Isolation and Proteomic Characterization of the Mouse Sperm Acrosomal Matrix

our AM proteome amyloidogenic cystatin CRES, within the epididymal lumen (59, into potential and


INTRODUCTION
An important step during fertilization is the sperm acrosome reaction in which the acrosome, an exocytotic vesicle overlying the sperm head, releases its contents allowing the spermatozoon to penetrate through the investments surrounding the oocyte. The acrosome is essential for normal fertilization since spermatozoa lacking these structures are infertile (1,2).
The acrosomal contents are compartmentalized into soluble and an insoluble/particulate material termed the acrosomal matrix (AM). The AM has been characterized as an electron dense and membrane-free insoluble material that remains following treatment of spermatozoa with Triton X-100 (3). Functionally it is thought to provide a stable scaffold for interactions between the sperm and oocyte and to allow the controlled and sequential release of matrix associated proteins important for fertilization. The importance of AM function during fertilization is emphasized by its conservation in spermatozoa across species including hamster, guinea pig, bull, stallion, boar, quail, water strider and human (4)(5)(6)(7)(8)(9)(10)(11). Proteases activated by increasing intraacrosomal pH as a result of the acrosome reaction are thought to contribute to the disassembly of the AM (3). However, the precise mechanism by which this occurs is not known.
The mechanism by which the AM forms is also not known but the self-assembly of proteins into a large complex has been proposed (3). Cytoskeletal proteins have also been found associated with the matrix and are thought to contribute to the scaffold structure (12).
Because of its critical role during fertilization, considerable effort has been put towards developing procedures for the isolation of the AM and thus far AM have been isolated from guinea pig, hamster, and bovine cauda epididymal spermatozoa (6,13,14). Several AM associated proteins have been identified either by biochemical analyses of the isolated structure or by immunolocalization of proteins to AM that remained associated with spermatozoa following Triton X-100 exposure. These proteins include Acr, proacrosin; Acrbp, proacrosin binding protein; Zpbp, zona pellucida binding protein; Zp3r, zona pellucida 3 receptor; zan, zonadhesin, and others (3). Although collectively these studies have identified a number of AM by guest on May 8, 2020 https://www.mcponline.org Downloaded from associated proteins, a full proteomic analysis of the AM has not been carried out. Unfortunately, the AM has also not been successfully isolated from mouse spermatozoa, the species in which fertilization is well-studied and in which gene knockout models are prevalent. Indeed, for some time it was questioned whether mouse sperm acrosomes even possessed an acrosomal matrix structure (15). The difficulty in isolating the mouse sperm AM may stem from the fusiform shape of the sperm head, the small size of its acrosome compared to that in guinea pig and hamster, and/or the general fragile nature of mouse spermatozoa compared to spermatozoa from other species. Also, to date, isolation of the AM has only been described for mature cauda epididymal spermatozoa and not for immature caput epididymal spermatozoa.
Herein, we describe a procedure for the isolation of the AM from caput and cauda mouse epididymal spermatozoa. Using mass spectrophotometric analyses we then carried out a proteomic characterization of the proteins present in the AM from these two sperm populations. These studies reveal the identity of 501 new proteins not previously found in spermatozoa by a proteomics approach. Furthermore, differences in AM protein composition were observed between the caput and cauda spermatozoa suggesting that the AM may undergo maturational changes during epididymal transit in preparation for downstream functions during fertilization. Together, these studies show the AM as a dynamic functional structure containing a diverse group of proteins including structural proteins, transporters, enzyme modulators, proteases, chaperones, kinases and others. To isolate AM from spermatozoa, a modification of previous methods used to expose but not extract the AM from mouse spermatozoa was followed (15,17). Because caput and cauda epididymal spermatozoa are in different maturational states, they required different percentages of Triton X-100 to extract the cell membranes and thus two different isolation procedures were developed to isolate the AM from these two populations of cells.
To isolate AM from caput epididymal mouse spermatozoa, 2.6-4 X 10 6 purified spermatozoa were incubated in 200µl 20 mM Tris-HCl, pH 7.4 containing 2% Triton X-100 (Surfact-Amps, cat.no.28314, Thermo Scientific, Rockford, IL) for 2 h on ice. After incubation, the cell suspension was centrifuged at 2000 x g for 5 min at 4°C to isolate a supernatant by guest on May 8, 2020 containing soluble proteins and membrane that was designated the Triton-soluble fraction and a pellet containing spermatozoa with exposed AM. The sperm pellet was resuspended in 200µl 20 mM sodium acetate pH 3 and vortexed for 2 min at RT using a Vortex Genie2 (ThermoScientific) set to position 4 to release the AM from spermatozoa. The sample was then centrifuged at 500 x g for 5 min at 4°C resulting in a supernatant containing released AM, designated as the AM fraction, and a pellet containing mainly spermatozoa without AM but also some that still had their AM attached. To increase the recovery of AM, the pellet was resuspended in 200 µl 20 mM sodium acetate pH 3, centrifuged at 500 x g for 5 min at 4°C and the resulting supernatant (AM fraction) pooled with the previous one. The pellet was washed once in 20 mM sodium acetate pH 3 by centrifugation (500 g, 5 min, 4°C) and the final pellet containing the extracted/washed spermatozoa without AM was resuspended in 20 mM Tris-HCl, pH 7.4 and designated the extracted sperm (Ext Spz) fraction.
To isolate AM from cauda spermatozoa, 1 X 10 7 purified spermatozoa were resuspended in 200 µl 20 mM Tris-HCl, pH 7.4 containing PIC and 0.625% Triton X-100 for 2 min on ice. The extraction buffer was supplemented with additional PIC to prevent dispersion of the AM. The remaining purification steps were the same as described for caput spermatozoa.
The number of isolated AM and the percent contamination by spermatozoa was calculated by examining an aliquot of AM fraction that was spread on a slide and stained with peanut agglutinin (lectin from Arachis hypogaea, cat.no.L7381, Sigma, Saint Louis, MO) conjugated to FITC that binds to glycoconjugates on the acrosomal matrix (18). The cauda acrosomal matrix preparation ranged from 98-100% pure while the caput acrosomal matrix preparation ranged from 89-95% pure. PNA was used to visualize the AM because of the ease of the staining procedure and because its staining correlated well with that of antibodies against known acrosomal matrix proteins such as zonadhesin. Coomassie Blue R-250 in 40% methanol, 10% glacial acetic acid. Gels were scanned prior to slicing of the gel into bands for subsequent MS analysis. Gels for caput epididymal AM were done in duplicate using AM purified in two different preparations and representing 1-2.5 X 10 6 purified AM while those for cauda AM were done in triplicate using AM purified from three different preparations representing 0.7-2.2 X 10 7 purified AM. Replicates were performed to improve our chances to identify low abundant proteins by loading increasing numbers of AM on the SDS-PAGE gels and are not technical or biological replicates. Due to fewer spermatozoa present in the caput compared to the cauda epididymis, fewer caput AM were analyzed by LC-

Tryptic in-gel digestion
Each SDS-PAGE gel lane containing purified AM was cut into 8-9 slices and each slice placed into a 0.5 ml microcentrifuge tube. In-gel digestion was performed on each slice as described previously (22). Briefly, the gel pieces were washed with 50/50 ACN/100 mM
Peptides were first injected onto the trapping column, which was equilibrated with 1% ACN, 0.1% formic acid in water and washed for 10 min with the same solvent at a flow rate of 300 nl/min. After washing, the trapping column was switched to the reverse-phase analytical column and bound peptides eluted using solvents A (2% ACN, 0.1% formic acid in water) and B (98% ACN, 2% water, 0.1% formic acid). The gradient was kept constant for first 10 min at 4% solvent B followed by a linear increase up to 30% solvent B for 20 minutes. Solvent B was further increased to 60% for 40 minutes followed by a fast increase of solvent B up to 90% over 5 minutes. The eluted peptides were directed into the nanospray ionization source of the LTQ-XL with a capillary voltage of ∼2 kV. The collected spectra were scanned over the mass/charge

Sequence database search
To get the maximum number of publicly available sequences for analyses, the mouse sequences from Ensembl (http://www.ensembl.org/info/data/ftp/index.html) were merged with those from NCBI (http://www.ncbi.nlm.nih.gov/guide/). The database was built using BioPerl modules installed on a MacBook Pro (OS X 10.6.8) using packages from the Fink project

Protein and peptide identification
Spectra obtained from the trypsin digestion products using the LTQ Orbitrap XL mass spectrometer were identified by the Proteome Discoverer ™ (version 1.3) program, based on SEQUEST cluster as a search engine (University of Washington, USA, licensed to Thermo Electron Corp., San Jose, CA) against our mouse database (203,220 nonredundant protein sequences). The search engine used the following parameters: precursor ion mass tolerance, 2.5 Da; fragment ion mass tolerance, 0.8 Da; fully tryptic enzyme specificity; two missed cleavages; dynamic modifications of cysteine carbamidomethylation and of methionine oxidation.
The proportion of false positive assignations among the tentative peptide identifications, also called false discovery rate (FDR), has been estimated by using decoy databases constructed from the target database (23) and was set at 1%. Spectra and search results may be downloaded from ProteomeCommons.org Tranche using the following hash: kLgCpMG61AQRPcKwA4JSCwedj9M0618nVfWyObpGuQBMAar0BuQlB77d/bkyUqCWrLnPDc uF6lKKRytyeV+rAb19oPsAAAAAAAAEiw==. Msf files can be visualized using the freelyavailable viewer thermo-msf-parser (24). For bioinformatic analyses, sequence accession by guest on May 8, 2020 numbers of identified proteins were converted to their corresponding gene ID from the Mouse Genome Database (25).

Mouse sperm proteome database
In order to compare our results with the mouse sperm proteome, a database was built by To decipher GO terms that were "specific" to the caput or cauda AM, we looked for the most representative terms of each protein list using two criteria: 1) at least six proteins had to be associated with the GO term; and 2) these proteins represented 70% or more of all proteins associated with the GO term. For example, if a GO term was represented by 10 proteins in the sperm AM proteome, 7 of which were detected only in the caput AM, then this GO term was considered as a being caput AM-specific. To compare proteins identified in the isolated AM with those associated with lysosome-related organelles (LRO), proteins associated with endosome Sequence alignment of the cystatin 2 family members including the CRES subgroup and lysozyme family members was done using Clustal W (version 2.1, (41)) and amyloid prediction was determined using Waltz (42) with parameters set at threshold= best overall performance and pH=2.6.

Low abundant proteins of interest
The data generated by shotgun proteomics experiments are highly redundant, i.e. a subset of the peptides present is repeatedly and preferentially selected for fragmentation and thus identified. In contrast, other subsets of peptides, e.g. those derived from low abundance proteins, are more difficult to detect, and a large number of fragment ion spectra have to be acquired to increase the likelihood of their detection (43, 44). To bypass this issue in the identification of type 2 cystatins and CRES subgroup members and since only one CRES member, CST13 was identified with an FDR < 1%, we decided to look for peptides that matched to this protein family without using an FDR filter but which were assigned by Sequest as being by guest on May 8, 2020 of medium and high confidence. For these peptides, the experimental and theoretical MS/MS spectra were confirmed visually (Supplemental Data 11).

Isolation of mouse sperm AM
To isolate AM from mouse spermatozoa we began by working with caput epididymal spermatozoa which are more resistant to extraction than cauda epididymal spermatozoa and therefore provide a more stable structure with which to work. Unlike cauda spermatozoa, caput epididymal spermatozoa have not undergone the maturation process which includes modifications of the sperm membrane and thus required increased amounts as well as longer incubations in Triton X-100 to remove the sperm membrane. A brief vortexing step allowed the release of the AM from the demembranated spermatozoa. Cauda epididymal spermatozoa required much lower concentrations and extremely short exposure times to Triton X-100, the presence of additional protease inhibitors as well as the brief vortexing step to allow the release of AM from the spermatozoa without full dispersion of the structure.
To demonstrate the isolated structures represented AM, AM from both caput and cauda spermatozoa were analyzed by indirect immunofluorescence using antibodies against known AM markers as well as the lectin peanut agglutinin, PNA, which binds to glycoconjugates on the acrosome and acrosomal matrix. As shown in Figure 1A, both caput and cauda isolated AM were immunostained with zonadhesin (ZAN), acrosin, (ACR) and acrosin binding protein, (ACRBP) antibodies as well as PNA supporting that the isolated structure indeed represents the AM. In contrast to caput AM that appeared to have a full crescent shape characteristic of the acrosome, cauda AM were of a more blunt shape suggesting that the cauda AM may be more fragile and that some dispersion of the AM may occur during the isolation procedure. Indeed, while zonadhesin was present in the majority of the isolated AM from both the caput and cauda by guest on May 8, 2020 spermatozoa, fewer of the isolated AM from both sperm populations contained ACRBP.
Because proteins are known to be differentially released from the AM (45), the absence of ACRBP from some of the isolated AM suggests that some proteins are starting to be released during the isolation procedure.
To confirm the immunofluorescence studies, the isolated AM as well as the Triton-soluble fraction and extracted spermatozoa from both the caput and cauda epididymis were examined for the pro and mature forms of acrosin by Western blot analysis. As shown in Figure 1B, as expected the proform of acrosin (53 kDa) was enriched in both the caput and cauda AM with some of the mature acrosin (39 kDa) released into the Triton soluble fraction during the isolation procedure. In addition to the 53 kDa proacrosin, the caput AM also contained a higher molecular weight form (~55 kDa) that may represent a second proform of acrosin. This is similar to the mixture of 53-55 kDa proacrosin forms found in ejaculated porcine spermatozoa (46, 47). A 25 kDa immunoreactive protein was also detected in the Triton soluble fraction which may represent a processed form of the mature acrosin. Little to no proacrosin/acrosin was detected in the spermatozoa following the removal of the AM confirming its localization within the matrix and the efficiency of the AM extraction procedure. To control for the very small number of spermatozoa that was present in the AM preparation, the same number of spermatozoa detected in the AM was loaded on the gel for analysis of acrosin. No detectable proacrosin/acrosin was detected in these samples (Spz con) supporting that the acrosin detected in the AM represented that of the isolated AM structure and not contamination by whole spermatozoa.

MS analysis of isolated sperm AM
The isolated sperm AM were next analyzed by MS to identify proteins associated with this structure. Both caput and cauda AM were isolated as described and proteins separated by SDS-PAGE followed by Coomassie staining (Figure 2A). Different electrophoretic patterns by guest on May 8, 2020 were observed between the caput and cauda AM samples which may reflect differences in the maturational status of the spermatozoa as well as that fewer caput spermatozoa were loaded on the gel compared to cauda spermatozoa due to fewer spermatozoa present in the caput region.
The AM lanes including the stacking gel were cut into 8-9 gel slices, protein digested with trypsin, and samples analyzed by MS. From the MS analysis of caput and cauda AM, 1026 proteins that matched to a gene in MGI were identified with a confidence of 99% proteins were similar to that identified in the 1026 AM proteins and included cytoskeletal proteins, nucleic acid binding, enzyme modulators, hydrolases, and others ( Figure 2D) (protein list in Supplemental Data 5).
As a confirmation that the structure analyzed by MS represented the AM, we next examined the 1026 proteins identified in the AM to determine if known AM markers were present. Currently there are twenty-two proteins that have been shown to be associated with the sperm AM including ACR, ACRBP, ZAN, ZP3R and others. Of these twenty-two proteins, seventeen including those mentioned above, were identified in our AM preparations with the majority of these proteins identified in both the caput and cauda sperm AM (Table 1). These observations support our previous immunofluorescence and Western blot studies that AM were successfully isolated from both the caput and cauda spermatozoa.

Biological networks and pathways
Having validated that the isolated structures represented AM, we next began to examine the 1026 AM proteome in greater detail by using biological network and pathway software to identify pathways enriched in the AM suggestive of AM functions. We also wanted to determine if these pathways differed between the functionally distinct caput and cauda spermatozoa.
Using the Gene Ontology (GO) database we first examined the 1026 AM proteome for proteins classified by the GO terms fertilization/sperm-egg recognition/acrosome reaction/fusion and identified twenty-three proteins including Catsper1 and 2, acrosin, Cd46, a regulatory component of the complement system; Park7, which belongs to the C56 family of peptidases with a putative role as a redox-sensitive chaperone and as a sensor of oxidative stress; Spa17, with proposed roles in cell adhesion; Pkdrej, a polycystin which may generate a calcium channel involved in the acrosome reaction; Zbp2, Zp3r, and Izumo1 ( Table 2). Several of these proteins have previously been shown to be involved in fertilization but not to be associated with the AM. by guest on May 8, 2020 To determine whether protein composition differed between the caput and cauda AM, we compared the AM proteins present in the caput spermatozoa (664 proteins) to that in the cauda spermatozoa (873 proteins) and found, not surprisingly that 511 were common as shown in the Venn diagram in Figure 3A (protein list in Supplemental Data 6). However, 153 AM proteins were unique to the caput spermatozoa while 362 proteins were unique to cauda spermatozoa suggesting that during epididymal transit some AM proteins may be lost or modified and other proteins may be added to the AM. Although the association of epididymal secretory proteins with the sperm surface including the acrosome is a common occurrence as spermatozoa pass through the epididymis, the addition/incorporation of epididymal secretory proteins into the AM has not been demonstrated. As shown in Table 3, analysis of the 1026 AM proteome showed that several known epididymal secretory proteins, that are known not to be expressed in the testis, were found in the isolated sperm AM suggesting they were added during sperm transit in the epididymis. These proteins included Adam7 and Gpx5 which are specifically expressed by the caput epithelium (48, 49) and detected in the caput AM, and defensin beta 30, expressed by all regions of the epididymis and detected in the cauda AM.
Several lipocalin family members including Lcn2 and Lcn5 expressed primarily by the caput region were also detected in the caput and cauda AM, respectively.

Protein functions enriched in the caput AM
Using the Gene Ontology (GO) database and a confidence of 99% (p<0.01), we compared the 664 proteins present in the caput AM with all proteins detected in the AM (1026 proteins) to determine whether there were specific functional classes of proteins that were enriched in the caput AM. As shown in Figure 3B Using the same analysis parameters as with the caput AM we examined whether the 873 proteins present in the cauda AM were enriched in distinct functional classes by comparing these proteins to those present in the 1026 AM proteome. However, no enriched groups were identified with p<0.01 or p<0.05. This may be because more proteins were identified in the cauda AM compared to that in the caput and thus the ability to detect enriched groups was decreased. Alternatively, the cauda AM is in a different functional state from that in the caput and many classes of protein are equally important.
by guest on May 8, 2020

Caput AM-specific proteins
To examine more closely protein functions that may be distinct to the caput and cauda AM, we carried out a GO analysis of the 153 proteins found only in the caput AM and the 362 proteins detected only in the cauda AM. We looked for the GO terms that were most representative of the proteins in each AM preparation and reflected caput or cauda "specific"AM functions based on the following criteria: 1) at least six proteins had to be associated with the GO term; and 2) these proteins represented 70% or more of all proteins associated with the GO term. For example, if a GO term was represented by 10 proteins in the sperm AM proteome, 7 of which were detected only in the caput AM, then this GO term was considered as a being caput AM-specific function.
As shown in Figure 3C, caput-specific GO terms included proteins involved in the biological processes of negative regulation of EGFR signaling pathway and synaptic transmission, associated with cellular compartments such as the endoplasmic reticulum and cortical cytoskeleton or with organelles such as melanosomes, and involved in molecular functions including structural constituent of cytoskeleton and actin binding. We chose to examine in greater detail an example of each GO class and selected synaptic transmission as the example of a biological process based on previous suggestions that spermatozoa have neuronal-like functions (50, 51). The caput AM proteins associated with the GO term synaptic transmission were all involved in cell signaling and protein transport and included camk4, calcium/calmodulin dependent protein kinase IV (a serine/threonine kinase); akap9, a kinase anchoring protein 9 (binds to regulatory subunit of protein kinase A); aldh2, aldehyde dehydrogenase; ap2a1, ap2a2, ap2b1, adaptor-related protein complex 2 subunits alpha1, alpha 2 and beta 1 (components of the AP-2 adaptor protein complex involved in protein transport via transport vesicles); Glu1, glutamate ammonia ligase which catalyzes synthesis of by guest on May 8, 2020 glutamine from glutamate and ammonia (glutamine is involved in inhibition of apoptosis and cell signaling); Gnb1, guanidine nucleotide binding protein 1, (G protein, cell signaling); and Myo6, myosin 6, involved in intracellular vesicle and organelle transport.
Under the GO class cellular compartment, we examined proteins that contributed to the GO term melanosome, based on studies suggesting that the sperm acrosome has some functional overlap with lysosome related organelles including melanosomes (52). Proteins that fell under this GO term included Pmel, premelanosome protein, that forms internal matrix fibers in melanosomes; Anxa2, annexin a2, a calcium-dependent phospholipid binding protein involved in signal transduction; Pdia6 and Pdia3, protein disulfide isomerase family members which are part of a large chaperone multiprotein complex and that inhibit aggregation of misfolded proteins and which play roles in the folding of disulfide-bonded proteins; P4hb, prolyl 4-hydroxylase beta polypeptide, also a member of the protein disulfide isomerase family and which at low concentrations facilitates protein aggregation (anti-chaperone) while at high concentrations inhibits protein aggregation (chaperone); Hsp90aa1, an inducible chaperone that promotes structural maintenance of proteins involved in signal transduction and which interacts with many different proteins including ion channels; and Canx, calnexin, a calcium binding molecular chaperone that assists in protein assembly and plays a role in quality control in the ER.
Under the GO class molecular function we examined the GO term structural components of the cytoskeleton because the acrosomal matrix has been shown to contain cytoskeletal proteins (12). Proteins that were listed under this term included the spectrins Spna2, Spnb1, Spnb2, scaffold proteins that organize intracellular organelles; Epb4.1, involved in cytoskeleton/plasma membrane interactions; and Plec1, Vim, and Dsp, involved in cell signaling and scaffolding with the cytoskeleton. Together these data suggest that within caput spermatozoa a fine tuning of the assembly and organization of the AM may be occurring by guest on May 8, 2020 allowing key multiprotein complexes to become oriented properly for downstream functions during fertilization.

Cauda AM-specific proteins
To examine the proteins that were present in the cauda AM but not the caput AM, a GO analysis of the 362 cauda AM proteins was carried out as described above for the caput AM.
Consistent with the idea that the cauda sperm AM is in a different functional state that the caput sperm AM, many GO terms falling under the classes of biological processes, cellular compartments, and molecular functions were identified and thus due to space limitations are shown in Supplemental Data 8. Shown in Figure 3E were grouped under the GO term serine-type peptidase activity included Rikb, a Riken clone (1810009J06Rik) with predicted serine peptidase activity; Abhd10, adhydrolase domain containing protein 10, Gm2663, predicted to have serine peptidase activity; Gzmn, granzyme H; Htra1, HtrA serine peptidase 1, a secretory peptidase that regulates availability of IGFs; Htra2, HtrA serine peptidase 2, involved in apoptosis; Immp1l, inner mitochondrial peptidase; Prcp, lysosomal prolylcarboxypeptidase involved in activation of cell matrix prekallikrein; Prss21, testisin, a serine protease; Prss52, a chymotrypsin-like serine protease primarily found in Leydig and Sertoli cells; and Tpp2, tripeptidyl peptidase II, a component of the proteolytic cascade acting downstream of the 26S proteasome.

Relationship of the sperm acrosome to lysosome-related organelles (LRO)
Several reports have suggested that the sperm acrosome is derived from and has functions similar to that of lysosomes while other studies have suggested relationships with endosomes or secretory granules. Indeed, the acidic pH of the sperm acrosome is not unlike that within lysosomes while several components of endocytic transport are proposed to be involved in acrosomal biogenesis (53, 54). Also within secretory granules, proteins are known to be compartmentalized and the acrosome reaction has been compared to the secretion process that occurs from secretory granules. Because cumulative studies do not strictly support the acrosome as being solely Golgi-derived or lysosomal in origin, (55) proposed the acrosome as a novel lysosome-related organelle (LRO), which are membrane-bound cytoplasmic organelles including melanosomes, endosomes, and synaptosomes that are restricted to specific cell types and that carry out functions unrelated to degradation. LROs utilize both the synthetic (derived from Golgi) as well as retrograde (endocytic) transport pathways during their biogenesis as seems to also occur during acrosome formation (52). To determine whether the AM showed similarities with LROs, we compared the 1026 AM proteome with those proteins that were classified by the NCBI GO cellular component term as being associated with by guest on May 8, 2020 endosomes, melanosomes, synaptosomes, lysosomes and secretory granules. Figure 4A shows the Venn diagram indicating the overlapping proteins between these cellular structures (protein list in Supplemental Data 9). Because the GO database considers the sperm acrosome a secretory granule or a lysosome, several proteins that were associated with these organelles were from the acrosome. While the AM proteome showed only a 2%, 4%, and 4% match to proteins in endosomes, lysosomes, and synaptosomes, respectively, there was a 16% and 21% overlap of proteins with secretory granules and melanosomes suggesting a more similar relationship between these organelles and the sperm AM. Also, while the overlap between the LRO and the AM was not large, it is important to realize that the sperm proteins represented those of the AM only and not the entire acrosome. Thus it is possible that proteins present in the sperm AM may be present within or associated with similar matrix-like structures in the related organelles.
The proteins that were common between the AM and LRO were categorized based on Panther protein classes as shown in Figure 4B (Supplemental Data 10) and included hydrolases (fucosidase, collagenase, Na/K transporting ATPase subunits 1 and 2, galactocerebrosidase, angiotensin converting enzyme, lysosomal pro-X carboxypeptidase) and chaperones (Hsp90 alpha, Hspa8, 14-3-3 protein, endoplasmin (Grp94),calnexin, T-complex protein 1) as being represented by the most number of proteins followed by enzyme modulators (Rab2A, Rab14, RhoB, son of sevenless homolog 2, AKAP3) proteases (pro-X carboxypeptidase, ADAM2, acrosin), transporters (Na/K transporting ATPases subunits 1,2,3 , Catsper4), transfer/carriers (annexin A2, hippocampal cholinergic neurostimulating peptide, secretory carrier-associated membrane protein 2), isomerases (protein disulfide isomerases A3, A6), membrane trafficking (clathrin heavy chain, vesicle-associated membrane proteins 2,3), and others. by guest on May 8, 2020 Amyloidogenic proteins in the AM The aggregation of proteins is a proposed mechanism by which cells sort proteins to the secretory pathway (56). Also, several recent reports suggest that amyloids, protein aggregates with a specific cross-β sheet structure, may contribute to the formation of stable structures that carry out biological functions within the cell. Specifically, the PMEL protein, known to be associated with a matrix-like structure within melanosomes, was shown to form amyloid in vitro and was associated with an amyloid structure within the lumen of the melanosome (57) With this background in mind, we next utilized our AM proteome data to determine whether known amyloidogenic proteins were present within the sperm AM. In this approach we either used a cutoff of 99% confidence (p<0.01) or no FDR (false discovery rate) and looked for peptides that fulfilled all the spectral requirements for a confident identification (visual comparison of experimental and theoretical MS/MS spectra). As shown in Table 4, besides PMEL and CRES, several proteins known to form amyloid were found in the AM. These included cystatin C, superoxide dismutase, and lysozyme 1 which typically are associated with amyloids that cause disease. Because CRES was identified by a single peptide match (full spectral data in Supplemental Data 11) we carried out Western blot analysis and confirmed its presence in AM isolated from both caput and cauda spermatozoa (Supplemental Data 12). by guest on May 8, 2020 We next expanded our analysis to examine the 1026 AM proteome for proteins related to the amyloidogenic proteins identified in Table 4 to determine whether additional putative amyloid forming proteins were present in the AM. We then used Waltz software to determine whether these related proteins contained domains predicted to form amyloid. Within the type 2 cystatin family, in addition to cystatin C and CRES, several other cystatins were detected in the sperm AM proteome. These included three cystatin proteins, CstR1, CstR2, and CstR1L, which are not well-characterized but which reside on the cystatin locus with the other cystatin family 2 members, as well as two other CRES subgroup family members including Cst13 (cystatin T) and Cst11 (CRES2) ( Figure 5A). The full spectral data for the identification of the cystatin peptides are shown in Supplemental Data 11. All cystatins contained several putative amyloidforming domains with most including one within their signal peptides as well as 2-4 other domains in the mature proteins, several of which were conserved between family members. Six members of the lysozyme family were also identified within the AM proteome including Lyz1, Lyzl1, Lyzl4, Lyz5, and SPACA3 and SPACA5 (sperm acrosome associated 3,5). Except for Lyz1 and Lyzl4, the lysozyme family members also had putative amyloid-forming domains in their signal sequences and in the mature proteins. Similar to the cystatins, the amyloid domains in the lysozyme family members were fairly well conserved between members ( Figure 5B). by guest on May 8, 2020

Isolation of AM from caput and cauda epididymal spermatozoa
The studies presented herein provide a protocol for the isolation of AM from mouse caput and cauda epididymal spermatozoa. Because the caput and cauda spermatozoa represent spermatozoa in different functional states, the isolation protocol had to be modified to successfully isolate an "intact" AM from the two different sperm populations. Our studies show that AM are easier to extract from immature caput spermatozoa than mature cauda spermatozoa. This initially was somewhat surprising given that cauda spermatozoa are the mature cells and known to be highly disulfide crosslinked which provides important structural stability necessary for its functions of progressive motility and the ability to fertilize. However, modifications of the sperm membrane are also known to occur during epididymal transit and as such it appears that this involves an increase in fluidity of the cauda sperm membrane perhaps as a critical step in preparation for fertilization (61). Therefore, extremely low concentrations and short incubation times in Triton X-100 were required to demembranate cauda spermatozoa compared to that used for caput spermatozoa.
A second noticeable difference between the two sperm populations was that the cauda sperm acrosome appeared to contain more protease activity at pH 7.4 compared to caput sperm acrosomes and thus additional protease inhibitors were needed during the AM isolation to prevent dispersion of the structure. The combination of appropriate Triton X-100 exposure, protease inhibitors, and a brief vortexing step allowed us to isolate AM from mouse spermatozoa. Using antibodies against known AM markers in immunofluorescence and Western blot analysis as well as LC-MS/MS identification of these markers in the AM preparations demonstrated our successful isolation of the AM from the caput and cauda spermatozoa.
by guest on May 8, 2020 Proteomic characterization of the AM Using nano-LC-MS/MS, we carried out a proteomic analysis of the AM isolated from the caput and cauda spermatozoa and generated an AM proteome composed of 1026 proteins, 501 of which have not been previously identified using a proteomic approach in mouse spermatozoa.
These proteins may represent lower abundant proteins not previously detected in proteomic studies of whole spermatozoa or proteins that are specific to AM functions and thus detectable only in a purified AM fraction. To identify the putative functions of these proteins, the 501 proteins were examined using the Panther database and approximately 54% of these proteins were present in the database and classified according to putative function. The absence of 46% of the sperm AM proteome from this analysis may be because proteins associated with the AM are rapidly diverging as a part of the process of adaptive evolution (62) and thus were not present in the Panther database generated during a study of evolutionarily-related proteins (37).
However, an overview of the proteins with assigned functions indicate the AM is a dynamic structure with proteins not only involved in zona/egg interactions but also those involved in quality control including protein folding and turnover, signaling, cell trafficking, proteases/hydrolyases, and those associated with the cytoskeleton. These proteins may carry out their functions either during spermatogenesis when the AM is formed, during AM maturation in the epididymis, and/or during the fertilization process.
The sperm AM proteome also contained what appeared to be some mitochondrial and sperm axonemal proteins. While it is possible that these proteins may also be present within the AM, we cannot rule contributions from the small percentage of Triton-extracted spermatozoa that remained in the AM preparations. Although protein was not detected by Coomassie staining in the gel lane containing a representative number of spermatozoa that were present in the isolated AM sample, because mitochondrial and axonemal proteins are highly abundant in a single spermatozoon, it is possible that peptides would be detected by MS/MS analysis. by guest on May 8, 2020 A number of proteins detected in the sperm AM proteome also contained transmembrane domains consistent with their association with the membrane, likely either the inner or outer acrosomal membrane. Although spermatozoa were treated with Triton-X-100 to remove membranes as part of the AM isolation procedure, studies suggest that some matrix proteins may interact with proteins present in the inner or outer acrosomal membrane. For example, zonadhesin, is initially synthesized as an acrosomal membrane protein that becomes part of the AM during sperm maturation and CD46, a membrane protein, may affect dispersion of the AM proteins during the acrosome reaction (63)(64)(65). It may be that an intimate association and crosstalk exists between the AM and its associated acrosomal membranes and the detection of these transmembrane domain containing proteins in the AM proteome may represent stable interactions/complexes that were maintained during the AM isolation procedure.
Alternatively, it may be that several proteins, like zonadhesin, start out in the membrane but later become associated with the AM.

AM maturation in the epididymis
In addition to a global overview of proteins in the sperm AM, studies were also carried out to determine whether there were differences in AM protein composition between caput and cauda epididymal spermatozoa. Region-dependent differences in AM proteins were detected suggesting that, similar to other sperm domains, the sperm AM undergoes maturational changes during epididymal transit. While it is possible that the inability to detect some proteins in the caput AM could be due to the fact that fewer caput AM were isolated and analyzed compared to that from cauda spermatozoa, this would not explain the inability to detect some AM proteins in the cauda that were present in the caput suggesting that some AM proteins may be lost or modified during epididymal transit. It is possible that during epididymal transit, some components of the AM are becoming more stabilized as part of the maturation process and thus are more difficult to solubilize during SDS-PAGE preventing them from being analyzed by LC-by guest on May 8, 2020 MS/MS since they do not enter into the electrophoretic gel. In addition to the potential loss of proteins from the AM during maturation, we observed that several epididymal secretory proteins, that are not expressed in the testis, were detected in the sperm AM. This suggests that as part of the sperm maturation process, some proteins in the epididymal lumen become associated with the sperm AM.
GO analysis of the proteins found only in the caput or cauda AM revealed functional groups of proteins that were specific to each sperm population. While the cauda AM contained a large list of proteins categorized under the GO classes of biological process, cellular compartment, and molecular function, the predominant proteins in each class represented those involved in transport, membranes, and hydrolase activity, respectively. These groups of proteins may reflect a mature sperm AM that has established signaling complexes and high enzymatic activity required for downstream fertilization events. In contrast, caput AM contained proteins that were involved in cell trafficking and transport, associations with the cytoskeleton including scaffolding proteins and control of protein folding. These groups of proteins are more consistent with a sperm AM that is immature and undergoing critical organization and establishment of its infrastructure.

AM: a structural scaffold
In addition to the global overview of the mouse sperm AM proteome and examination of maturational changes in the AM during epididymal transit, we utilized our AM proteome data to begin to address the possibility that the AM consists of a highly stable core scaffold structure that may form as the result of the self-aggregation of specific proteins. Indeed, previous investigators have suggested that the matrix structure may form as a result of protein aggregation (45, 65). Intriguingly, melanosomes have been shown to possess a matrix structure that serves as a functional scaffold for the synthesis of melanin. This scaffold structure was shown to contain self-aggregates of Pmel protein in an amyloid structure by guest on May 8, 2020 providing the first evidence in mammals of an amyloid carrying out a biological function (57).
Our studies examining the relationship of the sperm AM to LRO revealed the highest overlap of the sperm AM with proteins present in melanosomes suggesting that the sperm AM may be structurally and functionally similar to this LRO and may in fact represent a novel LRO as proposed by (55). Further investigation of our AM proteome showed the presence of several amyloidogenic proteins including the cystatin CRES, which we have previously shown forms amyloid in vitro and in vivo within the epididymal lumen (59, 60), as well as the related CRES subgroup members CRES2 and cystatin T (66,67), cystatin C, PMEL, SOD, and several lysozyme family members. Thus is may be that, similar to melanosomes, the matrix is composed of self-aggregated proteins organized into a high stable amyloidogenic structure that forms the core scaffold with which a number of proteins are associated to form a functional AM.
Studies are currently ongoing in the laboratory to address this. Amyloidogenesis may also play a role in sorting proteins to the acrosome, or in an even broader sense, to secretory granules, since the signal sequences in the majority of the amyloidogenic proteins detected in the AM were predicted to have the propensity to form amyloid.  Panther classes containing at least two proteins are shown. "synaptic transmission". Of these, 82% (9 proteins) were present in the caput AM and thus this GO term was classified as being a caput AM-specific term. D) Example of a GO term for each GO class that is specific to the caput AM. The proteins classified with the GO terms synaptic transmission, melanosome, and structural constituent of cytoskeleton are shown. E) Example of a GO term for each GO class that is specific to the cauda sperm AM. The proteins classified with the GO terms spermatogenesis, proteasome complex, and serine-type peptidase activity are shown. Rik, 1700019N12Rik; Rikb, 1810009J06Rik.