Rapid degradation of dominant-negative Rab27 proteins in vivo precludes their use in transgenic mouse models

Background Transgenic mice have proven to be a powerful system to study normal and pathological gene functions. Here we describe an attempt to generate a transgenic mouse model for choroideremia (CHM), a slow-onset X-linked retinal degeneration caused by mutations in the Rab Escort Protein-1 (REP1) gene. REP1 is part of the Rab geranylgeranylation machinery, a modification that is essential for Rab function in membrane traffic. The loss of REP1 in CHM patients may trigger retinal degeneration through its effects on Rab proteins. We have previously reported that Rab27a is the Rab most affected in CHM lymphoblasts and hypothesised that the selective dysfunction of Rab27a (and possibly a few other Rab GTPases) plays an essential role in the retinal degenerative process. Results To investigate this hypothesis, we generated several lines of dominant-negative, constitutively-active and wild-type Rab27a (and Rab27b) transgenic mice whose expression was driven either by the pigment cell-specific tyrosinase promoter or the ubiquitous β-actin promoter. High levels of mRNA and protein were observed in transgenic lines expressing wild-type or constitutively active Rab27a and Rab27b. However, only modest levels of transgenic protein were expressed. Pulse-chase experiments suggest that the dominant-negative proteins, but not the constitutively-active or wild type proteins, are rapidly degraded. Consistently, no significant phenotype was observed in our transgenic lines. Coat-colour was normal, indicating normal Rab27a activity. Retinal function as determined by fundoscopy, angiography, electroretinography and histology was also normal. Conclusions We suggest that the instability of the dominant-negative mutant Rab27 proteins in vivo precludes the use of this approach to generate mouse models of disease caused by Rab27 GTPases.


Background
Membrane traffic in eukaryotic cells is mediated by vesicular carriers, which bud from a donor compartment, are targeted to, and fuse with the appropriate acceptor membrane. The Rab family of Ras-like GTPases is known to play a crucial role controlling these mechanisms [1,2]. Among Rabs, the Rab27 subfamily consists of two isoforms, Rab27a (previously designated Ram) [3] and Rab27b (previously designated c25KG) [4].
In order to function in membrane traffic, Rabs require the covalent addition of one or two C20 geranylgeranyl groups [5,6]. Geranylgeranylation serves as a membraneanchoring device in Rab proteins [5]. This reaction is complex and occurs in several steps. Newly synthesised Rab proteins first associate with Rab Escort Protein (REP) and form a stable 1:1 complex [7,8]. The complex then serves as a substrate for Rab geranylgeranyl transferase to catalyse geranylgeranylation via thioether bonds to carboxyterminal cysteine residues in Rabs [9,10]. Alternatively, newly synthesised Rabs can associate with a pre-formed REP:Rab-geranylgeranyl-transferase complex [11]. After geranylgeranylation, the REP:Rab complex is competent for delivery of the newly prenylated protein to cellular membranes by a process that has not been characterised [12,13].
The human genome contains two related REP genes, REP1 and REP2 [14]. Loss-of-function mutations in REP1 result in choroideremia, an X-linked slow-onset retinal degeneration that affects photoreceptors, retinal pigment epithelium and the choroid [8,15]. We have speculated that the disease is caused by selective defects in Rab geranylgeranylation in the affected retinal cells and have identified Rab27a as one Rab that appears selectively unprenylated in choroideremia lymphoblasts [16]. Thus, Rab27a, which is expressed at high levels in the retinal pigment epithelium and choroid, could play an important role in triggering the degenerative process in choroideremia [16,17].
The rationale for the present study was to generate transgenic mice expressing dominant-negative mutant Rab27 proteins in order to interfere with the function of the wild type alleles and disrupt the Rab27 regulatory function. We found that dominant-negative Rab27 proteins are unstable and quickly degraded in vivo, and are thus unable to generate a dominant-interfering effect.

Construction of transgenic vectors
Previous studies indicated that selective point mutations in Ras-like proteins that interfere with the cycle of GTP binding and hydrolysis produce either dominant-negative or constitutively-active mutant proteins [18]. Equivalent mutations were introduced into Rab27a cDNA with the predicted effects indicated in Table 1. In order to study whether Rab27a played a role in the retinal degenerative process observed in choroideremia, we wished to generate transgenic mouse lines expressing dominant-negative Rab27a.
We constructed two sets of transgenic lines. For one set we took advantage of the tyrosinase tissue-specific promoter to drive expression of the transgene in pigmented cells including retinal pigment epithelium and choroidal melanocytes, given that these cell types are probably the primary site of degeneration in choroideremia [15,19,20]. cDNAs for Rab27a wild-type, dominant-negative mutants (Rab27a T23N and Rab27a N133I ) and constitutively-active mutant (Rab27a Q78L ) were cloned downstream of the tyrosinase promoter and upstream of the human growth hormone. Figure 1A provides a schematic diagram of the PTyr/Rab27a/hGH expression constructs. These constructs include a 2,200-bp fragment of the tyrosinase promoter, a 650-bp rat Rab27a coding sequence and 600-bp of sequence derived from the human growth factor gene, which provides a polyadenylation signal. To generate transgenic lines, a 3.4 kb XhoI -NotI fragment containing the expression cassette was injected into (C57BL/6JxCBA) embryos according to standard procedures [21]. For the second set, we used the chicken β-actin promoter to achieve ubiquitous and high level expression of the transgenes in the mice [22,23]. Figure 1B provides a schematic diagram of the PCAG/myc-Rab27/β-globin expression constructs, which include a 1,235-bp fragment of chicken β-actin promoter, a 365-bp CMV-IE enhancer, a 33-bp myc-tag fragment, a 650-bp rat Rab27a or human Rab27b coding sequence and 525-bp of the rabbit β-globin polyadenylation signal. To generate transgenic lines, a 3.4-kb SpeI -BamHI fragment containing each expression cassette was injected into (C57BL/6JxCBA) embryos according to standard procedures [21].

Production of transgenic lines
Transgenic lines generated following injection of the PTyr/Rab27a/hGH or PCAG/myc-Rab27/β-globin constructs were named according to the nomenclature indicated in Table 2. Mice containing the Rab27a transgenes were identified by PCR using primers specific to the tyrosinase promoter or myc-epitope (5' primer) and Rab27a cDNA (3' primer) (data not shown). In addition, each line was subjected to Southern blotting. Figure 2 shows a typical Southern blot screening following KpnI digestion and hybridisation using the entire coding sequence of Rab27a as a probe. Note that KpnI cuts the transgene once between the myc-tag and the coding sequence of Rab27a. Thus, a single transgenic integration will generate one hybridising fragment in addition to the endogenous bands. A tandem array of transgenes will generate two bands, one at 3.4 kb corresponding to the transgenic array and one of variable size corresponding to the Rab27a cDNA nearest the adjacent genomic DNA. As shown in Figure 2, DNA from mouse 21 showed a single extra band, suggesting a single integrated copy while DNA from mouse 25 exhibited one band at about 12 kb and one at 3.4 kb suggesting tandem integration. An accurate number of transgenic sequences per cell was not determined in part because the transgene expression was independent of transgene copy number, as previously reported [24,25]. For all the constructs injected, the germ line integration was approximately 20% and thus quite efficient. At least two founders that transmit each transgene were generated. All the founder animals (F0) obtained were successfully bred with C57BL/6J mice to establish transgenic lines. We also attempted to generate homozygous transgenic mice but this was only possible for a restricted number of lines.

Transgene mRNA expression
The expression of the transgenes was initially assessed using an RT-PCR assay on total RNA extracted from eye tissue (Fig. 3). We designed an assay where both transgenic and endogenous Rab27a mRNA could be amplified by selecting primers that would anneal equally well to the mouse and rat sequences. The RT-PCR products were then digested with EcoRI (cutting only the rat sequence) or SmaI (cutting only the mouse sequence) restriction enzymes to distinguish between endogenous and transgenic-derived products (Fig. 3A). This assay allowed us to distinguish between lines that do not express significant amounts of the transgene, such as A27aT25/1 from those that express high levels, such as A27aT25/2 (Fig. 3B). Al-most all actin promoter-driven transgenic lines expressed high levels of transgenic mRNA as revealed by this assay.
To quantify approximately the level of expression, the Rab27a or Rab27b product was compared to Hprt product ( Fig. 3C and Table 3). All actin promoter-driven transgenic lines tested, except A27aT25/1, appeared to express the transgene at significant levels, approximately 3-12 times higher than the endogenous counterpart. Conversely, the expression levels in tyrosinase promoter-driven transgenes were much lower, typically not exceeding 3-fold over the endogenous levels (data not shown).

Transgenic protein expression by immunoblotting
The pCAG/myc-Rab27/β-globin expression constructs carry a myc tag fused to the Rab27 protein, which allowed the transgenic protein to be easily quantified. We prepared spleen lysates from transgenic mice and subjected them to immunoblotting analysis. Spleen was chosen because it contains relatively high amounts of endogenous Rab27a, which could be used as an internal control [16]. We used initially a specific monoclonal anti-Rab27a antibody, 4B12 as a probe and observed a 27-kDa band corresponding to the endogenous Rab27a protein in all lysates tested (Fig. 4A). In addition, a slower migrating band was also detected in samples prepared from some of the transgenic mice, corresponding to the myc-tagged transgenic protein as confirmed by immunodetection using the antimyc antibody (Fig. 4B). To assess the range of tissues in which transgenic protein was expressed, twelve different organs (eye, spleen, liver, skin, lung, kidney, stomach, large intestine, small intestine, brain, testis and heart) were harvested and subjected to Western blotting analysis using anti-Rab27a or anti-myc specific probes. Examples where lines A27aQ24 (Rab27a Q78L ) and A27aT25/2 (Rab27a T23N ) were analysed are presented in Figure 4. The pCAG/myc-Rab27/β-globin constructs carrying the wild-type and constitutively-active mutant (Rab27a Q78L ) forms of Rab27a were found to express transgenic protein ubiquitously and at very high levels, usually more than ten times the endogenous levels ( Fig. 4B and data not shown). In contrast the pCAG/myc-Rab27/β-globin constructs carrying dominant negative mutant Rab27a T23N , Rab27a NI33I , Rab27b T23N and Rab27b NI33I gave much lower levels of protein. The highest levels of dominant negative protein were observed in the line A27aT25/2 as shown in Fig. 4A. The amount of transgenic protein expressed in A27aT25/2 was relatively constant from tissue to tissue. In tissues where endogenous Rab27a expression was high, the levels of Rab27a T23N transgenic protein were nearly equal to or lower than, but never higher, than endogenous protein (Fig. 4B). Lower levels of transgenic protein expression were observed for the other lines expressing dominant-negative mutant Rab27a and Rab27b (data not shown).

Figure 1
Organisation of transgenic constructs. Panel A depicts the PTyr/Rab27a/hGH constructs and Panel B depicts the pCAG/ myc-Rab27/β-globin constructs generated as described under "Methods". Point mutations are indicated by an asterisk.

*
The results for the tyrosinase promoter-driven transgenes were less clear as the transgenic protein did not contain an epitope tag and could not be distinguished from the endogenous protein. We have not been able to detect differences in the levels of Rab27a in any of the twelve tissues described above (data not shown).

Turn-over of Rab27a mutants
The discrepancy between mRNA levels and protein levels observed in the different transgenic lines suggested that some mutant proteins might be unstable and rapidly degraded. To explore this possibility, we performed a pulsechase experiment to determine the approximate half-lives of the transgenic proteins. COS-7 cells were transfected with pCAG/myc-Rab27/β-globin constructs expressing either Rab27a WT , Rab27a Q78L , Rab27a T23N or Rab27a N133I , then metabolically labelled with [ 35 S]methionine/ cysteine for 1 h, followed by different time intervals of chase in the presence of non-radiolabelled medium. The cells were then harvested, lysed and subjected to anti-Rab27a immunoprecipitations. Under these conditions, we observed that both myc-tagged wild type Rab27a and Rab27a Q78L decayed with a half-life of approximately 24 h (Fig. 5A,5C,5E). In contrast, the dominant-negative mutant proteins exhibited much shorter half-lives. Our results indicate that the half-life of myc-Rab27a T23N is approximately 2.5 h while that of myc-Rab27a N133I is approximately 6 h (Fig. 5B,5D,5E). These results suggest that dominant-negative Rab27a mutants are unstable and degrade rapidly in comparison with wild type Rab27a.

Phenotypic characterisation: coat colour
While this work was in progress, Jenkins and co-workers reported that a Rab27a loss-of-function mutation leads to

Figure 2
Screening of transgenic mice. Southern analysis was performed using mouse genomic DNA extracted from the tail of potential Rab27a T23N and Rab27a N133I transgenics. Each well contains genomic DNA extracted from an individual mouse tail biopsy and electrophoresed on an agarose gel as described under "Methods". The band around 3.4 kb suggests the presence of an array of transgenes. the phenotype observed in the ashen mouse [26]. The ashen mouse is a model for Griscelli Syndrome (GS), which is caused by Rab27a mutations [26,27]. The phenotype of GS is characterised by partial albinism (coat colour dilution in the mouse) and loss of cytotoxic T-lymphocyte (CTL) killing activity resulting in immunodeficiency. We therefore predicted that our transgenic mice expressing dominant-negative Rab27a mutants would show coat colour dilution if a dominant-negative effect had resulted in vivo. Figures 6A and 6B show representative photographs of dominant-negative mutant transgenic mice, demonstrating no apparent lightening of coat colour similar to that seen in ashen mice. Coat colour dilution in ashen mice is probably due to defects in distribution of melanosomes in skin melanocytes. However, microscopic examination of hair shafts from mutant Rab27a transgenic animals revealed no clusters of aggregated melanin in follicular melanocytes similar to those observed in ashen skin (data not shown). In addition, the Rab27a Q78L and wild-type transgenic mice expressing very high levels of transgenic protein did not present any obvious skin alterations (data not shown).
To test the functionality of the transgenic Rab27a proteins in vivo, we produced mice where the transgenic protein was the only source of Rab27a. This was accomplished by crossing the transgenic lines twice with ashen homozygous mice to obtain homozygous ashen mice carrying the transgenes. The offspring were genotyped [28] and examined for coat colour. As expected, the wild-type Rab27a transgene was able to rescue the ashen mice coat colour nearly completely (Fig. 6C) while the dominant-negative Rab27a T23N transgene did not (Fig. 6D). Mice that were homozygous for the ashen mutation and carried the constitutively-active mutant Rab27a Q78L transgene (ash/ash, -/tg Rab27aQ78L ) partially rescued the coat colour defect (Fig. 6D). This effect was strengthened by transgene homozygosity. Transgenic ash/ash, tg Rab27aQ78L /tg Rab27aQ78L mice were almost entirely rescued to wild type coat colour (data not shown).

Phenotypic characterisation: vision
Despite the lack of a coat colour phenotype, we decided to study retinal function in the transgenic mice. These studies included fundoscopy, angiography, electroretinography and histology and were performed at different ages, from 1 month to over 1 year old.
* T -T23N; Q -Q78L and N -N133I; when multiple lines were obtained for the same construct these are identified numerically.
Fundoscopy is a method that allows detection of gross abnormalities such as retinal or subretinal deposits, defects in the RPE, new vascularisation and detachments of the retina. The system allows rapid, inexpensive and serial examination of the retina and also obtaining very high quality images of the posterior segment of the mouse eye. The retinal vasculature was examined by fluorescein angiography, i.e., fundus photography combined with intravenous fluorescein injection [29]. Neovascularisation that is often present in retinal degeneration is easily identified. It is also possible to detect defects in the RPE, since the melanin in the RPE can attenuate the fluorescence from the

Figure 4
Expression of transgenic Rab27a proteins in mouse tissues. Protein extracts obtained from eyes, spleen, liver, skin, lung, kidney, stomach, large intestine, small intestine, brain, testis and heart (50 µg each) were subjected to SDS-PAGE and probed with monoclonal anti-myc-tag antibody (9E10) and/or a specific anti-Rab27a monoclonal antibody (4B12). Calnexin was used as a loading control and detected using a specific polyclonal antibody. A protein extract of HeLa cells transfected with myc-Rab27a T23N was used as a positive control.   Another important complementary method of eye examination is Ganzfeld electroretinography (ERG). In a Ganzfeld setup, the stimulus (usually a Xenon tube flash of less than 1 ms duration) passes through a diffusor that yields a relatively homogeneous distribution of light intensity on the inside of a bowl. The term 'Ganzfeld' meaning 'full field' denotes that the stimulus reaches practically all parts of the retina and its intensity is approximately equal across that area. For most applications, the ERG, which is an electric sum potential generated by retinal cells following exposure to light, is non-invasively measured at the corneal surface. Typically, the evoked response consists of an initial negative deflection (a-wave), fol-  assess whether potential functional changes in the transgenic lines could be picked up with this method, the lines that presented higher levels of transgenic protein were subjected to ERG at six months of age. Representative lines from each transgenic construct were analysed at six months and found to be indistinguishable from wild type in all but one case. In A27aT25/2 mice (Rab27a T23N ), the b-wave amplitudes were all in the low-normal range ( Fig.  9) but these differences were not statistically significant. Interestingly, this transgenic line showed the highest level of dominant-negative transgenic protein and thus the low-normal ERG may have resulted from a dominantnegative effect of mutant Rab27a. However, as only one of the lines studied showed this result, the disruption of a gene with important function in the retina upon transgenesis in the A27aT25/2 appears to be the most likely explanation.
Finally, we performed histological analysis. Retinal crosssections subjected to conventional histochemical staining taken from 10 month old Rab27a T23N (A27aT25/2 line) and littermate control are shown in Figure 10. Again, we observed no significant differences between the control and the Rab27 transgenic animals. Note the normal thickness of the outer nuclear layer, reflecting normal numbers of photoreceptors in the retina.

Discussion
At the outset, we wished to investigate the possible role of Rab27 proteins in human disease, in particular CHM. We decided to generate transgenic mice expressing dominantnegative forms of Rab27 for several reasons. Firstly, the strategy of generating Rab proteins defective in either GTP hydrolysis (constitutively-active) or GDP/GTP exchange (dominant-negative) to selectively alter the function of individual Rab proteins in cultured cells has been extensively used previously (for example [30,31]). In some cases,

A. B.
D.

C.
ash/ash -/tg Rab27aWT ash/ash +/+ these mutants have had profound effects on membrane trafficking such as with Rab5 [31], but in others such as Rab11a the effects were more subtle [32]. Secondly, a transgenic approach to express the equivalent Rab5 dominant-negative mutant, Rab5 N133I in immune cells reportedly resulted in the desired phenotype [33]. Thirdly, this approach was the best model for studying the partial dysfunction of Rab27a observed previously since dominantnegative proteins would unlikely inactivate 100% of wild type activity [15,16]. In addition, the dominant-negative approach could also result in inactivation of Rab27b activity, if not other related proteins. In contrast, a gene knock-out approach would result in complete absence of Rab27a while retaining full activity of the related protein Rab27b. Our recent finding that Rab27b shows functional redundancy with Rab27a further suggests that Rab27a-related Rabs may be involved in CHM [28]. Fourthly, we observed that these same Rab27a mutations were effective in cell culture models [34].
While this work was in progress, there was a considerable advance in understanding the function of Rab27a and its role in disease. Loss-of-function mutations in Rab27a were observed in patients with Griscelli syndrome (online Mendelian Inheritance in Man (OMIM) # 214450, [http:/ /www.ncbi.nlm.nih.gov/Omim/]) and in the corresponding mouse model, ashen [26,27]. GS is a rare, lethal autosomal recessive disorder, characterised by pigment dilution of the hair (and to a lesser extent of skin) together with early-onset immune deficiency with episodes of hemophagocytic (uncontrolled T lymphocyte and macrophage activation) syndrome [35,36]. The partial albinism in GS patients is due to clumping of melanin pigment in the hair shaft and accumulation of melanosomes around the nucleus within melanocytes [34][35][36][37][38]. The immune deficiency is due to loss of CTL killing activity [27,39,40]. If the Rab27a mutant proteins Rab27a T23N and Rab27a N133I acted as strong dominant-negative mutants, as they do in cell culture [34,38], the transgenic lines generated should exhibit a phenotype resembling ashen mice, possibly with additional phenotypes due to inactivation of Rab27b as well. Unfortunately, no phenotype could be detected in the transgenic lines. Coat colour was not affected (Fig. 5) and CTL activity could not be studied given that the β-actin promoter used in this study does not express in these cells [28]. We also performed detailed analysis of the retina in the transgenic lines, including fundoscopy, angiography, histology and ERG, and could not detect any significant pathological changes.
The lack of phenoytpe in the transgenic mice expressing dominant-negative mutant Rab27a and Rab27b could have been due to non-functional mutant proteins. However, this explanation seems unlikely as the same type of mutants are functional in cultured melanocytes [34,38]. It has been proposed that one mechanism for dominantnegative action by these mutations in Ras-like proteins is by competition for GDP/GTP exchange factors, required for activation of the endogenous proteins [18]. Thus, the absence of phenotype in these mice is probably due to inadequate levels of protein expression to elicit a dominantnegative effect. Indeed, out of more than twenty transgenic lines generated so far expressing dominant-negative mutant forms of Rab27a or Rab27b, only one (A27aT25/ 2) had transgenic protein levels close to the endogenous counterpart (Fig. 4B).
The low levels of dominant-negative transgenic protein are not due to poor expression of the transgenes as very high levels of mRNA was observed in many of the lines. In addition the wild-type and constitutively active proteins were present at high levels from similar constructs. The low levels of protein are probably due to more rapid turnover of the dominant-negative mutant proteins and we observed that the Rab27a T23N and Rab27a NI33I proteins have half-lives 4-8 times shorter than the wild type protein in tissue culture (Fig. 5). This fast turnover of the dominant-negative protein compared to the constitutively-active and wild-type proteins probably results from a combination of factors. The transgenic protein could be poorly prenylated, a possibility supported by the finding that only a small percentage of transgenic protein was membrane-associated (data not shown). Also, the muta- Our initial hypothesis that Rab27a plays a role in the retinal degenerative process of CHM remains untested given the lack of success of this approach. As we proposed originally, it is unlikely that Rab27a is the only dysfunctional Rab protein in this disorder and our inability to test retinal tissue from patients has hampered the identification of other Rabs that might be involved [16]. Nevertheless, the availability of Rab27a knock-out mice (ashen) may allow us to determine whether Rab27a plays an important role in retinal physiology and further elucidate the pathogenesis of CHM.

Conclusions
The results presented here suggest that the use of dominant-negative proteins in in vivo models should be carefully considered. We report the generation of several transgenic lines expressing dominant-negative mutants of Rab27a and Rab27b where no phenotype could be elicit-ed. The lack of a phenotype is likely due to the very modest levels of protein expression, probably caused by rapid degradation, despite very high levels of mRNA resulting from transcription driven by a strong promoter in the transgene.

Animals
All animals described here were maintained on 12-h light/ 12-h dark conditions at Imperial College, CBS Unit, London, UK, under Home Office Project Licence 70/5071.

Construction of mutant Rab27a transgenic vectors
Two general transgenic constructs were constructed, the first contained the pigment cell-specific tyrosinase promoter, Rab27a cDNA and the human growth hormone poly A signal and the second contained the strong ubiquitous chicken β-actin promoter and CMV-IE enhancer (PCAG), followed by an amino-terminal myc-epitopetagged version of Rab27a or Rab27b, followed by rabbit βglobin poly(A) signal (Fig. 1). The construction of the first transgenic vector was initiated by subcloning a BamHI -NotI fragment containing the human growth factor (hGH) termination sequence [41,42] (a generous gift from David Russell, University of Texas Southwestern Medical Center, Dallas, USA) into pBS containing a 2.2 kb insert corresponding to mouse tyrosinase promoter (a generous gift from Paul Overbeek, Baylor College of Medicine, Houston, USA) [19,20]. For the second transgenic construct, the promoter and poly A regions were from pCAGGS [22,23]. The point mutations, Rab27a T23N , Rab27a N133I , Rab27a Q78L , Rab27b T23N and Rab27b N133I were generated by PCR mutagenesis using the rat Rab27a or the human Rab27b cDNA as a template. The myc-epitope was inserted in frame into the Rab27 cDNA by subcloning the Rab27 cDNA into pCMV7-MYC [43] or pBS-MYC. pBS-MYC was generated by cloning the ClaI/SalI-fragment from pCMV7-MYC containing the myc-epitope into pBluescript SK. The fragments containing the myc-Rab27 cDNA were excised from pCMV7-MYC or pBS-MYC with XbaI and BamHI or XhoI and BamHI, respectively. After gel purification using QIAquick Gel Extraction Kit (Qiagen), the recessed 3' termini were filled with 0.5 U of Klenow fragment of Escherichia coli DNA polymerase I according to the manufacturer's instructions. The product was then subcloned into the blunted XhoI site of pCAGGS after Klenow fragment treatment (Fig. 1A).

Generation of the transgenic mice
A 3.4 kb XhoI -NotI fragment containing the Ptyr/ Rab27a/hGH or a 3.4 kb SpeI -BamHI fragment containing PCAG/myc-Rab27/β-globin were gel purified using QIAquick Gel Extraction Kit (Qiagen) as described by the manufacturer and eluted with 10 mM Tris-HCl (pH 8.5). DNA was then dissolved in sterile 0.1 mM EDTA, 10 mM The crosses indicate the median, the boxes the 25%-to 75%-quantiles, and the whiskers the 5%-to 95%-quantiles. The upper (95% quantile) and lower (5% quantile) red line indicate the normal range based on wild type data.
Tris-HCl (pH 8.0) at about 2 ng/µl and microinjected into the pronuclei of one-cell stage embryos from a C57BL/ 6JxCBA background collected from super-ovulated female mice [21]. Microinjected eggs were transferred at two-cell stage into the oviducts of pseudopregnant recipient females. Newborn mice were routinely screened for incorporation of the transgene by PCR and/or Southern blotting.

Screening and generation of transgenic lines
Mice were genotyped by PCR amplification and by Southern blot analysis using genomic DNA obtained from mouse-tail biopsy samples. The transgene was detected with sense oligonucleotide JR119 (5'-AT-GGAACAAAAACTCATCTCAGAAGAGG) corresponding to the myc-tag sequence of the transgene or JR13 (5'-ACATGTGATAGTCACTCCAGGGGTTGCT) corresponding to tyrosinase promoter, and antisense oligonucleotide JR14 (5'-AGTTGAACTTCCCATCAGTGTACTGGTA) corresponding to the Rab27a coding sequence. Tail biopsies (0 .5 cm) were digested with 40 µg/ml of proteinase K in 500 µl of extraction buffer containing 100 mM Tris-HCl (pH 8.0), 200 mM NaCl, 5 mM EDTA and 0.2% SDS overnight at 55°C followed by phenol/chloroform extraction. After precipitation with isopropanol, DNA was washed with 80% ethanol and resuspended in 50 µl of 10 mM Tris-HCl (pH 8.5). For Southern blotting, approximately 20 µg of tail genomic DNA was digested with KpnI overnight at 37°C. DNA was separated on a 0.8% TAE agarose gel overnight at 120 mA and blotted onto Hybond-N+ (Amersham) membrane. Membranes were prehybridised and hybridised with Rab27a or Rab27b probe, corresponding to the entire Rab27 coding sequence, radiolabelled with 20 µCi of [α-32 P]dCTP by Random Oligopriming (Amersham). In order to generate stable transgenic lines, animals shown to be transgenic were subsequently mated to wild-type C57BL/6J mice; transgenepositive offspring (black colour mice) from these crosses were likewise bred. Transgenic lines were maintained as heterozygotes. Transgenic mice (Rab27 T23N , Rab27a Q78L and Rab27a WT ) were crossed with ashen (ash/ash) on a C57BL/6J background obtained by repeated backcrossing of ash/+ mice (over five generations) with wild-type C57BL/6J mice [28]. The resulting heterozygous ashen (+/ ash) mice carrying the transgene were intercrossed to generate mice that carried the transgene on the ash/ash background. These mice were visually evaluated for the rescue of the coat colour. The ashen mutation was screened for by PCR using the primers ASH1 (5'-ACCTGACAAATGAGCA AAGTTTCCTCAATG) and ASH2 (5'-GGAGCAGGGCAGG GCTGGGGAAACCACTCG) followed by restriction enzyme analysis with TaqI and RsaI according to the procedure described previously [28].

RT-PCR analysis
RT-PCR was carried out on total RNA obtained from eyes and isolated using RNeasy (Qiagen). After total RNA isolation, samples were treated with 1 U RNase-free DNase for 45 min at 37°C to remove any transgenic DNA contamination. The DNase was inactivated by heating for 15 min at 70°C. cDNA synthesis and DNA amplification were carried out using 5 µg of total RNA. Labelling of the PCR amplification product was carried out by addition of 0.025 µCi of [α-33 P]dATP. DNA was amplified using only 23 cycles (exponential phase). PCR amplification of Rab27a cDNA was performed with oligonucleotides JR62 (5'-GCATTGATTTCAGGGAAAAGAGAG) and JR63 (5'- TTCTCCACACACCGCTCCATCCGC). To ensure that both transgenic and endogenous mRNA were amplified equally, these oligonucleotides were designed to be homologous to both endogenous (mouse) and transgenic (rat) Rab27 cDNA. After amplification, the PCR products were digested with Eco RI or Sma I to distinguish the mouse endogenous from rat transgenic cDNA. In most cases, the transgene expression was much higher than the endogenous expression making quantification impossible. Therefore, the transgene expression level was determined in each line by quantitating the Rab27 (endogenous plus mutant mRNA) relative to Hprt cDNA using all four primers, 0.025 µCi of [α-33 P] dATP and 23 cycles in PCR reaction. The oligonucleotides used for Hprt amplification were Hprt1 (5'-CCTGCTGGATTACATTAAAGCACTG) and Hprt2 (5'-GTCAAGGGCATATCCAACAACAAAC). Possible contamination of mRNA with transgenic DNA was excluded by control reactions without reverse transcriptase. The radiolabelled PCR products were separated on 2.5% agarose gels, stained with ethidium bromide, transferred onto 3 MM Whatman paper and quantified using a Cyclotron Storage Phosphor Screen (Packard).

Immunoblot analysis
Immunoblotting was performed according to procedures described elsewhere [16] with some modifications using total, cytosol or membrane protein fractions obtained from several C57BL/6J wild-type or transgenic tissues. All the transgenic and non-transgenic mice used for immunoblotting studies were thoroughly perfused with phosphate-buffered saline (PBS) (pH 7.4) prior to dissection of the tissue samples that included eye (the lens was always removed from the ocular globe), spleen, liver, lung, kidney, skin, stomach, large intestine, small intestine, brain, testis and heart. Tissue samples were washed with PBS, and thoroughly homogenised using a Polytron homogeniser in 3-5 volumes of homogenisation buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM phenyl-methylsulfonyl fluoride, 10 µg/ml leupeptin, 10 µg/ml aprotinin, 10 µg/ml pepstatin and 1 mM DTT. Homogenisation was followed by sonication for 5-10 seconds. To sediment unbroken cells and cell nuclei, the homogenates were centrifuged at 5,000 g for 30 min at 4°C. The postnuclear supernatant was then centrifuged at 100,000 g for 1 h at 4°C, resulting in the separation of cytosolic (supernatant) and membrane (pellet) protein fractions. The membrane fractions were resuspended into the same volume of homogenisation buffer containing 1% Triton X-100. The protein concentration was determined only in the total fraction by BCA method (Pierce) according to the manufacturer's instructions. For immunoblot analysis, samples containing 50 µg of protein were separated into soluble or membrane fractions and identical volumes were subjected to SDS gel electrophoresis. After being resuspended into loading buffer, protein samples were loaded onto 12.5% SDS-polyacrylamide electrophoresis gels, run at 35 mA and transferred to Immobilon-P polyvinylidene difluoride (Millipore) membrane in a minitrans-blot cell (Amersham) at room temperature in a buffer containing 25 mM Tris-HCl (pH 8.3), 192 mM glycine, and 10% methanol for 1 h and 30 min at 500 mA. To block non-specific binding sites, dried membranes were incubated in blocking solution containing 0.1% polyoxyethylenesorbitan monolaureate (Tween-20), 4% non-fat milk in PBS for 1 h at room temperature. Membranes were subsequently incubated with PBS/0.1% Tween-20 solution supplemented with the following preparation of antibodies: monoclonal 4B12 antibody (0.3 µg/ml) anti-rat Rab27a [34], monoclonal antibody anti-myc-tag (0.3 µg/ml) purchased from Oncogene (Cambridge, MA, USA) or polyclonal anti-calnexin antibody (1:5,000) purchased from StressGene (Victoria, BC, Canada). The mixture was incubated with the primary antibody with gentle agitation for 1 h at room temperature, rinsed and washed with PBS containing 0.1% Tween-20 for 15 min, 3 times. Membranes were then incubated with horseradish peroxidase-coupled secondary antibody (Dako). The blots were washed as described above and bands were visualised by chemiluminescence using SuperSignal West Pico Chemiluminescence Substrate (Pierce) according to the instructions supplied by the manufacturer. All blots were calibrated with prestained molecular weight markers (Bio-Rad Laboratories).
Cell culture, transfections and labelling COS-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Transfections were performed using lipofectamine (Qiagen). Total amounts of transfected DNA were 1 µg/6cm dish. For pulse-chase experiments, cells were washed with methionine/cysteine-free medium and subsequently incubated with 140 µCi of [ 35 S]methionine/cysteine per 6-cm dish for 2 h. After this, cells were washed three times in PBS, and further maintained in normal Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. At indicated time points (0, 30 min, 3, 5, 8 and 22 h), cells were lysed by incubation in buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM phenyl-methylsulfonyl fluoride, 10 µg/ml leupeptin, 10 µg/ml aprotinin, 10 µg/ml pepstatin, 1 mM DTT and 1% NP-40 for 20 min on ice followed by 5-10 s of sonication. After centrifugation at 14,000 rpm, equal amounts of radioactive lysates were incubated under agitation with 5 µg of 4B12 antibody for 1 h at RT, after lowering the concentration of NP-40 to 0.5% by dilution. Subsequently, 30 µl of protein G-Sepharose beads were added to the radioactive material and incubated overnight at 4°C, while rocking. Then beads were washed four times in PBS, denatured in SDS-containing sample buffer and loaded on a 12%