Osteoclast proton pump regulator Atp6v1c1 enhances breast cancer growth by activating the mTORC1 pathway and bone metastasis by increasing V-ATPase activity

It is known that V-ATPases (vacuolar H+-ATPase) are involved in breast cancer growth and metastasis. Part of this action is similar to their role in osteoclasts, where they’re involved in extracellular acidification and matrix destruction; however, the roles of their subunits in cancer cell proliferation, signaling, and other pro-tumor actions are not well established. Analysis of TCGA data shows that V-ATPase subunit Atp6v1c1 is overexpressed or amplified in 34% of human breast cancer cases, with a 2-fold decrease in survival at 12 years. Whereas other subunits, such as Atp6v1c2 and Atp6v0a3, are overexpressed or genomically amplified less often, 6% each respectively, and have less impact on survival. Experiments show that lentiviral-shRNA mediated ATP6v1c1 knockdown in 4T1 mouse mammary cancer cells significantly reduces orthotopic and intraosseous tumor growth. ATP6v1c1 knockdown also significantly reduces tumor stimulated bone resorption through osteoclastogenesis at the bone and metastasis in vivo, as well as V-ATPase activity, proliferation, and mTORC1 activation in vitro. To generalize the effects of ATP6v1c1 knockdown on proliferation and mTORC1 activation we used human cancer cell lines - MCF-7, MDA-MB-231, and MDA-MB-435s. ATP6V1C1 knockdown reduced cell proliferation and impaired mTORC1 pathway activation in cancer cells but not in the untransformed cell line C3H10T1/2. Our study reveals that V-ATPase activity may be mediated through mTORC1 and that ATP6v1c1 can be knocked down to block both V-ATPase and mTORC1 activity.


INTRODUCTION
The multi-subunit V-ATPase complex is composed of an ATP-hydrolytic domain (V1) and a membrane spanning proton-translocation domain (V0), along with two accessory subunits ac45 and M8-9 [1]. There are eight different subunits (A-H) in the 640kDa cytoplasmic domain, V1, and the V0 domain is an integral membrane-bound domain consisting of a, c, c", and d subunits in mammals [2][3][4][5]. The V-ATPase is a tightly coupled enzyme that only exhibits activity when it is fully assembled with the v1c subunit being the rate limiting component involved in the reversible dissociation of the V0 and V1 domains [6][7][8][9]. The V-ATPases have long been known to have roles in cancer [10][11][12][13], in addition to their originally characterized functions in acidifying lysosomal vacuoles. For instance, they are known to activate acid dependent lysosomal proteases and known for their role in osteoclast mediated www.impactjournals.com/oncotarget/ Oncotarget, 2017, Vol. 8, (No. 29), pp: 47675-47690

Research Paper
bone resorption, as well as a role that has been suggested to be similar to the formation of invadopodia of invasive tumor cells [9,14,15]. However, it remains unclear which subunits may be dysregulated in which cancers and which, if any, could be targeted for cancer therapy [9].
It has recently been shown that Atp6v1c1 (C1), an isoform of the v1c subunit, was the most strongly overexpressed V-ATPase subunit in metastatic oral squamous cell carcinoma suggesting that V-ATPase activity, and specifically ATP6v1c1, indeed facilitates tumor progression and metastasis [8,13,16,17]. However, little is known about the exact roles of C1 in cancer, even though a number of roles are known for the V-ATPase complexes in cancer including involvement in cell signaling, apoptosis resistance, drug resistance, and metastasis [9]. In recent studies, C1 knockdown by shRNA in 4T1 cells significantly inhibited breast cancer cell proliferation in vitro and in vivo with decreased metastasis to the lungs, liver, and bone, while a less efficient C1 knockdown of 60% in 4T1 cells showed a dose-response effect on tumor growth. These results provide some initial evidence for the importance of the roles of V-ATPases in mammary tumor development and progression to a more metastatic phenotype [9,18]. Similar evidence has been shown for the role of ATP6v0a3, where it is known to be localized to the plasma membrane. It is involved in metastasis through extracellular acidification, in a process reminiscent of osteoclast mediated bone resorption, as well as survival through internal pH homeostasis [19,20], indicating that further research to assess the roles of the various V-ATPase subunits is in order to determine which subunits are most responsible for these effects.
Accumulating evidence shows that V-ATPase inhibitors decreased the invasion and migration of highly metastatic cells through multiple means [9,21] including the secretion of H + , which allows tumor cells to survive in hypoxic conditions and in their obligatory, glycolysisinduced, acidic tumor microenvironment, thereby playing a major role in tumor growth and metastasis [11,22]. However, many classic V-ATPase inhibitors (e.g., bafilomycins) are somewhat non-specific and using them often results in the development of tumor tolerance [21,23,24]. Therefore, defining the exact mechanisms of V-ATPases and their subunits in breast cancer cell growth and metastasis is also very important for V-ATPase targeting drug development and it may reveal novel and specific drug candidates for overcoming V-ATPase targeted drug resistance [9]. For instance, Zoncu et al. have reported that V-ATPase function is required for an inside-out signaling mechanism that allows multiple lysosomal amino acids to activate mTORC1, a known target in cancer [25][26][27], indicating that this function of V-ATPase may be targeted for therapy [28].
The mechanistic target of rapamycin (mTOR) (originally "mammalian TOR," but now officially "mechanistic TOR" [29]) is a highly conserved serine/ threonine kinase that participates in at least two distinct multiprotein complexes, mTOR complex 1 (mTORC1) [30,31] and mTOR complex 2 (mTORC2) [32,33]. Compared to mTORC2, which has been shown to be an important regulator of the cytoskeleton [33], mTORC1 is characterized by the classic features of mTOR as a nutrient/energy/redox sensor [31,34]. Dysregulation of the mTOR pathway occurs in many human diseases, especially certain cancers such as breast cancer, where it is a known therapeutic target [26,35,36]. Recently, it has been found that mTORC1 can sense lysosomal amino acids through an "inside-out" mechanism that requires the V-ATPase [28]. These findings suggest that V-ATPases may be a potential target for attenuating the mTORC1 pathway dysfunction in cancer, in addition to being a therapeutic target in their own right [9]. Therefore, in this paper, we look into the role of ATP6v1c1 in tumor growth, and metastasis, as well as its role in mTOR signaling in both human and murine cancer cell lines to determine whether its knockdown can inhibit tumor growth.

Bioinformatic analysis of TCGA patient data indicates an important role for ATP6v1c1 in breast cancer clinical outcomes
In order to provide an initial assessment of the dysregulation of ATP6v1c1 we examined TCGA data on its expression and amplification, which we used as a proxy to indicate the potential for a clinically relevant role for those subunits and their dysregulation in human breast cancer, as through oncogene addiction [37]. We also examined the relationship between ATP6v1c1 dysregulation and other prognostic measures like survival time, time to metastasis, and time to relapse. First, we determined whether ATP6v1c1 was amplified or otherwise altered in patient tumors, using cBioportal, where we found that among 963 cases with gene sequencing data from the TCGA, 17.2% (163 of 963) of the tumors had an ATP6v1c1 gene amplification, while one tumor of the 963 had a gene deletion, and 2 had gene mutations; indicating that ATP6v1c1 gene amplification may be adaptive for breast tumors. Then we looked at the expression of ATP6v1c1 and found that 27% (260/963) of the tumors had gene overexpression relative to control tissue and 33.4% (322/963) of the tumors had either an amplification or overexpression of ATP6v1c1 gene or both ( Figure 1A) [38,39]. To further assess the potential for clinical relevance we examined whether there was a difference in clinical outcomes in patients segregated by amplification or overexpression of ATP6v1c1 and found that patients whose tumors had ATP6v1c1 overexpression or duplication had reduced survival time ( Figure 1B). Further, in separate studies that included relapse and metastasis data, there was a decrease in the time to relapse ( Figure 1C) or metastasis ( Figure 1D) for patients with tumors with greater than median ATP6v1c1 expression [40][41][42]. For comparison, other V-ATPase subunits, ATP6v1c2 or ATP6v0a3/TCIRG1, are only overexpressed or genomically amplified in 7% of breast cancers, respectively, and their overexpression or genomic amplification didn't significantly correspond with effects on patient survival (Supplementary Figure 4).

Breast cancer metastasis and growth is reduced by ATP6v1c1 depletion
We have shown that ATP6v1c1 has effects on tumor growth and metastasis with ATP6v1c1 knockdown reducing tumor growth and metastasis in vivo (Figure 2A-2C) [18]. We found that metastasis was reduced when assessed using multiple means, including bioluminescence imaging in luciferase labeled 4T1 cells ( Figure 2A) and epifluorescence in 4T1 cells expressing GFP ( Figure  2B, as quantified in Figure 2C) but we had not looked at all of the aspects of these effects; namely, whether these effects were purely a result of enhanced tumor invasion or an effect of the growth of tumors. We assessed the effects of ATP6v1c1 on local acidification of the environment by tumor cells, a known factor in breast cancer metastasis [19,[44][45][46], using acridine orange and found that knockdown blocked the local acidification of the cancer cell microenvironment ( Figure 2D-2E); a result in agreement with prior results in osteoclasts knocked down for ATP6v1c1 [18,43] and indicative of a loss of metastatic potential, where extracellular acidification has been previously shown to be involved in cancer metastatic potential [19,[44][45][46]. Given this direct effect on cancer cell acidification we performed a growth assay in the bone to see whether the difference in bone loss previously observed was primarily a function of tumor cell metastasis and growth at the bone, or whether the effect was a function of tumor cell signaling-inducing osteoclast formation to drive bone degradation.

Atp6v1c1-depleted 4T1 cells in the left femur
In breast cancer, bone metastasis is primarily osteolytic due to excessive osteoclastic activity despite the presence of a secondary increase in local bone formation [47]. Observing that the 4T1-c1s3-1 mice (90% C1 knockdown in 4T1 cells) have no bone metastasis and the 4T1-c1s3-2 mice (60% C1 knockdown in 4T1 cells) have significantly less bone metastasis [18], we sought to further define whether C1 deficiency has an effect on 4T1 cell osteolytic erosion in vivo when a metastasis develops in the bone-versus simply preventing bone metastasis to begin withby developing a local femur xenograft model. We inoculated into the bone and 16 days later harvested the samples. X-ray images showed that the left femur of 4T1 and 4T1-LacZ injected mice suffered severe osteolytic lesions, while the femurs in 4T1-c1s3-2 injected mice had less osteolysis, and 4T1-c1s3-1 injected mice, similar to the PBS mice, had no visible osteolytic lesions ( Figure 3A). 3D images of the whole left femur by micro-CT showed similar results with a loss of osteolysis in tumors formed with ATP6v1c1 knockdown cells ( Figure 3B). More detailed views of the distal end of these femurs had corresponding results as in Figure  3B, with the exception of the observation of a small lesion in the 4T1-c1s3-1 cell group when compared to PBS. H&E staining confirmed the Micro-CT results and further showed that tumor growth in the bone was decreased in the ATP6v1c1 knockdown groups ( Figure  3D). In addition, TRAP staining revealed a potential explanation for this phenomenon with 60% fewer osteoclasts in the osteolytic lesion of 4T1-c1s3-1 mice compared to 4T1 and 4T1-LacZ mice ( Figure 3C-3D). There were also fewer osteoclasts in the osteolytic lesion region of 4T1-c1s3-2 mice compared to 4T1-LacZ mice, but more than in the 4T1-c1s3-1 mice ( Figure 3C-3D). Together, these data indicate that, compared to the 4T1- LacZ control, C1 knockdown of 85% can significantly reduce the osteolytic lesions caused by 4T1 breast cancer, possibly due to the significant reduction in OCs. It also indicates that there is a corresponding, doseresponse-like reduction in osteoclasts and osteolysis in the 65% knocked down 4T1-c1s3-2.
Atp6v1c1 knockdown in the 4T1 mouse mammary cancer cell line inhibits mTORC1 pathway activation stimulated by amino acids mTORC1 is known to play important roles in the regulation of cell growth, cell proliferation and cell motility (2,6,7), and V-ATPases are necessary for mTORC1 activation stimulated by amino acids [28]. Thus, we sought to determine whether Atp6v1c1, as a component of the V-ATPase complex, is also required for mTORC1 activation stimulated by amino acids in breast cancer cells. It has been well characterized that the phosphorylation of p70 S6 Kinase (S6K or p70S6K), a critical mediator of cell growth in mammalian cells, is a downstream target of mTORC1 activity [28,48] where we used it as a proxy for mTORC1 activation.
In addition, mTORC1 translocation to the (LAMP-1+) lysosomal surface is a key event in mTORC1 activation which is known to require V-ATPase activity [28]. Therefore, we tested co-localization of mTOR and LAMP-1+ lysosomes and found it reduced considerably in response to amino acids in 4T1 cells with ATP6v1c1 knockdown compared to that in the control cells, indicating a loss of mTOR signaling with C1 knockdown ( Figure 4A-4D and Supplementary Figure 2). Consequently, we tested the Thr389 phosphorylation of p70S6K to confirm the role of ATP6v1c1 in mTOR signaling in transformed cells. We found that C1 knockdown in the 4T1 cell line significantly inhibited the phosphorylation of p70S6K stimulated by amino acids but had little effect on the phosphorylation of AKT and p44/p42 MAPK (ERK1/2) with amino acid stimulation ( Figure 4E-4F). Together, these results show that ATP6v1c1 knockdown in 4T1 cells inhibits activation of the mTOR pathway in response to amino acid stimulation and suggest that C1 knockdown's inhibition of 4T1 cell growth, migration and invasion may be related to C1's function in mTORC1 pathway.

ATP6V1C1 is expressed in human breast cancer cells and knockdown inhibits human breast cancer cell proliferation and mTORC1 pathway activation stimulated by amino acids
Our data in the mouse breast cancer cell line 4T1 suggests that Atp6v1c1 may facilitate breast cancer growth and metastasis through the mTORC1 pathway, which prompted us to further determine whether ATP6V1C1 has a similar effect on the proliferation of, and the amino acid-sensitive mTORC1 activation in, other human breast cancer cell lines. We used a variety of breast cancer cells to determine the extent of this effect, including MCF-7 cells to model ER/PR double positive cells with low metastatic potential, and MDA-MB-231 and MDA-MB-435s as more aggressive and metastatic triple negative models [49,50]. First, we used western blotting to confirm that cell lines expressed ATP6v1c1 and found that each of the cell lines expressed ATP6v1c1, with the more malignant cell lines (4T1, MDA-MB-231, and MDA-MB-435s) expressing more than the immortalized fibroblasts C3H10T1/2 ( Figure  5A-5B). Then, we used a RT-PCR assay with gene-specific   Figure 3A-3B). To knockdown ATP6V1C1 in these human cancer cells, we used lentiviruses expressing different shRNAs to target human C1, and lentiviruses expressing a scramble shRNA (as a control) to infect the human cancer cell line MDA-MB-435s in order to select the most efficient C1 targeting shRNA. Using western blotting, we found that TRCN0000029564, TRCN0000029565, TRCN0000029566, and TRCN0000029568 shRNAs can efficiently knockdown C1 expression compared to the control SHC002 shRNA ( Figure 5C-5D). We selected and developed two efficient shRNA lentiviral vectors, henceforth referred to as shRNA-1 (TRCN0000029566) and shRNA-2 (TRCN0000029568) ( Figure 5D), for use in other assays. We then assessed the effects of this knockdown on cell proliferation using bromodeoxyuridine (BrdU) incorporation assay and the effects on mTOR signaling using western blotting. The BrdU incorporation assay showed that both shRNA-1 and shRNA-2 mediated ATP6V1C1 knockdown significantly inhibited proliferation of the human breast cancer cell lines MDA-MB-231 ( Figure 5E-5F Figure 6F) cells was similar to that of the control cells expressing scramble shRNA. In order to confirm that these results were an effect of ATP6v1c1 knockdown we performed a rescue experiment, reintroducing ATP6v1c1 (human) to 4T1 cells knocked down for ATP6v1c1, and determining that after reintroduction of ATP6v1c1 expression we had a corresponding rescue of proliferation (Supplementary Figure  1A-1D). These results further show that C1 is required for the activation of the mTORC1 pathway and enhances proliferation in human cancer cell lines, which may explain the differences observed in metastatic potential in these lines [35].

Atp6v1c1 depletion mediated inhibition of mTORC1 activation by amino acids may be celltype specific and enhanced in transformed cells
Interestingly, when C1 was knocked down in the immortalized murine multipotential mesenchymal cell line C3H10T1/2 ( Figure 7A) there was no significant difference in cell proliferation compared to the control cells according to the BrdU incorporation assay ( Figure 7B-7C). Moreover, compared to the control cells, Atp6v1c1 knockdown in C3H10T1/2 resulted in virtually no change in phosphorylation of p70S6K, AKT and ERK1/2 at baseline or in response to amino acid stimulation ( Figure 7D). These results indicate that different cells have different responses to ATP6v1c1 knockdown and that the inhibitory effects of C1 depletion on mTORC1 activation and proliferation may be celltype specific. This result shows that ATP6v1c1 may have a specific functional role in human cancer, not seen in untransformed cells, in addition to being important for pathological functions of murine tumor cells in vivo and in vitro. It also shows that ATP6v1c1 expression mediates mTORC1 signaling in cancer specifically such that knockdown of ATP6v1c1 would knock down mTORC1 mediated signaling and cell growth as illustrated in our model ( Figure 7E).

DISCUSSION
We began by checking the human genomics data to find that ATP6v1c1 is a highly amplified and overexpressed V-ATPase subunit in breast tumors relative to other V-ATPase subunits, such as Atp6v1c2 and Atp6v0a3. We examined the extent of dysregulation and especially amplification of Atp6v1c2 and Atp6v0a3 in patient samples using data from TCGA (Supplementary Figure 1) and, relative to ATP6v1c1 (Figure 1), their amplification and overexpression was considerably less common and, even where present, didn't have a significant effect on patient survival. In fact, our model cell line MDA-MB-231 didn't express perceptible amounts of Atp6v1c2 (Supplementary Figure 3) nor did 4T1 [5]. Further, the role of ATP6v0a3 in cancers has been previously characterized by the Forgac lab [6,7]. ATP6v1c1's levels of overexpression and frequency of dysregulation are reminiscent of better known mutations, such as those in the MAPK pathway or p53, which encouraged us to examine its effects in vitro and in mouse models of breast cancer (Figure 1). We found that C1 depletion in 4T1 cells significantly inhibits growth and metastasis ( Figure  2), while C1 depletion in the murine multipotential mesenchymal cell line C3H10T1/2 has no effect on cell proliferation [51] (Figure 7). We further determined that the C1 preference is greater in breast cancer cell lines, such as MDA-MB-231, which only expresses C1a (C1) (Supplementary Figure 3). These results suggest that C1's involvement in growth and cell functions may be cell-type specific. Therefore, we expanded our study to see the effects of efficient C1 knockdown in human breast cancer cell lines, using MCF-7 as a low metastasis/grade cell line and MDA-MB-231 and MDA-MB-435s as high grade breast cancers in line with prior characterization [49]. Some recent literature has suggested that MDA-MB-435s may be a melanoma cell line, as it has remarkable similarity with the M14 melanoma cell line, but the evidence is unclear and indicates that it may be a breast cancer with lineage infidelity [50,52,53]. Specifically, M14 melanoma-the cell line with which MDA-MB-435s is so similarwas taken from a biopsy of a male patient [54], but the MDA-MB-435s cell line has no Y chromosome [55] indicating that MDA-MB-435s are most likely breast cancer cells with lineage infidelity and not melanoma.
The mTORC1 kinase complex evokes growth in response to growth factors, energy levels, and amino In brief, addition of amino acids to most cells will enhance their proliferation; however, with the inhibition of ATP6v1c1 this nutrient induced proliferation is attenuated in cancer cells as a result of the failure of the V-ATPase to assemble and the mTORC1 to form on the LAMP1+ lysosomes to receive amino acid signaling. acids, and its activity is dysregulated in many cancers, including breast cancer. Recent studies showed that the translational landscape of mTOR signaling steers cancer initiation and metastasis [35] and that mTORC1 senses lysosomal amino acids through an inside-out mechanism requiring V-ATPase activity [28]. C1 knockdown of 90% in 4T1 cells significantly inhibited proliferation of 4T1 cells [51]. Importantly, we found 4T1 cells with C1 knockdown have significantly less metastasis to the lungs, liver, and bone, in addition to significant reductions of osteolysis at the bone (Figure 2  and 3). We found that ATP6v1c1 can significantly inhibit cell proliferation and mTORC1 pathway activation in response to amino acids, without any significant change in the phosphorylation of AKT and ERK1/2 between the C1 depleted cells and the control cells when stimulated with amino acids, which was consistent with our observations in 4T1 cells (Figure 4). Conclusively, we suggest that C1 may be involved in mTORC1 pathway activation to facilitate breast cancer (and perhaps melanoma) growth and metastasis.
Eighty percent of breast cancer patients develop bone metastasis [56]. Bone marrow is a major site of metastasis, likely due to its vascularization, available growth factors, and the generally supportive microenvironment [56]. Once cancer cells lodge in the bone, tumor growth stimulates osteoclast-mediated bone loss, which causes extreme pain and osteolysis related fractures [56]. Knowing this, we sought to determine whether C1 deficiency has an effect on 4T1 cell osteolytic erosion when a tumor lodges in the bone. To do this, we developed a local femur inoculation model to mimic breast cancer bone growth and found that after 16 days mice, which were locally inoculated with 4T1 cells with c1-knockdown of 90%, had less bone erosion and fewer OCs in the lesion regions compared to control 4T1 and 4T1-LacZ mice (Figure 3). Mice with 4T1 cells with c1knockdown of 60% had considerably larger lesion sizes and OCs compared to those with a stronger knockout. This data demonstrates that elevated expression of C1 is involved in 4T1 breast cancer cell-mediated osteolytic lesion development in the bone microenvironment. It has been reported that breast cancer cells produce factors that directly or indirectly induce the formation of osteoclasts and that bone degradation by osteoclasts releases growth factors that enhance tumor growth and bone resorption in a vicious cycle [57]. Our data suggests that C1 deficiency of 90% may inhibit the vicious cycle between breast-cancer cells and bone degradation mediated by osteoclasts, and thereby decrease cancer's bone lesions and perhaps even latent disease.
Our study shows that knockdown of C1 reduces breast cancer growth, metastasis, and osteolytic lesion formation. We show that, aside from being an essential subunit of the V-ATPase and playing a role in breast cancer growth and metastasis, C1 knockdown can attenuate mTORC1 signaling and inhibit cancer cell proliferation [9,29]. Further studies will be required to clarify how C1 mediates its effects on breast cancer metastasis and osteolytic lesion formation. Our results suggest that C1 could be a fruitful pharmacological target for indirectly inhibiting the mTORC1 pathway in treating breast cancer and possibly cancers such as oral squamous cell carcinoma or melanoma [8].

MATERIALS AND METHODS
Generation of Atp6v1c1-depleted 4T1 cells 4T1 cells (ATCC) were cultured in RPMI medium 1640 (Invitrogen) supplemented with 10% FBS, 1% L-glutamine, 1% penicillin and streptomycin in 5% CO 2 at 37°C. 24 hours later lentiviruses (Lenti-LacZ or Lenti-c1s3 [43]) were added to the cells. Lentivirus preparation and infection, efficient shRNA targeting to human ATP6V1C1 selection were carried out as previously described [18]. Then, cells were trypsinized 24 hours after infection and resuspended in culture media, and single cell suspensions were seeded into 96-well culture plates. GFP + clones which express shRNA were selected as described [43]. We chose 4T1-LacZ as a control clone and 4T1-c1s3-1 and 4T1-c1s3-2 clones as c1-depleted 4T1 clones for experiments.

Cell growth and migration kinetics assay
Cells were cultured in 24-well plates (1×10 4 / well), after 24, 48, and 72 hours and cells were counted from three wells each. The experiment was performed in triplicate on three independent occasions (n=3). Migration was assessed in a wounded monolayer model as described [13]. 5 hours after injury, cell movement was captured. The experiment was performed in triplicate.

Cell invasion assay
Cell invasion was analyzed in 24-well Biocoat Matrigel invasion chambers (8 μm; BD Biosciences, Bedford, MA, USA) as described (Nam JS et al., 2006) with 10% FBS as a chemoattractant. After 20 hours of incubation in 37°C, 5% CO 2 , cells that had migrated through the membrane were fixed with methanol and stained with hematoxylin. Invasion cells per field were counted (n=10) in triplicate using a light microscope at an original 100× magnification.

In vivo metastasis
All animals were maintained according to the UAB IACUC regulations. For the spontaneous metastasis assay, anesthetized 7-week-old female BALB/c mice were divided into 5 groups with 6 animals per group, and surgically exposed so that PBS or 1×10 5 4T1, control vector infected (4T1-LacZ), or c1-depleted 4T1 cells (4T1-c1s3-1 or 4T1-c1s3-2) could be inoculated into the left thoracic (#2) mammary gland fat pad in a 50μl volume. From day 10 to day 26 after implantation, we monitored mean tumor diameter (TD) [60]. Mice were euthanized by carbon dioxide after being exposed to X-ray (Faxitron X-ray) on day 28, and then they perfused with 4% paraformaldehyde in PBS (pH 7.4). The primary tumors were surgically removed and weighed. The lungs, liver, femur, and tibia were removed and immersed in the same solution overnight at 4°C. The lungs were observed under a fluorescence microscope with a 490 nm excitation filter and a 525 nm emission filter to assess GFP+ metastases. The femur and tibia were scanned by Micro-CT at 16 μm voxel resolution in all three axes on a GE eXplore Locus SP Micro-CT scanner. The ROI began 0.1 mm from the lowest point of the growth plate and moved distally for ten slices at a 3D level.

In vivo osteolytic lesion assay
For assaying the osteolytic lesions by c1-depleted 4T1 cells, anesthetized 7-week-old female BALB/c mice were divided into 5 groups with 5 animals per group, and they were inoculated with PBS or 1×10 4 normal or c1depleted 4T1 cells in the left femur bone marrow cavity in a 10μl volume as described [62]. Sixteen days after implantation of tumor cells, mice were euthanized by carbon dioxide narcosis after exposure to X-ray, femurs and tibias were excised, fixed in 4% paraformaldehyde in PBS (pH 7.4), and then scanned by Micro-CT as described above.

AO staining, HE staining and TRAP staining
These were performed as described [44]. Mice whose metastatic breast cancer cells were found on any slide of lung, liver, and bone sections were considered positive for metastasis as described [63].

Amino acid starvation/stimulation
All cells were rinsed with and then incubated in amino acid-free RPMI 1640 for 50 minutes, and stimulated with the addition of normal RPMI 1640 containing amino acids for 10 minutes prior to staining [28]. RPMI 1640 used for these experiments was either amino acid free (US Biological, cat# R8999-04A) or conventional/amino acid containing (Life Technologies, cat# 31800-022); with the amino acid containing formulation having some of each of the 20 standard amino acids.

Cell immunofluorescence
Cells were grown on 8-well chamber, then cells were fixed after amino acid starvation/stimulation with 2% formaldehyde in phosphate-buffered saline (PBS) for 20 minutes, washed with PBS 3 times, then incubated in 0.2% Triton X-100 for 15 minutes and blocked for one hour with 10% normal donkey serum in PBS. Cells were incubated in the primary antibody (α-mTOR, 1:100; α-LAMP-1, 1:1.5) diluted in 1% normal serum in PBS overnight at 4°C; then washed three times with PBS for 5 minutes and incubated with secondary antibody Alexa Fluor® 647 Donkey Anti-Rabbit IgG (H+L) (1:200) and Alexa Fluor® 555 Goat Anti-Rat IgG (H+L) (1:200) for 1 hour. Cells were then washed with PBS and mounted with anti-fade mounting medium containing DAPI. Imaging was performed by a Zeiss LSM 510 confocal laserscanning microscope (Zeiss, Germany) using standard filter settings and sequential scanning to avoid crosstalk in UAB High Resolution Imaging Facility (Birmingham, AL). The experiments were done in triplicate.

Bromodeoxyuridine (BrdU) incorporation assay
Cells were seeded into a well of a 24-well plate then infected with lentivirus supernatant for 8 hours. The medium was replaced with fresh DMEM containing 10% FBS, 1μg/ml puromycin for another 72 hours and then cells were incubated with 10μM BrdU for 3 hours. Cells were fixed with Carnoy's fixative buffer, denatured by 2 M HCl, neutralized with borate buffer, and then incubated with anti-BrdU antibody (DSHB, USA) overnight at 4°C. To finally stain the BrdU positive cells, we used the VECTASTAIN Elite ABC kit and then ImmPACT™ DAB kit (Vector Laboratories, USA) according to the manufacturer's procedures. The experiment was performed in triplicate on three independent occasions.

Bioinformatics analysis
Data were obtained from the following sources, using standard methods for data extraction and analysis. ATP6v1c1 gene amplification, overexpression, and mutation analyses were performed using cBioportal [38,39] using data generated by the TCGA Research Network (cancergenome.nih.gov) [40][41][42]. Relapse and time to metastasis analysis were performed using ProgGeneV2, dividing high and low expressing groups at the median, with data coming from three different studies [40][41][42].

Authors' contributions
MM and SF contributed equally to this work, performed and designed experiments. GZ also performed additional experimental work. LD, DS, SP provided valuable insight and assisted in editing the manuscript, as well as assisting in data analysis. WC and YPL provided laboratory support, manuscript editing assistance, and directed experimental work performed.

ACKNOWLEDGMENTS
We thank Ms. Diep Nguyen and Manar Sakalla for their excellent assistance with this manuscript. We also thank Dr. Shawn R Williams for their technical assistance with confocal microscopy. We are grateful for the assistance of Small Animal Bone Phenotyping Core, Small Animal Imaging Shared Facility, Neuroscience Imaging Core, and Neuroscience Molecular Detection Core Laboratory at the University of Alabama at Birmingham