FLT3-ITD transduces autonomous growth signals during its biosynthetic trafficking in acute myelogenous leukemia cells

FMS-like tyrosine kinase 3 (FLT3) in hematopoietic cells binds to its ligand at the plasma membrane (PM), then transduces growth signals. FLT3 gene alterations that lead the kinase to assume its permanently active form, such as internal tandem duplication (ITD) and D835Y substitution, are found in 30–40% of acute myelogenous leukemia (AML) patients. Thus, drugs for molecular targeting of FLT3 mutants have been developed for the treatment of AML. Several groups have reported that compared with wild-type FLT3 (FLT3-wt), FLT3 mutants are retained in organelles, resulting in low levels of PM localization of the receptor. However, the precise subcellular localization of mutant FLT3 remains unclear, and the relationship between oncogenic signaling and the mislocalization is not completely understood. In this study, we show that in cell lines established from leukemia patients, endogenous FLT3-ITD but not FLT3-wt clearly accumulates in the perinuclear region. Our co-immunofluorescence assays demonstrate that Golgi markers are co-localized with the perinuclear region, indicating that FLT3-ITD mainly localizes to the Golgi region in AML cells. FLT3-ITD biosynthetically traffics to the Golgi apparatus and remains there in a manner dependent on its tyrosine kinase activity. Tyrosine kinase inhibitors, such as quizartinib (AC220) and midostaurin (PKC412), markedly decrease FLT3-ITD retention and increase PM levels of the mutant. FLT3-ITD activates downstream in the endoplasmic reticulum (ER) and the Golgi apparatus during its biosynthetic trafficking. Results of our trafficking inhibitor treatment assays show that FLT3-ITD in the ER activates STAT5, whereas that in the Golgi can cause the activation of AKT and ERK. We provide evidence that FLT3-ITD signals from the early secretory compartments before reaching the PM in AML cells.

FLT3 is a member of the type III receptor type tyrosine kinase (RTK) family and is expressed in the PM of hematopoietic cells [1][2][3] . Upon stimulation with FLT3 ligand, the receptor undergoes dimerization and autophosphorylates its tyrosine residues, such as Tyr591 and Tyr842 [3][4][5] . Subsequently, it activates downstream molecules, such as AKT, extracellular signal-regulated kinase (ERK), and transcription factors 3,6 . Activation of these cascades results in the growth and differentiation of host cells, leading to normal hematopoiesis 2 . Therefore, gain-of-function mutations of FLT3 cause autonomous proliferation of myeloid cells, resulting in the development of AML 2,7,8 .
FLT3 is composed of an N-terminal extracellular domain, a transmembrane region, a juxta-membrane (JM) domain, and a C-terminal cytoplasmic tyrosine kinase domain 1,3,6 (see Fig. 1a). Alterations of the FLT3 gene that lead the kinase to constitutive activation are seen in 30-40% of AML cases 6,8 . Internal tandem duplication (ITD) into the JM region of FLT3 interferes with its auto-inhibitory ability 9 . In addition, a D835Y substitution in the FLT3 activation loop stabilizes the tyrosine kinase domain in an active state 1,10 . Thus, signal transduction pathways from FLT3 mutants have been investigated 6,[11][12][13] , and molecular targeting drugs for blocking the www.nature.com/scientificreports/ mutants have been developed for the treatment of AML patients 7,8,14 . Previous studies showed that FLT3-ITD accumulates in the wrong compartments, resulting in low amounts of the mutant in the PM, compared with the allocation of FLT3-wt 4,5,[15][16][17] . Although FLT3-ITD is suggested to activate signal transducers and activators of transcription 5 (STAT5) soon after synthesis 4,[18][19][20] , the precise subcellular localization of the mutant and the relationship between the mislocalization and growth signals remain unclear.
Recently, we reported that KIT, a type III RTK, accumulates in intracellular compartments, such as endosomal/lysosomal membrane and the Golgi apparatus, in mast cell leukemia (MCL), gastrointestinal stromal tumor (GIST), and AML [21][22][23][24] . Mutant KIT in leukemia localizes to endosome-lysosome compartments through endocytosis, whereas that in GIST stops in the Golgi region during early secretory trafficking. We further showed that blockade of KIT trafficking to the signal platform inhibits oncogenic signals [24][25][26] , suggesting that trafficking suppression is a novel strategy for suppression of tyrosine phosphorylation signals.
In this study, we show that endogenous FLT3-ITD aberrantly accumulates in the perinuclear region in AML cells. In our co-staining assays, we found that the perinuclear region was consistent with the Golgi region but not with the ER, endosomes, or lysosomes. The Golgi retention of FLT3-ITD is decreased by tyrosine kinase inhibitors, such as AC220 and PKC412, suggesting that the mutant stays in the Golgi region in a manner that is dependent on its kinase activity. Interestingly, FLT3-ITD can activate AKT and ERK in the Golgi region before reaching the PM. Inhibiting the biosynthetic trafficking of FLT3-ITD from the ER to the Golgi by brefeldin A  www.nature.com/scientificreports/ (BFA) or 2-methylcoprophilinamide (M-COPA) can block the activation of AKT and ERK by FLT3-ITD. We also confirmed that STAT5 is activated by FLT3-ITD in the ER. Our findings provide evidence that FLT3-ITD signaling occurs at intracellular compartments, such as the Golgi apparatus and ER, in AML cells.

FLT3-ITD but not FLT3-wt localizes to the perinuclear Golgi region in leukemia cells. Next,
we investigated the perinuclear region, where FLT3-ITD is found, by examining AML cell lines using co-staining assays. First, we immunostained for FLT3 (green) in conjunction with trans-Golgi network protein 46 kDa (TGN46, Golgi marker, red), Golgi matrix protein 130 kDa (GM130, Golgi marker, red), lectin-HPA (Golgi marker, blue), calnexin (ER marker, red), transferrin receptor (TfR, endosome marker, red), or lysosome-associated membrane protein 1 (LAMP1, lysosome marker, red) in MOLM-14 cells. As shown in Fig. 2a, perinuclear FLT3 was co-localized with the Golgi markers but not with the ER marker calnexin. Furthermore, localization of endosomal/lysosomal vesicles was inconsistent with that of perinuclear FLT3 (Fig. 2a), indicating that FLT3-ITD localizes to the Golgi region in MOLM-14 cells. Similar results were obtained from immunofluorescence assays using both MV4-11 and Kasumi-6 cells ( Fig. 2b; Suppl. Fig. 2a,b). In these cells, a fraction of FLT3 was found outside the ER region ( Fig. 2a; Suppl. Fig. 2a,b), indicating that the receptor could localize in PM of ITD-bearing cells. We were unable to find co-localization of FLT3-wt with Golgi markers, such as lectin-HPA and GM130, in RS4-11 cells (Fig. 2c), indicating that ITD leads FLT3 to mislocalize to the Golgi region in leukemia cells.

FLT3-ITD remains at the Golgi region in a manner dependent on its tyrosine kinase activity in AML cells.
Recently, we reported that constitutively active KIT mutants in MCL, GIST, or AML accumulate in organelles in a manner dependent on their tyrosine kinase activity 21,22,24 . Thus, we asked whether FLT3-ITD tyrosine kinase activity was required for retention of the mutant in the Golgi region. To answer this, we treated AML cells with quizartinib or midostaurin, small molecule TKIs (hereafter, referred to as AC220 and PKC412, respectively), which block the activation of To check the effect of AC220 or PKC412 on FLT3 localization, we immunostained permeabilized MOLM-14 cells with an anti-FLT3 antibody. Interestingly, TKI treatment markedly decreased the FLT3 level in the Golgi region (Fig. 3e). Conversely, we found that the treatment increased the level of FLT3, probably within the PM region ( Fig. 3e, see insets). Previous reports showed that a kinase-dead mutation of FLT3-ITD or TKIs (AC220/ crenolanib) enhance PM distribution of the mutant receptors 15,17,35,36 . Therefore, we examined the PM levels of FLT3-ITD on non-permeabilized MOLM-14 cells by staining for the FLT3 extracellular domain (ECD). As shown in Fig. 3f, treatment with AC220 or PKC412 enhanced the PM staining of FLT3, similar to previous reports on crenolanib-treated MV4-11 35 . Taken together, these results suggest that FLT3-ITD remains in the Golgi region during secretory trafficking in a manner dependent on its kinase activity and that TKIs move the mutant receptor to the PM.
In AML cells, FLT3-ITD can activate STAT5, ATK, and ERK in early secretory compartments. Finally, we examined the relationship between FLT3-ITD localization and growth signals. To determine whether FLT3-ITD activated downstream molecules before reaching the PM, we treated AML cells with BFA, M-COPA (blockers of ER export to the Golgi 24-26,37 ), or monensin (an inhibitor of secretory trafficking thorough blocking Golgi export 21,22,33,38,39 ). Previous studies showed that BFA, an inhibitor of ER export 37 , suppresses the activation of AKT and ERK but not STAT5 in MV4-11 cells 4 www.nature.com/scientificreports/ as well as BFA to confirm the effect of blockade of ER export on FLT3 signaling. Our immunofluorescence assay on MOLM-14 cells showed that BFA/M-COPA treatment decreased FLT3 levels in the Golgi region within 8 h and greatly increased the co-localization of calnexin (ER marker) with FLT3 (Fig. 4a, see inset panels), confirming that these inhibitors block biosynthetic protein transport from the ER to the Golgi apparatus. Next, we performed immunoblotting. As shown in Fig. 4b, in the presence of BFA/M-COPA, FLT3-ITD was retained in a high mannose form (see upper panels), confirming blockade of complex glycosylation in the Golgi apparatus. Consistent with a previous report 4 , pFLT3 Y842 decreased with BFA/M-COPA treatment, indicating that the phosphorylation does not occur in the ER. On the other hand, pFLT3 Y591 was maintained in the ER (Fig. 4b, left, bottom panel). In MV4-11 and Kasumi-6 as well as MOLM-14, FLT3-ITD in the ER was unable to activate AKT and ERK ( Fig. 4c; Suppl. Fig. 4a,b). Blockade of ER export, however, did not inhibit STAT5 activation through FLT3-ITD (Fig. 4b,c; Suppl. Fig. 4a,b). These results indicate that ER-retained FLT3-ITD activates STAT5 but not AKT and ERK. Next, we asked whether FLT3-ITD in the Golgi apparatus can activate AKT or ERK by using monensin, a blocker of Golgi export 21,22,33,38,39 . As shown in Fig. 4d, our immunofluorescence assay showed that FLT3 distribution other than in the Golgi region was markedly decreased in MOLM-14 cells in the presence of 100 nM monensin for 8 h, confirming the expectation that the treatment blocks Golgi export of FLT3-ITD. Only the high-mannose form of the FLT3 bands was found in the presence of monensin (Fig. 4e, top panel). In the presence of monensin, pFLT3 Y842 was decreased (Fig. 4e), whereas pFLT3 Y591 remained unchanged (Fig. 4e, bottom  panel), suggesting that the treatment does not block all tyrosine phosphorylations in FLT3-ITD and that these www.nature.com/scientificreports/ tyrosine residues in FLT3 are regulated differently. Blocking the PM localization of FLT3-ITD with monensin in MOLM-14 caused it to be partially decreased in pAKT and pERK. Although monensin lowered the protein level of STAT5, pSTAT5 remained. We found FLT3 signals in the presence of monensin for 24 h (Suppl. Fig. 4c). The results obtained with this cell line indicate that the activation of AKT and ERK via FLT3-ITD occurs not only at the PM but also in the Golgi and support our finding that STAT5 is activated before FLT3-ITD moves to the PM. Although blocking Golgi export of FLT3 with monensin altered levels of pAKT, pERK, and STAT5, these effects were various among MOLM-14, MV4-11, and Kasumi-6 ( Fig. 4e,f), and the persistence of this phosphorylation in the presence of monensin was common to all these AML cell lines. These results indicate that FLT3-ITD can activate downstream before it reaches the PM. Taken together, these results suggest that in AML cells, FLT3-ITD can activate STAT5 and AKT/ERK on the ER and the Golgi apparatus, respectively.

Discussion
In this study, we demonstrated that unlike FLT3-wt (Fig. 5, left), FLT3-ITD accumulates in early secretory organelles, such as the Golgi apparatus, and in that location, causes tyrosine phosphorylation signaling in leukemia cells (Fig. 5, right). The Golgi retention of FLT3-ITD is dependent on the tyrosine kinase activity of the mutant. TKI increases PM levels of FLT3-ITD by releasing the mutant from the Golgi apparatus. FLT3-ITD in the Golgi region can activate AKT and ERK, whereas that in the ER triggers STAT5 phosphorylation, leading to autonomous cell proliferation.
Recently, we reported that in MCL, KIT D816V (human) or KIT D814Y (mouse) activates STAT5 and AKT on the ER and endolysosomes, respectively 21,25 , whereas KIT V560G in MCL activates them at the Golgi apparatus 24 . Furthermore, KIT mutants including KIT D816V in cells other than MCL, such as GIST and blood cells, cause oncogenic signals on the Golgi apparatus 22,24,33 . As previously described 4,18,19 , we confirmed that after biosynthesis in the ER, FLT3-ITD causes STAT5 tyrosine phosphorylation in a manner similar to that of KIT D816V in MCL. On the other hand, activation of AKT and ERK through FLT3-ITD is similar to activation through the KIT mutant in GIST in that it occurs on the Golgi apparatus. A recent report showed that FLT3 D835Y is also found in endomembranes 41 . As described above, since the signal platform for a kinase may be affected by its mutation site, there is great interest in carrying out a further investigation to determine whether FLT3 D835Y causes growth signaling on the ER, Golgi, or endosome/lysosomes.
Recently, novel protein interactions and downstream molecules for FLT3 which support cancer cell proliferation have been identified. One FLT3 mutant was found to activate Rho kinase through activation of RhoA small GTPase, resulting in myeloproliferative disease development 42 . GADS physically associates with FLT3-ITD, and the interaction enhances downstream activation 12 . Analysis of spatio-temporal associations of FLT3 mutants with these functional interactors is an attractive possibility.
Previous studies showed that other RTK mutants, such as FGFR3 K650E in multiple myeloma, RET multiple endocrine neoplasia type 2B (RET MEN2B ), and PDGFRA Y289C , are also tyrosine-phosphorylated via a secretory pathway 39,[43][44][45][46][47][48][49] . Signal transduction from the secretory compartments may be a characteristic feature of a large number of RTK mutants. Early secretory compartments can be subdivided into the ER, the ER-Golgi intermediate compartment, cis-, medial-Golgi cisternae, TGN, and others. It would be interesting to identify the sub-compartment at which RTK mutants are retained for precise understanding of the mechanism of growth signaling. Three-dimensional super-resolution confocal microscopic analysis of cancer cells is now under way.
Golgi retention of FLT3-ITD is dependent on receptor tyrosine kinase activity. As with previous reports 17, 36 , we confirmed that a TKI increased PM localization of FLT3-ITD, indicating that these inhibitors can release the mutant from the Golgi region for localization to the PM. Other reports together with the results of our studies showed that TKIs increase the PM levels of RTK mutants, such as EGFR(T790M), KIT(D816V), and PDGFRA(V 561G) 21,24,[50][51][52][53] . Enhancement of PM distribution with TKIs may be a common feature of RTK mutants. Furthermore, recent studies showed that the effect of chimeric antigen receptor T-cell therapy and antigen-dependent cell cytotoxicity using anti-FLT3 is enhanced by increasing the PM levels of FLT3-ITD through TKI treatment 30,35,36,53 . Combining TKIs together with immunotherapy will lead to improvements in the prognosis of cancer patients.
TKIs and antibodies against RTKs have been developed for suppression of growth signals in cancers. In this study, blockade of the ER export of FLT3-ITD with BFA/M-COPA greatly reduced tyrosine phosphorylation signals in AML cells. Since the bioavailability of M-COPA in vivo is higher than that of BFA and because it can be orally administered to animals, we will investigate the anti-cancer effect of the compound on AML-bearing mice. Together with the results of previous reports [24][25][26][54][55][56][57][58] , our findings suggest that an intracellular trafficking blockade of RTK mutants could be a third strategy for inhibition of oncogenic signaling.
In conclusion, we showed that in AML cells, the perinuclear region where FLT3-ITD accumulates is the Golgi apparatus. Similar to KIT mutants in GISTs, FLT3-ITD is retained at the Golgi region in a manner dependent on its kinase activity, but TKI releases the mutant to the PM. Our findings provide new insights into the role of FLT3-ITD in autonomous AML cell growth. Moreover, from a clinical point of view, our findings offer a new strategy for AML treatment through blocking the involvement of FLT3-ITD in secretory trafficking.

Materials and methods
Cell culture. RS4 Cell proliferation assay. Cells were cultured with PKC412 or AC220 for 48 h. Cell proliferation was quantified using the CellTiter-GLO Luminescent Cell Viability Assay (Promega, Madison, WI), according to the  Analysis of protein glycosylation. Following the manufacturer's instructions (New England Biolabs, Ipswich, MA), NP-40 cell lysates were treated with endoglycosidases for 1 h at 37 °C. Since the FLT3 expression level in THP-1 was low for this assay, FLT3 was concentrated by immunoprecipitation with anti-FLT3 (S-18), and then treated with endoglycosidases. The reactions were stopped with SDS-PAGE sample buffer, and products were resolved by SDS-PAGE and immunoblotted.

Data availability
All datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.