results revealed that the catalytic motif (DxxDxDxE) is conserved in all Chia2 proteins. However, we iScience Article discovered additional cysteine residues in Chia2: two in tree shrew, five in pig and goat, and six in buffalo and deer ( Figure 5D). These data suggest that Chia2 is inactivated in Cetartiodactyla regardless of the diet.

DISCUSSION
This study has identified a specific amino acid, H128, responsible for the evolutionary activity reduction of cattle Chia (Figures 1 and 2). This histidine residue is conserved in most bovids, whereas the insect-eating  Figures 3 and 4). This observation suggested that substituting histidine with arginine could alleviate the functional constraints of Chia. However, despite several Chia paralogs in mammals, there is no compensation for the low active Chia5 ( Figure 5).
The ruminants are one of the most successful mammalian lineages, exhibiting extensive morphological and ecological diversity. Recently, large-scale genome analysis of ruminants has been clarified, and genetic characteristics related to the diet and metabolism of ruminants have been reported. 47 However, more progress has yet to be made in analyzing how these changes affect the function of each molecule. Here, we showed that functional constraints to Chia had been relaxed in the branches of ruminants in Cetartiodactyla. In these lineages, we also showed that only herbivorous Chia had a marked decrease in chitinase activity. Our results provided novel insights into ruminants' evolution associated with feeding behavior.
Here we analyzed the enzymatic activity of Chia from eight ruminant species and showed that all the enzymes tested were reduced, but there were differences in activity levels. In Bovidae, the replacement of H128R significantly increased the activity of cattle Chia. Nevertheless, the sheep and goat Chia were not as active as that of cattle ( Figure 3C). In addition, ruminant Chia other than Bovidae and giraffe showed lower activity than a mouse, even though they retained R128 ( Figure 4B). In contrast, carnivorous whale and omnivorous peccary showed markedly higher activity than other ruminants, which was about 300% of that of the mouse ( Figure 4B). Since these animals include chitin-containing feeds such as insects and crustaceans in their diet, they are thought to possess high chitinase activity. These results indicate that changes in the functional sites of Chia in Cetartiodactyla have evolved in a diverse and complex manner in each clade.
Emerling et al. reported the presence of up to five Chia paralogs in mammalian genomes, and that Chia5 is the only functional paralog in some species such as cattle, mouse, and dog. 33 Although almost all other Chia paralogs have also been lost in most Cetartiodactyla, Chia2 is still functional at least in Bubalus bubalis, Capra hircus, and Sus scrofa. This gene may also be involved in chitin digestion in the species where it could potentially compensate for the generally reduced chitinolytic activity. However, although Chia2 genes are relatively intact, the proteins contain extra cysteines and show little activity (Figures 5C and  5D). These observations are consistent with our previous reports on the inhibitory effect of extra cysteine residues. 35 Thus, Chia2 is a pseudogenized and largely non-functional molecule in this phylogenetic group, regardless of the diet ( Figure 6A).
The ancestors of all placental mammals were small insect-eating organisms that evolved shortly after the extinction of dinosaurs. 1 It has been reported that Chia is a molecular record of the evolutionary process of the ancestors of insect-eating mammals. 33 Chia enzymes from omnivorous animals such as mouse, pig, marmoset, and crab-eating monkey show high chitinolytic activity ( Figure 6B, left). [23][24][25]30,31 Our recent analysis of 32 carnivoran Chia genes showed that in non-insect-eating species, Chia had been inactivated or pseudogenized. In contrast, insect-eating species preserved the complete ORF and high activity of the enzyme ( Figure 6B, middle). 35 As described previously, the insect-eating bovids have arginine, while most bovids have histidine at position 128 ( Figure 2; Figure 6B, right). The R128H substitution is also observed in some carnivorous Chia pseudogenes, including bears, mustelids, and pinnipeds. 35 Previous studies have shown that Chia in carnivorous and herbivorous animals, having chitin-free diets, underwent genetic alterations and evolved into a low-active molecule while, the chitin-consuming species have maintained Chia with high activity ( Figure 6B).
Chia expression and/or activity levels are markedly altered in various diseases, such as asthma and allergic inflammation. 10,11,18,19 Chia-deficient mice accumulate chitin and develop age-dependent lung fibrosis, which can be ameliorated by Chia supplementation. This observation suggests that enhancing chitinase activity in Chia has therapeutic potential for reducing environmentally derived chitin in the lungs. 3, 21 We have reported that Chia activity in humans and dogs can be increased by R61M 19 and F214L with A216G, respectively. 35 In this study, we activated cattle Chia by replacing a single amino acid residue. Generally, our strategy for enzyme activation combines biochemical and evolutionary approaches. Such a strategy could, e.g., create a highly active Chia that can treat lung diseases. 21,48 Limitations of the study

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Material availability
Materials generated in this study are available upon request. For further details contact the lead contact.
Data and code availability d All data reported in this paper will be shared by the lead contact upon request.
d This paper does not report original code.
d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS
We expressed proteins in E. coli as our experimental model instead of using living animals. We purchased total RNA products of bovine and porcine tissues from Zyagen (San Diego, CA, USA) to obtain the Chia cDNAs. These total RNA products are derived from normal healthy tissues for human consumption. The tissues were obtained from certified slaughterhouses in the USA and harvested from female donors at 30 months for bovine and 6 months for porcine. The animals used were domestic bovine and porcine animals specifically bred for meat production. We also purchased mouse stomach total RNA prepared from pooled healthy male/female BALB/c mice, ages 6 weeks, from Takara Bio (Mountain View, CA, USA). The use of animal-derived total RNAs and all procedures in this study were reviewed and approved by the Recombinant DNA Committee at Kogakuin University.

METHOD DETAILS
Total RNA and cDNA preparation The Cattle and Pig stomach Total RNA Panel were purchased from Zyagen. Mouse Total RNA Master Panel was also purchased from Takara Bio. Cattle, mouse, and pig stomach total RNAs were reverse-transcribed into cDNA essentially as described previously. 32

Construction of Chia expression vector
We expressed mouse and cattle Chia as recombinant fusion proteins with pre-Protein A (PA) and V5-His (pEZZ18/PA-Chia-V5-His). 32 In this report, PA-cattle Chia-V5-His and their derivative chimeric or mutant proteins were expressed by pET22b using the T7 promoter system (designated pET22b/Protein A-Chia-V5-His) as described recently. 35 The pEZZ18/PA-cattle Chia-V5-His was digested with EcoRI and XhoI, generating cattle Chia cDNA. Fragments were purified and subcloned into similarly digested pET22b/pre-Protein A-mouse Chia-V5-His to produce pET22b/pre-Protein A-cattle Chia-V5-His.

Construction of expression plasmids for chimeric and mutant proteins and preparation of recombinant proteins
Mouse and cattle Chia have similar exon structures at the nucleotide level. To create mouse/cattle chimeric proteins, we fused two units at the junctions among exons 3-5, exons 6-7, exons 8-10 and exon 11 using template DNAs and primers ( Figure S1; Tables S4 and S5) as described previously. 35 More chimeras were also produced by combining templates and primers (Tables S4 and S5). Chia mutant proteins were prepared by PCR using a template and primers (Tables S4 and S5) iScience Article E. coli BL21 (DE3) was transformed to express pre-Protein A-Chia-V5-His proteins using the plasmid DNAs. Transformed E. coli were grown in 250 mL of LB medium containing 100 mg/mL ampicillin at 37 C for 18 h. After induction with 0.1 mM isopropyl b-D-thiogalactopyranoside (IPTG), the bacteria were cultured for two h in an LB medium. Cells were harvested by centrifugation at 6,500 g for 20 min at 4 C. The recombinant protein was prepared from E. coli and purified by IgG Sepharose (Cytiva, Marlborough, MA, USA) chromatography as described previously. 35 The protein-containing fractions were desalted using PD MidiTrap G-25 (Cytiva) equilibrated with TS buffer [20 mM Tris-HCl (pH 7.6), 150 mM NaCl and a protease inhibitor (Complete, Roche, Basel, Switzerland)]. Western blot detected recombinant products using an anti-V5-HRP monoclonal antibody (Thermo Fisher Scientific, Waltham, MA, USA).
To determine the optimal pH for chitinase activity, the enzyme was incubated with the 4-MU-(GlcNAc) 2

QUANTIFICATION AND STATISTICAL ANALYSIS
We quantified the immune blots using the Luminescent Image Analyzer (Amersham ImageQuant 800 Western, Cytiva, Marlborough, MA, USA) according to the manufacturer's instructions. Welch's t-test was used to compare the biochemical data. We carried out experiments in triplicate for statistical analysis.

Sequence analysis
We conducted NCBI BLAST searches against whole genome assemblies of 33 Bovidae, 1 Moschidae, 14 Cervidae, 1 Giraffidae, 1 Tragulidae, 4 Tylopoda, 1 Tayassuidae, 1 Hippopotamidae, and 15 Cetacea genomes from the NCBI Genome Database using the cattle (NM_174699.2), deer (XM_043878569.1), camel (XM_031457795.1), pig (NM_001258377.1), and whale (XM_030869341.1) Chia gene sequences as a query. In addition to these sequences, we used annotated gene sequences available in GenBank. GenBank accession numbers and deduced Chia nucleotide sequences are described in Tables S1 and S2 and Data S1. We imported all sequences, including the mouse as an outgroup, into MEGA X 49 and aligned them using the MUSCLE algorithm. 50 The evolutionary relationships of the Chia genes in Cetartiodactyla were estimated by the maximum-likelihood method ( Figure S15).

Molecular evolution analysis
We performed positive selection analyses of genes based on a variation in the ratios of nonsynonymous to synonymous nucleotide substitutions (dN/dS or u) using the CODEML program of PAML 45 and RELAX implemented in HyPhy version 2.22 46 . The 88 Cetartiodactyla and mouse (outgroup) phylogenetic relationships were inferred using TimeTree (http://www.timetree.org/). 51 For the CODEML analysis, we used branch models, where u is assumed to be different between foreground and background branches. We first set up a null model estimating a single u across all branches (M0). We then classified the tree into two distinct classes (M2): those with herbivores (foreground branches) and those with non-herbivores (background branches). Its goodness-of-fit was analyzed using likelihood ratio tests (LRTs).
We also used the RELAX program to test for two different rates of u between lineages with herbivores versus all other branches and to distinguish positive from the relaxed selection because increased u may indicate either. 46 RELAX estimates u among three rate classes for each branch using a branch siterandom effects likelihood (BS-REL) model and then fits a parameter k indicating the strength of selection. Intensified selection is indicated by k > 1, whereas relaxed selection is indicated by k < 1. The goodness-offit for two given models was analyzed using the LRT by comparison with each null model whose k parameter was constrained to 1.