Growth Factors do not regulate Golgi Complex-to-ER relocation of GalNAc-Ts in HeLa cells

Mucin-type O-glycosylation is initiated by the UDP-GalNAc polypeptide:N-acetylgalactosaminyltransferase (GalNAc-T) family of enzymes. Their activity results in the GalNAc α1-O-Thr/Ser structure, termed the Tn antigen, which is further decorated with additional sugars. In neoplastic cells, the Tn antigen is often overexpressed. Because O-glycosylation is controlled by the activity of GalNAc-Ts, their regulation is of great interest. Previous reports suggest that growth factors, EGF or PDGF, induce Golgi complex-to-endoplasmic reticulum (ER) relocation of both GalNAc-Ts and Tn antigen in HeLa cells, offering a mechanism for Tn antigen overexpression termed “GALA”. However, we were unable to reproduce these findings. Upon treatment of HeLa cells with either EGF or PDGF we observed no change in the co-localization of endogenous GalNAc-T1, GalNAc-T2 or Tn antigen with the Golgi complex marker TGN46. There was also no enhancement of localization with the ER marker calnexin. We conclude that growth factors do not cause redistribution of GalNAc-Ts from the Golgi complex to the ER in HeLa cells.


Introduction
Mucin-type O-glycans decorate a wide range of secreted and transmembrane proteins. These molecules play a diverse set of physiological roles including structural roles within the extracellular matrix (Zhang et al., 2010), the function of the innate immune system (Tabak, 1995), and involvement in key signal transduction pathways (Kufe, 2013). In common with other secreted and trans-membrane proteins, the polypeptide core of mucins is synthesized in the endoplasmic reticulum (ER) and then transported to the Golgi complex via the ER-Golgi complex intermediate compartment (ERGIC). A family of enzymes, termed the UDP-GalNAc polypeptide:N acetylgalactosaminyltransferases (GalNAc-Ts, EC 2.4.1.41) initiates the addition of the first sugar in O-glycan biosynthesis in the Golgi complex by catalysing transfer of Nacetylgalactosamine (GalNAc) from the sugar donor UDP-GalNAc to the hydroxyl group of Thr/Ser residues of the protein core (Bennett et al., 2012). The resultant structure GalNAc a 1-O-Thr/Ser is termed "Tn antigen". Normally, additional sugars are added stepwise, resulting in an extraordinary array of diverse O-glycan structures (Bennett et al., 2012). In contrast, Tn antigen is often expressed in neoplastic cells (Brockhausen, 2006;Itzkowitz et al., 1991;Konno et al., 2002;Konska et al., 2006;Laack et al., 2002;Springer, 1984), and is thought to contribute to their "invasiveness" (Radhakrishnan et al., 2014).
Mucin-type O-glycans are often arrayed in clusters due to variably glycosylated repeating Thr/Ser-containing sequences that form the sites of sugar attachment. Further, recent Oglycoproteomic studies (e.g. Steentoft et al., 2013) reveal that many non-mucinous secretory proteins are O-glycosylated, often with isolated glycosites, and these sites may have profound effect on protein functions. Thus, the acquisition, pattern, and density of O-glycans may be regulated by both the substrate specificity and the spatiotemporal expression of the multiple GalNAc-Ts. Detailed studies of GalNAc-T substrate specificity point to a requirement for either unique GalNAc-Ts, or multiple GalNAc-Ts working in concert, to ensure that the acquisition of O-glycans is complete (Bennett et al., 2012;Gerken et al., 2013;Revoredo et al., 2016;. Gill et al. (2010) reported that either EGF or PDGF activation of the Src-kinase pathway stimulates O-glycosylation initiation in HeLa cells. Their model posits that recruitment of active Src family kinases (SFKs) to the Golgi complex modulates the formation of transport carriers formed by the action of the GTPases, Arf1 and its effector the COPI complex, the molecular machinery known to control retrograde transport of Golgi complex resident proteins to the ER (Spang, 2013). Using Helix pomatia (HPA) lectin to detect Tn antigen, Gill et al. (2010) observed progressive re-distribution of HPA staining from the Golgi complex to the surrounding cytoplasm following growth factor treatment. To account for these findings, they suggested that GalNAc-Ts (but not other glycosyltransferases) located in the Golgi complex are uniquely sorted into retrograde carriers that transport these enzymes to ER resulting in higher levels of the Tn antigen in the ER of cells stimulated with growth factors compared to control cells. They term this novel mechanism for Tn antigen overexpression "GALA". This model has important implications for the regulation of mucin biosynthesis. If O-glycan acquisition were initiated in the ER due to selective re-direction of GalNAc-Ts, this could result in a more densely glycosylated Tn antigen since the remaining O-glycosylation machinery is not available until the mucin traffics to the Golgi complex. O-glycan density underlies many biological roles including enhancement of protein resistance to proteolysis (van der Post et al., 2013;Zhang et al., 2014), as well as imparting important structural properties of some surface receptors (Jentoft, 1990;Ostuni et al., 2014;Paszek et al., 2014), and the selectivity with which the mucin-type protein interacts with lectins (Dam et al., 2007;Kato et al., 2008). Moreover, in vitro studies suggest that the presence of Tn antigen inhibits extension of neighbouring glycans by the relevant glycosyltransferases (Brockhausen et al., 2009). Moreover, it is known that ER localized O-mannosylation influences the subsequent addition of GalNAc by GalNAc-Ts (Tran et al., 2012); if GalNAc-Ts re-localize to the ER, it is plausible that this could influence the acquisition of O-mannose residues. Therefore, if GalNAc-Ts re-locate to the ER, this would offer an important mechanism to alter the sugar coat of cell surfaces and their functions. Given the potential implications suggested by Gill et al. (2010) we attempted to reproduce several of the key observations published by Gill and colleagues (Gill et al., 2010) as a starting point to conducting studies that would build upon these findings. However, when we performed colocalization studies with experimental approaches similar to those reported by Gill et al., 2010, and additional ones using standard ER and Golgi complex markers in combination with antibodies to GalNAc-T1 and -T2, in EGF-or PDGF-stimulated HeLa cells (Gill et al., 2010;Gill et al., 2013), we failed to obtain any evidence for changes in the normal localization of endogenous GalNAc-T1 and GalNAc-T2. Rather, GalNAc-Ts remain predominately localized to the Golgi complex. Moreover, we did not detect changes in either the location of HPA-lectin reactive materials. Taken together, our experiments lead us to conclude that in HeLa cells, activation by the growth factors EGF and PDGF does not stimulate O-glycosylation initiation of mucins via the relocation of GalNAc-T from the Golgi complex to the ER.

T2
To determine if growth factor treatment effects the localization of endogenous GalNAc-T1 and -T2 we treated HeLa CCL-2 cells (acquired from ATCC) with EGF or PDGF, using conditions previously reported as sufficient to cause redistribution of GalNAc-T from the Golgi complex (Gill et al., 2010). GalNAc-T1 and -T2 are the most abundant GalNAc-T transcripts expressed in HeLa cells and thus represented a logical choice for these studies (  (Dunn et al., 2011;Manders et al., 1993;McDonald and Dunn, 2013) of the GalNAc-T/TGN46 and GalNAc-T/calnexin staining, respectively (Fig. 1C). We found a strong coincidence of endogenous GalNAc-T1 or -T2 with TGN46 in both control, serum starved, cells and those stimulated with EGF or PDGF with a MCC of 0.9 ± 0.1 . In contrast, the MCC between either GalNAc-T1 or -T2 and calnexin was 0.4 ± 0.1 under all conditions tested. Together, our data demonstrates that EGF and PDGF treatment of serum starved HeLa cells does not alter the Golgi complex localization of GalNAc-T1 or -T2.

Proteins containing Tn antigen remain localized within the Golgi complex after growth factor stimulation
Because our growth factor treatment did not result in a change in GalNAc-T localization we next determined if EGF or PDGF treatment of serum starved HeLa cells resulted in a change in Tn antigen localization as was also reported by Gill et al. 2010. We similarly visualized the Tn antigen using a fluorescently labelled HPA (also referred to as HPL) lectin but additionally costained for both a Golgi complex (TGN46) and an ER (calnexin) marker. After treatment of serum starved cells with either EGF for 4 hours ( Fig Golgi complex or ER we calculated the MCC of HPA-reactive material/TGN46 or HPA-reactive material/calnexin, respectively. We found a strong coincidence of HPA-reactive material with the Golgi complex with a MCC of 0.8 ± 0.1 for untreated and cells treated with EGF or PDGF, while the coincidence of HPA-reactive material with ER was low with a MCC of 0.3 ± 0.1, that did not change after growth factors treatment (Fig. 2B). Together, our data indicate that EGF or PDGF treatment of serum starved HeLa cells has no effect on the Golgi complex localization of proteins carrying the Tn antigen.

Growth factor treatment results in rapid signalling in HeLa cells
Our results indicated that EGF or PDGF treatment of serum starved cells does not result in any changes in GalNAc-T or Tn antigen localization away from the Golgi complex ( Fig. 1 and 2).
Because these results were contrary to those reported by Gill et al. (2010), we sought to confirm that our growth factor treatments were in fact eliciting the appropriate cellular responses in HeLa cells. To this end we performed a time course of EGF or PDGF treatment following serum starvation of HeLa cells. Cell extracts were analysed by Western blot using antibodies specific to phosphorylated P44/P42 MAPK as an indicator of receptor tyrosine kinase activation and subsequent downstream signalling (Katz et al., 2007;Murphy and Blenis, 2006;Yoon and Seger, 2006). In serum starved cells treated with either EGF or PDGF, peak detection of phosphorylated P44/P42 occurred 10 minutes after treatment and then decreased over the remainder of the time course (Fig. 3A, 3B). These results are consistent with previously reported time frames for MAPK pathway activation (Gao et al., 2005;Pierce et al., 2001).
Our initial growth factor treatments were performed using conditions reported by Gill et al. (2010). Because 4 hour EGF or 3 hour PDGF treatment yielded no change in GalNAc-T Golgi complex localization (Fig 1. and 2) and the fact that there was little detectable MAPK activation at these time points (Fig 3A, 3B), we next determined if GalNAc-T or Tn antigen localization changed after 10 minutes of growth factor treatment, when the highest level of phosphorylated P44/P42 is detected (Fig. 3A, 3B). When serum starved cells were left untreated or treated with either EGF or PDGF for 10 minutes, GalNAc-T1, -T2 (

Discussion
Here, we demonstrate that growth factor treatment of serum starved HeLa cells causes no change in the Golgi complex localization of endogenous GalNAc-T1 and -T2, the predominant isoforms expressed in these cells, or Tn antigen, the immediate product of GalNAc-T enzymatic activity. These findings are in direct contrast to what was previously reported in the literature (Gill et al., 2010;Gill et al., 2011). Gill et al. utilized  tested. When comparing two entities, this approach relays the fraction of one channel that colocalizes with the second channel (Dunn et al., 2011;Manders et al., 1993;McDonald and Dunn, 2013). By this measure we confirmed that there was a strong degree of coincidence between both endogenous GalNAc-Ts and Tn antigen with TGN46 and a low degree of coincidence with calnexin under all conditions tested, demonstrating that growth factor treatment did not result in a change in the Golgi complex localization of either GalNAc-Ts or Tn antigen. Gill et al. (2010) did not report a quantitative measure of EGF or PDGF effect after addition to HeLa cells, except for the immunolocalization of GalNac-T1, HPA or Tn. We therefore cannot rule out the possibility that the effect of EGF and PDGF treatment is temporally different in the cells they employed relative to the HeLa CCL-2, cells obtained from the ATCC. However, to partially address this possibility we also performed our immunofluorescence experiments after 10 minutes of EGF or PDGF treatment, when we detect the highest levels of phosphorylated P44/P42. Our results indicate that regardless of when we measure the localization of GalNAc-Ts they are predominantly found co-localized with TGN46 and thus within the Golgi complex.
We are unable to rule out the possibility that there is an inherent difference between the HeLa cells used by Gill et al. (2010) and those purchased from ATCC. We did initial studies with one sample of cells graciously provided by Dr. F. Bard, and did not find any difference from the results reported here (data not shown). We subsequently learned that a second line of HeLa cells was also used by the Bard lab. At a minimum, our results indicate that the observations termed "GALA", reported by Gill et al. are not generalizable to all HeLa cells. Thus, care should be taken when exploring the possibility that this reported phenomenon is a global mechanism for regulating mucin-type protein O-glycosylation.

Immunofluorescence
For experiments in which un-transfected HeLa cells were used to quantify changes in the intracellular distribution of endogenous GalNAc-T1, T2 or HPA-labeled species, cells seeded at the desired density onto glass cover slips were grown in DMEM containing 10% FBS. Cells were washed twice with phosphate buffered saline (PBS) and left incubating in serum-free media for at least 16 h. Cells were washed once more and later incubated with or without DMEM containing 0.1% bovine serum albumin (BSA) and human recombinant EGF (100 ng/ml; Sigma Aldrich) or PDGF-bb (50 ng/ml; ThermoFisher Scientific). Cells were washed, fixed with 4% paraformaldehyde (PFA) in PBS for 10 min, washed with PBS, and later incubated in quench solution (NH4Cl 75 mM, glycine 20 mM) containing 0.1% Tx-100 for 10 min. Cells were blocked in 0.2 % BSA/fish skin gelatin (7mg/ml)/0.05% saponin in PBS for 30 min. Primary mouse antibodies against GalNAc-T1 and T2 were used non-diluted, complemented with 0.5% BSA and with antibodies against calnexin and TGN46 overnight at room temperature. For HPA labelling, cells were incubated with antibodies against calnexin and TGN46 in the blocking solution. Cells were then incubated with the respective secondary antibodies with or without HPA-conjugated to Alexa-488 (5 µg/ml) for 2 h at 37˚C. Cells were washed and later mounted with ProLong Gold (ThermoFisher Scientific) reagent and cured overnight before imaging.

Microscopy and image quantification analyses
Images shown in all figures were obtained with a Nikon A1R+ confocal microscope (Nikon Instruments Inc.) equipped with the following solid state lasers and filters: 488 nm (525/50), 561 nm (600/50) and 640 nm (650LP), using a 60x Plan Apochromat oil objective (NA 1.4) with an image pixel size of 70 nm and a pinhole size of 0.7 airy unit (AU). Image acquisition was performed at room temperature using the NIS-Element imaging software (Nikon Instrument Inc.).
Comparative co-localization analysis of HPA-reactive material or endogenous GalNAc-T1 and GalNAc-T2 with ER marker (calnexin) and Golgi complex marker (TGN46) of control and growth factor-stimulated cells was performed using Imaris (Bitplane, Oxford Instruments, Concord, MA). For each image, a mask is created using the HPA or GalNAc-Ts channel, and then an automatic threshold is applied for each marker channels. The Manders' coefficient A, corresponding to the proportion of intensity in channel A (ER or Golgi complex) coincident with intensity in channel B (HPA or GalNAc-Ts), is then used to quantify the co-localization of both markers with HPA or GalNAc-Ts.

Western blot analysis
HeLa cells were plated on 100mm dishes and allowed to recover for ~18 hours. Cells were then washed with PBS and incubated in serum-free DMEM for 16 h. Cells were then treated with DMEM containing 0.1% BSA and human recombinant EGF (100 ng/ml; Sigma Aldrich) or PDGF-BB (50 ng/ml; ThermoFisher Scientific) for 10 min, 30 min, 1 hours, 2 hours, 3 hours or 4 hours. After the indicated times, cells were trypsinized and collected by centrifugation.

RT-qPCR
Total RNA were extracted from HeLa ATCC cells using the RNeasy micro kit and the QIAshredder spin column (Qiagen, Valencia, CA) for homogenization, following manufacturer protocol. Then 1mg of RNA was used to produce cDNA using iScript™ Reverse Transcription Supermix (Biorad, Hercules, CA). RT-qPCR was performed using Taqman® Gene expression assay (ThermoFisher Scientific, Waltham, MA) following manufacturer procedure. The PCR was performed using TaqMan®      and Tn antigen found in Golgi complex or ER after 10 minutes of growth factor treatments.
Manders' correlation coefficient was calculated on the data set shown in Fig. S2 and Fig. S3.
Values represent the mean ± S.D.