Acid–base homeostasis is dysregulated in cancer cells and conditions for pH regulation are fundamentally different in 3D compared to 2D environments. Here, we show that localization and/or expression of four major net acid extruding transporters – NHE1, NBCn1, MCT1 and MCT4 – are regulated during growth of breast cancer spheroids and that these transporters contribute to breast cancer spheroid growth in a cell-type dependent manner.
General properties of MCF-7 spheroids
While in vivo tumors are obviously much more complex than cancer cell spheroids in terms of architectural and cellular diversity, spheroids resemble in vivo tumors much more closely than 2D cultures with respect to multiple parameters, and are excellent models of 3D growth [
4,
5]. Normal mammary epithelial cells can be induced to form acini in 3D culture, resembling the native mammary duct structure [
29,
31]. With increasing aggressiveness, organization is lost, resulting in the formation of solid spheroids [
31,
37] and at larger sizes often a necrotic core [
5]. The observation of a central cavity in such a spheroid can thus reflect lumen formation, generally characterized by apoptosis [
31] or necrotic core formation due to hypoxia. MCF-7 cell spheroids lack lumen formation after 1–2 days of growth ([
29] and APA and SFP, unpublished data), but we show here that continued spheroid growth of MCF-7 cells is associated with partial polarization and formation of a large lumen lined by apical markers and tight junctions, and a progressive increase in PARP cleavage which was most prominent close to the lumen. MDA-MB-231 spheroids were more loosely organized and lacked lumen and detectable PARP cleavage, yet both MCF-7 and MDA-MB-231 cell spheroids exhibited a clear gradient of increasing hypoxia towards the lumen.
Acid-extruding transporters are differentially distributed in MCF-7 spheroids, and their expression is altered in 3D compared to 2D
In both MCF-7 and MDA-MB-231 cells, expression of NHE1 and NBCn1 relative to total protein was reduced in 3D compared to 2D, while that of MCT1 and MCT4 was unaltered. The precise changes in the 3D setting underlying this difference require further investigation, yet it is well known that the expression of many transcription factors, cytokines, cytoskeletal proteins and other factors potentially relevant to the expression of the transporters studied here, differs profoundly between 2- and 3D conditions [
2,
38,
39], underscoring the importance of studying their roles in a 3D setting. Furthermore, whereas NHE1 and MCT4 appeared uniformly distributed throughout the MCF-7 cell spheroids, NBCn1 expression was most prominent in the spheroid periphery and, conversely, MCT1 expression exhibited an inward-directed gradient which largely coincided with the measured gradient of hypoxia. It was also notable that whereas MCT4 expression was essentially fully membrane-localized in MDA-MB-231 cell spheroids, MCT4 was partially localized to intracellular compartment(s) in MCF-7 cells. While not further pursued here, this may reflect differential expression of proteins contributing to MCT4 membrane localization in the two cell lines [
40,
41].
To our knowledge, ours is the first study to address how transporter localization is regulated by spheroid growth. In patient breast cancer tissue, we found NHE1 to be most highly expressed in well-perfused, peripheral tumor regions, while NBCn1 expression did not exhibit a detectable spatial gradient [
15]. In brain tumors, NHE1 was most highly expressed at the periphery, whereas MCT1 and −4 showed a broader distribution [
42]. In xenografts of colorectal and cervical cancer cells, MCT1 was found in the tumor periphery [
43]. In our hands, expression of MCT1, but not MCT4, followed the gradient of hypoxia in the spheroids. Functionally, the high MCT1 activity at the hypoxic spheroid core agrees well with the fact that glycolytic metabolism dominates in this region. What upregulates MCT1 expression in this region remains to be elucidated, since the majority of studies on this topic have shown that MCT4, yet not MCT1, is hypoxia-inducible (e.g. [
44,
45]). There are however reports of MCT1 upregulation by hypoxia in cancers [
46], possibly due to the additional presence of a glucose deprivation gradient [
47], a situation also present in spheroids [
5]. Interestingly, it was recently reported that while MCT1 expression was unaltered, its activity was increased under hypoxia due to the hypoxic upregulation of CAIX [
45].
It is tempting to suggest that the observed distribution of NHE1 and NBCn1 is reflected in different subcellular contributions to pH
i regulation and hence to growth, although it should be kept in mind that due to extensive posttranslational regulation, expression levels
per se say little about transporter function. Ser703 of human NHE1 has been widely implicated in regulation of NHE1 activity [
33], and we recently demonstrated its phosphorylation in breast cancer cells in response to prolactin [
48]. Relative NHE1 Ser703 phosphorylation was largely unaffected by 3D growth, except for a decrease in MDA-MB-231 cells with time of spheroid growth. Thus, Ser703-dependent NHE1 activity may be reduced in MDA-MB-231 spheroids during long-term growth. NHE1 was previously found to play a major functional role in pH
i regulation in the periphery of cancer spheroids, and HCO
3
− dependent mechanisms in the core [
26]. However, the role of HCO
3
− in the core at least in part reflects its role as a mobile buffer, rather than as a substrate for Na
+,HCO
3
− cotransporters [
26]. Given the importance of NBCn1 in breast cancer [
12,
15,
22] we focus on this isoform here. A full analysis of all HCO
3
− transporters is beyond the scope of this work, but would be needed to precisely map their contributions and activity, but roles of other isoforms are clearly also likely ([
53] and discussion below). Similar to MCT1, the mechanisms causing NBCn1 to be most strongly expressed in the spheroid periphery remain to be determined, but likely regulators would be the gradients of hypoxia, lactate, pH
e, pH
i, and ATP arising in spheroids [
5].
It should be noted that other acid–base transporters than the four studied here may play a role in 3D growth, depending on the cell type and conditions. For instance, pharmacological inhibitors or knockdown of proton ATPases have been shown to reduce growth of some cancer cells [
49], and such compounds are currently in clinical trials [
50]. Finally, while not further studied here, it is worth noting that the marked upregulation of CAIX as well as its specific localization to the inner regions of the spheroids, may also be important for the regulation of spheroid growth, given its known importance for pH homeostasis in the confined 3D space of spheroids [
51].
Growth of breast cancer spheroids is dependent on acid extruding ion transport proteins
A major conclusion of this work is that acid-extruding transporter(s) are important for spheroid growth yet that the specific transporters that play the predominant roles differ between breast cancer subtypes. This suggests that what is required to maintain 3D growth is the phenotype of acid extrusion rather than a given transporter protein, posing the challenge to therapeutic use that the relevant target(s) will likely differ between breast cancer subtypes, a notion corroborated by the differences between MCF-7- (luminal A) and MDA-MB-231 (triple-negative) cell spheroids revealed by the present work. Although complete knockout of NHE1 reduced spheroid growth in both cell lines, partial knockdown of NHE1 only reduced growth for MDA-MB-231 spheroids, and growth of MCF-7 spheroids was also delayed by knockdown of NBCn1 or MCT1 or by pharmacological inhibition of MCT, exacerbated by concomitant inhibition of NBCs. The role of MCT1 in spheroid growth is well in line with previous reports from in vivo studies of tumor growth [
13,
16,
21]. The role for NBCn1 corroborates previous reports from us and others demonstrating its upregulation in human breast cancer patients [
15] and the importance of NBCs in mammary tumor pH
i regulation and in vivo tumor formation [
14,
22]. In conjunction with GWAS reports linking NBCn1 to breast cancer risk [
20], this identifies NBCn1 as a target of potential therapeutic interest. However, the very marked differences in expression of the various NBC isoforms across different cancers [
52] suggests that the specific NBC isoform relevant is likely to differ, a notion substantiated by the recently reported role of another SLC4 family member, NBCe1 (SLC4A4), in proliferation of MDA-MB-231 cells as well as LS174 colon cancer cells [
53]. In congruence with our finding that NBCn1 expression did not follow the hypoxia gradient in the spheroids, this study furthermore showed that NBCe1, but not NBCn1, was upregulated by hypoxia [
53]. Importantly, the compound used to inhibit NBC activity, S0859, was recently shown to also inhibit MCTs [
54]. Since in our work, this compound had no effect on its own, but was additive to the effect of the MCT1 inhibitor, we favor the interpretation that NBCn1 is the main target of inhibition in our setup. This was confirmed by the knockdown data, however, a slight reduction in MCT1 expression was seen after NBCn1 knockdown (Fig.
5a), hence we cannot fully exclude a contribution from MCT1 to the observed effect.
In contrast to MCF-7 cell spheroids, MDA-MB-231 spheroids were not dependent on NBCn1 for growth, but depended only on NHE1 of the transporters studied here. This is supported by early experiments on xenograft growth of human bladder carcinoma cells [
55], and recent work demonstrating that NHE1 ablation in MDA-MB-231 cells reduces xenograft growth [
36]. Dependence on NHE1 may in part relate to glycolysis status: 50 % of tumors of CCL39 cells inoculated into nude mice underwent spontaneous regression if lacking NHE1 [
21], yet growth of non-glycolytic CCL39 cell tumors was unaffected by the absence of NHE1 [
56]. In line with this, MDA-MB-231 cells are more dependent on glycolysis than MCF-7 cells [
57]. It is furthermore intriguing that NHE1 has been proposed to be particularly dependent on glycolytically derived ATP [
58], suggesting that the link between metabolic profile and NHE1 dependence should be further explored.
Finally, despite the marked effects of transporter knockdown or knockout, pharmacological inhibition had no (NHE1) or limited (NBCn1) effect. The same concentration of cariporide strongly attenuated growth of BxPC-3 pancreatic spheroids (Noehr-Nielsen, A., and SFP, unpublished), and although S0859 is very lipophilic [
59], the concentration used was previously found effective in spheroids [
26] and indeed was additive to that of MCT1 inhibition in the present study. Hence, while they are likely less effective in spheroids than in 2D conditions, it seems unlikely that the inhibitors were not functional. An obvious difference between pharmacological inhibition and knockdown in the present work is that the inhibitors were only present from day 2 after spheroid formation. However, spheroids of knockdown cells were similar in size to controls at this time, hence, elucidation of this point requires further analysis.
We did not detect obvious changes in the core/lumen area in S0859- or AR-C-treated spheroids (
n = 2, data not shown), and there was no detectable increase in PARP cleavage in these spheroids compared to control, hence, although this remains to be directly addressed, we favor the interpretation that the decrease in spheroid size mainly reflects reduced growth/proliferation. Complete elucidation of the relation between pH
i regulation and 3D growth requires further studies. While pH
i recovery after an acid load in 2D-grown MCF-7 cells was dependent on both NHE1 and NBCs [
12], 3D growth of MCF-7 cells appeared to be more strongly dependent on NBCn1, and only full ablation of NHE1 reduced their growth. One interpretation of this is that hypoxia and strong extracellular acidity in the 3D setting limits contributions from NHE1 to pH
i regulation [
34], limiting its role in growth at least in the MCF-7 spheroids.
Our work thus corroborates and extends previous work pointing to the therapeutic potential of inhibiting acid extruding transporters in breast cancer. However, several open questions and challenges remain. It is noteworthy that the impact of NBCn1 knockdown on spheroid growth appears less dramatic than the strong inhibitory effect of NBCn1 knockout on growth of chemically induced tumors in vivo [
22], and the same appears to be true for NHE1 knockdown, the effect of which on spheroid growth of MDA-MB-231 cells appears to be smaller than that on their xenograft growth in vivo [
36]. This raises the exciting possibility that the role(s) of the transporters involves additional environmental factors present in vivo, a question which should be further addressed in future studies. A challenge is the limited specificity of some available pharmacological tools, especially problematic for NBCn1, for which currently available drugs are unspecific and/or unsuitable for tissue use (see [
14,
54,
59]). A second challenge illustrated by the present findings is to determine the relevant transporter(s) to target in a given cancer, and under which conditions. Clearly, transporter inhibition is likely to be most effective in combination with other therapeutic modalities, as previously suggested by findings by us and others [
12,
36,
60].