RESEARCH ARTICLE

ICA1L forms BAR-domain complexes with PICK1 and is crucial foracrosome formation in spermiogenesisJing He, Mengying Xia, Wai Hung Tsang, King Lau Chow and Jun Xia*

ABSTRACTMutations in the Pick1 gene cause globozoospermia, a male infertilitydisorder, in both mice and humans. PICK1 is crucial for vesicletrafficking, and its deficiency in sperm cells leads to abnormal vesicletrafficking from the Golgi to the acrosome. This eventually disruptsacrosome formation and leads to male infertility. Here, we identifiedICA1L, which has sequence similarities to ICA69 (also known asICA1), as a new BAR-domain binding partner of PICK1. ICA1L isexpressed in testes and brain, and is the major binding partnerfor PICK1 in testes. ICA1L and PICK1 are highly expressedin spermatids and trafficked together at different stages ofspermiogenesis. ICA1L-knockout mice were generated by CRISPR-Cas technology. PICK1 expression was reduced by 80% in the testesof male mice lacking ICA1L. Sperm from ICA1L-knockout mice hadabnormalities in the acrosome, nucleus and mitochondrial sheathformation. Both total and mobile sperm numbers were reduced,and about half of the remaining sperm had the characteristics ofglobozoospermia. These defects ultimately resulted in reducedfertility of male ICA1L-knockout mice, and ICA69/ICA1L-doubleknockout male mice were sterile.

KEY WORDS: Vesicle traffic, Acrosome, Globozoospermia,Infertility, ICA1L, PICK1

INTRODUCTIONInfertility is a health problem affecting 10–15% of couplesworldwide (Jarow et al., 1989; Dunson et al., 2004). Around halfof these cases are caused by male infertility (Jarow et al., 2002;Turek, 2005; Ferlin et al., 2006). Globozoospermia is a rare andsevere sperm morphology disorder characterized by round-headed,acrosomeless spermatozoa (Sen et al., 1971). The acrosome locatesat the head of mammalian sperm and contains various hydrolyticenzymes. The acrosome plays a crucial role during fertilization.When the sperm comes into contact with the zona pellucida of anegg, the acrosomal hydrolytic enzymes are released to facilitatesperm penetration into the zona pellucida prior to fusion with theegg (Ikawa et al., 2010). Human patients having a complete formof globozoospermia, characterized by 100% acrosomeless sperm,termed type I globozoospermia or total globozoospermia, areinfertile. There are also a larger group of people whose sperm arepartially acrosomeless, which is termed partial globozoospermia ortype II globozoospermia (Dam et al., 2011). Several genetic modelsphenocopy human globozoospermia, such as mice deficient inCK2α′, Hrb, GOPC, GBA2, ZPBP1 and ZPBP2, Hsp90b1, Vps54,

SPACA1, Dpy19l2, Atg7 and PICK1 (Xu et al., 1999; Kang-Deckeret al., 2001; Yao et al., 2002; Yildiz et al., 2006; Lin et al., 2007;Xiao et al., 2009; Audouard and Christians, 2011; Paiardi et al.,2011; Fujihara et al., 2012; Pierre et al., 2012; Wang et al., 2014).In addition, mutations in SPATA16, PICK1 and DPY19L2 havebeen identified in human globozoospermia patients, with deletionsand point mutations of DPY19L2 being a major cause ofglobozoospermia (Dam et al., 2007; Liu et al., 2010; Harbuzet al., 2011; Koscinski et al., 2011; Coutton et al., 2012; ElInatiet al., 2012; Zhu et al., 2013).

Protein interacting with C-kinase 1 (PICK1) is a peripheralmembrane protein that is important for protein and vesicletrafficking. PICK1 contains a PDZ (PSD-95, Dlg and ZO1)domain and a BAR (Bin, amphiphysin and Rvs) domain. TheBAR domain of PICK1, which forms banana-shaped dimers, iscapable of binding to the curved vesicle membrane (Jin et al., 2006).The PDZ domain of PICK1 interacts with α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor subunitsGluA2, GluA3 and GluA4c (Dev et al., 1999; Xia et al., 1999).PICK1 is involved in long-term depression (LTD) and long-termpotentiation (LTP) by regulating subcellular localization and surfaceexpression of AMPA receptors (Chung et al., 2000; Xia et al., 2000;Perez et al., 2001; Steinberg et al., 2006; Terashima et al., 2008).

Islet cell autoantigen 69 kDa (ICA69, also known as ICA1) wasfirst identified as an autoantigen from type 1 diabetes patients(Pietropaolo et al., 1993). ICA69 also contains an N-terminal BARdomain. ICA69 forms heteromeric BAR domain dimers withPICK1 and is the major binding partner for PICK1 in the brain andpancreas (Cao et al., 2007, 2013). ICA69 inhibits synaptic targetingof AMPA receptors by restricting the localization of PICK1 todendritic shafts (Cao et al., 2007; Xu et al., 2014). In addition, theBAR domain complex of PICK1 and ICA69 is associated withinsulin granule trafficking in pancreatic β-cells. Deficiency ofPICK1 leads to abnormal trafficking of insulin granules and glucoseintolerance (Cao et al., 2013; Holst et al., 2013). Moreover, thiscomplex has also been found to promote the budding of growthhormone granules from the Golgi apparatus in the pituitary gland(Holst et al., 2013). ICA69-knockout (KO) mice share similarinsulin trafficking defects and glucose intolerance to PICK1-KOmice (Cao et al., 2013). Interestingly, PICK1 and ICA69 arestabilized by each other in the complex, such that absence of oneleads to the disappearance of the other from pancreatic β-cells (Caoet al., 2013).

Our previous work has demonstrated that mutation of PICK1 inmice also leads to phenotypes resembling globozoospermia, withprimary defects in acrosome formation (Xiao et al., 2009). Theacrosome malformation in PICK1-KO mice was caused by theabnormal trafficking of vesicles from the Golgi to the acrosome.These defects could be rescued by testes-specific re-expression ofPICK1 (Li et al., 2013). In addition, mutation of PICK1 wasidentified in one human globozoospermia patient, providing furtherReceived 24 April 2015; Accepted 21 August 2015

Division of Life Science, Division of Biomedical Engineering and State KeyLaboratory of Molecular Neuroscience, The Hong Kong University of Science andTechnology, Clear Water Bay, Kowloon, Hong Kong, China.

*Author for correspondence ( jxia@ust.hk)

3822

© 2015. Published by The Company of Biologists Ltd | Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

support to the importance of PICK1 in acrosome formation andmale infertility (Liu et al., 2010). Interestingly, despite the ICA69-KO mice having very similar insulin trafficking defects to PICK1-KO mice, the ICA69-KO mice were very different from PICK1-KOmice in terms of acrosome formation and fertility, as no obviousdefects were observed. Here, we report that we have identifiedprotein islet cell autoantigen 1-like (ICA1L), which contains a BARdomain with high similarity to the BAR domain of ICA69, as themajor binding partner for PICK1 in testes. ICA1L and PICK1 werehighly expressed in spermatids and trafficked together at differentstages of spermiogenesis. Both the number of total and mobilesperm were reduced, and about half of the remaining sperm hadcharacteristics of globozoospermia in ICA1L-KO mice. Thesedefects ultimately resulted in reduced fertility of ICA1L-KO malemice, and ICA69/ICA1L-double knockout male mice were sterile.

RESULTSICA1L is abundant in testesICA69 is the major binding partner of PICK1 in brain and pancreas(Cao et al., 2007, 2013). ICA69-KO mice have very similarphenotypes to PICK1-KO mice in terms of insulin trafficking andglucose intolerance (Cao et al., 2013). However, unlike PICK1-KOmice, ICA69-KO mice have no obvious defect in fertility. Usinga BLAST search, we found a new protein ICA1L, which containsa BAR domain with high similarity to that of ICA69 and aunique C-terminal region (Fig. 1A,B). Interestingly, the BARdomain of ICA1L is highly conserved during evolution, implyingthat it has a conserved role in different species (Fig. S1). Therefore,we wondered whether ICA1L has a compensatory or differentrole to that of ICA69. To investigate their relationship, ICA69- andICA1L-specific antibodies, targeting their C-terminal regions,were generated (Fig. 1B–D). The ICA1L antibody specificallyrecognized ICA1L and the ICA69 antibody specifically recognizedICA69 (Fig. 1C). A strip test for the ICA1L antibody demonstratedthat it recognized a single band in testes homogenates (Fig. 1D). Apeptide antibody previously generated against ICA69 recognizedboth proteins (Cao et al., 2007) (Fig. 1C). However, this peptideantibody could also be used for western blotting because themolecular mass of ICA69 is bigger than that of ICA1L (Fig. 1C) andthis antibody could only weakly recognize endogenous ICA1L(data not shown).Using the ICA1L-specific antibody the tissue distribution of

ICA1L was determined. Interestingly, ICA1L was abundant intestes but only weakly expressed in brain (Fig. 1E). ICA1L wasabsent in other tissues. PICK1 was highly expressed in brain andtestes, followed by pancreas. ICA69 was mainly expressed in brainand pancreas but only weakly expressed in testes.

ICA1L interacts with PICK1 in vitro and in vivoICA69 forms a heteromeric BAR domain complex with PICK1(Cao et al., 2007). ICA1L contains a BAR domain with highsimilarity to that of ICA69. Next, we investigated whether ICA1Lcould interact with PICK1. We found that ICA1L strongly bound toPICK1 when co-expressed in a heterologous system (Fig. 2A). Inaddition, when using anti-PICK1 antibodies to immunoprecipitateendogenous PICK1, ICA1L was robustly co-immunoprecipitatedwith PICK1 from both brain and testes homogenates (Fig. 2B,C). Inaddition, anti-ICA1L antibodies were used to immunoprecipitateendogenous ICA1L in testes homogenates, and PICK1 was readilyco-immunoprecipitated with ICA1L (Fig. 2D). These resultssuggest that PICK1 and ICA1L strongly interact with each other.Interestingly, an additional higher band of ICA1L was observed in

brain and when it was overexpressed in HEK293T cells (Fig. 2A,B,indicated by asterisks).

To determine which domain of ICA1L interacted with PICK1,deletion mutants of ICA1L were generated (Fig. 2E). The ICA1LBAR domain strongly bound to PICK1, whereas the C-terminalregion weakly interacted with PICK1 (Fig. 2F). To map whichdomain of PICK1 interacted with ICA1L, deletion mutants ofPICK1 were generated (Fig. 2G). The BAR domain of PICK1bound to ICA1L, whereas the PDZ domain and the C-terminalregion did not (Fig. 2H). In addition, the BAR domain of PICK1interacted with the BAR domain of ICA1L (Fig. 2I). Taken together,these results suggest ICA1L interacts with PICK1mainly in a BAR-domain-dependent manner.

ICA1L is the major binding partner for PICK1 in testesInterestingly, we also found that ICA1L was absent in PICK1-KOmice, similar to ICA69 (Fig. 3A). We did not observe any differencein the presence of ICA69 and ICA1L mRNAs in the brain and testesof PICK1-KO mice, indicating that the disappearance of ICA69and ICA1L in PICK1-KO mice was not due to downregulatedgene transcription (Fig. 3B,C). The facts that ICA1L bound tightlyto PICK1 and ICA1L was highly expressed in testes suggestthat ICA1L could be the major binding partner for PICK1 in testes.To estimate the percentages of ICA1L and PICK1 that wereassociated with each other in testes, we performed quantitative co-immunoprecipitation from testes homogenates. First, anti-PICK1antibodies were used to immunoprecipitate endogenous PICK1 intestes twice consecutively and∼86% of ICA69 and∼85% of ICA1Lco-immunoprecipitated with PICK1 (Fig. 3D,E). Similarly, anti-ICA1L antibodies were used to immunoprecipitate endogenousICA1L in testes and ∼80% of PICK1 co-immunoprecipitatedwith ICA1L (Fig. 3F,G). No ICA69 co-immunoprecipitated withICA1L, suggesting that the three proteins did not form a complex(Fig. 3F,G). It also further confirmed the specificity of the ICA1Lantibody. After immunoprecipitation with anti-ICA1L antibodies,the ICA69 antibodies were used to immunoprecipitate endogenousICA69 in testes and ∼11% of PICK1 co-immunoprecipitated withICA69 (Fig. 3F,G). As control, GAPDH was immunoblotted and itwas not significantly changed by immunoprecipitation with anti-PICK1, -ICA69 or -ICA1L antibodies, further supporting thespecificities of these antibodies (Fig. 3D–G). Taken together, theabove results demonstrate that ICA1L is abundant and the majorbinding partner for PICK1 in testes.

ICA1L regulates PICK1-mediated AMPA receptor clusteringin a similarmanner to ICA69and is upregulated in the brain ofICA69-KO micePICK1 regulates subcellular localization and surface expression ofAMPA receptors and is consequently involved in LTD and LTP(Chung et al., 2000; Xia et al., 2000; Perez et al., 2001; Steinberget al., 2006; Terashima et al., 2008). ICA69 regulates AMPAreceptor trafficking by disrupting PICK1 homodimer formation(Cao et al., 2007). The fact that both ICA69 and ICA1L areexpressed in brain made us wonder whether ICA1L has acompensatory role for ICA69 (Figs 1E and 3A). GluA2 wasdiffuse in the cytoplasm when expressed in HEK293T cells, butupon co-expression of PICK1 GluA2 became clustered inHEK293T cells (Fig. 4A). Additional expression of ICA1Ldisrupted the PICK1-mediated GluA2 clustering (Fig. 4A),similar to what has been observed for ICA69 (Cao et al., 2007).PICK1 formed homodimers, which is important for the synaptictargeting of AMPA receptors (Fig. 4B). Both ICA1L and ICA69

3823

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

could compete for the PICK1 homodimer formation (Fig. 4B,C). Inaddition, increasing the relative ratio of ICA1L or ICA69 to PICK1further reduced PICK1 homodimer formation (Fig. 4D–G). ICA1L,ICA69 and PICK1 were expressed in similar regions of the brain

(Fig. 4H). More importantly, ICA1L expression was upregulated inthe brains of ICA69-KO mice (see Fig. 7G,J). Taken together, theseresults suggest ICA1L could have a compensatory role for ICA69 inbrain.

Fig. 1. ICA1L is abundant in testes. (A) Schematic representations of structures of ICA69 and ICA1L. Both ICA69 and ICA1L are composed of an N-terminalBAR domain and a C-terminal domain (ICAC and ICALC, respectively). (B) Similarity between rat ICA69 and rat ICA1L. Alignment was performed by ClustalOmega and presented by Boxshade. The red bar represents BAR domains of both ICA69 and ICA1L, the borders of which (amino acids 21–248 and 15–242,respectively) were determined by SMART. The upper and lower purple box shows the antigens for ICA69 (amino acids 275–459) and ICA1L (amino acids 266–411), respectively. The red box shows the peptide antigen (amino acids 468–480) for generation of ICA69 antibody. (C) Antibody specificity test. HEK293T cellswere transfected with Myc–ICA69 or Myc–ICA1L and cell lysates were used for western blotting. (D) A strip test for anti-ICA1L antibody. Testes sample wasimmunoblotted with following antibodies: lane1, serum, 1:1000; lane 2, flow through, 1:1000; lane 3, anti-ICA1L antibody preincubated with antigen, 1:300; lane 4,anti-ICA1L antibody, 1:300. (E) Different tissues were dissected frommouse and homogenized to obtain the total proteins. Equal amounts of proteins were loadedonto SDS-PAGE gels and analyzed by western blotting.

3824

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

ICA1L traffics together with PICK1 during spermiogenesisICA1L is mainly expressed in testes, so we wanted to focus onstudying the function of ICA1L in testes. PICK1 is mainlyexpressed in spermatids and is involved in acrosome trafficking(Xiao et al., 2009). Given that ICA1L was the major binding partner

for PICK1 in testes, this made us ask whether ICA1L was alsoexpressed in spermatids. Using immunohistochemistry, we foundthat both ICA1L and PICK1 were expressed in spermatids and theycolocalized very well with each other (Fig. 5A). The spermatidsundergo metamorphic and biochemical changes to produce mature

Fig. 2. ICA1L binds to PICK1 in vitro and in vivo. (A) HEK293T cells were transiently transfected with the indicated constructs. Anti-GFP antibodies were usedfor immunoprecipitation (IP). In vivo co-immunoprecipitation from mouse brain homogenates (B) and testes homogenates (C) using anti-PICK1 antibodies.The asterisk marks an additional band of ICA1L (see text). (D) In vivo co-immunoprecipitation from mouse testes homogenates using anti-ICA1L antibodies.(E) Schematic representation of structure of ICA1L and deletion mutants of ICA1L. (F) HEK293T cells were transiently transfected with cDNA as indicated.Anti-GFP antibodies were used for immunoprecipitation. PICK1 strongly bound to the BAR domain of ICA1L. (G) Schematic representation of structure of PICK1and deletion mutants of PICK1. NT, N-terminus; CT, C-terminus. (H) HEK293T cells were transiently transfected with indicated constructs. Anti-GFP antibodieswere used for immunoprecipitation. ICA1L potently bound to the BAR domain of PICK1. (I) HEK293T cells were transiently transfected with indicated constructs.Anti-GFP antibodies were used for immunoprecipitation. The BAR domain of PICK1 interacted with the BAR domain of ICA1L.

3825

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

sperm during spermiogenesis. To assess whether ICA1L and PICK1traffic together during different stages of spermiogenesis (the Golgi,cap, acrosome and maturation phases) dynamic locations of PICK1and ICA1L were characterized. Both PICK1 and ICA1L partiallycolocalized or were close to the acrosome during the Golgi phase

and started to move to the opposite ends of the acrosome during thecap and acrosome phases (Fig. 5B,C). PICK1 and ICA1L thencompletely located at the opposite ends of acrosome during thematuration phase (Fig. 5B,C). These results suggest ICA1L sharessimilar trafficking dynamics to PICK1 during spermiogenesis.

Fig. 3. ICA1L is themajor binding partner for PICK1 in testes. (A) Brains, pancreas and testes were dissected fromWTand PICK1-KOmice and homogenizedto obtain the total proteins. Equal amounts of proteins were loaded to SDS-PAGE gels and analyzed by western blotting. RNAs were extracted from brains(B) and testes (C) of WT and PICK1-KO mice and cDNAs were synthesized by reverse transcription and amplified by PCR. (D) PICK1 was immunoprecipitated(IP) with PICK1 antibodies consecutively twice from mouse testes homogenates. The testes homogenates after the first and second immunoprecipitation weredesignated as AIP1 and AIP2, respectively. (E) Quantification for D; n=4, error bar represents s.e.m. ***P<0.001; #, not significant (Student’s t-test). (F) ICA1L andICA69 were immunoprecipitated with anti-ICA1L and -ICA69 antibodies, respectively, from mouse testes homogenates. The testes homogenates afterimmunoprecipitation with anti-ICA1L and -ICA69 antibodies were designated as AIP1 and AIP2, respectively. (G) Quantification for F; n=4, error bar represents s.e.m. ***P<0.001; #, not significant (Student’s t-test). M1–M4 in D and F represent samples from four different mice.

3826

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

Fig. 4. See next page for legend.

3827

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

Notably, and consistent with western blotting data, the ICA1Lsignal was absent in PICK1-KO mice (Fig. 5C). The acrosome wasfragmented and nucleus formation was impaired in PICK1-KOmice, as previously reported (Xiao et al., 2009) (Fig. 5B,C). Takentogether, these results suggest that ICA1L and PICK1 traffictogether during spermiogenesis.

Generation of ICA1L-KO mice by CRISPR-CasGiven that ICA1L is the major binding partner for PICK1 in testesand they traffic together during spermiogenesis, we asked whetherdeletion of Ica1l would affect spermiogenesis. CRISPR-Cas is afast, efficient and reliable technology that can be used for multiplexgenome editing (Ran et al., 2013; Wang et al., 2013; Yang et al.,2013). ICA1L-KO mice were therefore generated by CRISPR-Castechnology. Exon 2 was targeted and a restriction enzyme site forNspI was chosen for restriction fragment length polymorphism(RFLP) analysis (Fig. 6A). The PCR product containing the targetregion was gel purified, and NspI digestion gave the two predictedbands (Fig. 6B). The mutation rate was about 28% in vitrowhen theconstruct pX330 expressing single-guided RNA (sgRNA) targetingIca1l was expressed in N2a cells as demonstrated by RFLP analysis(Fig. 6C). After synthesizing Cas9 mRNA and sgRNA,microinjection, transferring embryos into foster mothers, in total12 founders from three foster mothers were identified and verifiedby sequencing (Fig. 6D,E; Fig. S2). Two injection parameters werechosen – large volume (‘L’) and relatively small volume (‘S’)(Fig. 6E). Injection of a large volume gave a higher mutation rate,whereas injection of a low volume resulted in a lower mutation rate(Fig. 6E; Table S1). Interestingly, we found that most insertionshappened at sites between the third and fourth base pair (bp) beforethe protospacer adjacent motif (PAM) (Fig. S2). No off-targetingeffects were found in our case (Table S2) and the mosaic ratewas about 50%, similar to the previous report (Yang et al., 2013)(Fig. S2). A founder with a mutant allele of a 23 bp deletion(Founder 2 in Fig. S2) was back-crossed to C57BL/6J for thefollowing experiments. For genotyping, the PCR products for wild-type (WT) and ICA1L-KO mice are 112 bp and 89 bp, respectively,and heterozygous mice contain both fragments (Fig. 6F).

PICK1 is reduced by 80% in the testes of ICA1L-KO miceAfter ICA1L-KO mice were generated, protein expression levelswere investigated in testes, brain and pancreas. Interestingly, wefound that PICK1 was reduced by ∼80% in the testes of ICA1L-KO mice (Fig. 7A,B), which correlates to the earlier finding that80% of PICK1 was associated with ICA1L in testes (Fig. 3F,G).This suggests that PICK1 was degraded without binding to ICA1Lin testes. However, the levels of PICK1 were not significantlychanged in the testes of ICA69-KO mice, implying that ICA69does not play an important role in testes (Fig. 7A,B). The levels ofICA69 were not significantly altered in the testes of ICA1L-KOmice (Fig. 7A,C) and, similarly, the levels of ICA1L were alsonot significantly changed in the testes of ICA69-KO mice(Fig. 7A,D). In addition, we mated ICA69/ICA1L double-knockout (DKO) mice and found that PICK1 was reduced by∼82% in the testes of ICA69/ICA1L DKO mice, which is similarto the 80% reduction of PICK1 found in ICA1L-KO mice, againconfirming that the loss of ICA69 did not contribute much to theloss of PICK1 in testes (Fig. 7E,F).

PICK1 was reduced by ∼73% in the brains of ICA69-KO mice,whereas PICK1 was slightly reduced in the brains of ICA1L-KOmice (Fig. 7G,H). This result suggests that ICA69 is the majorbinding partner for PICK1 in brain whereas ICA1L is a minorbinding partner. ICA69 was not changed in the brains of ICA1L-KOmice (Fig. 7G,I). PICK1 was reduced to almost undetectable levelsin the pancreas of ICA69-KO mice (Fig. 7K,L). ICA1L was absentfrom the pancreas (Fig. 7K). The levels of ICA69 and PICK1 werenot significantly altered in the pancreas of ICA1L-KO mice(Fig. 7K–M).

Defective sperm and fertility in ICA1L-KO and ICA69/ICA1LDKO miceAlthough 80% of PICK1 was lost in the testes of ICA1L-KO mice,ICA1L-KO male mice were still fertile (Fig. 8A). However, thefertility of ICA1L-KO male mice was significantly reduced. Thelitter size of ICA1L-KOmale mice was reduced by∼36% comparedtoWTmice, whereas no significant changewas found in ICA69-KOmale mice (average litter size, mean±s.e.m., male WT, 7.4±0.4,n=8; male ICA69-KO, 6.0±0.5, n=8; male ICA1L-KO, 4.7±0.7,n=16) (Fig. 8A). More importantly, male ICA69/ICA1L DKO micewere sterile (Fig. 8A). To investigate how deficiency of ICA1L ledto reduced fertility and why ICA69/ICA1L DKO male mice weresterile, sperm from the cauda epididymis were examined. The totalsperm number from the cauda epididymis of ICA1L-KO mice wasreduced by ∼46% (Fig. 8B). The numbers of mobile and linearmobile sperm from the cauda epididymis were also reduced by∼54% and ∼60%, respectively (Fig. 8C,D). Although there weretrends of reduction in the total, mobile and linear mobile spermnumbers for ICA69-KO mice, they did not reach significancecompared to WT controls (Fig. 8B–D). By contrast, ICA69/ICA1LDKO mice had severe defects. Total sperm number was reduced by∼98% (Fig. 8B). The mobile sperm number was reduced by ∼99%compared toWTmice, and no linear mobile spermwere observed inICA69/ICA1L DKO mice (Fig. 8C,D). Hematoxylin and Eosin(H&E) staining of the cauda epididymis confirmed that the numbersof mature sperm in ICA1L-KO and ICA69/ICA1L DKO mice werereduced (Fig. 8E). From the H&E staining of seminiferous tubulesof testes, it could be seen that the layers of cells (spermatogonia,spermatocytes and round spermatids) were preserved in ICA69-KO,ICA1L-KO and ICA69/ICA1L DKO mice (Fig. 8F). However,sperm with round nuclei could be detected in ICA1L-KO andICA69/ICA1L DKOmice. This suggests that an abnormal transition

Fig. 4. ICA1L compensates for ICA69 in brain. (A) HEK293T cells weretransfected with the indicated constructs. When PICK1 (green) and GluA2(red) were co-expressed in cells, they formed many co-clusters in the cytosol.Co-expression with ICA1L (blue) disrupts the PICK1–GluA2 clusters as doesICA69 (blue). When PICK1, GluA2, ICA69 and ICA1L were expressed alone,they were all diffusely localized in the cytosol (lower panels). Scale bars:10 μm. (B) HEK293T cells were transfected with the indicated constructs. Anti-GFP antibodies were used for immunoprecipitation (IP). Both ICA69 andICA1L could compete for PICK1 homodimer formation. (C) Quantification for B.The amount of co-immunoprecipitated Myc–PICK1 normalized to totalimmunoprecipitated YFP–PICK1 represents the relative level of PICK1homodimers, n=5, error bar represents s.e.m. (D) HEK293T cells weretransfected with the indicated constructs. Anti-GFP antibodies were used forimmunoprecipitation. Increasing the relative transfection ratio of ICA1L toPICK1 (from 1:1 to 3:1) could further reduce PICK1 homodimer formation.(E) Quantification for D. The amount of co-immunoprecipitated Myc–PICK1normalized to total immunoprecipitated YFP–PICK1 represents the relativelevel of PICK1 homodimers, n=4, error bar represents s.e.m. (F) HEK293Tcells were transfected with the indicated constructs. Anti-GFP antibodies wereused for immunoprecipitation. Increasing the relative transfection ratio ofICA69 to PICK1 (from 1:1 to 3:1) could further reduce PICK1 homodimerformation. (G)Quantification for F. The amount of co-immunoprecipitatedMyc–PICK1 normalized to total immunoprecipitated YFP–PICK1 represents therelative level of PICK1 homodimers, n=5, error bar represents s.e.m. (H) Theindicated brain parts were dissected and homogenized for western blotting.ICA1L, ICA69 and PICK1 were expressed in similar regions of the brain.*P<0.05, **P<0.01, ***P<0.001 (Student’s t-test).

3828

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

from round spermatids to mature sperm during spermiogenesiscould result in defective sperm in ICA1L-KO and ICA69/ICA1LDKO mice.When carefully examining the morphology of the sperm from

the cauda epididymis, we observed that a large number of sperm

from ICA1L-KO and ICA69/ICA1L DKO mice did not have thetypical hook-like morphology, but had abnormal heads resemblingirregularly shaped balls (Fig. 8G). Next, we examined the spermfrom the cauda epididymis by labeling nuclei with DAPI and theacrosomes with sp56 (also known as ZP3R). The sperm from WT

Fig. 5. ICA1L traffics together with PICK1 during spermiogenesis. (A) ICA1L colocalized with PICK1 in spermatids. Rabbit anti-ICA1L (green) and guinea piganti-PICK1 (red) antibodies were used for immunolabeling. Lower panels showenlarged images of box areas in upper panels. Colocalization of ICA1L and PICK1is marked by arrows. Dynamic locations of PICK1 (arrowheads) (B) and ICA1L (arrowheads) (C) during acrosome formation. Guinea pig anti-PICK1 antibodieswere used to label endogenous PICK1 (red). Rabbit anti-ICA1L antibodies were used to label endogenous ICA1L (red). Mouse anti-sp56 antibodies were used tolabel acrosomes (green, arrows). DAPI was used to label the nucleus (blue). Scale bars: 10 μm.

3829

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

and ICA69-KO mice have typical crescent moon shape acrosomesmarked by sp56 whereas the acrosomes of some sperm inICA1L-KO mice have defects including fragmentation,mislocalization and deformation (Fig. 8H). The acrosomes in

ICA69/ICA1L DKOmicewere either fragmented or completely lost(Fig. 8H). Smaller and irregular nuclei could also be observed inICA1L-KO and ICA69/ICA1L DKO mice (Fig. 8H).Immunostaining with Tomm20, which labels mitochondria,

Fig. 6. Generation of ICA1L-KO mice by CRISPR-Cas. (A) Targeting region for Ica1l. The red sequence is the PAM motif. The uppercase letters are the NspIrestriction site. The targeting sequence is underlined. (B) A fragment containing the targeting region was PCR amplified from mouse genomic DNA and gelpurified. The PCRproduct was left untreated or incubatedwith NspI. The fragment (703 bp) could be cut into two bands (308 bp and 395 bp). (C) Testing CRISPR-Cas efficiency in vitro. N2a cells were transfected with empty vector (EV) or pX330 expressing the sgRNA targeting Ica1l. Gel purified PCR products weredigested by NspI. Fragments carrying mutations could not be digested by NspI. (D) Work flow of microinjection. (E) Screening of founders. PCR products fromnewborns were gel purified and digested by NspI. Three littermates from three foster mothers were screened. Embryos in recipient 1 were injected at a relativelyhigher volume, whereas embryos in recipient 2 and 3 were injected at a lower volume. Injection of a higher volume resulted in a higher mutation rate.(F) Genotyping of ICA1L-KO mice. The PCR product for WT mice is 112 bp. The PCR product for KO mice is 89 bp. Heterozygous mice contain both products.

3830

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

Fig. 7. PICK1 is reduced by 80% in the testes of ICA1L-KO mice. The testes (A), brain (G) and pancreas (K) from WT, ICA69-KO and ICA1L-KO mice weredissected and homogenized for western blotting. Quantification of relative expression levels of PICK1 (B), ICA69 (C), ICA1L (D) normalized to GAPDH in A; WT,n=3; ICA69-KO, n=3; ICA1L-KO, n=3, error bar represents s.e.m. The testes (E) from WT and ICA69/ICA1L DKO mice were dissected and homogenized forwestern blotting. Quantification of relative expression levels of PICK1 (F) normalized toGAPDH in E;WT, n=4; ICA69/ICA1LDKO, n=3, error bar represents s.e.m.Quantification of relative expression levels of PICK1 (H), ICA69 (I), ICA1L (J) normalized to GAPDH in G; WT, n=6; ICA69-KO, n=8; ICA1L-KO, n=5, error barrepresents s.e.m. Quantification of relative expression levels of PICK1 (L) and ICA69 (M) normalized to GAPDH in (K), WT, n=3; ICA69-KO, n=3; ICA1L-KO, n=3,error bar represents s.e.m. *P<0.05; **P<0.01; ***P<0.001; #, not significant (Student’s t-test).

3831

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

Fig. 8. See next page for legend.

3832

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

revealed that the mitochondrial sheaths, which are responsible forsperm movement, had a variety of defects in the sperm of ICA1L-KO and ICA69/ICA1L DKO mice, including wrapping around thedeformed nucleus, aggregating near the deformed nucleus andsplitting into separate aggregates (Fig. 8I). Quantification of thenumbers of sperm from the cauda epididymis ofWT and ICA69-KOmice showed that the percentages of abnormal sperm with defects inthe nucleus, acrosome and mitochondrial sheath were only 4% and9% inWT and ICA69-KOmice, respectively (Fig. 8J). However, forICA1L-KO and ICA69/ICA1L DKO mice, the percentages ofabnormal sperm with defects in the nucleus, acrosome andmitochondrial sheath were ∼51% and ∼95% in ICA1L-KO andICA69/ICA1L DKO mice, respectively (Fig. 8J). Theseabnormalities are reminiscent of the defects seen inglobozoospermia, a human infertility disorder characterized byround-headed sperm with deformed nuclei, abnormal acrosomesand malformed mitochondrial sheaths. Taken together, our resultsdemonstrate that deficiency of ICA1L leads to reduced male fertilityin a mouse model with characteristics of type II globozoospermia,whereas deficiency of both ICA1L and ICA69 lead to completeinfertility in male mice.

DISCUSSIONIn this study, we identified ICA1L as a new binding partner ofPICK1. ICA1L is homologous to ICA69, another BAR-domain-containing protein that we have previously reported as the majorbinding partner for PICK1 in brain and pancreas (Cao et al., 2007,2013). Our previous results suggested that ICA69 associates with76% of PICK1 in brain by forming a tight BAR-domain complex.Around 73% of PICK1 is lost in the brains of ICA69-KO mice. Inthis study, we found that ICA1L also forms BAR domain complexeswith PICK1. ICA1L interacts with PICK1 in brain and ICA1Ldisrupts PICK1-mediated GluA2 clustering and competes withPICK1 for homodimer formation, similar to ICA69. In addition, theexpression levels of ICA1L is similar to that of ICA69 and PICK1 indifferent brain regions and ICA1L is increased in the brains ofICA69-KO mice. Taken together, ICA1L has a compensatory rolefor ICA69 in the brain.Despite the similarity, the tissue distribution of ICA1L is quite

different from that of ICA69. Expression of ICA69 is high in brain

where ICA1L is low. Expression of ICA69 is low in testes whereICA1L is high. In addition, ICA1L is completely absent in pancreas,where ICA69 expression is high. ICA1L accounts for binding to80% of PICK1 whereas ICA69 accounts for only 11% in testes.PICK1 was reduced by 80% in the testes of ICA1L-KO mice.ICA69 and ICA1L are completely absent in PICK1-KO mice. Wedid not observe any difference in the presence of ICA69 and ICA1LmRNAs in the brain and testes of PICK1-KO mice. One possibleexplanation is that ICA69 and ICA1L form complexes with PICK1and stabilize each other in the complexes. Loss of one protein wouldlead to the degradation of the other partner in the complexes.

In western blotting, although the ICA69 band of the expectedfull-length size was gone in ICA69-KO mice, a smaller band wasdetected by our ICA69 antibodies in the testes of ICA69-KO micebut not in the brains and pancreas. The antibodies we used wereC-terminal antibodies. The smaller size of this additional bandsuggests that it could be a truncated form or abnormal spliced formof ICA69. The ICA69-KO mice used here were generated bydeleting exon 2 of Ica1 gene (Winer et al., 2002). It is possible thatabnormal splicing isoform could be generated in ICA69-KO miceby joining the exons before and after exon 2. Alternatively, atruncated form of ICA69 could be produced by using alternativestart codon downstream of the original start codon. The fact that thisadditional band was only found in testes, but not in brain or pancreassuggests that this abnormal splicing or alternative start codon wasonly utilized in testes. This truncated protein, however, could notcompensate for the loss of ICA69 in the double knockout mice,suggesting that it is not functional.

Although PICK1 was reduced by 80% in the testes of ICA1L-KO mice, male mice deficient in ICA1L were still fertile. Themice became sterile only after both ICA1L and ICA69 wereknocked out. This suggests that there might be two independentpathways or types of vesicles that contribute to the biogenesis ofacrosomes. The major one is mediated by ICA1L and the minorone depends on ICA69. Despite the minor role of ICA69 in testes,it is still able to form heteromeric complex with PICK1 andcompensate for the loss of ICA1L to a certain degree in acrosomebiogenesis. In the absence of both ICA69 and ICA1L, PICK1could only form homodimers, which is not sufficient to supportthe normal vesicle trafficking needed for acrosome formation. As aresult, acrosome formation was severely impaired in ICA69/ICA1L DKO mice.

Mice deficient in PICK1 are sterile and have a globozoospermia-like phenotype (Xiao et al., 2009). In addition, mutation of PICK1causes globozoospermia in humans, emphasizing the importance ofPICK1 in acrosomal trafficking (Liu et al., 2010). The mutation ofPICK1 (G393R), which causes globozoospermia in human, is in theC-terminal acidic region of PICK1. PICK1 G393R does not affectthe interaction with ICA1L (our unpublished data). The acidicregion of PICK1 negatively regulates the lipid-binding ability of theBAR domain of PICK1, which is also important for proteintrafficking (Jin et al., 2006). Hence, it is possible that the lipid-binding ability of the BAR domain is affected by the mutation of theacidic region, which would influence normal acrosome trafficking(Liu et al., 2010).

There are two different types of globozoospermia in human.Type I globozoospermia (total globozoospermia) is characterizedby 100% acrosomeless sperm. Type II globozoospermia (partialglobozoospermia) presents an increased proportion of round-headed sperm and acrosome malformations compared tonormozoospermia (Dam et al., 2011). The sperm in ICA1L-KOmice are not completely acrosomeless. Some of the sperm still have

Fig. 8. Sperm in ICA1L-KO mice have globozoospermia-like phenotypes.(A) Average pup numbers from WT, ICA69-KO, ICA1L-KO and ICA69/ICA1LDKOmale mice mated with female mice. WT, n=8; ICA69-KO, n=8; ICA1L-KO,n=16; ICA69/ICA1L DKO, n=6, error bar represents s.e.m. Total spermnumbers (B), mobile sperm numbers (C) and linear mobile sperm numbers (D)of WT, ICA69-KO, ICA1L-KO and ICA69/ICA1L DKO mice. WT, n=8; ICA69-KO, n=11; ICA1L-KO, n=5; ICA69/ICA1LDKO, n=3, error bar represents s.e.m.(E) H&E staining of the cauda epididymis of WT, ICA69-KO, ICA1L-KO andICA69/ICA1L DKO mice. The numbers of mature sperm in ICA1L-KO andICA69/ICA1L DKO mice were reduced. Insets, enlarged views of the heads ofthe sperm. (F) H&E staining of the testes of WT, ICA69-KO, ICA1L-KO andICA69/ICA1L DKO mice. The morphology of spermatogonia, spermatocytes,round spermatids and sperm is labeled by white arrows, white arrowheads,black arrows and black arrowheads, respectively. The lower panel showshigher magnification views of the boxed regions in the upper panels. (G) Brightfield images of sperm from the cauda epididymis of WT, ICA69-KO, ICA1L-KOand ICA69/ICA1L DKOmice. Sperm from the cauda epididymis of WT, ICA69-KO, ICA1L-KO and ICA69/ICA1L DKO mice were labeled for the acrosomemarker sp56 (green) (H) and themitochondrial marker Tomm20 (red) (I); nuclei(blue) were labeled by DAPI. Scale bars: 5 μm. (J) The percentage ofglobozoospermia-like sperm in WT, ICA69-KO, ICA1L-KO and ICA69/ICA1LDKO mice. WT, n=8; ICA69-KO, n=11; ICA1L-KO, n=4; ICA69/ICA1L DKO,n=3, error bar represents s.e.m. nu, nucleus; ms, mitochondrial sheath.*P<0.05; **P<0.01; ***P<0.001; #, not significant (Student’s t-test).

3833

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

some acrosomes and mitochondria in their heads, characteristics oftype II globozoospermia, suggesting ICA1L could be involved inthe pathogenesis of type II globozoospermia. It would be interestingto search for potential mutations of ICA1L in human patients withtype II globozoospermia.In some animal models of globozoospermia, such as mice

deficient in GOPC, or ZPBP1 and ZPBP2, the sperm concentrationsare normal (Yao et al., 2002; Lin et al., 2007). In other animalmodels, such as mice deficient in CK2α′, Hrb, GBA2, Vps54,Dpy19l2 or Atg7, sperm concentrations are reduced (Xu et al.,1999; Kang-Decker et al., 2001; Yildiz et al., 2006; Paiardi et al.,2011; Pierre et al., 2012;Wang et al., 2014). Interestingly, the spermconcentrations of total globozoospermia patients are normal, but thesperm concentrations of partial globozoospermia patients arereduced (Dam et al., 2011). Sperm numbers are reduced inICA1L-KO and ICA69/ICA1L DKO mice as well as in PICK1-KOmice. PICK1, ICA69 and ICA1L are expressed in hypothalamus(Fig. 4H). This raises the possibility that the reduction in spermconcentration could be a consequence of impaired hypothalamic–pituitary–gonadal axis function. However, the defects inspermiogenesis are the primary effect of PICK1 deficiency intestes and are not caused by altered hypothalamic–pituitary–gonadal axis function because re-expression of PICK1 bylentiviral infection into the seminiferous tubules can rescue theabnormal spermiogenesis and male infertility of PICK1-KO mice(Li et al., 2013). The functions of ICA1L, ICA69 and PICK1depend on each other, so it is unlikely that defective spermiogenesisis caused by impaired neuroendocrine function in ICA1L-KO orICA69/ICA1L DKO mice.

MATERIALS AND METHODSAntibodiesICA1L-specific antibody was generated against amino acid residues 266–411 of rat ICA1L. ICA69-specific antibody used in immunoprecipitationwas generated against amino acid residues 275–459 of rat ICA69. Apeptide antibody against ICA69 which could recognize both ICA69 andICA1L was previously generated against amino acid residues 468–480(IGKTDKEHELLNA) of rat ICA69 (Cao et al., 2007). The guinea pig anti-PICK1 antibody used in immunostaining was previously generated againstthe C-terminal 100 amino acid residues of mouse PICK1 (Cao et al., 2007).Rabbit anti-PICK1 antibody (PC100) used in immunoblotting waspreviously generated against the C-terminal 100 amino acid residues ofmouse PICK1 (Cao et al., 2007). Rabbit anti-PICK1 antibody previouslygenerated against the N-terminal 29 amino acid residues of mouse PICK1was used in immunoprecipitation (Cao et al., 2007). Rabbit anti-GFPantibody was generated using GFP fusion protein as the antigen. Primaryantibodies used include the following: mouse anti-PICK1 antibody (antigenfrom amino acid residues 10–130) was purchased from NeuroMab. Mouseanti-Myc antibody was purchased from the Developmental StudiesHybridoma Bank (9E10). Mouse anti-GAPDH antibody was purchasedfrom Abcam. Mouse anti-GluA2 antibody was purchased from Millipore.Mouse anti-sp56 antibody was purchased from QED Bioscience. Rabbitanti-Tomm20 antibody was purchased from Santa Cruz Biotechnology.

cDNA cloningRat PICK1, Rat ICA69 and mouse ICA1L cDNA were subcloned intocorresponding expression vectors in frames by restriction enzyme SalI/NotI.All constructs were verified by sequencing.

Preparation of tissue samples for western blottingTissues were dissected and homogenized using homogenate buffer [10 mMTris-HCl and 320 mM sucrose supplemented with protease inhibitorcocktail tablet (Roche), pH 7.4] to obtain the total proteins. Homogenatewas centrifuged at 700 g for 10 min at 4°C and the supernatant was retained.

Protein concentration was determined by a Bradford protein assay (Pierce,Rockford, IL). Equal amounts of proteins were loaded to SDS-PAGE gelsand analyzed by western blotting.

Genotyping PCR and RT-PCRPICK1 mice were genotyped by PCR using the following three primers: 5′-TCACTTGCCAGAGGAGAAAACTG-3′, 5′-AAAAATAGGCGTATCA-CGAGGC-3′ and 5′-CACTCGCAGCTTGTTCTGATCTG-3′. The WTPCR product is a 400-bp band whereas the mutant band is a 200-bp band.The brains and testes fromWT and PICK1-KO mice were used for RT-PCRanalysis. Total RNAs were extracted with TRIzol reagent (Invitrogen)according to manufacturer’s instructions, and cDNAs were synthesized byreverse transcription using a first-strand cDNA Synthesis kit (Fermentas).PCR was performed with Q5 High-Fidelity DNA Polymerase (NewEngland Biolabs). The primers used for PCR amplification of ICA69 were:forward 5′-AAGGATGACCTCTTGCTGTTGAATG-3′ and reverse 5′-A-TAGCGATAGAAACAGGGCCTTGAC-3′. The primers used for PCRamplification of ICA1L were: forward 5′-GCATCCGATGCAGAACTG-GACGCTAAGTTGGA-3′ and reverse 5′-TCCATCATTTCGCCAGCTT-GAGTCGAATCCCGCT-3′. The primers used for PCR amplification ofPICK1 were: forward 5′-GTCACCCTACAGAAGGATGCCCAGAACC-TGATTG-3′ and reverse 5′-GTCCGCCTGCAGCTTGTTGTAATGGAT-GGTC-3′. The primers used for PCR amplification of β-actin were: forward5′-TGAGAGGGAAATCGTGCGTG-3′ and reverse 5′-TGCTTGCTGAT-CCACATCTGC-3′.

HEK293T cell culture and transfectionHuman embryonic kidney 293T cells (HEK 293T) were cultured in ahumidified atmosphere containing 5% CO2 in MEM (Gibco) supplementedwith 10% fetal bovine serum, 1 mM sodium pyruvate, 100 U/ml penicillin,and 100 µg/ml streptomycin and passaged every 2–3 days when the cellconfluence reached 80–90%.Calcium phosphate co-precipitationmethodwasused for transient transfection andmediumwas changed 9 h after transfection.

ImmunocytochemistryCells were fixed 48 h after transfection by 4% paraformaldehyde plus 4%sucrose in PBS for 20 min at room temperature. The cells were thenpermeabilized by 0.2% Triton X-100 in PBS for 10 min at roomtemperature. After blocking with 10% normal donkey serum (NDS) inPBS for 1 h, the cells were incubated with primary antibody in 3% NDS for1 h at room temperature, followed by a 1-h incubation with fluorescent-dye-conjugated secondary antibodies (Jackson Immuno). After washing withPBS three times at 10-min interval, the coverslips were mounted withPermafluor (Immunon).

ImmunohistochemistryTestes were fixed by cardiac perfusion through the left ventricle with4% paraformaldehyde and 4% sucrose in PBS followed by post fixationwith 4% paraformaldehyde and 4% sucrose in PBS for 4 h at 4°C. Cryo-protection was performed by incubating testes in gradients of sucrose in PBSsolution (10% sucrose, 20% sucrose and 30% sucrose in PBS) at 4°C.Cryosections of 5 µm thickness were used for immunohistochemistry. Allanimal procedures were approved by the Animal Ethic Committee of theHong Kong University of Science and Technology.

CRISPR-Cas-mediated gene targetingICA1L-KO mice were generated as previously reported (Wang et al., 2013).Briefly, a DNA template for sgRNAwas cloned into pX330 by BbsI (NewEngland Biolabs). The sgRNA oligonucleotides used for generating sgRNAexpression vector are: forward 5′-CACCGGGAAAGAAGGAGGACGA-GCATG-3′ and reverse 5′-AAACCATGCTCGTCCTCCTTCTTTCCC-3′.Empty vector and the construct expressing the sgRNAwere transfected intoN2a cells respectively and RFLP analysis was used for the efficiency testing.Primers used for PCR amplification of fragment containing the targetingregion are: forward 5′-AGCAGGCTGAGCAAGCCACTAAGCAGTAC-TTCCTATG-3′ and reverse 5′-TAGAATTATTTCTAGCCATAATTATA-AAATCTTAACAAT-3′. The PCR products were digested with NspI

3834

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

(New England Biolabs) and separated on a GelRed-stained agarose gel(1%). Fragments carrying mutations could not be cut by NspI. Afterefficiency testing in vitro, T7 promoter was added to Cas9 template by PCRamplification using the following primers: forward 5′-TAATACGACTC-ACTATAGGGAGAATGGACTATAAGGACCACGAC-3′ and reverse 5′-GCGAGCTCTAGGAATTCTTAC-3′. The PCR product was gel purifiedand used as the template for in vitro transcription using the mMESSAGEmMACHINE T7 Ultra kit (Life Technologies). T7 promoter was added intosgRNA template by PCR amplification using the following primers:forward 5′-TTAATACGACTCACTATAGGGAAAGAAGGAGGACGA-GCATG-3′ and reverse 5′-AAAAGCACCGACTCGGTGCC-3′. The PCRproduct was gel purified and used as the template for in vitro transcriptionusing MEGAshortscript T7 kit (Life Technologies). Both the Cas9 mRNAand the sgRNA were purified using MEGAclear kit (Life Technologies).Cas9 mRNA (100 ng/μl) and sgRNA (20 ng/μl) were microinjected into thecytoplasm of B6CBAF2 zygotes in M2 medium (Millipore). Thereafter, 20embryos were transferred into the oviducts of each pseudopregnantB6CBAF1 female. In total 60 embryos were transferred into three fostermothers.

Off-targeting analysisPotential off-targets were predicted by searching themouse genome (mm10)for matches to the 23-nucleotide sgRNA sequence containing up to threemismatches followed by the NGG PAM motif. Fragments containing thepotential off-targeting sites were PCR amplified from genomic DNA offounders and cloned into pEGFP-C3 vector and sequenced. More than 12clones were sequenced for each site. Primers used for cloning are listed inTable S3.

Epididymal sperm count and morphology classificationThe number of epididymal sperm was determined as previously described(Cooper and Castilla, 2009). Briefly, the cauda epididymis was dissectedand incised. Sperm were exuded into 5 ml sperm extract medium for 30 minat 37°C under 5% CO2 and counted by using a hemocytometer or spread onslides for morphological observation.

HistologyThe cauda epididymis and testes were fixed by cardiac perfusion through theleft ventricle with 4% paraformaldehyde and 4% sucrose in PBS followedby post fixation with 4% paraformaldehyde and 4% sucrose in PBS forovernight at 4°C. The cauda epididymis and testes were dehydrated througha graded ethanol series, and then embedded in paraffin. Sections (5 μm)were cut on a microtome (Shandon Finesse; Thermo Fisher Scientific).H&E staining was performed to stain the sections.

Quantitative co-immunoprecipitationTestes fromWT mice were homogenized using homogenate buffer [10 mMTris-HCl and 320 mM sucrose supplemented with protease inhibitorcocktail tablet (Roche), pH 7.4]. Homogenates were centrifuged at 700 gfor 10 min at 4°C, and the supernatant was kept. 2%Triton X-100 was addedto the supernatant to solubilize proteins for 2 h at 4°C. The solution wascentrifuged at the maximum speed (16,873 g) for 20 min. The supernatantwas kept, and protein concentrations were determined by performing aBradford protein assay (Pierce, Rockford, IL). The final concentration wasadjusted to 1 mg/ml. To 0.5 ml of the testes samples, immunoprecipitationwas performed by using protein A beads (GE Healthcare) previouslyincubated with antibodies. The testes homogenates beforeimmunoprecipitation was designated as ‘input’. The testes homogenatesafter the first and second immunoprecipitation were designated as AIP1 andAIP2, respectively. The immunoprecipitation and co-immunoprecipitationefficiencies were calculated by determining the amount of proteins as aAIP:input ratio by densitometry analysis.

Confocal microscopyFixed cells or testes slices were imaged using inverted microscopesLSM510 or LSM710Meta (Carl Zeiss, Inc.) with a 63×1.4 NA oil DIC PlanApo objective. Images were acquired by using the software LSM or ZEN2009. Images were processed by Adobe Photoshop to adjust intensity and

contrast, to select a region of interest and to overlay images. All images weretaken in grayscale and artificially colored for presentation.

Statistical analysesStatistical analyses of group data were performed using unpaired, two-tailedStudent’s t-tests. Asterisks indicate a significant difference: *P<0.05,**P<0.01, ***P<0.001.

AcknowledgementsWe thank Dr Chung Nga Tam (Transgenic Service, HKUST) for assistant inmicroinjection of Cas9 mRNA and sgRNA and Mr Ho-chun Lai for drawing themouse cartoon. Special thanks are given to Miss Shui Wa Yun for assistance intaking care of mice and genotyping.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsJ.X. conceived, directed this project and wrote the manuscript. M.X. purified theICA1L-specific antibody. W.H.T. performed the microinjection of Cas9 mRNA andsgRNA into one-cell embryos. K.L.C. revised this manuscript. J.H. conducted theremaining experiments and wrote the manuscript.

FundingThis work was supported in part by the Research Grants Council of the Hong KongSAR, China [grant numbers 16102914, 663613, HKUST10/CRF/12R, C4011-14R,T13-607/12R and AoE/M-05/12].

Supplementary informationSupplementary information available online athttp://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.173534/-/DC1

ReferencesAudouard, C. and Christians, E. (2011). Hsp90b1 knockout targeted to male

germline: a mouse model for globozoospermia. Fertil. Steril. 95, 1475-U1326.e4.Cao,M., Xu, J., Shen, C., Kam, C., Huganir, R. L. and Xia, J. (2007). PICK1-ICA69

heteromeric BAR domain complex regulates synaptic targeting and surfaceexpression of AMPA receptors. J. Neurosci. 27, 12945-12956.

Cao, M., Mao, Z., Kam, C., Xiao, N., Cao, X., Shen, C., Cheng, K. K. Y., Xu, A.,Lee, K.-M., Jiang, L. et al. (2013). PICK1 and ICA69 control insulin granuletrafficking and their deficiencies lead to impaired glucose tolerance.PLoSBiol. 11,e1001541.

Chung, H. J., Xia, J., Scannevin, R. H., Zhang, X. Q. and Huganir, R. L. (2000).Phosphorylation of the AMPA receptor subunit GluR2 differentially regulates itsinteraction with PDZ domain-containing proteins. J. Neurosci. 20, 7258-7267.

Cooper, T. G. and Castilla, J. A. (2009). WHO laboratory manual for theexamination and processing of human semen. J. Androl. 30, 9.

Coutton, C., Zouari, R., Abada, F., Ben Khelifa, M., Merdassi, G., Triki, C.,Escalier, D., Hesters, L., Mitchell, V., Levy, R. et al. (2012). MLPA andsequence analysis of DPY19L2 reveals point mutations causingglobozoospermia. Hum. Reprod. 27, 2549-2558.

Dam, A. H. D. M., Koscinski, I., Kremer, J. A. M., Moutou, C., Jaeger, A.-S.,Oudakker, A. R., Tournaye, H., Charlet, N., Lagier-Tourenne, C.,van Bokhoven, H. et al. (2007). Homozygous mutation in SPATA16 isassociated with male infertility in human globozoospermia. Am. J. Hum. Genet.81, 813-820.

Dam, A. H., Ramos, L., Dijkman, H. B., Woestenenk, R., Robben, H., van denHoven, L. and Kremer, J. A. (2011). Morphology of partial globozoospermia.J. Androl. 32, 199-206.

Dev, K. K., Nishimune, A., Henley, J. M. and Nakanishi, S. (1999). The proteinkinase C alpha binding protein PICK1 interacts with short but not long formalternative splice variants of AMPA receptor subunits. Neuropharmacology 38,635-644.

Dunson, D. B., Baird, D. D. and Colombo, B. (2004). Increased infertility with agein men and women. Obstet. Gynecol. 103, 51-56.

ElInati, E., Kuentz, P., Redin, C., Jaber, S., VandenMeerschaut, F., Makarian, J.,Koscinski, I., Nasr-Esfahani, M. H., Demirol, A., Gurgan, T. et al. (2012).Globozoospermia is mainly due to DPY19L2 deletion via non-allelic homologousrecombination involving two recombination hotspots. Hum. Mol. Genet. 21,3695-3702.

Ferlin, A., Arredi, B. and Foresta, C. (2006). Genetic causes of male infertility.Reprod. Toxicol. 22, 133-141.

Fujihara, Y., Satouh, Y., Inoue, N., Isotani, A., Ikawa, M. and Okabe, M. (2012).SPACA1-deficient male mice are infertile with abnormally shaped sperm headsreminiscent of globozoospermia. Development 139, 3583-3589.

3835

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience

Harbuz, R., Zouari, R., Pierre, V., Ben Khelifa, M., Kharouf, M., Coutton, C.,Merdassi, G., Abada, F., Escoffier, J., Nikas, Y. et al. (2011). A recurrentdeletion of DPY19L2 causes infertility in man by blocking sperm head elongationand acrosome formation. Am. J. Hum. Genet. 88, 351-361.

Holst, B., Madsen, K. L., Jansen, A. M., Jin, C., Rickhag, M., Lund, V. K., Jensen,M., Bhatia, V., Sorensen, G., Madsen, A. N. et al. (2013). PICK1 deficiencyimpairs secretory vesicle biogenesis and leads to growth retardation anddecreased glucose tolerance. PLoS Biol. 11, e1001542.

Ikawa, M., Inoue, N., Benham, A. M. and Okabe, M. (2010). Fertilization: a sperm’sjourney to and interaction with the oocyte. J. Clin. Invest. 120, 984-994.

Jarow, J. P., Espeland, M. A. and Lipshultz, L. I. (1989). Evaluation of theazoospermic patient. J. Urol. 142, 62-65.

Jarow, J. P., Sharlip, I. D., Belker, A. M., Lipshultz, L. I., Sigman, M., Thomas,A. J., Schlegel, P. N., Howards, S. S., Nehra, A., Damewood,M. D. et al. (2002).Best practice policies for male infertility. J. Urol. 167, 2138-2144.

Jin, W., Ge, W.-P., Xu, J., Cao, M., Peng, L., Yung, W., Liao, D., Duan, S., Zhang,M. and Xia, J. (2006). Lipid binding regulates synaptic targeting of PICK1, AMPAreceptor trafficking, and synaptic plasticity. J. Neurosci. 26, 2380-2390.

Kang-Decker, N., Mantchev, G. T., Juneja, S. C., McNiven, M. A. and vanDeursen, J. M. A. (2001). Lack of acrosome formation in Hrb-deficient mice.Science 294, 1531-1533.

Koscinski, I., ElInati, E., Fossard, C., Redin, C., Muller, J., de la Calle, J. V.,Schmitt, F., Ben Khelifa, M., Ray, P., Kilani, Z. et al. (2011). DPY19L2 deletionas a major cause of globozoospermia. Am. J. Hum. Genet. 88, 344-350.

Li, X. M., Mao, Z., Wu, M. and Xia, J. (2013). Rescuing infertility of Pick1 knockoutmice by generating testis-specific transgenic mice via testicular infection. Sci.Rep.UK 3, 2842.

Lin, Y.-N., Roy, A., Yan, W., Burns, K. H. and Matzuk, M. M. (2007). Loss of zonapellucida binding proteins in the acrosomal matrix disrupts acrosome biogenesisand sperm morphogenesis. Mol. Cell. Biol. 27, 6794-6805.

Liu, G., Shi, Q.-W. and Lu, G.-X. (2010). A newly discovered mutation in PICK1 in ahuman with globozoospermia. Asian J. Androl. 12, 556-560.

Paiardi, C., Pasini, M. E., Gioria, M. and Berruti, G. (2011). Failure of acrosomeformation and globozoospermia in the wobbler mouse, a Vps54 spontaneousrecessive mutant. Spermatogenesis 1, 52-62.

Perez, J. L., Khatri, L., Chang, C., Srivastava, S., Osten, P. and Ziff, E. B. (2001).PICK1 targets activated protein kinase C alpha to AMPA receptor clusters inspines of hippocampal neurons and reduces surface levels of the AMPA-typeglutamate receptor subunit 2. J. Neurosci. 21, 5417-5428.

Pierre, V., Martinez, G., Coutton, C., Delaroche, J., Yassine, S., Novella, C.,Pernet-Gallay, K., Hennebicq, S., Ray, P. F. and Arnoult, C. (2012). Absence ofDpy19l2, a new inner nuclear membrane protein, causes globozoospermia inmice by preventing the anchoring of the acrosome to the nucleus. Development139, 2955-2965.

Pietropaolo, M., Castano, L., Babu, S., Buelow, R., Kuo, Y. L., Martin, S., Martin,A., Powers, A. C., Prochazka, M., Naggert, J. et al. (1993). Islet-cell autoantigen69-Kd (Ica69) - molecular cloning and characterization of a novel diabetes-associated autoantigen. J. Clin. Invest. 92, 359-371.

Ran, F. A., Hsu, P. D., Lin, C.-Y., Gootenberg, J. S., Konermann, S., Trevino,A. E., Scott, D. A., Inoue, A., Matoba, S., Zhang, Y. et al. (2013). Double nicking

by RNA-guided CRISPR Cas9 for enhanced genome editing specificity. Cell 154,1380-1389.

Sen, C. G. S., Holstein, A. F. and Schirren, C. (1971). uber die Morphogeneserundkopfiger Spermatozoen des Menschen. Andrologia 3, 117-125.

Steinberg, J. P., Takamiya, K., Shen, Y., Xia, J., Rubio, M. E., Yu, S., Jin, W.,Thomas, G. M., Linden, D. J. and Huganir, R. L. (2006). Targeted in vivomutations of the AMPA receptor subunit GluR2 and its interacting protein PICK1eliminate cerebellar long-term depression. Neuron 49, 845-860.

Terashima, A., Pelkey, K. A., Rah, J.-C., Suh, Y. H., Roche, K. W., Collingridge,G. L., McBain, C. J. and Isaac, J. T. R. (2008). An essential role for PICK1 inNMDA receptor-dependent bidirectional synaptic plasticity. Neuron 57, 872-882.

Turek, P. J. (2005). Practical approaches to the diagnosis and management of maleinfertility. Nat. Clin. Pract. Urol. 2, 226-238.

Wang, H., Yang, H., Shivalila, C. S., Dawlaty, M. M., Cheng, A. W., Zhang, F. andJaenisch, R. (2013). One-step generation of mice carrying mutations in multiplegenes by CRISPR/Cas-mediated genome engineering. Cell 153, 910-918.

Wang, H., Wan, H., Li, X., Liu, W., Chen, Q., Wang, Y., Yang, L., Tang, H., Zhang,X., Duan, E. et al. (2014). Atg7 is required for acrosome biogenesis duringspermatogenesis in mice. Cell Res. 24, 852-869.

Winer, S., Astsaturov, I., Gaedigk, R., Hammond-McKibben, D., Pilon, M., Song,A., Kubiak, V., Karges, W., Arpaia, E., McKerlie, C. et al. (2002). ICA69(null)nonobese diabetic mice develop diabetes, but resist disease acceleration bycyclophosphamide. J. Immunol. 168, 475-482.

Xia, J., Zhang, X., Staudinger, J. and Huganir, R. L. (1999). Clustering of AMPAreceptors by the synaptic PDZ domain-containing protein PICK1. Neuron 22,179-187.

Xia, J., Chung, H. J., Wihler, C., Huganir, R. L. and Linden, D. J. (2000).Cerebellar long-term depression requires PKC-regulated interactions betweenGluR2/3 and PDZ domain-containing proteins. Neuron 28, 499-510.

Xiao, N., Kam, C., Shen, C., Jin, W., Wang, J., Lee, K. M., Jiang, L. and Xia, J.(2009). PICK1 deficiency causes male infertility in mice by disrupting acrosomeformation. J. Clin. Invest. 119, 802-812.

Xu, X., Toselli, P. A., Russell, L. D. and Seldin, D. C. (1999). Globozoospermia inmice lacking the casein kinase II alpha′ catalytic subunit.Nat. Genet. 23, 118-121.

Xu, J., Kam, C., Luo, J.-H. and Xia, J. (2014). PICK1mediates synaptic recruitmentof AMPA receptors at neurexin-induced postsynaptic sites. J. Neurosci. 34,15415-15424.

Yang, H., Wang, H., Shivalila, C. S., Cheng, A. W., Shi, L. and Jaenisch, R.(2013). One-step generation of mice carrying reporter and conditional alleles byCRISPR/Cas-mediated genome engineering. Cell 154, 1370-1379.

Yao, R., Ito, C., Natsume, Y., Sugitani, Y., Yamanaka, H., Kuretake, S., Yanagida,K., Sato, A., Toshimori, K. and Noda, T. (2002). Lack of acrosome formation inmice lacking a Golgi protein, GOPC. Proc. Natl. Acad. Sci. USA 99, 11211-11216.

Yildiz, Y., Matern, H., Thompson, B., Allegood, J. C., Warren, R. L., Ramirez,D. M. O., Hammer, R. E., Hamra, F. K., Matern, S. and Russell, D. W. (2006).Mutation of beta-glucosidase 2 causes glycolipid storage disease and impairedmale fertility. J. Clin. Invest. 116, 2985-2994.

Zhu, F., Gong, F., Lin, G. and Lu, G. (2013). DPY19L2 gene mutations are a majorcause of globozoospermia: identification of three novel point mutations.Mol. Hum.Reprod. 19, 395-404.

3836

RESEARCH ARTICLE Journal of Cell Science (2015) 128, 3822-3836 doi:10.1242/jcs.173534

Journal

ofCe

llScience