Dietary phosphatidylcholine promotes breast cancer in the hyperlipidemic E0771 murine model

Dietary phosphatidylcholine promotes breast cancer in the hyperlipidemic E0771 murine model 1 2 Esias Bedingar1,*, Baoxiang Yan1,2, Ebubechi Adindu1, Lakshman Chelvarajan1 and Fredrick O. Onono1,2, 3 4 1Division of Cardiovascular Medicine and The Gill Heart and Vascular Institute, 5 2Markey Cancer Center, 6 University of Kentucky Colleges of Medicine, Lexington, KY 40536-0200 7 *Current address; Harvard T.H. Chan School of Public Health, Harvard University, 677 Huntington Ave, 8 Boston, MA 02115 9 10 Running title: Phosphatidylcholine-rich dietary casein promotes breast cancer 11


Introduction
Animal models are invaluable resources for cancer research and drug efficacy testing. However, many 93 studies using animal models draw conclusions about dietary effects from comparisons of different diets without 94 making considerations for PC composition in the diets. Typically, the PC content of laboratory animal diets is 95 not reported in the literature or provided by the manufacturers. Therefore, we examined the PC composition of 96 different fat sources in rodent diets and the effects of manipulating dietary PC levels on breast cancer 97 development. This was extended to studies in which ATX inhibition was combined with dietary PC depletion. 98 We used the hyperlipidemic C57BL/6 LDL receptor-null (LDLR-/-) mice since these mice are reported to 99 exhibit an accumulation of ATX-derived LPA when fed diets containing high fat contents [27,28]. Our studies 100 show that milk casein is the main source of PC when present in rodent diets. Stripping of casein and other 101 sources of fat with denatured alcohol reduces PC levels contained in these constituents by up to 70%. 102 Furthermore, replacing casein in diets with amino acids decreases PC contents by nearly 90%. The results of 103 our study show that casein in rodent diets promotes greater tumor development compared to diets in which 104 casein is replaced with amino acids. However, inhibition of ATX was without effect on primary tumor growth 105 in PC-rich diet fed mice and reversed the beneficial effects of PC depletion on primary tumor growth. Feeding 106 mice amino acid-defined diets or inhibiting ATX significantly reduced plasma inflammatory cytokine levels. 107 Our findings suggest that the beneficial effects of replacing milk casein with amino acids is through a 108 mechanism that is independent of the ATX-derived LPA signaling.   121 To test the hypothesis that casein-containing diets could promote tumor development in hyperlipidemic mice, 122 we formulated western and high fat diets in collaboration with Envigo diet manufacturers (Madison, WI). 123 Western diets (42% Kcal from fat) were modeled around the diet TD.88137 [29] while high fat diets (60% kcal 124 from fat) were modeled on TD.06414 [30]. Casein diets were formulated with casein or "Vitamin-free" Test 125 (VFT) Casein (Envigo, Madison, WI). The amino acid defined diets (isonitrogenous diets) were manufactured 126 using the amino acid profile of casein as formulated for diet TD.140088 with modifications equivalent to 127 casein-containing diets. Aspartic acid (Asp) and glutamine (Gln) are non-essential amino acids and not 128 included in the amino acid (AA) profile of casein. These amino acids were added to the diets at similar levels 129 found in other AA-defined diets (such as TD.99366). Asp was added at approximately half of the level of Asp 130 in other AA defined diets. Gln was added at a similar rate as Glu. Cysteine was added to the casein diet to 131 balance the amino acid profile -casein is limiting in sulfur AA. Since casein is a significant source of calcium 132 and phosphorus the levels of these minerals were also adjusted in the diets. VFT  week-old female LDL receptor deficient (LDLr -/-) C57BL/6 mice were purchased from Jackson laboratory (Bar 145 Harbor, Maine) and maintained on 12-hour light/dark cycle at the University of Kentucky Animal Care 146 Facility. The mice were fed water and standard rodent chow diet (2014 Teklad Harlan Rodent Diet) ad libitum 147 for 1 week to acclimatize to the laboratory conditions before being assigned randomly into groups to receive the 148 diets containing either labeled or unlabeled choline. After 8 weeks of ad libitum feeding, blood was drawn from 149 mice, plasma prepared and frozen at -80°C for further lipid analysis as described below. Kentucky. Five-week-old female LDLr -/-C57BL/6 mice were purchased from Jackson laboratory (Bar Harbor,153 Maine) and maintained on 12-hour light/dark cycle at the University of Kentucky Animal Care Facility. The 154 mice were fed water and standard rodent chow diet (2014 Teklad Harlan Rodent Diet) ad libitum for 1 week to 155 acclimatize to the laboratory conditions before being assigned randomly into groups to receive the assigned 156 diets. Food consumption was monitored by weighing food pellets before and after replacement. Blood was 157 drawn via submandibular bleeding into Ethylenediaminetetraacetic acid (EDTA) coated tubes (BD Microtainer) 158 at the onset of high fat feeding and biweekly thereafter. Plasma was prepared immediately and stored frozen at -159 80°C until processing for measurements of lipids (PC, LPC and LPA), ATX inhibitors and other markers as 160 needed. Twelve weeks after initiating feeding, mice were anesthetized, a small incision made and syngeneic 161 estrogen receptor positive (ER+) E0771 tumor cells (500,000 cells suspended in 100 µl HBSS) were injected 162 into the inguinal mammary fat pad using an insulin syringe. The incision was then closed with a wound clip.

163
When tumors became palpable after 4-5 days, two orthogonal caliper measurements were taken and tumor 164 volumes estimated using the formula width 2 × length/2. Primary tumors were allowed to grow until they 165 reached the maximum allowable size of 1 cm 3 . The primary tumors were resected and animals kept for a further 166 2 weeks for spontaneous metastasis when the animals were euthanized and various organs harvested. To    Results were expressed as mean ± standard error of the means (SEM). Statistical analyses were performed 235 using the Prism 7 package (GraphPad, La Jolla, CA) and SigmaStat (San Jose, CA, USA). Unpaired two-236 tailed Student's t-test or analysis of variance (one way or multiple comparisons) was performed for 237 comparisons between the groups. Comparisons in tumor growth kinetics between groups were performed 238 using 2-way ANOVA with post hoc Bonferroni-corrected t test. Values of p < 0.05 were considered 239 statistically significant.

241
Casein is the major source of Phosphatidylcholine when added as a constituent of rodent diets 242 We obtained the ingredients commonly used as fat sources to formulate rodent diets from Envigo (Madison, 243 WI) and quantitated PC in the separate components. As shown in Table 1, PC was detected in all of the 244 preparations tested. Lard and anhydrous milk, which are common fat sources in high fat diets, are rich in PC.

245
Other components used in rodent diets to substitute for lard and anhydrous milk also contain detectable PC 246 amounts. The most striking observation however is that PC is highly abundant in casein. Casein is usually 247 added to diets as a protein source. In our analysis, the quantities of PC in casein were approximately 100-fold 248 higher compared to traditional fat sources. We also observed that PC levels can be significantly reduced in the 249 respective sources by stripping with organic solvents. Levels of PC were reduced in "Vitamin-Free" Test (VFT) 250 Casein by about 70%. VFT-casein is produced by extracting casein obtained from lactic acid precipitation of 251 skim milk with hot denatured ethanol. Compared to regular casein, VFT-casein is reported by the manufacturer 252 to contain less fat, less moisture and more protein. As with hot denatured ethanol, we also observed that 253 tochopherol stripping of lard and corn oil reduced PC levels in these constituents substantially by up to 50-75% 254 ( PCs of different carbon lengths and fatty acyl chain saturations, with PC (34:1) being the most abundant (Table   256 2).

257
Having established that casein is the major source of PC we formulated different types of rodent diets 258 based on the composition of the widely used western diet TD.88137 (42% Kcal from fat) and high fat diet 259 TD.06414 (60% Kcal from fat), respectively. The casein content in the western diet and high fat diet were kept 260 at 195 g/kg and 295 g/kg, respectively. Western diets were formulated with 210 g/kg anhydrous milkfat while 261 high fat diets contained 310 g/kg lard. Three different types of western diet were designed based on the amino 262 acid composition of casein. One of the diets contained casein (Casein Diet). The second diet was made with 263 VTF-Casein while the third diet was constituted by replacing casein with amino acids (AA diets) based on the 264 amino acid profile of casein as provided Envigo. Amino acid composition was adjusted to be equivalent to the 265 levels found in casein. Care was taken to also match the levels of calcium, phosphorus and sucrose in both diets.

266
High fat diets on the other hand were formulated with either casein or amino acids. To determine whether the 267 expected alteration of PC levels occurred in the different formulated diets, lipids were extracted from these diets Casein promotes breast cancer in E0771 hyperlipidemic mice fed high fat diet. 296 To understand the effects of dietary casein and inhibition of autotaxin on breast cancer development in 297 hyperlipidemic mice, we fed age-matched LDLR-/-mice with high fat diets (60% kcal from fat) formulated 298 with casein or defined amino acids in place of casein (AA diets). After 12 weeks of feeding, breast cancer was 299 induced in these mice by injection of the syngeneic E0771 mammary tumor cells into the inguinal mammary fat 300 pads while treating with either 30 mg/kg ATX inhibitor PAT-505 or methyl cellulose vehicle control twice 301 daily. In the liver fibrosis mouse models, PAT-505 was efficacious and shown to inhibit ATX activity in 302 plasma and liver tissues after oral administration [39]. We observed reduced primary tumor growth in mice that 303 were fed amino acid defined diets compared to casein-containing diet (Fig. 3A). There was no significant 304 difference in primary tumor volumes and excised tumor masses between PAT-505-and vehicle-treated mice 305 when the animals were fed with casein-containing diet. However, treatment with PAT-505 abolished the tumor 306 growth reducing effects of amino acid-defined diet since tumor volume (Fig. 3A) and excised tumor masses 307 (Fig. 3B) were higher in PAT-505 treated mice compared to the vehicle-treated mice. Previous studies using a 308 potent ATX inhibitor ONO-8430506 and the highly metastatic 4T1 breast cancer models reported that long term 309 treatment with the ATX inhibitor did not result in significant difference in primary tumor size [26]. The authors 310 observed an initial reduction of tumor growth in ONO-8430506-treated mice which eventually caught up to the 311 vehicle group resulting in comparable primary tumor development. In our studies, treatment with PAT-505 312 increased the number of metastatic nodules in the lungs of mice that received casein-containing diet but did not 313 have significant metastatic effects in mice that were fed amino acid-defined diet (Fig. 3C).

314
In vivo Bioavailability of PAT-505 in breast cancer mouse model of cancer and in vitro drug potency. 315 Because of the results in Fig. 3 showing a lack of breast tumor growth inhibition by PAT-505, we decided to 316 develop methods to determine the bioavailability of PAT-505 in harvested plasma and tissues and also validate 317 the drug's potency in vitro. The concentration of PAT-505 was assessed in plasma obtained from mice (n=7-318 9/group) implanted with tumor and receiving twice daily oral administration of the drug. Using tandem MS 319 methods we confirmed the bioavailability of PAT-505 in plasma and tissues harvested from our mice (Fig. 4A). 320 We observed very high levels of circulating PAT-505 in plasma, reaching concentration levels of 60 µM after 321 one week of drug administration (Fig. 4B). These levels are consistent with the reported pharmacokinetic (PK) 322 of PAT-505 as assessed in mice and rats [39]. We also determined PAT-505 levels in tissues that were harvested 323 after the animals were sacrificed. PAT-505 was detected in the all the organs analyzed with the highest levels 324 detected in the intestines and liver, while the lowest levels were present in the brain and spleen (Fig. 4C). PAT-  were significantly reduced in mice that received amino acid-defined diets compared to casein diets ( Figure 6).

348
Treatment with PAT-505 reduced the levels of both TGF-β1 and TGF-β2 in casein mice to levels observed in 349 AA-mice but did not have further reduction in AA-mice. In addition, feeding mice with amino acid-defined  [42][43][44][45]. This is because casein, loaded with PC, would effectively be a rich precursor that replenishes 370 the omitted choline in the choline-deficient diets.

371
Our results reveal that formulating rodent diets with amino acids in place of casein reduce breast cancer 372 development. The results are consistent with previous findings that suggest an association between higher 373 dietary casein intake and increased risk of cancer [46][47][48][49][50][51][52]. Our study provides further evidence that the tumor 374 promoting effects of casein is due to nutrients contained in casein since supplementing diets with all amino 375 acids that are present in casein as replacement for casein decreases breast tumor development in mice. Casein is 376 a significant source of calcium, phosphorus and vitamins. Therefore, we adjusted the levels of these nutrients by 377 adding vitamin premix with choline and the two minerals to minimize variability among the diets we 378 formulated. Dietary PC has been associated with increased cardiovascular disease risk [53]. We were interested 379 in establishing whether casein also can promote tumor development by being a rich precursor source of PC 380 metabolites. Orally administered PC is hydrolyzed in the intestinal lumen by phospholipase A2 to form LPC 381 which is absorbed by enterocytes [54]. Some of the absorbed LPC is reacylated to form PC while the rest is 382 hydrolyzed to LPA majorly by the enzyme ATX. Because of the potential of ATX to contribute to many 383 diseases, a number of inhibitors have been described [24,40,[55][56][57][58]. Many ATX inhibitors have been developed 384 some with better in vivo bioavailability. We used the ATX inhibitor PAT-505 which has demonstrated 24-hour 385 maximal inhibition of ATX activity with once daily dosing [39]. We reasoned that targeting LPA production by 386 depleting PC in diet and inhibiting ATX activity with a potent drug would decrease tumor growth. We used 387 LDLR knockout mice as they have higher circulating levels of LPA [27,28] substantially reduced yet only LPA 18:2 levels were lowered to levels which were statistically significant [15]. has been replaced by amino acids. We also showed that casein promotes primary tumor growth in the 438 hyperlipidemic murine breast cancer models through a mechanism that involves pro-inflammatory signaling.

439
The cancer-promoting effects of casein does not depend on production of LPA by ATX as inhibition of ATX 440 did not attenuate primary tumor growth either singly or in combination with casein-replacement in these mice.    treated twice daily with 30 mg/kg PAT-505 and blood was collected after 7 days of continuous treatment.