Evolution of the egg glue protein in great ape lice inferred from its sequence comparisons across five louse types

Abstract

Humans host three types of parasitic louse: the conspecific head and body lice differing in habitat (scalp and clothing), and the pubic louse which is a distinct species adapted to coarser body hair. Chimpanzees and gorillas host only a single louse species respectively related to the human head/body lice and the pubic louse. Louse nit sheath proteins (LNSPs) are the egg glue of human head (or body) that function to secure the eggs to the hair (or clothing). In this study, comparisons of protein sequences at various regions of LNSP1/2 from five types of lice infesting the great apes were done, which showed species-dependent variations. Estimations of the Unit Evolutionary Period (UEP, time required for 1% difference in amino acid sequence) performed using the known times of divergence from their common ancestors suggested that both the LNSP1 (0.37 MY (million years)) and the LNSP2 (0.48 MY) evolved ~3-fold faster than the cytochrome c of these louse species (with UEP of 1.3 MY). Furthermore, the observation inferred from the dN/dS analysis that evolutionary pressure acts differently on various regions of LNSP1/2 may be related to the homologous LNSP1 and LNSP2 having differing detailed functionalities (desiccation resistance and glue platform formation, respectively).

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-24011-w.

Introduction

Lice are wingless insects belonging to the Phthiraptera infraorder (under the Psocodea order) that parasitize birds and mammals. The Phthiraptera infraorder is further divided into four parvorders, in which chewing lice are classified into three parvorders and blood-sucking lice into only one Anoplura parvorder. In particular, two species of Anoplura are parasites of humans – head louse (Pediculus humanus capitis) and body louse (P. h. humanus or formerly called P. h. corporis) regarded as two sub-species (of insect family Pediculidae), and pubic lice (or crab lice, Pthirus pubis of insect family Pthiridae). Among them, only the body louse is known to be a vector of various bacteria-mediated diseases. The divergence of body louse from the ancestral head louse is thought to have occurred at least by 83,000 and possibly as early as 170,000 years ago when humans started wearing clothes¹. Also in the genus Pediculus are Pediculus schaeffi (parasitizing chimpanzees and bonobos) and Pediculus mjobergi, a New World monkeys’ louse which is thought to be transmitted to monkeys from the first humans that had reached the American continent thousands of years ago. Of note, a species of lice belonging to another insect family of Pedicinidae and genus Pedicinus are found on Old World monkeys (for example, P. badii is known to parasitize red colobus monkeys). Along with the human P. pubis, another louse in the genus Pthirus is Pthirus gorillae which parasitizes gorillas. It is interesting to note that humans are host to two genera of lice (i.e. Pediculus and Pthirus, or head/body louse and pubic louse) while other great apes are host to only one species of louse. While the split between gorillas and the lineage leading to chimpanzees and humans is anywhere between 7 to 9 million years (MY) ago, small genetic differences between P. pubis and P. gorillae suggest that the ancestor of P. pubis was transferred from gorilla to the early humans as recently as 3.3 MY ago likely from close contact between the two great ape species^2–4.

The lice parasitizing human lay their eggs on hair (for head louse), clothing (for body louse), or pubic hair (for pubic louse). To attach the eggs, the egg-laying adult female secretes a liquid glue-like material from their accessory gland, which solidifies into a sheath covering the embryo except for the operculum through where the developing embryo breathes. After analyzing the amino acid composition of the human head louse nits (empty hatched egg) and matching the result against the proteins encoded by the complete genome of the louse, two homologous proteins making up the glue-like material were identified and named the louse nit sheath proteins 1 and 2 (LNSP1/2)⁵. Subsequent studies suggest that transglutaminase-mediated cross-linking is functionally critical for the solidification of the protein into a glue^6–8. The LNSP is divided mainly into three repeat sequence containing domains after the very N-terminal signal sequence which targets the glue protein into the ER for extra-cellular secretion. Following the signal sequence, there is the N-terminal domain (N-domain) characterized by the Gln-Ala (QA) repeats. Next, the middle domain (M-domain) has repeating sequences of the Gly-Ala (GA) followed by the C-terminal domain (C-domain) containing the tandem occurrences of Gln residues (PolyQ). There are also regions connecting these three repeat domains (namely, loop 1 between the N- and M-domains and loop 2 between the M- and C-domains) as well as a sequence after the C-domain (termed loop 3).

An egg is the most vulnerable stage of an insect life and it is needless to say that to a louse that attaches the egg to the environment the evolutionary success of a species depends on the glue used. In the GenBank Sequence Read Archive (SRA), information on short DNA read sequences generated by high-throughput sequencing of P. pubis, P. gorillae, and P. schaeffi were available and by de novo assembling these reads, we were able to determine the protein encoding sequences of LNSP1 and LNSP2 in these louse species parasitic to the great ape host. In this study, we compare the LNSP1/2 of the five types including P. h. capitis and P. h. humanus in relation to the lice’s known phylogeny.

Materials and methods

De novo assembly of louse genomes using SRA short reads

The GenBank Sequence Read Archive (SRA) data of P. pubis (accession number SRX2405464), P. gorillae (accession number SRX2405463) and P. schaeffi (accession number SRX2405462) were used for the de novo assembly of the individual louse genomes. The raw data were passed through the de novo assembly pipeline originally designed by Feng Zhang⁹ modified for this study. Firstly, the duplicated reads were removed and the similar read mates were grouped using Clumpify included in the BBTools suites¹⁰. Next, BBDuk was used to handle contamination in the read library, in which the low-quality bases were trimmed from both ends of reads, and both the ambiguous bases and the long polyA sequences were discarded to result in only the reads longer than 15 base-pair. In order to reduce the computational resource demand, the read coverage was also constrained by read normalization using BBNorm, and the erroneous bases further corrected by Lighter¹¹. The SPAdes, a multiple k-mer de novo assembler, was used subsequently for the assembly via Shovill which is the wrapper for various de novo assemblers^12,13. Lastly, the qualities of the assemblies were evaluated with the BUSCO and Insecta gene catalog¹⁴. When this identical approach was done to assemble the whole genome of P. badii (one of Old World monkey lice) using the SRA accession number SRX2609261, the BUSCO score (61.1% in Insecta gene set) turned out to be significantly lower than that of the P. pubis, P. gorillae, and P. schaeffi (>95%). This result is likely due to the low sequence coverage of the P. badii library. The number of reads are half of that of P. pubis with the reads being ~2/3 in average length.

LNSP1 and LNSP2 search in the assembled louse genomes

As the estimated gene sizes of LNSP1 and LNSP2 are between 700 bp and 2100 bp respectively, the ORFs within the corresponding sizes were selected¹⁵ and compared to the previously reported LNSP protein sequences of the P. h. capitis and P. h. humanus using BLASTp on Galaxy platform^16–19. In the end, the candidate sequences found by BLASTp were manually distinguished as LNSP1 or LNSP2 by using the signal sequences at the N-terminus of the encoded genes. In the low quality assembled contigs of P. badii, only a sequence covering ~66% of LNSP2 was found without any presence of LNSP1. Hence, further analysis on the LNSPs was performed without that of P. badii. In contrast to the LNSP, the sequence of P. badii cytochrome c was successfully found and included during its analysis (see below).

Sequence identity, amino acid change per 100 residues, and mutation rate calculations

Using the aligned LNSP sequences, the percent sequence identities for the full-length and the individual regions of LNSP1/2 were calculated between two sequences (pairwise) where the total number of matched residues was divided by the total number of matched and unmatched (including the substitution as well as gaps (insertions and deletions)) residues of the pair in each case (i.e. commonly known as the BLAST identity) (Table S1, Fig. 1). In order to calculate the mutation rates for the full-length LNSP1/2 and for the regions within, the amino acid changes per 100 residues were calculated by counting the unmatched (substitution and gaps; for gaps see below) and dividing it by the total number of matched and unmatched residues counted over the alignments of the two LNSP1s (or LNSP2s). The calculations were done for P. h. capitis and P. h. humanus (noted as Group 1 throughout the text) and for P. pubis and P. gorillae (noted as Group 2). Also, three LNSP1s (or LNSP2s) of the P. humanus (capitis and humanus) and P. schaeffi (noted as Group 3), or the five entire LNSP1s (or LNSP2s) (for all genera of Pediculus and Pthirus, noted as Group 4) were compared. Importantly, consecutive gaps were regarded as only one difference because insertion or deletion in evolution is created by one event (i.e. gap-compressed identity). The mutation rates were obtained from the slopes for the graph plotted using the times of divergence from the common ancestor between the different louse species^2,4 and the amino acid changes per 100 residues (Fig. 2). The Unit Evolutionary Period (UEP) which is the average time required for 1% difference in amino acid sequence to arise was calculated by the inverse of the slope giving the unit in million years (MY) as defined in other studies^20–22.

Fig. 1.

Sequence comparison of LNSP1 (top) and LNSP2 (bottom) from different great ape louse types. Full-length and individual regions of LNSP1 and LNSP2 were all pairwise compared using sequences from five great ape louse types (Table S1). The comparison pairs having closer time of divergence from the common ancestor are shown from the left. For convenience, comparison between the closest two P. humanus of capitis and humanus is noted as Group 1, and between the genus Pthirus (P. pubis and P. gorillae) is noted as Group 2 throughout the text. Two pair comparisons were done for P. humanus (both capitis and humanus) and P. schaeffi (noted as Group 3), and the identity results are shown using the average values with the error bars noting the lowest and the highest values. Six pair comparisons (P. h. capitis, P. h. humanus and P. schaeffi vs. P. pubis and P. gorillae) were done for the comparisons of Pediculus and Pthirus genera LSNPs (noted as Group 4). Similarly, the average sequence identities are shown with error bars noting the lowest and the highest values.

Fig. 2.

The rates of mutation for LNSP1 and LNSP2 in five great ape louse types. The amino acid changes per 100 amino acids of full-length (FL) LNSP1 and LNSP2 from five louse types are plotted with the known years of divergence (in million years (MY)) from common ancestors. The pairwise comparison between P. h. capitis and P. h. humanus (Group 1), and P. pubis and P. gorillae (Group 2), the two pair comparisons between the Pediculus (P. h. (capitis and humanus) to P. schaeffi, Group 3) and the six pair comparisons between the Pediculus and the Pthirus (P. h. capitis, P. h. humanus and P. schaeffi vs. P. pubis and P. gorillae, Group 4) are represented in the same ways as in Fig. 1 with circle symbols indicating the average sequence identities and the error bars noting the lowest and the highest values. The same amino acid changes for the individual regions of LNSP were calculated as well, and those that show more than 3-fold or less than 1/3-fold of the full-length identity values which were either the signal sequence or the loop 1~3 are indicated. The slope which is the rate of protein evolution was calculated by fitting the amino acid changes per 100 amino acids for the full-length (FL) LNSPs against the four different timeline of divergence. The Unit Evolutionary Period (UEP) in MY which is the average time required for 1% difference in amino acid sequence to arise is the inverse of the slope.

Cytochrome c search in the assembled louse genomes

For comparison of the divergence in the LNSP1/2 sequences with that of the cytochrome c, orthologues corresponding to the putative cytochrome c of the P. h. humanus were searched in the de novo assembled genome of the four louse species. Of the two putative P. h. humanus cytochrome c sequences (XP_002430460.1 and XP_002426706.1) in the NCBI database, all orthologues of XP_002426706.1 in P. pubis, P. gorillae, and P. schaeffi had a stop codon in the middle of the DNA sequences. Also, this orthologue was not found in the low quality P. badii genome. Hence, analysis was performed using the cytochrome c orthologues of XP_002430460.1 which were all successfully found in the genomes of P. h. capitis, P. pubis, P. gorillae, P. schaeffi and P. badii.

LNSP1/2 sequence confirmation in P. pubis

Three dead P. pubis adults in 75% ethanol were provided by a private urology clinic in Seoul, South Korea. Genomic DNA (gDNA) was extracted from P. pubis adults using the DNeasy Blood & Tissue kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The LNSP1 and LNSP2 genes were PCR-amplified in 10 μL reaction mixtures contained 9 ng of template gDNA, 0.8 μL of dNTPs (2.5 mM each), 1 μL of 10x buffer, 0.5 μL of each primer (Table S2), and 0.2 μL of 50x Advantage 2 polymerase mix (Takara, Kusatsu, Japan). The reaction was performed in a T100™ Thermal Cycler (Bio-Rad, CA, USA) under the following conditions: initial denaturation at 95 °C for 5 min; 35 cycles of denaturation at 95 °C for 15 s, annealing at 58 °C (for LNSP1) or 62 °C (for LNSP2) for 20 s, and extension at 68 °C for 2 min; followed by a final extension at 68 °C for 5 min. The PCR products were purified using Expin™ PCR SV Kit (GeneAll, Seoul, South Korea), cloned into pGEM-T Easy vector (Promega, WI, USA) and then transformed into DH5α competent cells (Enzynomics, Daejeon, South Korea). The plasmids were extracted using the Plasmid Mini Extraction Kit (Bioneer, Daejeon, South Korea) and sequenced (Macrogen, Seoul, South Korea).

Pairwise (MEGA) and site-model based (CodeML in PAML) dN and dS calculations for various regions of great ape lice LNSP1/2, and for cytochrome c

As before, the five types of great ape lice were grouped based on their evolutionary relationships (Group 1, P. h. capitis and P. h. humanus; Group 2, P. pubis and P. gorillae; Group 3, P. h. capitis and P. schaeffi; Group 4, P. h. capitis and P. pubis, P. h. capitis and P. gorillae, P. schaeffi and P. pubis, and P. schaeffi and P. gorillae), and the DNA sequences taken for each LNSP1/2 regions based on the aligned protein sequences. The dN and dS values for the LNSP1 and LNSP2 regions were calculated for sequence pairs within each lice group using the distance estimation function with Kumar (Kimura 2-parameter) model²³ and pairwise deletion treatment for the gaps/missing data in MEGA version 11²⁴ via 1000 iterations of Bootstrap analysis²⁵. The dN and dS values were subsequently used to calculate the ratio of dN and dS (dN/dS). In some cases, non-calculable values (N/A) were observed due to the absence of certain regions in the sequence (Table 1). For Group 4 having identical divergence times of 13 MY, pairwise dN and dS values were averaged and new errors calculated using error propagation. Because dS values were very low in several of the pairwise comparisons, dN/dS consequently turned out to be numerically unstable yielding uninformative standard errors. We therefore report only the dN and dS values with bootstrap errors, and interpret the dN/dS with less emphasis on the exact values but focusing on the differences between various LNSP1/2 regions. The dN and dS values for cytochrome c were calculated in a similar manner. Unlike in the case for LNSP1/2, P. badii. cytochrome c was also included in the analysis as Group 5.

Table 1.

The pairwise dN and dS calculations between LNSP1/2 regions from various great ape lice.

	Group 1		Group 2		Group 3		Group 4
	Head louse vs Body louse		P. pubis vs P. gorillae		P. h. capitis vs P. schaeffi		Pediculus vs Pthirus
	LNSP1	LNSP2	LNSP1	LNSP2	LNSP1	LNSP2	LNSP1	LNSP2
dN
Signal sequence	0.07 ± 0.05	0	0	0.05 ± 0.03	0.03 ± 0.03	0	0.23 ± 0.06	0.03 ± 0.02
N-domain	0.03 ± 0.02	0.03 ± 0.01	0.02 ± 0.01	0.08 ± 0.02	0.04 ± 0.02	0.16 ± 0.03	0.18 ± 0.04	0.18 ± 0.03
M-domain	0.04 ± 0.02	0.02 ± 0.01	N/Aa	0	0.05 ± 0.03	0.16 ± 0.09	0.15 ± 0.07	0.15 ± 0.07
C-domain	0.02 ± 0.01	0.02 ± 0.01	0.02 ± 0.02	0.01 ± 0.01	0.15 ± 0.04	0.14 ± 0.04	0.40 ± 0.13	0.15 ± 0.04
Loop 1	0	0.03 ± 0.01	0.01 ± 0.01	0.02 ± 0.01	0.07 ± 0.02	0.06 ± 0.02	0.24 ± 0.03	0.18 ± 0.02
Loop 2	0.02 ± 0.02	0.03 ± 0.02	0.11 ± 0.06	0.01 ± 0.01	0.07 ± 0.03	0.15 ± 0.04	0.13 ± 0.05	0.13 ± 0.03
Loop 3	0	0.05 ± 0.02	N/Ab	0.01 ± 0.01	0.02 ± 0.02	0.04 ± 0.03	N/Ab	0.48 ± 0.12
dS
Signal sequence	0	0	0	0.06 ± 0.08	0.35 ± 0.24	0.10 ± 0.12	0.31 ± 0.12	0.75 ± 0.20
N-domain	0.01 ± 0.01	0.06 ± 0.02	0.00 ± 0.00	0.19 ± 0.05	0.12 ± 0.06	0.29 ± 0.07	0.68 ± 0.15	0.60 ± 0.11
M-domain	0.11 ± 0.05	0.08 ± 0.04	N/Aa	0.30 ± 0.13	0.35 ± 0.12	0.36 ± 0.19	0.65 ± 0.35	0.76 ± 0.25
C-domain	0.03 ± 0.03	0	0.09 ± 0.06	0.08 ± 0.04	0.33 ± 0.11	0.33 ± 0.09	1.20 ± 0.37	1.11 ± 0.26
Loop 1	0	0.02 ± 0.02	0.04 ± 0.02	0.01 ± 0.01	0.07 ± 0.04	0.19 ± 0.06	0.53 ± 0.08	0.44 ± 0.06
Loop 2	0.02 ± 0.03	0.08 ± 0.04	0.06 ± 0.05	0.03 ± 0.02	0.32 ± 0.14	0.35 ± 0.14	0.33 ± 0.16	0.58 ± 0.13
Loop 3	0	0.08 ± 0.04	N/Ab	0.02 ± 0.02	0.03 ± 0.04	0.10 ± 0.06	N/Ab	1.39 ± 0.33
dN/dS
Signal sequence	–**	–*	–*	0.83	0.08	0***	0.73	0.04
N-domain	2.07	0.51	–**	0.43	0.33	0.55	0.26	0.30
M-domain	0.37	0.23	N/Aa	0***	0.14	0.43	0.23	0.20
C-domain	0.58	–**	0.24	0.12	0.44	0.42	0.33	0.13
Loop 1	–*	1.35	0.14	1.48	0.94	0.33	0.46	0.41
Loop 2	0.77	0.45	1.73	0.45	0.21	0.44	0.40	0.22
Loop 3	–*	0.58	N/Ab	0.45	0.53	0.44	N/Ab	0.34

^*dN=0, dS=0.

^**dN≠0, dS=0.

^***dN=0, dS≠0.

^aNo M-domain for P. gorillae.

^bNo loop 3 for P. pubis and P. gorillae.

As a complementary approach to the pairwise dN and dS calculations using MEGA, codon-based site models implemented in CodeML (PAML v4.9j)²⁶ were used to estimate the dN/dS among codon sites. Likelihood ratio tests (LRTs) were applied to assess evidence for positive selection, in which two pairs of models were compared: M1a (nearly neutral) vs. M2a (positive selection), and M7 (beta) vs. M8 (beta & ω (= dN/dS) > 1). For each comparison, the test statistic (2ΔlnL) was calculated from the maximum log-likelihood (lnL) values of the nested models, and the significance was assessed using a chi-square test with 2 degrees of freedom. Additionally, positively selected sites (PSS) were estimated using Bayes Empirical Bayes (BEB) analysis under the M2a and M8 models²⁷. Statistical significance was defined as P < 0.05.

Results and discussion

General overview

Amino acid sequences of the egg glue proteins, LNSP1 and LNSP2, of human head louse (P. h. capitis) and body lice (and P. h. humanus) have been already reported⁵. In this study, the sequences of the orthologous proteins in three other species of lice (P. schaeffi, P. pubis, and P. gorillae) were obtained from the whole genome of these lice assembled using the GenBank SRA sequences. With these results, the protein and DNA sequences of LNSP1/2 from five types of lice parasitic to three great ape species of human, chimpanzee (or bonobo), and gorilla are analyzed and compared.

The validity of the LNSP1/2 sequences obtained from our de novo assembled genome was checked by comparing them to our Sanger sequencing results using our isolated P. pubis genomic DNA (Fig. S1). The LNSP1 sequences between the two were ~98% identical having a few single residue switches and slight differences in the numbers of glutamine residues within the PolyQ C-domain (Fig. S1A). However, more prominent differences between the two LNSP2 sequences mainly resulting from insertions of repeating sequences were observed (Fig. S1B). Such differences in insertions have been observed before when different strains of body lice (i.e. Culpepper and San Francisco body louse strain) were compared⁵. Because the P. pubis used to obtain the SRA and our Sanger sequencing data are from different regions, we regard that this discrepancy results from genetic variations existing in different populations of pubic lice. The LNSP1 and LNSP2 sequences from the two different Culpepper and San Francisco strains of body lice studied before have ~2% amino acid differences in the aligned region apart from the observed insertions⁵. However, in the two P. pubis LNSP2 sequences, no single residue switches were seen (Fig. S1B). Further validations of the LNSP1/2 sequences in P. schaeffi and P. gorillae were not possible due to the difficulty in obtaining their genomic DNA. However, the general agreement in the LNSP1/2 sequences derived from our assembled P. pubis genome and the targeted Sanger sequencing suggested that the P. schaeffi and P. gorillae LNSP1/2 sequences from our assembled genomes are also reliable enough to be further analyzed.

The domain boundaries of the LNSP1/2 in the five types of lice were defined (Fig. S2), and the regional sequences aligned for analysis and comparison (Fig. S3). As paralogous proteins, the LNSP1 and LNSP2 are also very similar in sequence⁵. It is interesting to point out that when the LNSP1 and LNSP2 within the single species of lice were compared, the two in the genus Pediculus showed 55−57% identities. However, the identities between the LNSP1 and LNSP2 were lower in the genus Pthirus (40% for P. pubis, and 22% for P. gorillae) (Fig. S4).

LNSP1 comparisons across five type of great ape lice

As secreted glue proteins, both LNSP1 and LNSP2 have a N-terminal tandem hydrophobic amino acid sequences located at the very N-terminus, which are predicted as the signal sequences to target the nascent peptide chains into the endoplasmic reticulum for proper trafficking. This 18-residue signal sequence in both cases of the LNSP1 and LNSP2 are very conserved that they themselves can act as the marker for distinguishing the homologous LNSP1 and LNSP2 (Fig. S3). The signal sequences of LNSP1 were 50−94% identical among five louse types.

Between the two closely related human lice of P. h. capitis and P. h. humanus (Group 1), the three loops happened to be more conserved than both the signal sequence and the repeat sequence (Fig. 1). When the repeat sequence containing domains of LNSP1 were compared (Fig. S3), it is notable that the M-domain is completely missing only in the P. gorillae LNSP1. Also in only the P. gorillae LNSP1, the PolyQ glutamine repeat C-domain was shortened in half the size compared to the LNSP1s of other lice. Additionally, for the LNSP1s of genus Pthirus (P. gorillae and P. pubis of Group 2), the M- and C-domains were connected by only five residues while a ~50 residue loop 2 connects the two domains in the other LNSP1s of the genus Pediculus (P. h. capitis, P. h. humanus and P. schaeffi) (Fig. S3). Moreover, a 28-residue loop after the C-domain (loop 3) did not exist in the LNSP1s of genus Pthirus (Fig. S3). Also in Group 2, the N-domain had high 95% sequence identity compared to the C-domain (42%), and loop 1 had 99% sequence identity.

Among the three LNSP1s of the genus Pediculus (Group 3), the sequence identities of the signal sequence (78−94%), loop 1(82−100%), loop 2 (83−96%) and loop 3 (96−100%) were more conserved than the repeat sequence containing domains of the N-domain (51−79%), M-domain (67−84%) and C-domain (58−72%) (Fig. 1 and Table S1).

Between the two genera of Pediculus and Pthirus (Group 4), the signal sequence had 50−61% sequence identities, and the loop 1 had 56−59% sequence identities (Fig. 1). As mentioned above, neither the long loop 2 nor the entire loop 3 were observed in the Pthirus LNSP1s. Also between the two genera of Pediculus and Pthirus LNSP1s, the M-domain (59−75%) seemed to be more conserved than the N-domain (41−61%) or the C-domain (28−53%) (Fig. 1).

It is notable that the N-domain of LNSP1 in all species of lice is half the size of that in LNSP2. Another interesting fact to note is that in the C-domain of P. schaeffi there was one PolyQ sequence which had 35 tandem glutamine (Q) residues (Fig. S2). This likely results from gene level combination of three PolyQ sequence blocks which normally have 6−11 Qs in a row (Fig. S3).

LNSP2 comparisons across five type of great ape lice

The signal sequences of LNSP2 were 100% identical in four louse types (P. h. capitis, P. h. humanus, P. schaeffi and P. pubis except in P. gorillae which was still 89% identical), hence being much more conserved than that of LNSP1 (Fig. S3). In Group 1 of P. h. capitis and P. h. humanus, the repeat sequence domain of the M- and C-domains were more conserved (96−97% identical) than the N-domain (89% identical) (Fig. 1). In general, these repeat sequence domains are more conserved in LNSP2 than in LNSP1, while all three loops were more conserved in LNSP1 than in LNSP2 (Fig. 1).

In the two Pthirus genus lice (Group 2), the most identical repeating sequence containing domain was the N-domain with 89% identity (M-domain, 53%; C-domain, 77%) (Fig. 1). The low identity of the M-domain in this case resulted from the ~30 residue shorter M-domain of the P. pubis. Also, the loops of LNSP2 in Group 2 lice were more conserved than the repeat sequence containing domains (loop 1, 88%; loop 2, 90%; loop 3, 96%)

Among the three lice in the genus Pediculus (Group 3), the most conserved repeat sequence containing domain within LNSP2 was the C-domain with 73−97% identity (N-domain, 71−89%; M-domain, 66−96%) (Fig. 1). In these three lice, loop 1 was most conserved (loop 1, 88−93%; loop 2, 69−89%; loop 3, 67−92%). Between the two genera of Pediculus and Pthirus (Group 4), the most conserved repeat sequence containing domain of LNSP2 was the N-domain with 72−77% identity (M-domain, 46−68%; C-domain, 60−72%) (Fig. 1). Among these two different genera of lice in Group 4, loop 2 was most conserved (loop 1, 59−70%; loop 2, 68−80%; loop 3, 38−48%).

Mutation rates of LNSP1/2 across five types of great ape lice

While the previous analysis of LNSP1/2 was done using percent identities, amino acid changes per 100 residues were applied for relating the differences in LNSP1/2 to protein molecular evolution by calculating the protein mutation rates. However, unlike in the case for calculating the percent identities, the amino acid changes per 100 residues were determined with consecutive gaps regarded wholly as one change due to the fact that insertion or deletion in evolution is considered to be created by a single event.

For the cases of LNSP1 and LNSP2 in the two human P. humanus lice (P. h. capitis and P. h. humanus, Group 1), the values of the amino acid changes per 100 residues in the full-length forms were very similar between the two (Fig. 2). However, when the amino acid changes were determined for the various regions of LNSP1/2, the LNSP1’s signal sequence between the two human P. humanus lice seemed to have significantly more amino acid changes (~3-fold of the full-length value). The loops 1 and 3 of LNSP1 and the signal sequence of LNSP2 were 100% identical in the two species (Fig. 1) giving the value 0 amino acid change (Fig. 2). When the amino acid changes in LNSP1/2 for the two Pthirus species (P. pubis and P. gorillae, Group 2) were calculated, the values for the full-length LNSP1 and LNSP2 were also very similar. The only region that had significantly higher changes compared to the full-length forms were the LNSP1’s loop 2 (~4-fold) (Fig. 2). Because the signal sequence of LNSP1 is 100% conserved in the two species, the amino acid change value was 0. The loop 1 of LNSP1 is also highly conserved compared to the other regions of the protein (Figs. 1 and 2).

When the amino acid changes in LNSP1/2 between the Pediculus (P. h. (capitis and humanus) vs. P. schaeffi, Group 3) were calculated, the values for the full-length LNSP1 and LNSP2 were also close in range (Fig. 2). As expected from the high sequence identities in the previous analysis (Fig. 1), the loop 3 of LNSP1 and the signal sequence of LNSP2 had the least varied amino acids (the signal sequence of LNSP2 is 100% conserved). For the LNSP1/2 of the genera Pediculus and Pthirus (Group 4), the amino acid changes for the full-length LNSP1 and LNSP2 deviated from each other likely from the fact that their time of divergence from the common ancestor was longer than the previous comparisons. Interestingly, however, it is notable that the conservation of the signal sequence among the various LNSP2s was very high compared to other regions of the protein as mentioned before (Fig. 1).

From this analysis, it is notable that the signal sequences (except for that of the LNSP1 in the two human P. humanus lice) are mostly conserved among all types of lice. This may originate from the fact that the signal sequence of ~23 amino acids are comparatively short compared to other regions, but the importance of the glue protein to be secreted may give certain pressure against any changes during evolution. Along with the signal sequences, the loops of the LNSP1/2 also constitute the outliers of the amino acid changes when compared to the overall full-length protein (only the loop 2 of the LNSP1 in the two Pthirus lice have significant differences, but for other cases they are much conserved). The observation that the repeat sequence containing domains (N-, M-, and C-domains) which are predicted to have defined secondary structure (likely β-stranded for N- and M-domains) and hence presumed to be important for the function of LNSP1/2 do not have changes that are significantly higher or lower than the full-length values. Hence, the important functionality credited to these domains may allow a certain window of evolutionary modification necessary in the individual lice types for adaptation to their unique environment.

The mutation rate of a protein refers to the frequency at which changes (substitution) occur in the DNA sequence of the gene encoding that protein over a given period. The mutation rates of LNSP1 and LNSP2 can be calculated from the slopes for the graph plotted using the times of divergence from the common ancestor between the different types of lice and the amino acid changes per 100 residues. Furthermore, the Unit Evolutionary Period (UEP) which is the average time required for 1% difference in amino acid sequence to arise is calculated by the inverse of the slope giving the unit in million years (MY). The determined UEPs for LNSP1 (0.37 MY) and LNSP2 (0.48 MY) suggested that it took more time for changes to happen in LNSP2 compared to LNSP1 (Fig. 2), which was also consistent with the fact that LNSP2 in general is more conserved amongst different louse types than LNSP1 (Fig. 1).

The positive relationship between divergence times and the amino acid changes per 100 residues (i.e. percent of sequence differences) has been well-documented with the slope of the regression lines being markedly different for the three proteins: cytochrome c, hemoglobin, and fibrinopeptides^12–14. The estimated UEPs for these proteins were 20.0 MY, 5.8 MY, and 1.1 MY, respectively, indicating a slow pace of evolution for cytochrome c, a fast evolution for fibrinopeptides, and an intermediate evolutionary rate for hemoglobin. Such calculations suggest fibrinopeptides evolving ~18 times faster than cytochrome c and hemoglobin evolving ~3.4 times faster than cytochrome c¹². Other studies report faster evolving proteins of immunoglobulin (UEP = 0.7) and snake venom toxins (UEP = 0.8) at the fastest limit¹³, and it was very intriguing to see our glue protein LNSP1/2 having a faster rate of evolution than these two.

Because the previous studies of molecular evolution have been focusing on species across greater time scale (~600 MY) than our interest within the louse types having divergence of 13 MY at most (Group 4), we sought out to compare the sequence differences of cytochrome c in the six different louse types which included the louse parasitizing one of the Old World monkey (P. badii) in addition to the five louse types considered so far (Fig. S5). When the UEP of cytochrome c among these louse was calculated, the value was 1.3 MY meaning a ~15 times faster evolutionary rate than the so far known cytochrome c. This incredibly fast evolutionary rate in the lice may result from shorter generation time or high number of offspring produced in insects, which merits further investigation in the future. We do not know of any studies of molecular evolution particularly centered in insect species. In any case, taking account of this fast evolution rate of cytochrome c in lice allowed the estimation of the LNSP1/2 evolution to be ~3 times faster than the cytochrome c in lice. Such rate positioned the glue protein LNSP1/2 among proteins having an intermediate evolutionary rate as would be the case for the hemoglobin of the vertebrates (evolving ~3.4 times faster than the vertebrate cytochrome c).

The dN/dS analysis of LNSP1/2 across five type of great ape lice

The ratio of non-synonymous to synonymous substitutions (dN/dS) can be used to measure selective pressures acting on protein-coding genes, in which it is meant to compare the rate of non-synonymous substitution (dN) to the rate of synonymous substitution (dS) across different species. Generally, the non-synonymous substitutions in the DNA sequence which result in a change to the amino acid sequence of the protein are considered to have the potential to affect the protein’s function, and the synonymous substitutions which do not change the amino acid sequence because they occur in the redundant part of the codon to encode the same amino acid are considered neutral. The dN/dS analysis is a powerful tool for understanding molecular evolution and selective pressures acting on genes, helping researchers draw conclusions about the evolutionary history and functional constraints of genes across species. If the dN/dS ratio for a specific gene is greater than one, this likely indicates that the gene had undergone adaptive changes possibly in response to an environmental challenge. On the other hand, the ratio being less than one suggests that the gene is under strong purifying selection, meaning that any substitution that results in altering the amino acid would be detrimental to the organism.

In order to determine whether the selective pressures that had acted on the different regions (N-, M-, C-domains, and loop 1~3) of great ape LNSP1/2 during their diversification and co-evolution with the host, pairwise dN/dS values were calculated via MEGA for all the regions between species (Table 1). The two closely related P. h. capitis (head louse) and P. h. humanus (body louse), noted as Group 1 in this study, have diverged from their common ancestor 0.19 MY ago, and while the head lice lay their eggs directly on hair, the body lice lays their eggs in clothing fibers. Hence, we expected that any region in LNSP1/2 that necessarily changed for the organism’s adaptation to the different egg attaching environment would have resulted in dN/dS >1, while the region important for the general functional properties (i.e. adhesive or anti-desiccant) would have resulted in dN/dS <1. Between the two human head and body louse (Group 1) LNSP1, N-domain had dN/dS of 2.07 suggesting that it underwent adaptive changes (positive selection), in which non-synonymous mutations were favored and spread through the population because they confer some sort of advantage, leading to a higher rate of non-synonymous substitutions compared to synonymous ones (Table 1). Because either the N-domain or the M-domain by itself had similar measured adhesion properties compared to the N- and M-domains together in tandem (with loop 1 in-between)⁷, the adaptive changes in the N-domain alone may not have had critical overall adhesiveness in LNSP1 as a glue protein. Other regions of LNSP1 all had dN/dS <1 suggesting negative (purifying) selection, in which substitutions that would have changed the protein sequences are disadvantageous and are selected against, leading to lower rates of non-synonymous substitutions. Interestingly for LNSP2 of Group 1, LNSP1’s tendencies were not followed: the loop 1 had dN/dS >1 (value of 1.35) and the other region all had dN/dS <1 (Table 1). This fact supports the idea that although being homologous proteins, LNSP1 and LNSP2 possibly have different detailed functions independent of each other. Different physiological functions of LNSP1 and LNSP2 have been observed in the previous study reporting their functional characterization⁶. When LNSP1 was knocked down in head louse, most eggs became nonviable via desiccation, suggesting LNSP1’s role in desiccation resistance. However, knockdown of LNSP2, resulted in another phenotype of oviposition failure, suggesting that LNSP2 on the other hand may serve as more of a basic platform for further deposition of glue and may have an additional function of lubrication. For comparison, there was no non-synonymous substitution in the cytochrome c of Group 1 (dN = 0) (Table S3).

For LNSPs in the Pthirus genus (Group 2 in which they diverged from their common ancestor 3.3 MY ago to parasitize different hosts of human vs. gorilla), except for the loop 2 of LNSP1 (dN/dS = 1.35) and loop 3 of LNSP2 (dN/dS = 1.73), all the region had dN/dS <1. Again, this finding illustrated that the regions of positive selection (dN/dS >1) conferring certain advantage upon change in the amino acids are not conserved between LNSP1 and LNSP2. As noted above, it may again mean that the LNSP1 and LNSP2 have more detailed dedicated functions in Group 2 during the egg attachment as seen for Group 1. As for the case of Group 1, there was no non-synonymous substitution in the cytochrome c of Group 2 (dN = 0) (Table S3).

For further distanced Pediculus genus (Group 3, P. humanus capitis and P. schaeffi which are 5.6 MY from their common ancestor), except for loop 1 in LNSP1 with dN/dS value of 0.94, all the regions in LNSP1 and LNSP2 have dN/dS <1. The dN/dS near one implies that this region is evolving under neutral selection (no selective pressure), meaning that resulting changes in the protein sequences did not significantly affect the fitness of the organism (i.e. changes in the protein sequence are neither advantageous nor disadvantageous), in which the synonymous and non-synonymous substitutions occurred at similar rates, and changes in the protein sequence did not drastically impact the organism’s survival or reproduction. Between the most distanced Pediculus and Pthirus genera (Group 4 of 13 MY from their common ancestor), all regions of LNSP1 and LNSP2 had dN/dS <1. This probably results from the rate of non-synonymous substitution in the functionally essential genes becoming smaller for distantly related organisms with time (evolutionary distance). Because essential genes have to be conserved due to their crucial roles in cellular processes, accumulated non-synonymous substitutions with enough time would be deleterious to lead in their removal from the population. Hence, it is natural for dN/dS in all regions of conserved essential genes like LNSPs to converge to a value <1 over time. For example, cytochrome c of Group 3~5 (Group 5 comparing the cytochrome c of Pediculus/Pthirus vs Pedicinus) had dN/dS values of 0.06~0.07 (Table S3).

Although pairwise calculation of dN/dS (via MEGA) is useful for comparing the average evolutionary rate difference between two sequences, site model-based analysis has its advantages in detecting statistically significant positive selection signals localized to the codon level²⁸. For this purpose, multiple sequences of LNSP1/2 for all the seven individual regions in five louse types were combined as simultaneous codon alignments to perform a site model-based LRT analysis via CodeML (in PAML) (Table 2).

Table 2.

Site model-based likelihood ratio tests (LRTs) and positively selected sites (PSS) identified in LNSP1 and LNSP2

Gene	Region	Models	lnL	Parameter estimates	PSS	P value
LNSP1	Signal sequence	M1a	− 133.18	P0a=0.23833, P1=0.76167, ω0b=0.00000, ω1=1.00000		0.98
		M2a	− 133.15	P0=0.26246, P1=0.00000, P2=0.73754, ω0=0.00000, ω1=1.00000, ω2=1.16505	–
		M7	− 133.70	pc=1.4685, qc=0.005		0.58
		M8	− 133.15	p=0.005, q=1.4052, P0=0.26246, P1=0.73754	6, 16
	N-domain	M1a	− 635.87	P0=0.65988, P1=0.34012, ω0=0.00000, ω1=1.00000		0.0004
		M2a	− 628.09	P0=0.75490, P1=0.00000, P2=0.24510, ω0=0.03476, ω1=1.00000, ω2=6.69602	15, 16**, 24*, 25**, 26, 27, 30, 31, 34, 35, 62*, 82*
		M7	− 636.78	p=0.01723, q=0.03541		0.0002
		M8	− 628.10	p=3.6719, q=99, P0=0.75474, P1=0.24526	15, 16**, 24**, 25**, 26, 27*, 30, 31, 34, 35, 62*, 82**
	M-domain	M1a	− 357.04	P0=0.24805, P1=0.75195, ω0=0.00000, ω1=1.00000		1.68×10-5
		M2a	− 346.05	P0=0.71713, P1=0.00000, P2=0.28287, ω0=0.20308, ω1=1.00000, ω2=29.49191	9*, 13, 19, 26, 27, 29, 32, 35, 38, 40, 43, 52**, 58
		M7	− 357.80	p=0.91525, q=0.005		0.0001
		M8	− 348.60	p=0.005, q=1.5543, P0=0.61801, P1=0.38199	7, 9,** 13, 16, 19, 26, 27, 28, 29, 32, 35*, 38, 40, 43, 52**, 58
	C-domain	M1a	− 816.68	P0=0.51458, P1=0.48542, ω0=0.00000, ω1=1.00000		1
		M2a	− 816.68	P0=0.51458, P1=0.42920, P2=0.05622, ω0=0.00000, ω1=1.00000, ω2=1.00000	12, 92, 94, 96, 124
		M7	− 808.44	p=0.10292, q=0.64287		1
		M8	− 816.68	p=0.005, q=6.4168, P0=0.51454, P1=0.48546	12, 92, 94, 95, 96, 98, 124
	Loop 1	M1a	− 677.07	P0=0.71483, P1=0.28517, ω0=0.10455, ω1=1.00000		0.13
		M2a	− 674.99	P0=0.87197, P1=0.00000, P2=0.12803, ω0=0.19638, ω1=1.00000, ω2=3.39964	17, 33, 62, 100
		M7	− 677.73	p=0.21218, q=0.35714		0.07
		M8	− 675.00	p=24.511, q=99, P0=0.87288, P1=0.12712	17, 33, 62, 100*
	Loop 2	M1a	− 305.86	P0=0.99999, P1=0.00001, ω0=0.35804, ω1=1.00000		1
		M2a	− 305.86	P0=1.00000, P1=0.00000, P2=0.00000, ω0=0.35805, ω1=1.00000, ω2=1.00000	–
		M7	− 305.87	p=55.428, q=99		1
		M8	− 305.87	p=55.427, q=99, P0=0.99999, P1=0.00001	–
	Loop 3	M1a	− 103.43	P0=0.00001, P1=0.99999, ω0=0.92444, ω1=1.00000		1
		M2a	− 103.43	P0=0.99583, P1=0.00033, P2=0.00384, ω0=0.99407, ω1=1.00000, ω2=1.00000	9
		M7	− 103.43	p=2.2071, q=0.005		1
		M8	− 103.43	p=2.2478, q=0.005, P0=0.69901, P1=0.30099	9
LNSP2	Signal sequence	M1a	− 102.67	P0=0.99999, P1=0.00001, ω0=0.06919, ω1=1.00000		1
		M2a	− 102.67	P0=1.00000, P1=0.00000, P2=0.00000, ω0=0.06919, ω1=1.00000, ω2=1.00000	–
		M7	− 102.67	p=7.4976, q=99		1
		M8	− 102.67	p=7.4973, q=99, P0=0.99999, P1=0.00001	–
	N-domain	M1a	− 1324.80	P0=0.52763, P1=0.47237, ω0=0.00625, ω1=1.00000		6.58×10-6
		M2a	− 1312.87	P0=0.50563, P1=0.40668, P2=0.08769, ω0=0.00195, ω1=1.00000, ω2=18.10647	13, 49, 62, 63, 65*, 69**, 71, 72, 83**, 85, 86, 91, 140, 141, 156*, 157*
		M7	− 1324.90	p=0.01264, q=0.00972		9.29×10-6
		M8	− 1313.31	p=0.00607, q=0.00528, P0=0.91840, P1=0.08160	13, 43, 49, 62, 63, 65**, 66, 67, 69**, 71, 72*, 82, 83**, 84, 85, 86, 87, 89, 90, 91, 127, 128, 136, 140, 141, 142, 156**, 157**
	M-domain	M1a	− 400.66	P0=0.44732, P1=0.55268, ω0=0.20206, ω1=1.00000		0.18
		M2a	− 398.93	P0=0.94684, P1=0.00000, P2=0.05316, ω0=0.48704, ω1=1.00000, ω2=22.72604	53
		M7	− 400.76	p=0.5426, q=0.29072		0.16
		M8	− 398.93	p=20.162, q=21.099, P0=0.94689, P1=0.05311	53
	C-domain	M1a	− 795.07	P0=0.70269, P1=0.29731, ω0=0.00000, ω1=1.00000		0.02
		M2a	− 790.99	P0=0.70434, P1=0.26973, P2=0.02592, ω0=0.00000, ω1=1.00000, ω2=76.39935	34, 47, 58*, 59, 65, 98, 101, 117, 127**, 140, 143
		M7	− 795.76	p=0.00597, q=0.01855		0.01
		M8	− 791.19	p=0.005, q=0.01535, P0=0.97166, P1=0.02834	34, 47, 58*, 59, 65, 98, 101, 117, 127**, 140, 143
	Loop 1	M1a	− 683.96	P0=0.61828, P1=0.38172, ω0=0.06373, ω1=1.00000		0.005
		M2a	− 678.69	P0=0.88283, P1=0.00000, P2=0.11717, ω0=0.21696, ω1=1.00000, ω2=5.54259	1*, 14, 15, 51, 52, 80, 92*
		M7	− 684.26	p=0.005, q=0.005		0.004
		M8	− 678.70	p=27.688, q=99, P0=0.88316, P1=0.11684	1*, 14, 15, 51, 52, 80, 92*
	Loop 2	M1a	− 528.33	P0=0.54351, P1=0.45649, ω0=0.00000, ω1=1.00000		0.57
		M2a	− 527.77	P0=0.59440, P1=0.00000, P2=0.40560, ω0=0.01318, ω1=1.00000, ω2=1.72299	17, 19, 21, 28, 36
		M7	− 528.49	p=0.00528, q=0.005		0.49
		M8	− 527.78	p=0.005, q=2.543, P0=0.57333, P1=0.42667	17, 19, 21, 27, 28, 30, 33, 34, 36*, 37, 42, 44, 47, 51, 53, 65, 69
	Loop 3	M1a	− 447.19	P0=0.27692, P1=0.72308, ω0=0.00000, ω1=1.00000		0.62
		M2a	− 446.70	P0=0.27550, P1=0.66130, P2=0.06321, ω0=0.00000, ω1=1.00000, ω2=6.70100	10, 32
		M7	− 447.23	p=0.02188, q=0.00766		0.64
		M8	− 446.78	p=0.02728, q=0.01573, P0=0.93382, P1=0.06618	10, 17, 28, 32, 35

^*, **Pr, posterior probability; *Pr ≥ 0.95, **Pr ≥ 0.99.

^aP_n, Proportion of sites in site class n.

^bω_n, dN/dS ratio for site class n.

^cParameters of the beta distribution.

For LNSP1, significant differences were observed in both model comparisons, particularly in the N- and M-domains, suggesting strong positive selections (10^-5 < P < 10^-3) in these regions with no additional significant signals elsewhere (Table 2). However, in LNSP2, the N-domain only exhibited strong positive selection (P~10^-5) with the C-domain and loop 1 having moderate positive selections (10^-3 < P < 10^-2). No significant positive selection was detected in other regions of LNSP2. In summary, both LNSP1 and LNSP2 again showed positive selection patterns in specific domains as observed in the pairwise analysis, indicating that the two genes had evolved under functionally distinct selection pressures. In particular, the strong positive selection signal taking place both at the N-domain of LNSP1/2 supported that this region must have been critical in making necessary changes during environmental adaptation. As anticipated, the site-model analysis of the cytochrome c gene revealed no evidence of significant positive selection (Table S4), consistent with its key functional role in cellular respiration, subject to strong purifying selection.

Concluding remarks

In this study, amino acid sequences of the louse egg glue protein − the paralogous LNSP1 and LNSP2 − in five great ape louse types (three in human (head/body louse of the Pediculus genus and pubic louse of the Pthirus genus), one in gorilla (Pthirus genus), and one in chimpanzee (Pediculus genus)) were aligned and comprehensively compared (Fig. 1). To our knowledge, this study is the first to compare orthologous proteins across multiple closely related species separated by up to only ~13 MY from their common ancestor, and it provides a framework that can be extended to other survival-critical genes in diverse biological systems.

The comparisons were done throughout the seven regions in the sequence: the signal sequence, the N-, M-, C-domains and the loops (loop 1~3). The signal sequences of the LNSP1 and LNSP2 were very conserved allowing them as markers for distinguishing the two paralogous proteins. Especially, the signal sequence of LNSP2 was 100% identical even across species ~13 MY apart (Pediculus and Pthirus genera). Another characteristic that differentiates the LNSP1 and LNSP2 was the N-domain which is half sized in all LNSP1 compared to that of LNSP2. The M-domain varied mostly in Pthirus in that it was completely missing in the P. gorillae LNSP1, and that its size decreased in half in the P. pubis LNSP2. It was also notable that the C-domain of the P. gorillae LNSP1 is half the size when compared to other LNSP1/2s. These observations clearly show that there had been variations allowed in LNSP2 depending on species, which has not resulted in altering the important overall function as an egg glue protein. While such variations are inevitably expected to result in structural and functional consequences, they may have been tolerable due to the presence of two paralogous forms of LNSP1 and LNSP2.

Further analysis suggested that LNSP2 in all the domains and loops were generally more conserved across various louse types than LNSP1. Alteration in the loops had happened mostly in the LNSP1 of the two Pthirus genus lice. Namely in Pthirus LNSP1, loop 2 was only a third in size compared to that in the Pediculus genus, and loop 3 was completely missing. All things considered, detailed characterization of the different domains and the loops within LNSP1/2 in relation to its function would be necessary to gain further meaning on the various changes which had happened during the course of evolution. The changes in amino acids analyzed with evolutionary distances among various louse types further suggested that the signal sequence and the loops either had significantly faster or slower mutation rates with time compared to the M-, N-, and C-domains (Fig. 2). Again, whether this has any meaning related to the functionality of LNSP1/2 needs further investigation.

The UEP (average time required for 1% difference in amino acid sequence to arise) values for LNSP1/2 calculated by taking the inverse of the mutation rates (Fig. 2) were 0.37 MY (LNSP1) and 0.48 MY (LNSP2). When compared to the UEP for cytochrome c (1.3 MY) of the louse types in this study, it demonstrates that the egg glue proteins evolve ~3 fold faster than cytochrome c which is known to be one of the slowest evolving proteins. Still, the very fast mutation rate of cytochrome c in louse species revealed in this study, which is ~15-fold faster than that of the cytochrome c (UEP = 20 MY) studied by comparing cytochrome c sequences across more distantly related species requires in-depth investigation. Systematic analysis of cytochrome c sequences derived from the rapidly growing whole genome data accumulated across many diverse insect species will shed light on this issue.

In order to see whether selective evolutionary pressures had acted differently on the various regions of great ape LNSP1/2 due to their diversification and host-specific co-evolution, dN/dS values were calculated pairwise in different four groups for all seven regions. From this analysis, we expected that it would point out where the critically important region of function in LNSP1/2 would reside. Most of the regions in LNSP1/2 had dN/dS <1 suggesting that non-synonymous mutations in these parts that altered the protein function by resulting changes in the amino acid had been detrimental (negative selection). However, some regions such as the N-domain of LNSP1 when compared between P. h. capitis and P. h. humanus (dN/dS = 2.07), the loop 2 of LNSP1 when compared between P. pubis and P. gorillae (dN/dS = 1.73), and the loop 1 of LNSP2 when compared between either P. h. capitis and P. h. humanus (dN/dS = 1.35) or P. pubis and P. gorillae (dN/dS = 1.48) had dN/dS >1 suggesting that non-synonymous mutations and the resulting amino acid changes in these regions conferred some advantage during evolution (positive selection). Such positive selections may have been necessary for adaption to different environments in the two lineages of lice: for example, the P. h. capitis (head louse) laying their eggs on the hair while the P. h. humanus (body louse) laying their eggs in the clothing fibers. Also for the P. pubis (pubic louse) and the P. gorillae (gorilla louse), different environments rooting from the host difference may have led to the positive selection.

Complementary to the pairwise dN/dS estimates, the LRT analysis using combined sequences of LNSP1/2 in five types of lice allowed more precise detection of positive selection signals in LNSP1/2. Both LNSP1 and LNSP2 again showed positive selection patterns at specific domains supporting that the two genes had evolved distinctly under different environmental pressures. The analysis further suggested that the N-domain of both LNSP1 and LNSP2 had undergone most changes during environmental adaptation. In any case, both methods of dN/dS calculation clearly showed that selective pressures not only had occurred differently in various regions of LNSP1/2, but also had happened differently between LNSP1 and LNSP2. This may originate from the fact that the two homologous proteins have different elaborate functionalities in addition to being glue proteins⁶, which would in turn respond differently to various environmental factors during the course of evolution.

Abbreviations

LNSP

Louse nit sheath protein

SRA

Sequence read archive

UEP

Unit evolutionary period

Author contributions

S.P. devised the experiment. J-Y.Y. performed the bioinformatics studies and produced the DNA and protein sequences of LNSPs. J.H.K. verified the sequenced of P. pubis LNSP1 and LNSP2 by Sanger sequencing. C.W.P. and S.P. analyzed the LNSP sequences. All the authors co-wrote the manuscript.

Funding

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2021-NR060140 and RS-2023-00243435).

Data availability

The LNSP1 and LNSP2 sequences reported in this paper have been deposited in the GenBank database. The GenBank accession number for P. humanus capitis (head louse) LNSP1 which results from Sanger sequencing previously not reported is “GenBank: PQ815049”. The GenBank accession numbers for other LNSP1 and LNSP2 are as follows: “GenBank: BK072090” and “GenBank: BK072093” respectively for P. pubis (pubic louse); “GenBank: BK072091” and “GenBank: BK072094” respectively for P. gorillae (gorilla louse); “GenBank: BK072089” and “GenBank: BK072092” respectively for P. schaeffi (chimpanzee louse). The cytochrome c sequences used for analysis have been also deposited in the GenBank database with the following numbers: “GenBank: PQ819564” for P. h. capitis (head louse); “GenBank: BK072085” for P. pubis (pubic louse); “GenBank: BK072086” for P. gorillae (gorilla louse); “GenBank: BK072087” for P. schaeffi (chimpanzee louse); “GenBank: BK072088” for the P. badii (Old World monkey red colobus louse).

Competing Interests

The authors declare no competing interests.