PEPT1 is essential for the growth of pancreatic cancer cells: a viable drug target

PEPT1 is a proton-coupled peptide transporter that is up-regulated in PDAC cell lines and PDXs, with little expression in the normal pancreas. However, the relevance of this up-regulation to cancer progression and the mechanism of up-regulation have not been investigated. Herein, we show that PEPT1 is not just up-regulated in a large panel of PDAC cell lines and PDXs but is also functional and transport-competent. PEPT2, another proton-coupled peptide transporter, is also overexpressed in PDAC cell lines and PDXs, but is not functional due to its intracellular localization. Using glibenclamide as a pharmacological inhibitor of PEPT1, we demonstrate in cell lines in vitro and mouse xenografts in vivo that inhibition of PEPT1 reduces the proliferation of the cancer cells. These findings are supported by genetic knockdown of PEPT1 with shRNA, wherein the absence of the transporter significantly attenuates the growth of cancer cells, both in vitro and in vivo, suggesting that PEPT1 is critical for the survival of cancer cells. We also establish that the tumor-derived lactic acid (Warburg effect) in the tumor microenvironment supports the transport function of PEPT1 in the maintenance of amino acid nutrition in cancer cells by inducing MMPs and DPPIV to generate peptide substrates for PEPT1 and by generating a H+ gradient across the plasma membrane to energize PEPT1. Taken collectively, these studies demonstrate a functional link between PEPT1 and extracellular protein breakdown in the tumor microenvironment as a key determinant of pancreatic cancer growth, thus identifying PEPT1 as a potential therapeutic target for PDAC.


Introduction 46
Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal forms of cancer with a 5year overall survival rate of 7% [1][2][3][4][5][6]. PDAC is currently the third leading cause of cancer-related 48 deaths in the U.S., trailing only lung and colorectal cancer, and is projected to outpace colorectal 49 cancer to become the second leading cause of cancer-related deaths by 2030 [2,6]. With such a 50 bleak prognosis, there is a need for new, hitherto untested, therapies. Recognizing that maintenance 51 of amino acid nutrition in cancer cells is obligatory for the rapid tumor growth, several studies 52 focused on amino acid transporters as a key component of this phenomenon and hence potentially 53 therapeutic targets [7,8]. Peptide transporters belonging to the SLC15 family could also support 54 amino acid nutrition by providing di-and tripeptides to the cells, but little is known on this group 55 of transporters for their role in cancer. PEPT1, a proton-coupled peptide transporter, is expressed 56 primarily in small intestinal epithelium and is known to transport over 8 000 di-and tripeptides as 57 well as peptidomimetic drugs [9][10][11][12][13]. The other three members in the SLC15 family include 58 PEPT2, PHT1, and PHT2. PEPT2 is predominantly expressed in the kidneys; though very similar 59 to PEPT1 in terms of function and substrate selectivity, PEPT2 has much higher substrate affinity 60 than PEPT1. PHT1 and PHT2, unlike the other two proteins, transport histidine and only a limited 61 number of di-and tripeptides [9]. There is very little in the published literature on the association between PEPT1 and 72 pancreatic cancer. It is known that PEPT1 is upregulated in pancreatic cancer cell lines and that its 73 inhibition suppresses the growth of cancer cells, but these findings were only limited to a few cell 74 lines [23,24]. In the present study, we characterized the expression profile of PEPT1 in a large 75 panel of PDAC cell lines and PDXs, and also monitored its transport function. Using 76 Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210377/921191/bcj-2021-0377.pdf by guest on 28 September 2021 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20210377

Plasmids and Transfection 112
All PEPT1 shRNA variants (TRCN0000043298, TRCN0000043299, TRCN0000043300, 113 TRCN0000043301, TRCN0000043302) were purchased from Millipore Sigma. pLK0.1 puro 114 (8453) was purchased from addgene. PLP1, PLP2, and VSVG (Invitrogen, K497500) were 115 purchased from Fisher Scientific. Lipofectamine 3000 (L3000015) was purchased from 116 ThermoFisher Scientific. 117 For plasmid transfection, SU.86.86 cells were plated to form a 30 % to 50 % confluent culture. 118 The 293FT cells were transfected using Lipofectamine 3000, according to the ThermoFisher 119 protocol. The shRNA viruses were harvested from 293FT cells after 42 h and applied to SU.86.86 120 cells after filtration using 0.45 m syringe filter. Viral medium was replaced with complete 121 medium after 16 h, and cells were maintained in complete medium for 24 h. After 24 h, cells 122 underwent selection for 10-14 days with 0.5 g/mL puromycin added to the medium. Thereafter, 123 all transfected cell lines were cultured in complete medium with 0.5 g/mL puromycin. shRNA 124 knockdown of PEPT1 was confirmed by qPCR, western blot analysis, and [ 3 H]Gly-Sar uptake 125 assay to assess PEPT1 mRNA, protein expression, and transport function. 126 qPCR and Western Blot 127 RNA isolation, cDNA synthesis, and qPCR were performed as previously described [25]. The 128 following PCR primers were used: PEPT1/SLC15A1 forward: 5'-129 CTCCCAATGTTCTGGGCCTT-3' and reverse: 5'-CGTTCACGGTCTGCATCTGA-3'; 130 PEPT2/SLC15A2 forward: 5'-ATCAGCAGGGTTCACGATGG-3' and reverse: 5'-131 CCACACTTGGAGACCAGACG-3'. Cell line and PDX protein lysates were prepared, and 132 western blot analysis was performed, as previously described [25]. In short, 30 g of protein was 133 loaded per lane in each of the western blots and electrophoretically separated on 10 % SDS-Page 134 gels. All proteins were transferred at 4 C overnight at 20V to a nitrocellulose membrane (Bio-135 Rad), blocked with 5 % Blotting Grade Blocker (Bio-Rad) at room temperature for 90 minutes, 136 then incubated with primary antibody at 4 C overnight. All primary antibodies were diluted 1,000-137 fold, with the exception of anti--actin, which was diluted 10,000-fold. After washing with TBST, 138 were then grown in three separate conditions, with glibenclamide added in a dose responsive 170 manner: complete culture media, complete culture media with Gly-Pro added, or complete culture 171 media with Gly-Pro added and pH adjusted to pH 6.5. After being allowed to form colonies for 2 172 weeks, cells were fixed, stained, and evaluated as previously described, using Enhanced Gram 173 Crystal Violet (Remel). 1, demonstrated a significant upregulation of PEPT1 at the mRNA level (Fig. 1A). Both the normal 214 cell lines expressed negligible PEPT1. We corroborated these data using quantitative real-time 215 PCR and showed upregulation in PEPT1 mRNA 10-10 000-fold in the majority of the PDAC cell 216 lines tested, relative to hTERT-HPNE (Fig. 1B). Western blotting data convincingly showed that 217 the upregulated mRNA expression correlated with PEPT1 protein expression (Fig. 1C). We further 218 validated our cell line data using PDX samples. PEPT1 mRNA was significantly upregulated in 219 all the PDXs tested ( Fig. 1D) with an average fold increase of 100, compared to normal human 220 pancreas (Fig. 1E). Likewise, majority of the PDXs also showed an increase in PEPT1 protein 221 ( Fig. 1F). Taken together, these data indicated that PEPT1 is significantly upregulated in 222 pancreatic cancer at both mRNA and protein levels, with little or no expression in normal pancreas. 223

PEPT2 is also upregulated in PDAC cell lines and PDXs 224
Since PEPT2 is functionally very similar to PEPT1, we were curious to check its expression status 225 in PDAC cell lines and PDXs. For this, we performed a regular RT-PCR and real-time PCR 226 utilizing two normal cell lines (hTERT-HPNE and HPDE) and ten PDAC cell lines (AsPC-1, normal cell line, did express PEPT2. Similar results were seen in real-time PCR (Fig. 2B). PEPT2 231 protein was also expressed as evident from western blotting data, but there was no difference in 232 the protein levels between the normal and PDAC cell lines (Fig. 2C). Similar results were also 233 seen in PDXs, with results establishing PEPT2 mRNA levels of 100-fold or higher in PDXs as 234 compared to normal pancreas ( Fig. 2D & 2E). 235

PEPT1, but not PEPT2, is functional in PDAC cell lines 236
To test whether the expression of PEPT1 and PEPT2 correlated with transport function, we 237 monitored the uptake of Gly-Sar, a hydrolysis-resistant dipeptide substrate for PEPTs, in these cell 238 lines. Uptake buffers containing NMDG chloride in place of NaCl with two pH conditions (pH 5.5 239 and pH 7.5) were used to monitor H + -coupled Gly-Sar uptake. All PDAC cell lines, with the 240 exception of BxPC-3, MIA PaCa-2, and PANC-1, demonstrated a significantly higher Gly-Sar 241 uptake at pH 5.5 than at pH 7.5, providing evidence for H + -coupled dipeptide uptake (Fig. 3A). As 242 expected, normal cell lines did not exhibit H + -coupled Gly-Sar uptake. PEPT1 as well as PEPT2 243 are capable of H + -coupled dipeptide transport; therefore, the data in Fig. 3A simply indicate the 244 presence of PEPT transport function in PDAC cell lines but do not reveal whether the observed 245 transport function is due to PEPT1 or PEPT2 or both. One of the revealing differentiating features 246 of these two transporters is the difference in their substrate affinities. PEPT1 is a low-affinity 247 transporter with almost a 10-fold less affinity for its substrates than PEPT2 for the same substrates. 248 In order to check which PEPT transporter is functional in the PDAC cell lines, we performed Gly-249 Sar kinetics experiment in AsPC-1 cells to determine the kinetic parameters (Michaelis constant 250 Kt and maximum velocity Vmax). If PEPT2 is functional in this cell line, then the Kt value should 251 be in lower micromolar (0.005 -0.5 M) concentrations as opposed to PEPT1. Based on the 252 kinetics experiment, the transport system was saturable and the uptake data fit well to a model 253 describing a single saturable transport system. However, the values for Kt and Vmax were 859.3  254 22.46 M and 8751  22.46 pmol/mg protein/15 min, respectively, indicating PEPT1 and not 255 PEPT2 to be functional in this cell line (Fig. 3B). This was puzzling because PEPT2 protein is 256 indeed expressed in these cells. To investigate as to why PEPT2 transport function was not 257 detectable in the PDAC cell lines, we performed immunofluorescence studies to determine 258 whether the PEPT2 protein localizes to the plasma membrane. For this, we used HPDE, AsPC-1, 259 BxPC-3, Capan-1, and CFPAC-1 cells, which showed high level of PEPT2 protein expression. 260 with little evidence for its presence in the plasma membrane or lysosomes (Fig. 3C). In contrast, 262 immunofluorescence for PEPT1 in AsPC-1 and SU.86.86 showed majority of PEPT1 to be 263 localized in the plasma membrane and lysosomes (Fig. 3D). Since the technique of Gly-Sar uptake 264 used in the present study monitors only the uptake across the plasma membrane, the 265 immunofluorescence studies support the conclusion that PEPT1, not PEPT2, is solely responsible 266 for the observed dipeptide uptake in the PDAC cell lines. While AsPC-1 had the highest PEPT1 267 protein expression, it did not have the highest PEPT functionality as shown by the Gly-Sar uptake. 268 In fact, SU.86.86 maintained highest functionality. Next, Capan-1 despite having lower protein 269 expression than AsPC-1 was functionally as good as AsPC-1 and in fact exhibited a very clean 270 Gly-Sar uptake at pH 5.5 with almost a negligible uptake at pH 7.5. Therefore, SU.86.86 and 271 Capan-1 were utilized for the majority of remaining experiments, including in vivo murine 272 experiments. MIA PaCa-2 was used as a PEPT1-negative cell line. 273

Glibenclamide inhibits PEPT1 function 274
Studies have already shown that glibenclamide, a sulfonylurea used for the treatment of non-275 insulin-dependent diabetes mellitus (type 2 diabetes), has the ability to inhibit PEPT1 and PEPT2-276 transporter, the uptake function is detectable at pH 6.5 as well as at pH 7.5, and as expected, the 284 uptake at pH 6.5 is greater than the uptake at pH 7.5. Glibenclamide had no effect on Gly-Sar 285 uptake in MIA PaCa-2 cells (Fig. 4C) because there is no PEPT1 in these cells and the uptake 286  in the presence of Gly-Pro and at pH 6.5. SU.86.86 cells showed a more robust sensitivity to 303 glibenclamide than Capan-1 cells, with cell proliferation being significantly stunted in the presence 304 of 50 M glibenclamide (Fig. 5B). MIA PaCa-2, though PEPT1-negative, also demonstrated a 305 slight sensitivity to glibenclamide (Fig. 5C); but the effects were comparable in all three culture 306 conditions, indicating a PEPT1-independent mechanism for the observed effect. These data clearly 307 suggested that glibenclamide by inhibiting PEPT1 had the ability to attenuate the growth of PDAC 308 cells, thereby implicating that glibenclamide could be used as an anticancer agent for pancreatic 309 cancer. 310

Genetic knockdown of PEPT1 reduces the colony formation ability of PDAC cells 311
To test if genetic knockdown of PEPT1 would have similar effects as that of pharmacological 312 inhibition, we used SU.86.86 cells as the model and silenced PEPT1 using PEPT1-specific shRNA. 313 Among the five shRNAs tested, TRCN0000043299 (299, Millipore Sigma) demonstrated the most 314 downregulation of PEPT1 (~83 %), evident both at the transcriptional as well as translational levels 315 (Fig. 6A & B). We also found a corresponding decrease in Gly-Sar uptake (~63 %) in shRNA-316 cells (Fig. 6C). We then examined whether the genetic deletion of PEPT1 would also affect its 317 proliferation capacity. The colony-formation assay using SU.86.86/WT and SU.86.86/shRNA-318 PEPT1 clearly showed that PEPT1 loss does significantly impact the growth of SU.86.86 cells 319 (Fig. 6D). 320  (Fig. 7A, B, E & F) and the tumor weight (Fig. 7C & G). 329 The body weight of the mice did not change following glibenclamide treatment (Fig. 7D & H), 330 suggesting no obvious toxicity of glibenclamide. To investigate whether shRNA-mediated 331 knockdown of PEPT1 would have similar effects, SU.86.86/WT and SU.86.86/shRNA-PEPT1 332 cells were injected subcutaneously in athymic nude mice and followed for 30 days. PEPT1 333 knockdown led to a robust reduction in tumor size (6.7  0.1 mg) compared to untreated controls 334 (94.1   mg) (Fig. 7I-K). Again, there was no significant change in body weight due to 335 treatment (Fig. 7L). 336

Relevance of tumor-derived lactic acid in the TME to the transport function of PEPT1 337
The TME in PDAC is highly desmoplastic and consists of the immune cells, pancreatic stellate 338 cell-derived myofibroblast-like cells, and ECM with collagen. Of relevance to PEPT1 is collagen, 339 which upon its proteolytic breakdown could generate di-and tripeptides as substrates for PEPT1 340 in vivo. MMPs in the TME would hydrolyze collagen into large peptides, which can be further 341 hydrolyzed into dipeptides by DPPIV. The unique amino acid composition of collagen 342 [Gly/Pro/Ala-X)n] is the most appropriate for the action of DPPIV which prefers peptides 343 containing the Gly-Pro/Ala sequence at the N-terminus to release Gly-Pro or Gly-Ala dipeptides. 344 In addition, the TME is acidic, which will provide the driving force for PEPT1. Since tumor cells 345 generate lactic acid and release into the TME, we asked whether lactate in the extracellular medium 346 acts as a signaling molecule to influence the generation of peptide substrates for PEPT1. To address 347 this question, we treated SU.86.86 cells with 10 mM lactate for various time points and examined 348 the expression levels of MMPs and DPPIV. We found that lactate treatment led to a significant 349 increase in the mRNA levels of MMP9, MMP13, and MMP16 ( Fig. 8A-C) and DPPIV (Fig. 8D). 350 Similar results were also seen in AsPC-1 cells (Fig. S1A-D). Our next question was then to check restimulation of the cells to the amino acid phenylalanine or the dipeptide Gly-Pro restored PEPT1 365 mRNA expression (Fig. 8F & 8G). Similar results were also seen in AsPC-1 (Fig. S1E & F). That 366 said, in SU.86.86 cells that underwent amino acid deprivation with 1% FBS medium (12 h) did 367 show a downregulation in PEPT1 mRNA expression, however, we did not see full restoration of 368 PEPT1 expression following restimulation of the cells to the amino acid phenylalanine or the 369 dipeptide Gly-Pro (Fig. 8H-I). It is possible that for SU.86.86 a longer restimulation time is 370 required as opposed to Capan-1. Likewise, we also performed [ 3 H]Gly-Sar uptake in both the cell 371 lines using similar deprivation conditions but a prolonged restimulation time of 2 h. It was 372 interesting to see that the transport data corroborated with the Real-time PCR data for Capan-1 373 ( Fig. 8J-M). More interestingly, with 2 h of re-exposure time, SU.86.86 showed a better PEPT1 374 functional restoration (Fig. 8N-Q). Even though lactate induced the expression of MMPs and 375 DPPIV, it had no effect on PEPT1 expression. 376

Discussion 377
Cancer cells reprogram their metabolism to support their rapid proliferation and growth. This 378 metabolic change necessitates the need for increased supply of various nutrients to feed into these 379 accelerated metabolic pathways. Therefore, starving the tumor cells of their essential nutrients 380 seems a logical strategy to suppress their growth and hence a plausible therapeutic paradigm for 381 the treatment of cancer. As most of these nutrients are water-soluble and cannot traverse the plasma 382 Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210377/921191/bcj-2021-0377.pdf by guest on 28 September 2021 membrane by simple diffusion, cancer cells upregulate specific transporters to meet their increased 383 demands for nutrients [28]. These transporters represent potential drug targets for cancer therapy. 384 The recent emphasis on tumor-selective metabolic pathways such as glutaminolysis, reductive 385 carboxylation, and one-carbon metabolism places particularly a special focus on amino acid 386 transporters for development of a new class of anticancer therapeutics [7,8,29,30]. In addition to 387 the amino acid transporters, peptide transporters also could make potential contributions to amino 388 acid nutrition in cancer cells, but this aspect has received very little attention. This is surprising 389 because the acidic pH in the TME is ideal to provide the necessary driving force for the peptide 390 transporters due to their transport function coupled to a transmembrane H + gradient [31-34]. One 391 of the reasons for the relative lack of interest in the potential role of the peptide transporters in 392 cancer is the fact that amino acids are present at several-fold higher concentrations than peptides 393 in the circulation. This casts doubt as to the physiological relevance of the peptide transporters to 394 amino acid nutrition in cancer cells. There have been some studies examining the therapeutic 395 potential of peptide transporters, but the focus was on the exploitation of these transporters for the 396 delivery of peptide drugs or non-peptide drugs in the form of peptide prodrugs [35,36]. Obviously, 397 what has been overlooked is the possibility that the TME might contain small peptides at 398 concentrations much higher than found in the circulation and that peptide transporters could 399 actually play a significant role in the maintenance of amino acid nutrition in cancer cells in vivo. cells. Further, these findings provide a framework for the biological importance of the observed 412 induction of PEPT1 in PDAC (Fig. 9). These results demonstrate upregulation of upstream markers 413 Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210377/921191/bcj-2021-0377.pdf by guest on 28 September 2021 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20210377 after exposure to sodium lactate in vitro and tumor derived lactate in vivo, but no upregulation in 414 PEPT1 expression. While these results on their own do not seem to indicate PEPT1 regulation, 415 previous research has demonstrated that PEPT1 has an amino acid response element (AARE) in 416 the promoter region that causes transcriptional upregulation of PEPT1 in response to available 417 amino acids and dipeptides [28]. As current research understands various MMPs and DPPIV to 418 cleave TME collagen into tripeptides, and further to dipeptides, this provides a stable framework 419 through which PEPT1 upregulation occurs in PDAC [37]. 420 In a recent review, we speculated that the acidic pH in the TME might provide the driving 421 force for a number of nutrient transporters whose function is coupled to a transmembrane H + Glibenclamide is used at a dose not more than 20 mg/day for the treatment of type 2 diabetes. 445 In the present study, we found the drug to be effective in causing a significant reduction in tumor 446 growth in vivo in mouse xenografts at a dose of 166 mg/kg/day. This translates into more than 1 g 447 per day for humans with ~70 kg body weight. However, even with this high dose, there was no 448 evidence of toxicity seen in the mice used in the present study as evidenced by normal body 449 weights compared to untreated control mice and the absence of drug-related morbidity. This is 450 supported by a published report that glibenclamide did not lead to toxicity in mice even at a higher PEPT2 is also a H + -coupled transporter with higher affinity for peptide substrates than PEPT1. 459 It is expressed in the plasma membrane in normal tissues such as the kidneys, lungs, and brain 460 where it plays a role in the transport of peptides into cells [48,49]. In the present study we have 461 shown that PEPT2 expression is markedly upregulated in PDAC cell lines and PDXs. However, 462 we were unable to find evidence for the expression of PEPT2 in the plasma membrane of the 463 cancer cells; this is obvious from the lack of a high-affinity peptide transport activity in PDAC cell dietary proteins produces predominantly small peptides and, only to a small extent, free amino 476 acids [50,51]. There is no reason to believe that the lysosomal protein breakdown should be any 477 different. Nuclear membrane also possesses a pH gradient with H + concentration in the cytoplasm 478 higher than in the nucleus, which might be of relevance to the findings in the present study that 479 PEPT2 might be located in the nuclear membrane. The biological function of PEPT2 in this 480 intracellular location and its potential connection to cancer cell biology remain to be determined. 481 In summary, our data suggests PEPT1 to have an essential role for pancreatic cancer cell 482 growth and therefore could prove to be a viable drug target.  activation of MMPs and DPPIV, which can degrade collagen within the extracellular matrix to di-723 and tripeptides; how these small peptides and the amino acids induce PEPT1 expression, which by 724 the virtue of the acidic pH in the TME becomes functional and brings in dipeptide substrate inside 725 the cancer cells, gets hydrolyzed into amino acids and helps in DNA and protein synthesis and 726 ultimately tumor growth; and how pharmacological inhibition of PEPT1 with glibenclamide can 727 starve the tumor cells of its substrates, eventually affecting its growth and proliferation. 728