Localization of the Drosophila Rad9 Protein to the Nuclear Membrane Is Regulated by the C-Terminal Region and Is Affected in the Meiotic Checkpoint

Rad9, Rad1, and Hus1 (9-1-1) are part of the DNA integrity checkpoint control system. It was shown previously that the C-terminal end of the human Rad9 protein, which contains a nuclear localization sequence (NLS) nearby, is critical for the nuclear transport of Rad1 and Hus1. In this study, we show that in Drosophila, Hus1 is found in the cytoplasm, Rad1 is found throughout the entire cell and that Rad9 (DmRad9) is a nuclear protein. More specifically, DmRad9 exists in two alternatively spliced forms, DmRad9A and DmRad9B, where DmRad9B is localized at the cell nucleus, and DmRad9A is found on the nuclear membrane both in Drosophila tissues and also when expressed in mammalian cells. Whereas both alternatively spliced forms of DmRad9 contain a common NLS near the C terminus, the 32 C-terminal residues of DmRad9A, specific to this alternative splice form, are required for targeting the protein to the nuclear membrane. We further show that activation of a meiotic checkpoint by a DNA repair gene defect but not defects in the anchoring of meiotic chromosomes to the oocyte nuclear envelope upon ectopic expression of non-phosphorylatable Barrier to Autointegration Factor (BAF) dramatically affects DmRad9A localization. Thus, by studying the localization pattern of DmRad9, our study reveals that the DmRad9A C-terminal region targets the protein to the nuclear membrane, where it might play a role in response to the activation of the meiotic checkpoint.

The crystal structure of the 9-1-1 complex has been determined shows that 9-1-1 proteins share high structural resemblance to the proliferating cell nuclear antigen [PCNA], despite low sequence identity (14%), as was predicted by earlier bioinformatics analysis [14]. A comparison of each 9-1-1 subunit to PCNA revealed that Rad1 shares the highest structural resemblance to each monomer of PCNA. It was also found that the major differences between the two complexes are assigned to the inter-domain connecting (IDC) loop [15][16][17].
Previous studies in human cell lines revealed that human Rad9 (hRad9) contains a nuclear localization signal (NLS) near the Cterminus of the protein and that this NLS is essential for hRad9 localization to the nucleus. Furthermore, co-expression of hRad9 with either hRad1 or hHus1 resulted in the nuclear localization of these otherwise cytoplasmic proteins, indicating the importance of the NLS in nuclear localization of the human 9-1-1 complex [18]. It was also found that human Rad1 (hRad1) but not hRad9 stabilizes the expression of human Hus1 (hHus1) in vitro and acts as a chaperone, stabilizing hHus1 in the cytoplasm [19]. hHus1 was found to be degraded by the ubiquitin-proteasome pathway, with such degradation being suppressed by hRad1 but not by hRad9 [19].
Focusing our initial analysis on the hus1 gene, we have begun to investigate the function of the 9-1-1 complex in Drosophila [20][21]. Mutations in Drosophila hus1 (DmHus1) lead to female sterility, suggesting that DmHus1 plays a role in the meiotic program. DmHus1 mutation suppresses the dorsal-ventral patterning defects caused by mutations in DNA repair enzymes, suggesting a role for hus1 in regulating the meiotic DNA damage checkpoint. We also demonstrated that DmHus1 is required for homologous recombination repair during meiosis [21]. In mitotic cells, we determined that DmHus1-mutant flies are sensitive to hydroxyurea and methyl methanesulfonate but not to X-ray irradiation, suggesting that DmHus1 is required for the activation of an S-phase checkpoint. On the other hand, DmHus1 is not required for the G2-M checkpoint or for post-irradiation induction of apoptosis [20].
In this study, we addressed the localization pattern of the Drosophila 9-1-1 complex and analyzed the importance of the localization pattern of DmRad9A during activation of the meiotic checkpoint.

Results
Drosophila Rad9A protein is localized to the nuclear membrane in Schneider cells (S 2 R+) and in ovarian follicle cells To better understand the function of the 9-1-1 complex during Drosophila development and in DNA damage checkpoint responses, we analyzed the localization pattern of each of the proteins alone. First, polyclonal antibodies against DmRad9 were raised, however, these antibodies did not work for both western blot and for immunolocalization. Next, a construct in which the endogenous DmRad9 gene was tagged with GFP was generated; however, the tagged protein could not be detected in S 2 R+ or S2 cells. Thus, an alternative approach for studying the localization pattern of the Drosophila 9-1-1 complex was chosen. Since previous studies on the localization pattern of the human 9-1-1 complex relied on the expression of tagged protein in mammalian cells [18][19], the same approach was adopted here. Thus, tagged versions of each protein that could be expressed in S 2 R+ or S2 cells (Drosophila-embryo derived cells) or in transgenic flies using the UAS/Gal4 binary system were created. All of the proteins used in this study were tagged at their N-terminus (see Material and Methods) and were over-expressed using actin-Gal4 promoter. Expression of HAtagged DmHus1 in S 2 R+ cells and showed that the protein is evenly distributed throughout the cytoplasm ( Figure 1A). We also found that GFP-tagged DmRad1 was localized throughout the entire S 2 R+ cell ( Figure 1B). Based on the Drosophila genome annotation in flybase, it was suggested that the Drosophila rad9 gene transcribed two alternative splicing forms, DmRad9A and DmRad9B. Expression of a GFP-tagged versions of these two alternatively spliced variants showed that whereas DmRad9B is concentrated in the S 2 R+ nucleus ( Figure 1F), DmRad9A is localized to the S 2 R+ nuclear membrane ( Figure 1C), as revealed by co-localization with lamin ( Figure 1D, G), a structural component of the nuclear membrane.
To analyze the localization pattern of the 9-1-1 complex in flies, transgenic insects in which a tagged form of each protein could be expressed using the UAS/Gal4 system were created. These genes were cloned into the pUASp vector, which allows for expression in somatic cells, as well as in germline-derived tissues. We first overexpressed the protein in the ovarian somatic follicle cells, since it could be easily detected in these cells, It was found that DmHus1 is localized to the cytoplasm ( Figure 1I), that DmRad1 is evenly distributed in the follicle cells ( Figure 1J) and that DmRad9A is found at the follicle cell nuclear membrane ( Figure 1K).
The DmRad9A transcript is more abundant than is the DmRad9B transcript during oogenesis Our preliminary results showed differences in the localization pattern of the two alternatively spliced DmRad9 proteins. Based on flybase, DmRad9A and DmRad9B transcripts encode for proteins containing 456 amino acids each. These two alternatively spliced forms share the same first 424 amino acids, with the last 32 amino acids differing between the two DmRad9 forms. Interestingly, Drosophila DmRad9A and DmRad9B share homology over the first 268 amino acids with their human counterpart, whereas that the C-terminal region of the DmRad9 protein is unique to Drosophila and was not identified in the human protein.
First, to check whether the two predicated alternative splice forms are indeed transcribed, we tried to amplify them by PCR from cDNA of flies ovaries using the same forward primer and a reverse primer specific to each transcript. Both DmRad9 alternative splice forms were amplified, demonstrating that the DmRad9 gene generates two splice variants. To determine the expression level of each of the transcripts during oogenesis, real time RT-PCR using primers specific to each transcript was performed. The real time RT-PCR data was analyzed by the 2 2DDCt method [22]. Our results revealed that the DmRad9A transcript is more abundant than is the DmRad9B transcript (6.4260.27-fold), suggesting that DmRad9A in the major transcribed alternative splice during oogenesis. It is worth mentioning that even though the DmRad9A transcript predominates during oogenesis, this does not necessary imply that DmRad9B plays no role during oogenesis.
DmRad9A is also localized to the nuclear membrane in cultured mammalian cells Since it was found that DmRad9A is localized to the cell nucleus membrane ( Figure 1C), the localization pattern of this Drosophila protein in mammalian cells was determined. For that purpose, each of the 9-1-1 genes was cloned into a mammalian expression vector to yield the protein tagged to GFP. We found that similarly to their localization pattern in Drosophila cell line; these proteins show the same localization pattern in mammalian cells. Specifically, DmHus1 protein is localized to the cell cytoplasm ( Figure 2A), DmRad1 is evenly distributed in the cell ( Figure 2D) and DmRad9A is localized to the cell nucleus membrane ( Figure 2G). Interestingly, an almost similar localization pattern was shown for the human homologues; both hRad1 and hHus1 are cytoplasmic while hRad9 is a nuclear protein [19]. The only difference between the Drosophila Rad9 gene and its human counterpart is that the DmRad9 gene, but no the human gene, contains an alternative splice form that is localized to the nucleus membrane.

DmRad9, DmRad1 and DmHus1 physically interact
Next, the interaction between the Drosophila Rad9, Rad1 and Hus1 proteins was analyzed. Previously, using a yeast two-hybrid system, we reported the direct interaction of these proteins and found that Hus1 interacts with DmRad9A and Rad1, although no interaction between Rad1 and DmRad9A was detected by in this assay [20]. To more directly test whether these proteins form a complex, HA-DmHus1, FLAG-DmRad9A and GFP-DmRad1 proteins were expressed in S2 cell. Cell lysates was subjected to immunoprecipitation using anti-GFP antibodies or normal rabbit IgG as a negative control ( Figure 3). Captured complexes were analyzed using antibodies specific for FLAG, HA, or GFP. Our results demonstrate for the first time that it is possible to coprecipitate DmRad9 with DmRad1 ( Figure 3) and Hus1, indicating that DmRad9 forms a complex with DmRad1 and DmHus1.
DmRad9 determines the localization of the 9-1-1 complex It was shown that the human Rad9 protein is localized to the nucleus and that the Rad9 protein also determined the localization of the 9-1-1 complex [18]. Thus, the localization of the Drosophila complex when tagged versions of the three polypeptides are coexpressed, was assessed. For this purpose, transgenic flies expressing tagged versions of all three proteins were created. Expressing all three proteins in somatic follicle tissues cells lead to their accumulation at the nuclear membrane ( Figure 3J), i.e. where DmRad9A was detected when expressed alone. Similar results were obtained when all three proteins were expressed in S 2 R+ cells ( Figure 3F). These results confirm that DmRad9 can determine localization of the 9-1-1 complex.

Identification of a DmRad9 nuclear localization signal
It was shown that human Rad9 contains a predicted NLS that plays an essential role in the nuclear transport of not only Rad9 but also of human Rad1 and Hus1 [18]. As our results showed that DmRad9A is also localized to the nuclear membrane and that DmRad9B is concentrated in the nucleus, it appears that both DmRad9A and DmRa9B contain a NLS. To explore this possibility, the DmRad9A and DmRad9B proteins were scanned for NLSs using the PSORT II algorithm [23]. This search revealed three potential NLSs, the first found between amino acids 287 and 289, the second between amino acids 300 and 302 and the third between amino acids 314 and 316. To determine the importance of each potential NLS, the suspected lysine and arginine resides were mutated to alanine. We found that mutating amino acids either in DmRad9A (Figure 4) or DmRad9B ( Figure 5) at sites 287-289 ( Figure 4C and 5C) and 314-316 ( Figure 4I and Figure 5I) to alanines had no effect on the localization of the proteins. However, mutating residues 300-302 to alanine had a dramatic effect on the localization of DmRad9A ( Figure 4F) and DmRad9B ( Figure 5F). In this case, the protein is no longer detected in the cell nucleus and is instead found in the cytoplasm. These results suggest that Drosophila DmRad9 contains a monopartite NLS between amino acids 300 and 302.
DmRad9A localization at the oocyte nuclear membrane is not affected by over-expression of non-phosphorylatable Barrier-to-Autointegration Factor (BAF3A) but is affected by a failure to repair DSBs To study the physiological importance of DmRad9A nuclear localization, we addressed the localization of DmRad9A is response to activation of a meiotic checkpoint. It was suggested that the interaction between DNA, protein and the nuclear membrane plays an important role during Drosophila meiosis [24][25]. It was shown that after recombination has completed, NHK-1 directly phosphorylates BAF that anchors meiotic chromosomes to the nuclear envelope [24]. Thus, mutation of NHK-1 or overexpression of non-phosphorylatable BAF (called BAF3A) resulted in an association of the chromosomes with the nuclear envelope and prevented the proper organization of the highly packed chromatin of the oocyte nucleus (also called the karyosome) [24]. BAF protein was found to link chromatin DNA to LEM-domaincontaining inner nuclear envelope proteins (i.e. Drosophila otefin) by binding to both simultaneously. In cells expressing nonphosphorylatable BAF-3A, otefin often accumulated in a region of the nuclear envelope in close contact with meiotic chromosomes [24].
Taken into account that DmHus1, as part of the 9-1-1 complex, is involved in the meiotic checkpoint pathway, together with the fact that DmRad9A is localized to the nuclear membrane, we determined if DmRad9A localization is affected by over-expression of BAF3A. We first found that expression of GFP-DmRad9A in the germline under the control of nos Gal4-VP16 led to the accumulation of the protein at the nuclear membrane in both nurse cells and in the developing oocyte ( Figure 6B). Next, BAF3A was over-expressed in the ovaries and the localization of GFP-DmRad9A was analyzed. We found that affecting karyosome formation ( [24]; Figure 6I) by over-expression of BAF3A had no effect on the localization of DmRad9, with the protein remaining at the oocyte nuclear envelope ( Figure 6J).
Finally, we asked whether activation of the meiotic checkpoint in response to the presence of unrepaired DSBs would affect DmRad9A localization. Accordingly, the localization of DmRad9A in a DNA repair mutant background was studied. Thus, flies that are homozygous for okra (okr), the Drosophila Rad54 homologue, and that express GFP-DmRad9 under the control of nos Gal4-VP16 were generated. We found, as described before, that the karyosome of the okr mutant is fragmented ( [26]; Figure 6M). Moreover, GFP-DmRad9 was no longer located at the nuclear membrane ( Figure 6N), suggesting that persistence of DSBs during meiosis affects DmRad9A localization.

Discussion
In this study, we showed that the Drosophila 9-1-1 proteins present an almost similar localization pattern as do their mammalian homologues. Both hRad1 and hHus1 are cytoplas- mic, while hRad9 is a nuclear protein [19]. In Drosophila the two alternatively spliced forms of DmRad9 are both localized to the nucleus, yet DmRad9A (but not the human homologue) is a nuclear membrane-bound protein. Moreover, when all three proteins were co-over-expressed, DmRad9A and DmRad9B determined the localization of the two other proteins, namely DmRad1 and DmHus1. The same is also true in mammalian cells, where it was suggested that genotoxic stress induces the expression of hRad1, which in turn stabilizes hHus1. Once all proteins are present, hRad9 transports the complex into the nucleus [19].
We have shown that DmRad9 possesses a NLS near the Cterminus of the protein. Much like its human counterpart [18], the DmRad9 NLS was found to be crucial for localization of the protein to the nucleus. Interestingly, the NLS of hRad9 and DmRad9 reside in the C-terminal regions of both proteins, no other similarity exists between these two proteins in this region. The first 274 amino acids of the human Rad9 protein show relatively high similarity (52%) to the first 268 amino acids of DmRad9. Moreover, both the human Rad9 NLS motif, which lies between amino acids 356 to 364, and the DmRad9 NLS motif, found between amino acids 300 to 302, are not conserved. Despite these differences, DmRad9A is localized to nuclear membrane when expressed in mammalian cells, suggesting that the mechanism by which the protein is targeted to the nuclear membrane is likely conserved. Previously, it had been shown that DmHus1 is involved in activation of a meiotic checkpoint [20]. Moreover, as described in this study, the DmRad9A transcript is more abundant than is the DmRad9B transcript during oogenesis. Thus, the physiological function of DmRad9A nuclear membrane localization during activation of the meiotic checkpoint was studied. The Drosophila meiotic checkpoint was first revealed upon study of a class of mutant genes that required for the repair of recombinationinduced DSBs during Drosophila oogenesis [26][27][28][29][30]. Mutations in these genes lead to activation of a meiotic checkpoint [28,[31][32], leading to the appearance of several defects during oogenesis. The most obvious phenotypes manifested are the dorsal-ventral patterning defects of the egg and the organization of the oocyte nucleus karyosome. Recent studies have offered some insight into the connection between activation of the meiotic checkpoint and the karyosome. It was found that nucleosomal histone kinase-1 (NHK-1) is essential for karyosome formation [33]. NHK-1 phosphorylates the linker, BAF, to release meiotic chromosomes from the oocyte nuclear envelope during karyosome formation [24]. Expression of a non-phosphorylatable BAF3A mutant prevented the release of meiotic chromosomes and resulted in a karyosome defect, as was observed in NHK-1 mutants [25]. Based on the above, we decided to analyze whether DmRad9A oocyte nuclear localization could be affected in response to an inability to repair DSBs or due to failure in releasing meiotic chromosomes from the oocyte nuclear envelope. For this purpose, we considered the localization of GFP-tagged DmRad9A in okr mutants, a Rad54-like protein, a double-strand DNA breaks repair enzyme [26], and in flies expressing a non-phosphorylatable form of BAF (BAF3A). DmRad9A localization is not affected in the background of over-expression of BAF3A. Were DmRad9A involved in the physical connection between the chromosomes and the oocyte nuclear envelope, we would have expected to get results similar to what was shown for otefin. In the wild type oocyte nucleus, otefin is found at the nuclear membrane. However, upon over-expression of BAF3A, otefin accumulated in a region of the nuclear envelope in close contact with meiotic chromosomes and was absent from other region of the oocyte nuclear membrane [24]. On the other hand, persistence of DSBs, as observed in okr mutants, dramatically affected DmRad9A oocyte nuclear membrane localization. Thus, the displacement of DmRad9A from the oocyte nuclear membrane due to activation of a meiotic checkpoint is probably part of the oocyte response to DSBs, rather than reflecting a step in the process of attachment of the meiotic chromosome to the nuclear membrane.

Cloning of rad1 and rad9 into pUASp vectors and creation of transgenic flies
To create GFP-tagged DmRad1, DmRad1 was amplified from cDNA using modified primers to create an XbaI restriction site at the 5' end (5' TCTAGAATGACTGATGTGGA GCCATCGCCC 3'), and a NotI restriction site at the 3' end (5'GCGGCCGCTTAATCAGTGTGAG TGTGAGCAAAG-GAATTATG 3'). The resulting PCR product was digested with Xba and NotI and cloned into the pUASp vector containing GFP. To create GFP-tagged DmRad9, DmRad9 was amplified from cDNA using modified primers to create an XbaI restriction site at the 5' end (5' TCTAGAGTTTGTTTACAAATTTCAGC3'), and an XbaI restriction site at the 3' end (5'TCTAGATGCTTT-TAAAATTTATGTTT 3'). The resulting PCR product was digested with XbaI and cloned into the pUASp vector containing GFP. P-element-mediated germline transfection of these constructs was carried out according to standard protocols [36].
Cloning of rad9, rad1 and hus1 in a mammalian vector 9-1-1 proteins were cloned into the pEGFP-C3 vector (28). The full-length coding sequences of all 9-1-1 proteins were amplified from cDNA by PCR. The DmRad9A sequence was amplified using modified primers to create an XhoI restriction site at the 5' end (5' CTCGAGATGAAATACACTTTAGAGGG 3') and a KpnI site at the 3' end (5' GGTACCTCA AAGCAGCTCGTAACC 3') of the gene. The full-length DNA sequence of DmHus1 was amplified by PCR using modified primers to create an XhoI restriction site at the 5' end (5' CTCG AGATGAAGTTCCGCGCA CTGATGC 3') and a KpnI site at the 3' end (5' GGTACCCTA-CATACAAAC AGCTGGC 3') of the gene. To express DmRad1, the coding sequence was amplified using modified primers to introduce a HindIII restriction site at the 5' end (5' AAGCT-TATGACTGATGTGGAGCCATCGC 3') and a KpnI site at the 3' end (5' GG TACCTTAATCAGTGTTGAGCAAAGG 3') of the gene. The resulting DmRad9A and DmHus1 PCR products were digested with XhoI and KpnI and cloned into the pEGFP-C3 vector. DmRad1 was digested with HindIII and KpnI and cloned into the pEGFP-C3 vector. Real-time reverse-transcription (RT)-PCR Total RNA was extracted from ovaries using the NucleoSpin RNA II kit, including DNase treatment, according to the manufacturer's instructions (Macherey-Nagel). cDNA was transcribed from 1-5 mg total RNA using reverse transcriptase and oligo(dT) (Bio-Lab, Beit Haemek, Israel), again according to the manufacturer's instructions (ABgene). Reverse-transcribed total RNA (100 ng) was amplified in a 20 ml reaction containing 100 nM of each primer and 10 ml of SYBR Green PCR Master Mix (Stratagene). RT-PCR was performed to amplify DmRad9A and DmRad9B cDNA using the same forward primer (5'-GCACGGAGGTTTGCTTTATC-3') and the DmRad9A (5'-CAACATAGTCTTCAGTCGGC-3') or DmRad9B (5'-GTAGGTCCTCTGAAAGCAAC -3') reverse primers. To normalize differences in total cDNA between samples, ribosomal protein 49 cDNA was amplified using primers Rp49 Fwd (5'-CCGCTTCAAGGGACAGTATCTG-3') and Rp49 Rev (5'-CACGTTGTGCACCAGGAACTT-3'). PCR conditions were as follows: The reaction mixtures were first kept at 95uC for 15 min, then 40 cycles of PCR (95uC for 30 s, 55uC for 1 min, and 72uC for 1 min) were performed, and finally, the mixtures were incubated at 95uC for 1 min, 55uC for 30 s, and 95uC for 30 s. All quantitative PCR analyses were performed in triplicate. Real-time PCR was performed using the Mx3000p cycler (Stratagene, La Jolla, CA) and the amount of gene product in each sample was determined by the comparative quantification method, using MxPro software (Stratagene).

Transformation of Human Embryonic Kidney 293 (HEK293) cells
T-REx-293 cells (HEK cells stably containing the pcDNA6/TR regulatory vector and thus expressing the tetracycline repressor) (Invitrogen) were cultured in DMEM medium (Gibco) enriched with glucose, 10% fetal calf serum (FCS), 2 mM L-glutamine and a 1% antibiotic mixture comprising penicillin and streptomycin (Biological Industries). Cells were then incubated at 37uC and 92% humidity in the presence of 5% CO 2 . 5610 6 cells were transfected with 1 mg of the expression vector and 1 mg of a plasmid containing the actin-Gal4 driver using the TransIT-LT1 reagent (Mirus), according to the manufacturer's protocol. Prior to transfection, the growth medium was replaced with fresh DMEM. 24 h post-transfection, the medium was again replaced.
Cell fixation and staining S 2 or S 2 R+ cells were cultured in Drosophila Schneider's medium (Biolabs Industries, Israel) containing 10% fetal calf serum and PSA solution containing penicillin (10,000 U/ml), streptomycin (10 mg/ml) and amphotericin B (0.025 mg/ml) (1:100;, Biolabs Industries). Cells were maintained at 25uC under normal atmospheric conditions. Prior to transfection, the cells were cultured in fresh Schneider's medium. 4610 6 cells were transfected with 1 mg of tej pUASp-based expression vector and the Actin-Gal4 driver using Escort IV (Sigma), according to the manufacturer's protocol. 24 h post-transfection, the medium was replaced.

Western blot analysis
Proteins were loaded onto a 10% polyacrylamide gel. Following electrophoresis, proteins were transferred to nitrocellulose membranes (PROTRAN, Schleicher & Schuell) for 1 h at 300 mA. The nitrocellulose membranes were blocked by incubation in TTBS (0.2 M Tris, pH? 1.5 M NaCl, 9 mM Tween 20) containing 5% non-fat dry milk for 30 min at room temperature, followed by a 1 h incubation with primary antibodies. The membranes were washed in TTBS and incubated for 30 min with labeled a-mouse antibodies (Amersham) at a 1:2000 dilution. Antibody binding was visualized using an enhanced chemiluminesence detection kit (Biological Industries). Primary antibodies used were mouse a-HA (1:500; Santa Cruz Biotechnology), mouse a-GFP (1:1000; Roche Diagnostics) and mouse a-tubulin (1:1000; Sigma) antibodies.

Ovary antibody-staining
Ovaries were dissected in PBS, fixed for 20 min in 3.8% formaldehyde in PBS and heptane and washed 3610-min in PBST. The ovaries were incubated for 1 h in PBS, 1% Triton X-100, and blocked for 1 h in 3% BSA in PBST. After overnight incubation at 4uC with primary antibodies at appropriate dilutions followed by PBST washes, the ovaries were incubated with secondary antibodies for 1 h, washed, and mounted in 50% glycerol. As primary antibodies, we used mouse a-Flag M2 (1:250; Sigma) and mouse a-Flag M2 (1:250; Sigma) antibodies. The secondary antibodies, Cy3 goat a-mouse antibodies (Jackson Immunoresearch), were used at a 1:100 dilution. Egg chambers were imaged on a Zeiss LSM510 laser-scanning confocal microscope.