Cell-derived xenograft tumors and prostatic tissues derived from wild type and TRAMP mice were from our archival collection. De-identified human tissues were obtained from LSU Health-Shreveport Biorepository Core, Overton Brooks VA Medical Center and Tissue for Research organization. The tissues include BPH, low-grade AdPCa, high-grade AdPCa, NEPCa, and NE tumors arising from pancreas, skin, lung, and bladder. Most tissues were from surgical resection except for five NEPCa cases that were from biopsies. All experiments were carried out in accordance with approved protocols by the Institutional Animal Care and Use Committee and Institutional Review Board of LSU Health-Shreveport.

IHC staining was performed using Vectastain elite ABC peroxidase kit (Vector Laboratories, Burlingame, CA) as described previously (37). Primary antibodies include PKM1 and PKM2 (7067S, dilution 1:500, and 4053S, dilution 1:1000, respectively, Cell Signaling Technology, Danvers, MA), P63 and FOXA2 (ab735 and ab108422, respectively, dilution 1:500, Abcam, Cambridge, MA), HOXB13 and chromogranin A (CHGA) (sc-28333 and sc-1488, respectively, dilution 1:200, Santa Cruz Biotechnology, Dallas, TX), FOXA1 (A15278, dilution 1:500, Abclonal, Woburn, MA), Synaptophysin (SYP) (611880, dilution 1:1000, BD biosciences, San Jose, CA), and NKX3.1 (0314, dilution 1:500, Athena Enzyme Systems, Baltimore, MD). Images were taken using a Zeiss microscope (White Plains, NY). The intensity score was evaluated by a semiquantitative blinded manner and graded as 0 (negative), 1⁺ (low), 2⁺ (moderate), and 3⁺ (high). For IF staining, primary antibodies include CHGA (AB_1553436, dilution 1:50, Developmental Studies Hybridoma Bank, Iowa City, IA), PKM1, PKM2, P63, and SYP (source was the same as mentioned above, dilution 1:100). The IF staining was imaged with a Nikon fluorescence microscope (Melville, NY) as reported previously.

PCa cell lines (VCap, LNCaP, C4-2B, 22RV1, PC3, and DU145) were obtained from ATCC. These cells were cultured in RPMI 1640 supplemented with 10% FBS, and 1% penicillin-streptomycin. H660 cells were cultured in Prostalife media (LifeLine Cell Technology, Oceanside, CA). C2C12 cells were cultured in DMEM supplemented with 10% FBS, and 1% penicillin-streptomycin. Cells were maintained in an incubator at 37°C and 5% CO2.

Cells were lysed in Laemmli SDS sample buffer followed by SDS-PAGE and Western blotting. Primary antibodies are beta-actin (sc-47778, dilution 1:500, Santa Cruz Biotechnology, Dallas, TX), PKM1 and PKM2 (the same source as mentioned above, dilution 1:1000), and HRP-conjugated secondary antibody (Cell Signaling, Beverly, MA). Protein bands were visualized by using ProSignal^® Dura ECL Reagent (Genesee Scientific, San Diego, CA) and Chemidoc™ Touch Imaging System (Bio-Rad). For the cross-linking experiments, cells were washed with ice-cold PBS three times and treated with 5mM disuccinimidyl suberate (DSS, A39267, Thermo Scientific, Waltham, MA) for 30 min at room temperature. The cross-linking reaction was stopped by adding the quenching solution (1M Tris, PH 7.5) to the final concentration of 20mM for 15 min. Then, cell lysates were used for WB as described above.

RNAseq data collected in the CTPC dataset were analyzed to assess the expression of PKM isoforms using STAR-Salmon based transcript level quantification method (38–41). The raw Fastq files were downloaded from NCBI SRA database and dbGaP (phs000909.v1.p1 (42), phs000915.v2.p2 (43)). The raw count and TPM data of TCGA data were acquired from GDC Data Portal. NCBI human GRCh38 (release 106) and associated GFT files were acquired through “AWS iGenomes” (https://ewels.github.io/AWS-iGenomes/) and were used for alignment. FastQC (https://github.com/s-andrews/FastQC), Trim Galore (https://github.com/FelixKrueger/TrimGalore), STAR (38), Salmon (39) software were used for sequencing quality control, alignment and quantification in the Linux Ubuntu environment. The transcript expression was visualized in R (V4.3.0) environment. The TPM values were first log2 transformed and quantile normalized using “preprocessCore” R package (44). The data was visualized by boxplot using “ggplot2” R package (45).

Based on the inclusion of exon 9 or exon 10, the transcripts were classified into “PKM1-like” and “PKM2-like” isoforms, respectively. Among the PKM1-like transcripts, NM_182470.2 and NM_182471.2 transcripts encode the same protein since the only difference in them lies within their 5’ UTRs. The protein encoded by these two isoforms is known as PKM1. NM_002654.4 transcript encodes a protein that is known as PKM2. To coordinate our bioinformatics analysis of the ten PKM isoforms with the well-known PKM1 and PKM2, we summed the normalized log2-transformed TPM values of NM_182470.2 and NM_182471.2 as PKM1 transcript. The expression data of NM_002654.4 was directly used to reflect the levels of PKM2 transcript. Additionally, to complement our bioinformatics analysis with the PKM1/2 antibodies’ staining results, we summed the values of all exon 9-retained transcripts as PKM1-like expression data and all exon 10-retained transcripts as PKM2-like expression data.

The Chi-Square test was used to compare the protein expression of PKM isoforms between AdPCa and NEPCa of human PCa and student t-test was used to compare the levels of PKM transcripts between these groups.

Using IHC, we examined the expression of PKM1 and PKM2 across murine prostate developmental stages ranging from the embryonic prostate undergoing budding morphogenesis (urogenital sinus, UGS), to the postnatal 2- and 3-week prostate, and the fully developed 8-week prostate. Our analysis revealed distinct expression patterns of both proteins. Specifically, we observed that PKM1 expression was largely absent during prostate development (including embryonic and postnatal 2–3-week prostate, Figures 1A, E, I ) but minimal expression was detected in the fully developed 8-week prostate ( Figure 1M ). On serial sections, the embryonic prostate buds were highlighted by positive expression of FOXA2 and P63 ( Figures 1C, D ), and prostate luminal epithelial cells were marked by positive staining of HOXB13, FOXA1, and NKX3.1 ( Figures 1G, H, K, L, O, P ). In contrast, robust PKM2 expression was consistently detected in prostate epithelial cells throughout all developmental stages ( Figures 1B, F, J, N ).

Next, we evaluated the expression of PKM1 and PKM2 in TRAMP tumors. Similar to the previously observed pattern in fully developed 8-week prostates, low to moderate levels of PKM1 expression were noted in the luminal epithelial cells of normal prostates as well as the normal adjacent epithelia of TRAMP tumors ( Figures 2A, B ). However, PKM1 expression was absent in the PIN lesions ( Figures 2B, C ). Interestingly, PKM1 expression was detected in the NEPCa cells ( Figure 2C ). This finding was further corroborated by dual-IF staining. As illustrated in Figures 2D–F , PKM1 expression was detected in the NEPCa cells (as indicated by the expression of NEPCa marker SYP), while it was absent in the adjacent PIN lesions.

Robust PKM2 expression was observed in both normal ( Figure 2G ) and transformed prostate epithelial cells ( Figures 2H, I ). It is noteworthy that NEPCa cells exhibited a slightly weaker expression of PKM2 compared to adjacent PIN cells ( Figure 2I ). This finding was further supported by dual IF staining results. As shown in Figures 2J–L , PKM2 expression was more pronounced in the PIN lesions compared to the adjacent NEPCa areas, as indicated by the positive staining of SYP.

We extended our investigation to characterize the expression of PKM1 and PKM2 proteins in human patient samples, including BPH, low-grade AdPCa, high-grade AdPCa, and NEPCa.

Consistent with its expression profile in murine prostate, PKM1 expression in human BPH samples was minimally detected in prostate luminal epithelial cells but exhibited high expression in basal epithelia and stromal cells ( Figure 3A , n=13). This observation was further confirmed by dual-IF staining. As depicted in Figures 3E–H , PKM1 was highly expressed in stromal cells ( Figure 3E ) and basal epithelial cells, as evidenced by the co-expression of basal marker P63.

In human AdPCa (including both low- and high-grade tumors), PKM1 expression was generally sparse ( Figures 3B, C , n=13 and 12, respectively). However, in the AdPCa with NE differentiation (NED, Figure 3I , n=2), PKM1 expression was detected in the scattered NE cells, as indicated by the co-expression of NEPCa marker CHGA ( Figures 3I–L ). Additionally, PKM1 expression was detected in 5 out of 12 human NEPCa specimens ( Figure 3D ). The overall quantification of PKM1 expression in different groups of PCa samples was summarized in a bar plot ( Figure 3Q ). Notably, the expression of PKM1 correlated with NE phenotype (Chi-Square test, p<0.01).

Considering the limited sample number of NEPCa, we expanded our study by examining PKM1 expression in the NE tumors arising from other anatomical sites. We found that PKM1 was expressed in 4 out of 4 pancreatic carcinomas with NED ( Figure 3M ), 2 out of 3 Merkel cell carcinomas ( Figure 3N ), 1 out of 2 small cell lung cancers ( Figure 3O ), and 2 out of 3 urothelial carcinomas with NED ( Figure 3P ). These results suggest a potential role of PKM1 in the pathogenesis of NE tumors.

On the other hand, the expression of PKM2 was detected in the majority of human prostatic tissues. In the BPH samples, PKM2 expression was prominent in basal epithelial cells and comparatively moderate in luminal epithelial cells ( Figures 4A, B , n=14). This expression pattern was confirmed by dual IF staining, which demonstrated that PKM2 was highly expressed in basal epithelial cells (as indicated by the co-expression of basal marker P63) with less staining in luminal epithelial cells ( Figures 4I–L ).

Human AdPCa samples exhibited a diverse PKM2 expression pattern, varying from negative to high levels in both low-grade ( Figures 4C, D , n=12) and high-grade AdPCa ( Figures 4E, F , n=15). Its expression level was higher in high-grade AdPCa compared to low-grade AdPCa (Chi-Square test, P<0.01). Unlike the decreased PKM2 expression observed in murine NEPCa, PKM2 expression in human NEPCa did not differ significantly from that of AdPCa ( Figures 4G, H , n=9). The intensity of PKM2 staining in PCa cells, including both AdPCa and NEPCa, was quantified, and its distribution among PCa samples is presented in Figure 4M .

Furthermore, we assessed the expression of PKM1 and PKM2 in cell lines representing various stages of PCa, including androgen receptor (AR) positive (VCaP, LNCaP, C4-2B and 22RV1), AR negative (PC3 and DU145), and NEPCa (H660) cells. Both in vitro cultured cells and cell-derived xenograft tumors were examined. Our IHC staining revealed that the protein expression of PKM1 was lowest in 22RV1- and highest in PC3- derived xenograft tumors ( Figures 5A–G ). This was consistent with the results of Western blot analysis of in vitro-cultured cells ( Figure 5H ). NEPCa H660 cells exhibited low to moderate levels of PKM1 expression ( Figures 5G, H ).

Consistent with its expression in human patient samples and murine TRAMP tumors, we observed high protein expression of PKM2 in all the AdPCa cell lines examined ( Figures 5I–N, P ). Its level was relatively lower in NEPCa cell line H660 cells ( Figures 5O, P ).

Using publicly available RNAseq data derived from human PCa cell lines and patient samples, we quantified the transcripts of ten PKM isoforms that have been identified on GRCh38 (release 106). Based on the inclusion of exon 9 or exon 10, these isoforms were classified into “PKM1-like” and “PKM2-like” transcripts, respectively. They include the well-studied PKM1 (NM_182470.2 and NM_182471.2) and PKM2 (NM_002654.4) isoforms, as well as seven under-studied isoforms ( Figure 6A ). The expression levels of these isoforms were summarized in Supplementary Figure 1 . Notably, the previously mentioned PKM1 and PKM2 antibodies, targeting exon 9 or 10, detect the protein products of PKM1-like or PKM2-like transcripts.

As shown in Figures 6B, C , there was a wide variation in the expression of PKM1 transcripts among PCa cell lines and human prostatic tissues. Its level was higher in human NEPCa specimens compared with AdPCa samples ( Figure 6C ).

Overall, the levels of PKM2-like transcripts were higher than that of PKM1-like transcripts in both PCa cell lines and human PCa specimens ( Figures 6B, C ). Among the PKM2-like isoforms, NM_002654.4 transcript was the most abundant one expressed in PCa cell lines and human prostatic tissues ( Supplementary Figure 1 ), and its level was lower in NEPCa cell line H660 compared to the aggressive PCa cell lines PC3 and DU145 ( Figure 6B , t-test, p<0.05).

Additionally, we analyzed the prostatic expression of PKL/R, the other two PK isoforms (41). We found that the levels of PKL/R transcripts were generally low in human PCa cell lines and human PCa samples. However, these transcripts were detected in a subset of human NEPCa samples ( Supplementary Figure 2 ). This finding is consistent with a recent study indicating that PKL/R expression is increased in castrate-resistant PCa, and the upregulated PKL/R, along with elevated MYCN, may drive metabolic reprogramming during NE differentiation (46).

To study the stoichiometry of PKM in PCa, we conducted a cross-linking reaction. While PKM1 was hardly detected after cross-linking through Western blotting due to the low expression level, the multi-unit formation of PKM2 was visible. C2C12 myoblast cells were used as a positive control for the detection of PKM2 tetramers ( Figure 7 ). We found that in PCa cells, PKM2 primarily formed inactive monomers/dimers rather than active tetramers ( Figure 7 ).