Blocking muscle wasting via deletion of the muscle specific E3 ubiquitin ligase MuRF1 impedes pancreatic tumor growth

Cancer-induced muscle wasting reduces quality of life, complicates or precludes cancer treatments, and predicts early mortality. Herein, we investigated the requirement of the muscle-specific E3 ubiquitin ligase, MuRF1, for muscle wasting induced by pancreatic cancer. Murine pancreatic cancer (KPC) cells, or saline, were injected into the pancreas of WT and MuRF1−/− mice, and tissues analyzed throughout tumor progression. KPC tumors induced progressive wasting of skeletal muscle and systemic metabolic reprogramming in WT mice, but not MuRF1−/− mice. KPC tumors from MuRF1−/− mice also grew slower, and showed an accumulation of metabolites normally depleted by rapidly growing tumors. Mechanistically, MuRF1 was necessary for the KPC-induced increases in cytoskeletal and muscle contractile protein ubiquitination, and the depression of proteins that support protein synthesis. Together, these data demonstrate that MuRF1 is required for KPC-induced skeletal muscle wasting, whose deletion reprograms the systemic and tumor metabolome and delays tumor growth.


INTRODUCTION
separation between genotypes, as well as between time points for WT mice (Fig. 4A). Across both 156 genotypes and all time points, a total of 815 ubiquitination sites, mapping to 302 proteins, were detected 157 (Supplementary file 1). Figure 4B-C depicts the number of ubiquitination sites and the number of 158 proteins whose ubiquitination status changed during tumor growth for both WT and MuRF1 -/mice.

159
Interestingly, the number of identified sites showing increased ubiquitination almost doubled in 160 skeletal muscle of WT mice between D8 (206 sites), when there is no muscle wasting, and D12 (367 161 sites), when muscle wasting is first detectable (Fig. 3), and then plateaued. This demonstrates that a  Fig. 4C). This suggests that some proteins ubiquitinated early 166 during tumor growth may be ubiquitinated on additional sites as the tumor grows and cachexia 167 develops (Fig 4D). We also identified a significant number of ubiquitination sites and proteins that 168 showed reduced levels of ubiquitination in response to KPC tumors, that similarly increased in number 169 throughout tumor progression. Importantly, while muscles from MuRF1 -/mice also showed early

193
To further identify proteins whose ubiquitination level increased in tumor-bearing mice through Of these twenty-seven sites, twenty-one (78%) were ubiquitinated in response to KPC tumors 209 through a MuRF1-dependent manner. Of these MuRF1 targets, five were related to the sarcomere 210 (Desmin,MYH4,TNNT3,MYLPF and Titin), three were related to protein degradation/ubiquitination 211 (SQSTM1, MuRF1 and VCP) and two were related to glycolysis (GAPDH and LDHA).

213
Reductions in proteins that support muscle protein synthesis parallel KPC-induced muscle 214 wasting and are mediated through MuRF1

215
To further identify the proteomic signature associated with the development of cancer-induced muscle 216 wasting in WT mice, and the requirement of MuRF1 for such a signature, we next performed an 217 unbiased TMT 10plex analysis on the tibialis anterior, using the same sample homogenates used for 218 the ubiquitinome profiling (Fig. 3). Overall, we detected 58,088 peptides mapping to 1,728 proteins, of 219 which 1,496 proteins were considered high confidence matches with at least two peptides mapping to 220 them. Principal component analysis (PCA) revealed good separation between genotypes and between 221 time points (Fig. 5A). The numbers of proteins that showed an alteration in their relative abundance 222 throughout tumor progression for WT and MuRF1 -/mice is shown in Fig. 5B.

223
To identify key biological pathways associated with KPC-induced muscle wasting and 224 dysfunction, we first performed bioinformatic enrichment analysis using findings from the muscles of 232 signaling and mTOR signaling from D12 onwards -the time point at which muscle wasting is first 233 detectable (Fig. 3).

234
A follow-up comparative analysis was conducted using Ingenuity Pathway Analysis (IPA) to 235 identify canonical pathways that show differences in response to KPC tumors between genotypes. This 236 analysis revealed that MuRF1 is required for the KPC-induced reduction in the relative abundance of 237 proteins annotating to EIF2 signaling -including multiple ribosomal proteins -as well as mTOR 238 signaling, oxidative phosphorylation, and tRNA charging processes. In contrast, MuRF1 deletion had 239 minimal effects on the KPC-induced increase in proteins related to Acute Phase Response, Coagulation

250
Of note, many of the myofibrillar and structural proteins that showed increased ubiquitination 251 in response to KPC tumors in WT mice, whose proteolytic degradation likely contribute directly to 252 muscle fiber atrophy, were not decreased in their relative abundance when compared to Sham mice 253 (Supplementary file 4). This was an expected finding, since normalization steps utilized for the 254 proteomics assay included use of equal amounts of starting tissue per group, and normalization of data 255 to total protein signal.

257
Preventing muscle wasting via MuRF1 deletion blunts KPC-induced changes in muscle, serum 258 and tumor metabolome

259
Our finding that blocking muscle wasting via the deletion of MuRF1, which is muscle-specific, impedes 260 KPC tumor growth suggests that muscle catabolism is required to fuel tumor growth. Based on this 261 observation, we explored the effects of MuRF1 deletion on the tumor metabolome by conducting global 262 metabolomic profiling via mass spectrometry on tumors from WT KPC mice at endpoint (day 16) and 263 on tumors from MuRF1 -/on day 16 and their endpoint (D35) (Supplementary file 5). These analyses revealed a differential metabolic profile in tumors from cachectic mice (WT KPC) and non-cachectic 265 mice (MuRF1 -/-KPC) ( Fig. 6A-B, Supplementary file 6). On day 16, tumors from MuRF1 -/mice -266 which were 84% smaller than tumors from WT mice -showed a preferential accumulation of 267 metabolites compared to tumors from WT mice. Indeed, 45 metabolites showed a higher abundance, 268 and only 14 showed a lower abundance, in tumors from MuRF1 -/mice compared to tumors from WT 269 mice (Fig. 6B, Supplementary file 6). Alternatively stated -metabolites are more depleted in tumors 270 from WT mice undergoing wasting than in MuRF1 -/mice spared from wasting. This aligns with 271 recently published work showing that metabolites are more frequently depleted in PDAC tumors 272 compared to adjacent non-tumor tissue 32 . Among the metabolites higher in abundance in tumors from 273 MuRF1 -/mice were alpha-D-glucose, several amino acids and amino acid derivatives, and lipidic 274 compounds (Table 1). These findings suggest that, at this time point, tumors from MuRF1 -/mice may 275 be less metabolically active than tumors from WT mice. As tumors from MuRF1 -/mice grew and 276 reached a size comparable to tumors from WT KPC END mice, the tendency for tumors from MuRF1 -277 /mice to accumulate metabolites remained, with 26 metabolites showing a higher abundance, compared 278 to 11 metabolites showing a lower abundance, in tumors of MuRF1 -/vs. WT mice (Fig. 6B).

279
To obtain a better understanding of the muscle -tumor crosstalk, we also profiled the muscle

291
In serum from WT KPC mice, 36 metabolites showed an altered abundance on day 12, when 292 cachexia is first measurable, which increased to 90 metabolites as WT KPC mice developed severe 293 cachexia and reached endpoint (Fig. 6E, Supplementary file 10). Among the metabolites decreased in 294 serum in response to KPC tumors, several were related to lipid and carbohydrate metabolism (Table   295 3). We also observed changes in the abundance of several amino acids and amino acid derivatives, with 296 a majority showing an increased abundance in response to KPC tumors, especially in severely cachectic 297 mice (Table 3). Of note, amongst the 20 proteogenic amino acids, 11 showed an altered abundance in 298 response to KPC tumors. KPC tumors also altered vitamin B and vitamin C metabolism, which are 299 involved in antioxidative defense and energy production ( Table 3). In relation to this, serum from 300 severely cachectic mice showed higher levels of circulating 2-hydroxyphenylalanine, which is a marker 301 of oxidative stress, and of 3-(4-hydroxyphenyl)lactate, which is an endogenous antioxidant 33 (Table   302 3). Similar to muscle, the serum metabolome of MuRF1 -/mice remained largely unchanged in response 303 to tumor burden (Table 3).

304
Lastly, to obtain a better understanding of the muscle-tumor crosstalk, we integrated our 305 metabolomic analyses from skeletal muscle, serum and tumor, and identified metabolites that showed 306 altered abundance across tissues (Fig. 6F). The majority of these metabolites, including glycine, acyl-307 carnitine, serine and several carbohydrates were decreased in skeletal muscle and/or serum from 308 cachectic WT KPC mice compared to WT Sham, suggesting a systemic depletion of these metabolites.

309
These same metabolites were also reduced in KPC tumors from WT mice when compared to the slower-310 growing tumors from MuRF1 -/mice, suggesting that KPC tumors from WT mice may deplete these 311 metabolites more rapidly. In support of this, glycine has been shown to be rapidly consumed by 312 proliferating cancer cells, in order to sustain de novo purine synthesis 34 . Since purines and pyrimidines 313 are formed simultaneously, our finding that thymidine is higher in abundance in tumors from WT mice 314 compared to MuRF1 -/mice, aligns with an increase in purine synthesis.

316
MuRF1 deletion also protects against KPC-induced wasting, slows tumor growth and extends 317 survival in female mice mice is similarly conferred in female mice, we conducted a final experiment in which female WT and 320 MuRF1 -/mice were inoculated with KPC cells and tracked until each mouse reached its humane 321 endpoint (Fig. 7). Importantly, these experiments were conducted in a different lab (Doles Lab at Mayo 322 Clinic) to those in male mice (Judge Lab at University of Florida) and utilized a different PDAC cell 323 line (KPC T42D versus KPC 1245). Thus, this experiment not only considers sex as a biological 324 variable, but also accounts for geographical location/lab specificity and KPC cell line used. Our findings 325 show that, similar to male mice, MuRF1 deletion in female mice extended maximum survival -to 64 326 days post inoculation, from 39 days in WT mice (Fig. 7A), and slowed tumor growth (Fig. 7B). Further, 327 as tumors grew, both lean mass and fat mass were greater in MuRF1 -/vs. WT mice (Fig. 7C-D), and 328 tissue mass of the tibialis anterior, gastrocnemius and hearts were all significantly greater in MuRF1 -/-329 mice compared to WT mice at their respective endpoints ( Fig. 7E-G). Moreover, muscle fiber cross 330 sectional area was significantly larger in tibialis anterior muscle from MuRF1 -/-KPC compared to those 331 from WT KPC mice ( Fig. 7H-I). Thus, overall, these findings in female mice, using a different KPC 332 cell line and conducted in a different lab, lend strong support for the role of MuRF1 in mediating cancer-333 associated muscle wasting and tumor growth.

335
Overall, our findings demonstrate that, in the context of KPC tumors, loss of the muscle-specific E3

382
Ex vivo skeletal muscle function assessment 383 Ex vivo skeletal muscle contractile properties were assessed as previously described 38 at the 384 Physiological Assessment Core of the University of Florida. Briefly, freshly isolated soleus muscles 385 and diaphragm strips were mounted on a force transducer (dual-mode lever system -Aurora Scientific,

387
After determination of muscle optimal length, maximum isometric twitch and tetanic forces were

416
Female experiments. Tibialis anterior harvested from female mice were placed in a sucrose sink (30%) 417 overnight prior to freezing, sectioning and laminin staining as previously described 37 . Briefly, sections 418 (8 μm) were fixed in 4% paraformaldehyde (PFA) for 5 minutes at room temperature prior to 419 immunostaining. Once fixed, tissues were stained with rat anti-laminin (MilliporeSigma, 4HB-2). Next, 420 the Alexa Fluor 488 (Invitrogen) conjugates were used as secondary antibodies. Three non-overlapping 421 fields per tissue were imaged at 200X. Myovision software was utilized to measure tibialis anterior 422 minimum feret diameters as previously described 39 .

424 425
Blood collection Franklin Lakes, New Jersey, USA), incubated at room temperature for at least 30 min and centrifuged 428 at 2,500g for 10 min at 4°C. Collected serum samples were then stored at -80°C until analysis.

430
Proteomics 431 Sample preparation. Freshly harvested tibialis anterior muscles were quickly rinsed in PBS and snaped-432 frozen in liquid nitrogen before being stored at -80°C until analysis. For each group, approximately 433 equal amount of starting tissue (~200 mg, N = 3-6 samples/group) were pooled and sent to Cell

458
Tribridmass spectrometer (Thermo Fisher Scientific). The same methods as described above for the 459 global proteome profiling were used to assign spectra to peptides and proteins. Only diGly 460 modifications annotating to unique proteins were retained for further analyses. For these modifications,

461
FC were computed between WT tumor-bearing vs. WT Sham mice and MuRF1 -/tumor-bearing vs.

484
Separation was achieved on an ACE 18-pfp 100 x 2.1 mm, 2 µm column with mobile phase A as 0.1% 485 formic acid in water and mobile phase B as acetonitrile. This is a polar embedded stationary phase that 486 provides comprehensive coverage, but does have some limitation is the coverage of very polar species.

487
The flow rate was 350 µL/min with a column temperature of 25°C. 4 µL was injected for negative ions 488 and 2 µL for positive ions. MZmine (freeware) was used to identify features, deisotope, align features 489 and perform gap filling to fill in any features that may have been missed in the first alignment algorithm.

500
Statistical analyses 501 Data normality was tested with Shapiro-Wilk test and parametric or non-parametric tests were used.

502
Separate t-tests were conducted to examine the effect of KPC tumor burden on body and tissue masses.

503
T-tests were also performed to evaluate the effect of tumor burden on twitch and tetanic force in the 504 WT mice, while one-way ANOVAs were used to examine this effect in the MuRF1 -/mice. One-way 505 ANOVAs were also used to compare tumor mass across genotype as well as body mass and tissue mass 506 throughout the progression of cancer in WT mice. Two-way ANOVAs were performed to examine the 507 effect of tumor burden on fiber type distribution and force-frequency relationship. When statistically 508 significant differences were detected by ANOVAs, Tukey (for one-way ANOVA), Dunn's (for 509 Kruskal-Wallis one-way ANOVA), and Sidak (two-way ANOVA) post hoc analyses were used to test 510 for differences among pairs of means. For, skeletal muscle fiber cross-sectional area, data were binned, 511 fit with a Gaussian least squares regression and significance was determined by calculating the extra 512 sum-of-squares F test. Lastly, an exponential one-phase decay model was used to compare fatigue 513 resistance. The alpha level for significance was set to 0.05 and data are expressed as mean ± standard 514 error. Except for the metabolomics data, all statistical analyses were performed with Prism (version    Bold values indicate the change was statistically significant.   tumor burden induces reductions in maximal (Ai-Bi) and relative (Aii-Bii) soleus peak twicth, and maximal (Ci-Di) and relative (Cii-Dii) tetanic force in WT but not MuRF1 -/mice, while it did not affect twitch relaxation rate (E-F). Reductions in forces evoked at frequencies > 15 Hz (G-H) were also observed in response to tumor burden in WT but not MuRF1 -/mice. (I-P) KPC tumor burden induces reductions in diaphragm peak twitch (I-J) and maximal tetanic force (K-L), slows twicth relaxation (M-N) and leads to reduced forces in response to stimulations delivered at frequencies > 15 Hz (O-P) in WT but not MuRF1 -/mice.       showing the relative abundance of all known metabolites detected in skeletal muscle (gastrocnemius-plantaris, C) and serum (D) of WT and MuRF1 -/mice. (E) Number of metabolites presenting altered abundance (p < 0.05 and FC > |1.2|) in skeletal muscle (gastrocnemiusplantaris) and serum of WT and MuRF1 -/mice. (F) Metabolites showing altered abundance in response to tumor burden in at least two tissues in the WT mice. In A, C, D and F, red indicates increased abundance and blue indicates reduced abundance compared to the averaged abundance of the respective Sham groups. In F, the average value is presented when metabolites were detected in both negative and positive phases.