The murine fibroblast-like cell lines Balb3T3 (American Type Culture Collection, clone A31, ATCC CCL-163, Manassas, VA, USA) and NIH3T3 (ATCC CRL-163) were used for the experiments and primary murine embryonic fibroblasts (MEFs) from 17-day-old BALB/c mouse embryos were used for virus production and plaque assay. The cells were cultured in 10-cm dishes for propagation and divided into appropriate plates for the experiments once they were 80-90% confluent. For cell culture at 37°C and 5% CO₂, Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% (5% for MEF) fetal bovine serum (FBS), 2 mM L-glutamine, 100 mg/ml streptomycin and 100 U/ml penicillin (all reagents from Gibco/Invitrogen, Grand Island, NY, USA) was used.

The recombinant virus Δm138-MCMV (ΔMC95.15) with the deletion of fcr1 (m138) gene (Crnković-Mertens et al., 1998) was regularly used in the experiments to avoid FcR-mediated non-specific binding of antibody reagents. Wild-type (wt) MCMV (strain Smith, ATCC VR-194) was used to calibrate the plaque assay. To monitor MCMV replication by flow cytometry, we used C3X-GFP MCMV (MCMV-GFP), a recombinant virus expressing green fluorescent protein (GFP) in the early phase of infection (Angulo et al., 2000). MCMV stocks were produced, and cells were infected according to standard procedures (Brizić et al., 2018). Cells were infected with 1 PFU/cell at a multiplicity of infection (MOI) of 10 after increasing infectivity by centrifugation. The efficiency of infection was monitored by intracellular detection of Immediate-Early-1 protein (pIE1) as previously described (Lučin et al., 2020).

Antibodies against proteins regulating endocytic transport were monoclonal (mAb) or polyclonal (pAb) as follows: rabbit monoclonal IgG anti Rab10 (Cell Signaling Inc., Danvers, MA, USA; Cat. No. 8127), rabbit pAbs against SNX27 (Abcam, Cambridge, UK; Cat. No. ab241128) and GRASP65 (Thermo Fisher Scientific, Waltham, MA, USA; Cat. No. PA3-910), mouse mAbs from Santa Cruz Biotechnology (Dallas, USA) against SNX1 (IgG₁, Cat. No. sc-376376), SNX2 (IgG_2b, Cat. No. sc-390510), SNX27 (IgG_2b, Cat. No. sc-515707) and Vps35 (IgG_2b, Cat. No. sc-374372), and mouse mAbs against actin (IgG₁, Millipore, Burlington, Massachusetts, USA; Cat. No. MAB150).

Monoclonal antibodies against MCMV proteins were produced, purified, and verified by the University of Rijeka Center for Proteomics, (https://products.capri.com.hr/shop/?swoof=1&pa_reactivity=murine-cytomegalovirus; accessed on February 11, 2024). These antibodies included: mouse mAbs IgG₁ (clone CROMA101) and IgG_2a (clone IE1.01) against pm123/pIE1, mAb IgG₁ (clone CROMA103) against pM112-113 (E1), mAb IgG₁ (clone CROMA 229) against pm06, mAb IgG₁ (clone M55.01 for Western blot) against pM55/gB, mouse monoclonal anti M57 (clone M57.02), mAb IgG₁ against pM74 (clone 74.01), and mAb IgG₁ (clone M116.02) against pM116.

Alexa Fluor (AF)⁴⁸⁸-, AF⁵⁵⁵-, AF⁵⁹⁴-conjugated (Molecular Probes; Leiden, The Netherlands) and AF⁶⁸⁰-conjugated (Jacksons Laboratory, Bar Harbor, ME, USA) antibodies against mouse IgG₁, mouse IgG_2a, mouse IgG_2b and rabbit Ig were used as secondary antibodies for immunofluorescence analysis. Goat anti-rabbit and goat anti-mouse antibodies conjugated to HRP were used for Western blot analysis (Jackson Laboratories, Bar Harbor, ME, USA). DAPI (4,6-diamidino-2-phenylindole dihydrochloride) was from Thermo Fisher Scientific (Waltham, MA, USA; Cat. No. D1306). Puromycin was obtained from Santa Cruz Biotechology Inc., Dallas, USA. Propidium iodide and other chemicals were obtained from Sigma-Aldrich Chemie GmbH (Schnelldorf, Germany).

The cells were cultured on coverslips in 24-well plates to 60-70% confluency for the experiment. Fixation was performed in 4% paraformaldehyde (20 min at r.t.) and permeabilization in 0.5% Tween 20 (20 min at 37°C). After incubation with appropriate primary antibodies and AF-conjugated secondary antibodies (60 min at 37°C), samples were embedded in Mowiol (Fluka Chemicals, Selzee, Germany)-DABCO (Sigma Chemical Co, Steinheim, Germany) in PBS containing 50% glycerol and analyzed by confocal microscopy (Leica DMI8 inverted confocal microscope (confocal part: TCS SP8; Leica Microsystems GmbH, Wetzlar, Germany) and HC PLAPO CS2 objective (63×1.40 oil). Lasers were used as follows: UV (Diode 405) for DAPI, Ar 488 for AF⁴⁸⁸, DPSS 561 for AF⁵⁵⁵ and AF⁵⁹⁵, and He/Ne 633 for AF⁶⁸⁰. The microscope was also equipped with 4 detectors, two of which are PMT and two are HyD. Images were acquired in sequential mode (515x515 pixels, z-series of 0.5 μm) using LAS (Leica Application Suite) X version 3.5.6.21594 software (Leica Microsystems GmbH, Wetzlar, Germany). The zoom factors were as follows: 0.75× (pixel size 481.47 × 481.47 nm), 1.5× (pixel size 240.74 × 240.74 nm), 3× (pixel size 120.37 × 120.37 nm), and 6× (pixel size 60.18 × 60.18 nm). The offset was set to 0-1.5% depending on the background. All samples that were compared within an experiment were imaged with the same parameters.

The presence of AC in infected cells was defined as a concentrated fluorescent signal within the angle of α ≤ 90° (Štimac et al., 2021). The ratio of AC-positive cells per one microscopic sample was counted in at least 10 fields of view. The direct counting was performed on an epifluorescence Olympus BX51 microscope equipped with a DP71CCD camera (Olympus, Tokyo, Japan) with UPlanFL N 40×/0.75 objective.

For the colocalization analysis, we calculated the Manders’ overlap coefficients (M1 and M2) of the entire z-stack (8-12 confocal sections) of images (120.37 × 120.57 nm pixel size) using the JACoP plugin (https://imagej.net/ij/plugins/track/jacop2.html, accessed on February 11, 2024) (Bolte and Cordelières, 2006). Briefly, red, green, and blue channels were split and colocalization of pixels between two selected channels was determined after background subtraction, as described previously (Marcelić et al., 2022). At least 15-20 cells were analyzed in each experiment.

Small interfering (si)RNA sequences were purchased as follows: non-targeting negative siRNA (1022076) and Mm_Snx27_7 sequence (SI04939543) were from Qiagen (Hilden, Germany); siRNA for SNX1 (sc-41346), siRNA for SNX2 (sc-41350) and siRNA for Vps35 (sc-63219) were from Santa Cruz Biotechnology Inc. (Dallas, USA). Cells were transfected according to the manufacturer’s guidelines: siRNA and RNAiMAX Lipofectamine Reagent (Invitrogen, Carlsbad, CA, USA) were mixed and added dropwise to the cells. The final concentration of siRNA for SNX1 was 20 nM, for SNX2 80 nM, for SNX27 60 nM and for Vps35 80 nM. After 48 hours, the cells were analyzed or used for infection.

Balb3T3 cells, treated with the corresponding siRNA for 48 h, were infected with C3X-GFP-MCMV (MOI of 10). Samples were collected at 0, 6, and 16 h post-infection (hpi) and the GFP signal was quantified by flow cytometry (FACSCalibur flow cytometer; Becton Dickinson & Co, San Jose, CA, USA) on 5000 viable cells (dead cells were excluded by propidium iodide staining). The signal at 0 hpi was considered a negative signal and the percentage of GFP-positive cells was determined at 6 and 16 hpi.

MCMV DNA replication was detected at 16-24 hpi by incorporation of 5-ethynyl-2′-deoxyuridine (EdU) and its detection by a fluorescent azide using a Cu(I)-catalyzed [3 + 2]-cycloaddition reaction (Salic and Mitchison, 2008), as described previously (Mahmutefendić Lučin et al., 2023). The infection conditions of 1 PFU/cell block the incorporation of EdU into host cell DNA and allow visualization of replicated MCMV DNA starting at 15-16 hpi (Mahmutefendić Lučin et al., 2023). Cells treated with siRNA were infected with MCMV 48 h after transfection, and the infected cells were incubated with EdU at 16-24 hpi. The fixed and permeabilized cells were subjected to a click reaction with AF⁵⁵⁵ fluorescent azide, washed in PBS containing 3% BSA and incubated with the primary antibodies anti-IE1 and anti-SNX27, followed by AF⁶⁸⁰-conjugated anti-mouse IgG₁ and AF⁴⁸⁸-conjugated anti-mouse IgG_2a, respectively. Cells were analyzed by confocal imaging using a Leica DMI8 inverted confocal microscope, and the percentage of EdU-labeled cells was determined by quantification of IE1- and EdU-positive cells using an Olympus BX52 fluorescence microscope (DP72CCD camera, CellF software, 400x magnification).

To obtain whole cell lysates (WCL), cells were lysed with RIPA lysis buffer supplemented with protease inhibitors (Roche Diagnostics GmbH, Unterhaching, Germany; Cat. No. 11697498001) and mixed with sample buffer (50% glycerol, 10% SDS, 0.05% bromophenol blue, 0.3M Tris, pH 6.8). Proteins were separated by SDS-PAGE (Bio-Rad PowerPac Universal, Hercules, CA, USA) and blotted onto a polyvinylidene difluoride membrane (PVDF-P WB membrane; Millipore, Burlington, MA, USA) at 80 V for two hours using Bio-Rad Trans-Blot Turbo Transfer System (Hercules, CA, USA). Membranes were blocked for 1 to 2 hours in 1% blocking reagent (Roche Diagnostics GmbH, Mannheim, Germany) and incubated with the appropriate primary antibody (overnight at 4°C), washed three times in T-TBS (TBS with 0.05% Tween 20; pH = 7.5) and labeled with peroxidase (POD)-conjugated secondary antibody in T-TBS containing 0.5% blocking reagent for 45-60 min. The membranes were washed again in T-TBS and the signal was detected by chemiluminescence (SignalFire [TM] Plus ECL Reagent or SignalFire [TM] Elite ECL Reagent; Cell Signaling, Cat. No. 12630S or 12757P, respectively) using Transilluminator Alliance 4.7 (Uvitec Ltd., Cambridge, UK) and ImageQuant LAS 500 (GE Healthcare Bio-Sciences AB, Upsala, Sweden). In each experiment, the expression of relevant protein(s) and β-actin as loading control, was detected at the same membrane. Quantitative analysis of the chemiluminescence signal was performed using ImageJ 1.53 software and ImageQuantTL, version 10.2., Cytiva. All values were normalized to the signal from β-actin, which was used as a loading control. First, the β-actin signals were normalized according to the following formula: Lane normalization factor = Observed actin signal for each lane/Highest observed actin signal for the blot. Second, normalized actin value was then used to normalize the experimental signal (raw signal value/normalized actin index). The following formulas were used: (1) to calculate the kinetics of host-cell protein expression during MCMV infection = Normalized experimental signal (t_{x hpi})/Normalized experimental signal (t_{0 hpi}); (2) to calculate the fold change in MCMV protein expression after siRNA treatment = Normalized experimental signal (t_{x siRNA hpi})/Normalized experimental signal (t_{x scr-siRNA hpi}).

pEGFP-SNX27 constructs were gifts from Josef Kittler (University College London, UK): pEGFP-N1-mSNX27 (Addgene plasmid #163617; http://n2t.net/addgene:163617; RRID: Addgene_163617); pEGFP-N1-mSNX27-H112A (Addgene plasmid #163619; http://n2t.net/addgene:163619; RRID: Addgene_163619); and pEGFP-N1-mSNX27-ΔF3 (Addgene plasmid #163618; http://n2t.net/addgene:163618; RRID: Addgene_163618). For transient transfection, Balb3T3 fibroblasts were cultured on coverslips in 12- or 24-well plates and transfected with EGFP-mSNX27 constructs using Lipofectamine 3000 transfection reagent (TR) (Invitrogen, Carlsbad, CA, USA; Cat. No. L3000001) according to the manufacturer’s guidelines: Solutions containing Lipofectamine 3000 TR (1 μl) and DNA solution (1 μg plasmid DNA with 1 μl P3000 reagent) were mixed, incubated at r.t. for 10-20 min, and added dropwise to cells (70% confluent). Cells were infected with MCMV (MOI 10) 30 h after transfection, fixed 16 hpi, and permeabilized with Tween 20 (0.1%). After labeling with appropriate primary and secondary antibodies, the cells were analyzed by immunofluorescence microscopy.

EGFP-mSNX27 ORF was subcloned from the pEGFP-N1-mSNX27 plasmid (Addgene plasmid #163617; http://n2t.net/addgene:163617; RRID: Addgene_163617) into the pLIX_Kan-PstI lentiviral vector that allows doxycycline-inducible expression of the transgene (Kutle et al., 2020). In the pLIX_Kan_Pst vector, PstI and BamHI sites were used to replace the kanamycin resistance cassette (kan) with the PCR amplified EGFP-mSNX27 ORF [forward primer: 5’-CGCCTGGAGAATTGGCTGCAGTCCGCTAGCGCTACCGGA-3’ (providing PstI site) and reverse primer 5’-AAGGCGCAACCCCAACCCCGTTACTTGTACAGCTCGTCCATGC-3’] using NEBuilder HiFi DNA Assembly Cloning Kit (New England Biolabs (NEB)Mass, Ippswich, MA, USA). PstI and BamH1 restriction endonucleases were from NEB.

The constructed lentiviral plasmid pLIX-Kan-PstI_EGFP-mSNX27 was used to produce the NIH3T3-pEGFP-mSNX27 cell line with doxycycline-inducible expression of EGFP-mSNX27. For lentivirus production, 5 μg of pLIX-Kan-PstI_EGFP-mSNX27 was mixed with 10 μg of p8.91 (a gift from Simon Davis; Addgene plasmid #187441; http://n2t.net/addgene:187441; RRID: Addgene_187441) and 0.5 μg of plasmid p-CMV-VSV-G (a gift from Bob Weinberg; Addgene plasmid #8454; http://n2t.net/addgene:8454; RRID: Addgene_8454) in 1.5 ml of Optimem (Thermo Fisher Scientific, Waltham, MA, USA; Cat. No. 31-985-070) and with 41 μl of Lipofectamin 3000 in 1.5 ml of Optimem. The solution was added dropwise to 10-cm dishes containing 80-85% confluent HEK 393T cells (ATCC clone A31, ATCC CRL-3216, Manassas, VA, USA) in 10% FCS DMEM without antibiotic. The supernatants containing the lentiviruses were collected after 24, 30, and 48 h, centrifuged (5 min at 2,000 rpm), filtered (0.45 μM filter), and used for transduction of NIH3T3 cells with puromycin (2.5 μg/ml) as a selection marker. Finally, the puromycin-resistant EGFP-positive cells were additionally sorted using the FACSAria cell sorter (Becton Dickinson & Co, San Jose, CA, USA).

Balb3T3 fibroblasts grown in 24-well plates were treated with SNX27 siRNA or scr-siRNA for 48 h or were non-transfected. Cells were then infected with MCMV (MOI of 10) for an additional 48 and 72 h. Cell lysates and supernatants were then collected, and the production of released virions was determined using the standard plaque assay, as described previously (Štimac et al., 2021).

Data comparison was performed using a two-tailed Student’s t-test when two samples were compared and one-way ANOVA analysis of variance for data with more than two experimental groups. Differences were considered significant when p values were < 0.05 (*p ≤ 0.05; **p < 0.01; ***p < 0.001).

Many cytoplasmic proteins that regulate endosomal flux are retained on membranes within the inner AC of MCMV-infected cells (Lučin et al., 2018, 2020). The retention develops gradually early in the infection, prior to viral DNA replication and expression of late viral proteins, in a structure designated as pre-AC (Taisne et al., 2019; Lučin et al., 2020). Thus, we first used confocal imaging to investigate the distribution of SNX27, Vps35 and SNX1 in the early phase of infection. In uninfected fibroblast-like cells, SNX27 was mainly dispersed in the cytosol in barely detectable punctate structures ( Figure 1A ). In MCMV-infected cells, SNX27-positive structures gradually accumulated in the perinuclear area ( Figure 1A , indicated by arrows) of 49.5 ± 11.8% of cells at 6 hpi and 72.9 ± 11.9% of cells at 16 hpi ( Figure 1A ). This pattern was not associated with a change in the total cellular SNX27 amount ( Figure 1B ). In contrast to SNX27, Vps35, an essential component of the Retromer complex (Vps26:Vps29:Vps35), and SNX1, an essential component of ESCPE-1 that interacts with SNX27 to promote tubulation and cargo exit (Simonetti et al., 2023), were recruited to clearly visible cytoplasmic membrane structures of uninfected cells ( Figure 1A ), consistent with their localization at endosomes. Similar to SNX27, Vps35- and SNX1-positive membranous organelles concentrated in the perinuclear area of MCMV-infected cells at 6 and 16 hpi ( Figure 1A , arrows), and MCMV infection did not alter their expression ( Figure 1B ).

As expected from the data from the literature, Vps35 and SNX1 colocalized mainly on endosomes of uninfected cells ( Supplementary Figure S1 ) as well as at membranous structures within the inner pre-AC of MCMV-infected cells ( Supplementary Figure S1 ). Since endogenous SNX27 was barely detectable by Abs at distinct membrane elements of uninfected cells, we established the NIH3T3 cell line with inducible expression of EGFP-SNX27. In NIH3T3 cells, which are more suitable for transfection, the detection of endogenous SNX27 on membrane elements by Abs was as low as in Balb3T3 cells. After induction, EGFP-SNX27 was recruited to Vps35-positive and SNX1-positive structures ( Figure 1C ). These data suggest that both Retromer- and ESCPE-1-associated functions operate in fibroblast-like cells used for infection, consistent with their known localization at EEs and their distribution in recycling carrier-generating EE microdomains as components of large (SNX27:Retromer:ESCPE-1) or smaller (SNX-BAR or SNX27:Retromer) complexes (Lauffer et al., 2010; Cullen and Steinberg, 2018). In the fully developed pre-AC of MCMV-infected cells (at 16 hpi), SNX27, Vps35, and SNX1 localized in the inner pre-AC ( Figures 1C , Supplementary Figure S1 , arrows) and exhibited a similar level of colocalization as in uninfected cells ( Figure 1C , Supplementary Figure S1 , Supplementary Table S1 ). A similar expression pattern to SNX27 was observed for SNX2 ( Supplementary Figure S2 ), a component of ESCPE-1 that is redundant to SNX1 (Simonetti et al., 2023).

Overall, reorganization of the membrane system during the E phase of MCMV infection is also associated with retention of SNX27, ESCPE-1, and Retromer suggesting that the expanded membranes within the inner pre-AC are capable of cargo retrieval and tubulation. Essentially, the same results were obtained in the NIH3T3 cell line with inducible expression of EGFP-SNX27 ( Figure 1C ) as in Balb3T3 cells ( Figure 1A , Supplementary Figure S1 ), since infection in Balb3T3 cells increased the visualization of endogenous SNX27 through its retention on membranous structures. Thus, to study endogenous expression conditions, we performed further experiments on infected Balb 3T3 cells.

The development of the pre-AC is initiated at 4-6 hpi by the simultaneous displacement of the Golgi and the expansion of the EE, ERC, and TGN membranes within the inner pre-AC (Karleuša et al., 2018; Lučin et al., 2020). The accumulation of Rab10 is a hallmark of this reorganization and represents the earliest event that can be used for monitoring pre-AC development by immunofluorescence imaging, together with markers of the Golgi. Therefore, to investigate the contribution of SNX27:Retromer:ESCPE-1-associated functions in pre-AC development, we monitored pericentriolar Rab10 accumulation after knockdown of SNX1, SNX2, SNX27, and Vps35 expression with siRNA. After successful suppression of these proteins in uninfected cells ( Figures 2A , Supplementary Figure S3 ), cells were infected with MCMV for 16 h. The knockdown state was confirmed by simultaneous Western blot detection of these proteins and viral pIE1 ( Figure 2B ) and cell viability by flow cytometry ( Supplementary Table S2 ). None of the knockdown procedures resulted in the pericentriolar accumulation of endogenous Rab10 in uninfected cells ( Supplementary Figure S3 ). Both untreated and control scr-siRNA-treated cells developed pericentriolar accumulation of Rab10 in MCMV-infected cells, indicating that the transfection procedure and the introduction of foreign irrelevant RNA do not affect membrane reorganization during MCMV infection ( Figures 2C-F ). By contrast, knockdown of SNX27 ( Figure 2C ), Vps35 ( Figure 2D ), and SNX1 ( Figure 2E ) significantly reduced the percentage of cells with pericentriolar Rab10 accumulation. A similar effect was observed after the simultaneous knockdown of SNX1 and SNX2 (SNX1 + 2, Figure 2F ), as they act redundantly in the SNX-BAR (ESCPE-1) coat complex; however, depletion of SNX2 alone had no effect ( Supplementary Figure S4A ). Collectively, these data suggest that SNX27:Retromer:ESCPE-1 complexes are required for Rab10-associated membrane reorganization events in the pre-AC.

Suppression of the SNX27:Retromer:ESCPE-1 complexes did not prevent compaction and displacement of the Golgi into the ring-like configuration that forms the outer pre-AC, as demonstrated by visualization of the cis-Golgi marker GRASP65. In untreated cells, SNX27-, SNX1/2-, and Vps35-positive membranous structures localized mainly within the GRASP65-positive ring representing the inner pre-AC ( Supplementary Figure S5 , arrows), whereas suppression of SNX27, Vps35, SNX1, and SNX1+2 had no effect on the distribution of GRASP65-positive membranes in uninfected cells ( Supplementary Figure S6 ) and did not prevent Golgi compaction in MCMV-infected cells at 16 hpi ( Supplementary Figures S5, S4B ).

To gain further insight into the role of SNX27:Retromer:ESCPE-1 in the development of the Rab10-associated components of the pre-AC, we investigated the role of cargo retrieval and complex formation of SNX27 via its PDZ domain and F3-FERM domain, respectively. The PDZ domain recognizes cargo and interacts with Vps26 in the Retromer (Temkin et al., 2011; Cullen and Steinberg, 2018). The F3-FERM subdomain binds SNX1 or SNX2 to form full SNX27:Retromer:ESCPE-1 complexes (Yong et al., 2021; Simonetti et al., 2022), regulates interactions with endosomal cargoes, and serves as a scaffold for signaling complexes (Chandra et al., 2021). We transfected cells to overexpress wt-SNX27 (pEGFP-N1-mSNX27), SNX27 with the point mutation H112A that disables cargo but not Retromer binding (Gallon et al., 2014) to the PDZ domain (pEGFP-N1-mSNX27 H112A), and SNX27 with a deleted F3-FERM subdomain (pEGFP-N1-mSNX27 ΔF3), in which the interaction of SNX27 with SNX1/SNX2 and signaling complexes is disabled (Halff et al., 2019). In uninfected cells, wt-SNX27 and the mutants SNX27 H112A and SNX27 ΔF3 showed vesicular patterns ( Figure 3A ) that corresponded to EEs, as previously described (Halff et al., 2019; Simonetti et al., 2022). After MCMV infection (16 hpi) of transfected cells, all three forms of SNX27 accumulated in the pericentriolar region of the pre-AC ( Figure 3B ). In wt-EGFP-SNX27- and EFGP-SNX27 H112A-transfected cells, Rab10 accumulated in the pericentriolar region ( Figure 3B , arrowheads) to a comparable degree as in non-transfected cells ( Figure 3B , NT), whereas in cells overexpressing the EGFP-SNX27-ΔF3 Rab10 enrichment was low or absent ( Figure 3B , indicated by arrowheads at the bottom row). These data suggest that the F3-FERM subdomain is required for downstream events associated with Rab10 accumulation. Accordingly, SNX1 accumulated in pre-AC of cells transfected with wt-EGFP-SNX27 or EGFP-SNX27 H112A, but very little in cells expressing EGFP-SNX27 ΔF3 ( Figures 3D, E ). In EGFP-SNX27 ΔF3-expressing cells, SNX1 was dispersed in the cytosol and occasionally exhibited scattered dots ( Figure 3D ). These data are consistent with the activity of the F3-FERM subdomain in recruiting the ESCPE-1 proteins SNX1/SNX2, which are required for the expansion of the subset of Rab10-associated membranes within the inner pre-AC. As Rab10 is an essential component of tubular endosomes (Etoh and Fukuda, 2019), these data suggest that SNX27:ESCPE-1 complexes are required for the initiation of tubulation and Rab10 recruitment at very early stages of tubule biogenesis. Moreover, these data indicate that the binding of cargo proteins to the PDZ domain is not required for Rab10-associated membrane reorganization.

Previous experiments show that depletion of SNX27 and SNX27-associated complexes abolishes an important step in the MCMV replication cycle that contributes to membrane reorganization and the formation of the pre-AC. Although all immunofluorescence experiments shown above included pIE1 expression as a control of infection at the cell population level, we consistently observed a degree of attenuation of the pIE1 signal by fluorescence microscopy. Thus, to clarify the observed effects of SNX27:Retromer:ESCPE-1 depletion, we performed a more thorough analysis of the progression of the MCMV replication cycle. We analyzed seven MCMV-encoded proteins and viral DNA replication to monitor various stages of the MCMV replication cycle (Lacaze et al., 2011; Marcinowski et al., 2012; Ružić et al., 2022; Mahmutefendić Lučin et al., 2023), as schematically illustrated in Figure 4A .

The percentage of cells expressing pIE1 and pE1, key indicators for entry into the E phase of infection, was similar in cells transfected with scr-siRNA ( Figures 4B , Supplementary Table S3 ) and untreated cells (Lučin et al., 2020). However, infection of SNX27-depleted cells consistently resulted in a small but significant decrease in the percentage of cells expressing pIE1 ( Figure 4B , Supplementary Table S3 ). Western blot analysis of pIE1 and pE1 also showed a consistent decrease in their expression in the E (6 hpi), late-E (16 hpi), and L (40 hpi) phases of infection ( Figures 4C , Supplementary Figure S7 , Supplementary Table S4 ). As these data suggest that SNX27-dependent functions are already required in the E phase of infection, we extended our analysis to all necessary components of SNX27:Retromer:ESCPE-1 complexes using well-established MCMV-GFP infection and flow cytometric quantification of the viral protein expression program (Angulo et al., 2000). This analysis confirmed a mild reduction of the fluorescence intensity and percentage of GFP-positive cells after SNX27-depletion ( Supplementary Figures S8A, B ), suggesting that SNX27-associated functions contribute to the control of virus replication prior to the establishment of the Rab10-associated part of pre-AC and that expression of E proteins that mediate membrane system reorganization and pre-AC development appears to be sufficient, as the Golgi displacement also occurs in SNX27-depleted cells ( Supplementary Figure S5 ).

As CMV gene expression kinetics can be divided into at least seven temporal classes (Rozman et al., 2022), we extended the analysis of the effects of SNX27 suppression on the expression of pm06 and pM57. These two MCMV proteins are expressed with delayed early kinetics after the first set of E genes at 2-3 and 5-6 hpi, respectively, suggesting two distinct temporal classes in MCMV-infected cells ( Figure 4A ). Immunofluorescence ( Figure 4B ) and Western blot analysis ( Figure 4C , Supplementary Figure S7, Supplementary Table S4 ) showed a substantial decrease in the expression of pm06 and pM57 in SNX27-depleted cells, already in the E phase of infection (6 hpi) as well as at later time points (16 and 40 hpi). Immunofluorescence analysis of pM57 showed a reduction in the number of expressing cells ( Figures 4B , Supplementary Table S3 ), associated with reduced accumulation of pM57 in nuclear pre-replication centers at 6 and 16 hpi ( Supplementary Figure S10 , arrowheads) and limited pM57 condensation in nuclear replication centers at 40 hpi ( Supplementary Figure S10 , arrows). The effect of SNX27 depletion on pM57, the ssDNA-binding protein involved in viral DNA synthesis (Strang et al., 2012) and viral gene transcription (Chang et al., 2011), may have an effect on viral DNA replication. Therefore, we next examined the effect of SNX27 depletion on viral DNA replication using 5-ethynyl-2-deoxyuridine (EdU) labeling (Strang et al., 2012; Mahmutefendić Lučin et al., 2023). This protocol allows visualization of replicated viral DNA in the nucleus of the infected cell at 16-24 hpi, as MCMV infection completely blocks host cell DNA replication in the E phase of infection at 6-12 hpi ( Figure 4A , Mahmutefendić Lučin et al., 2023). As expected, EdU labeling at 16-24 hpi visualized replicated viral DNA in 53.8 ± 8.78% of cells transfected with scr-siRNA ( Figures 4D , Supplementary Figure S10B ), which is comparable to untreated cells (Mahmutefendić Lučin et al., 2023). In SNX27-depleted cells, the number of cells showing an EdU-labeled signal in the nucleus was reduced to 33.8 ± 6.18% ( Figure 4D ). These data suggest that SNX27-associated functions contribute to the overall capacity of the cell to replicate viral DNA and, subsequently, the expression of viral L genes.

To assess the expression of L genes, we examined the expression of pM55 and pM74, two late glycoproteins that localize in the AC and form the virion envelope (Kattenhorn et al., 2004; Lučin et al., 2020), and pM116.1, an abundant L protein that localizes in the AC but is not integrated into the virion particles (Ružić et al., 2022). As in untreated cells (Lučin et al., 2020), pM74 perinuclear cytoplasmic staining was detected in 49.05 ± 14.8% of scr-siRNA-transfected cells at 40 hpi ( Figures 4B , Supplementary Figure S11 , Supplementary Table S3 ). In SNX27-depleted cells, the percentage of pM74-positive cells was reduced to 29.07 ± 5.84% expressing cells ( Figure 4B , Supplementary Table S3 ), resulting in the reduction of pM74 expression to 30% of the control level at 24 and 40 hpi ( Figures 4E , Supplementary Figure S7 , Supplementary Table S4 ). In addition to pM74, 55 kDa and 130 kDa forms of pM55 were reduced to approximately 15-30% of the control level after 40 hpi in SNX27-depleted cells ( Figures 4E , Supplementary Figure S7 , Supplementary Table S4 ). A similar observation was made for pM116.1, which was absent at 24 hpi and reduced to 45% of the control level at 40 hpi ( Figures 4E , Supplementary Figure S7 ; Supplementary Table S4 ). These data demonstrate that depletion of SNX27 reduces the expression of viral late proteins, including viral glycoproteins that form the virion envelope.

To distinguish whether the observed effects of SNX27 depletion are associated with its contribution to the SNX27:Retromer:ESCPE-1 complexes, we examined the effect of Vps35 and SNX1 + 2 depletion on the expression of pM74 in the L phase of infection. Western blot analysis showed that the expression of pM74 was significantly reduced in both Vps35- and SNX1+2-depleted cells at 40 hpi compared to scr-siRNA-treated cells ( Figures 4F ), suggesting that the observed effects on the expression of viral proteins are associated with the function of the SNX27:Retromer:ESCPE-1 complexes.

Inhibition of the Rab10-associated membranous events in the inner pre-AC, viral DNA replication, and expression of late genes in SNX27-depleted cells suggest that SNX27-associated processes are required for virion assembly and release. Therefore, we next investigated the production and release of infectious virions in SNX27-depleted cells. Nontransfected Balb3T3 cells and cells transfected with control or SNX27 siRNAs were infected with 1 PFU/cell and the amount of infectious viral particles was determined at 48 and 72 hpi using a standard plaque assay. Cells treated with scr-siRNA released similar amounts of infectious particles as untreated cells ( Figure 4G ). As expected, SNX27-depleted cells released 7- and 9-fold fewer virions after 48 and 72 hpi, respectively ( Figure 4G ). A similar pattern was observed when cellular lysates (i.e., the presence of intracellular virions) were analyzed ( Figure 4G ), suggesting that the lower release of virions is related to the reduced assembly capacity rather than the inability to release assembled virions. These data confirm that SNX27-associated processes are essential in virion biogenesis.

Downstream effector functions of the SNX27 are associated with the retrieval functions mediated by its PDZ and F3-FERM domains, including PDZ-mediated recruitment of the Retromer and F3-FERM-mediated recruitment of the SNX1 or SNX2 component of ESCPE-1 (Seaman, 2018). To elucidate the contribution of SNX27 domains, we monitored the expression and subcellular distribution of pM74 by immunofluorescence and confocal imaging in pEGFP-N1-mSNX27, pEGFP-N1-mSNX27 H112A, and pEGFP-N1-mSNX27 ΔF3 transfected and MCMV-infected cells. A ring-shaped perinuclear expression of pM74 ( Figure 5A , arrow, top row) was detected in 48.96 ± 5.69% wt-SNX27 transfected cells after 48 h of MCMV infection ( Figure 5B ), similar to previous studies with untreated cells (Lučin et al., 2020). EGFP SNX27 wt was concentrated within the pM74 ring, consistent with its accumulation in the inner AC ( Figure 5A ). Infection of cells overexpressing SNX27 with the H112A mutation resulted in a similar number of pM74-expressing cells (44.76 ± 6.81% of transfected cells; Figure 5B ), including a similar ring-shaped distribution of pM74 and accumulation of EGFP SNX27 H112A in the inner AC ( Figure 5A , middle row). However, overexpression of the SNX27 mutant lacking the F3-FERM subdomain almost completely prevented the expression of pM74 after MCMV infection, resulting in only 6.13 ± 5.54% pM74-positive transfected cells ( Figure 5A , bottom row, and 5B ). These data suggest that the mechanism driving MCMV gene expression does not rely on PDZ domain-mediated recruitment of signaling receptors but rather on SNX27:Retromer:ESCPE-1 complex formation, as the H112A mutation does not prevent association with Retromer. The complex formation also requires a functional F3-FERM domain that binds ESCPE-1, consistent with the effect of SNX1 + 2 depletion on pM74 expression shown above ( Figure 4F ).