Ser71 Phosphorylation Inhibits Actin-Binding of Profilin-1 and Its Apoptosis-Sensitizing Activity

The essential actin-binding factor profilin-1 (Pfn1) is a non-classical tumor suppressor with the abilities toboth inhibit cellular proliferation and augment chemotherapy-induced apoptosis. Besides actin, Pfn1 interacts with proteins harboring the poly-L-proline (PLP) motifs. Our recent work demonstrated that both nuclear localization and PLP-binding are required for tumor growth inhibition by Pfn1, and this is at least partially due to Pfn1 association with the PLP-containing ENL protein in the Super Elongation Complex (SEC) and the transcriptional inhibition of pro-cancer genes. In this paper, by identifying a phosphorylation event of Pfn1 at Ser71 capable of inhibiting its actin-binding and nuclear export, we provide in vitro and in vivo evidence that chemotherapy-induced apoptotic sensitization by Pfn1 requires its cytoplasmic localization and actin-binding. With regard to tumor growth inhibition byPfn1, our data indicate a requirement for dynamic actin association and dissociation rendered by reversible Ser71phosphorylation and dephosphorylation. Furthermore, genetic and pharmacological experiments showed that Ser71 of Pfn1 can be phosphorylated by protein kinase A (PKA). Taken together, our data provide novel mechanistic insights into the multifaceted anticancer activities of Pfn1 and how they are spatially-defined in the cell and differentially regulated by ligand-binding.


INTRODUCTION
As the first actin-binding protein identified more than four decades ago (Carlsson et al., 1977), profilin-1 (Pfn1) has been extensively studied in the context of actin regulation. By binding monomeric G-actin, Pfn1 exchanges ADP for ATP and facilitates the addition of ATP-bound G-actin to the barbed ends of filamentous actin (Haarer and Brown, 1990;Witke, 2004;Jockusch et al., 2007;Birbach, 2008). In addition, Pfn1 interacts with a wide range of poly-L-proline (PLP)containing proteins many of which are actin-regulatory factors and cooperate with Pfn1 to control actin polymerization (Haarer and Brown, 1990;Witke, 2004;Jockusch et al., 2007;Birbach, 2008). Pfn1 is essential for the development and survival of multiple eukaryotic organisms including mice, Drosophila and yeast (Balasubramanian et al., 1994;Verheyen and Cooley, 1994;Witke et al., 2001;Bottcher et al., 2009). Paradoxically, Pfn1 also shows anti-tumor and anti-metastatic activities for various types of cancer (breast, pancreatic, and liver) (Janke et al., 2000;Roy and Jacobson, 2004;Wittenmayer et al., 2004;Ding et al., 2006;Wu et al., 2006;Zou et al., 2007Zou et al., , 2009Zou et al., , 2010Bae et al., 2009Bae et al., , 2010Das et al., 2009;Yao et al., 2014;Diamond et al., 2015). Our prior study suggested that some of these anticancer activities may stem from nuclear Pfn1 (Diamond et al., 2015). Our more recent work supported this theory and demonstrated that nuclear Pfn1 functions as a transcriptional repressor by binding and inhibiting the Super Elongation Complex (SEC), a positive regulator of transcriptional elongation of many pro-cancer genes (Zhu et al., 2021). Furthermore, we provided evidence that Pfn1 undergoes spatial deregulation in a broad range of cancer due to overexpression of its nuclear exporter exportin-6. This explains, to some extent, how anticancer activity of nuclear Pfn1 can be inhibited while its essential cytoplasmic functions are sustained. However, given that Pfn1 influences distinct cancer phenotypes including proliferation, metastasis, and survival upon chemotherapy treatments (Zou et al., 2010;Yao et al., 2013;, it remains unclear whether these activities stem from the same or different subcellular locations and whether there are additional regulatory mechanisms besides exportin-6-dependent nuclear export. We have found in prior and recent studies that PLP-binding is important for the tumor-inhibitory function of Pfn1 (Diamond et al., 2015), at least partially due to the direct interaction of nuclear Pfn1 with ENL, a PLP-containing protein, in the SEC complex (Zhu et al., 2021). We and others found that PLPbinding of Pfn1 can be abolished by Ser 137 phosphorylation in its C-terminus (Shao et al., 2008b;Diamond et al., 2015). In addition to PLP-binding, actin-binding was also suggested to be important for tumor inhibition by Pfn1. This was based on the loss-offunction effect of Y59A, an actin-binding mutation of Pfn1 (Schluter et al., 1998;Wittenmayer et al., 2004). However, despite being the first actin-binding protein identified several decades ago, it remains unknown whether phosphorylation events exist in Pfn1 which can inhibit its actin-binding and modulate its anticancer activities.
In this paper, we identified a protein kinase A (PKA)dependent phosphorylation site in Pfn1 at Ser 71 . Residing in the actin-binding site of Pfn1, Ser 71 phosphorylation abolishes the Pfn1/actin interaction and causes nuclear retention of Pfn1. Functional characterization using breast cancer cell lines revealed that Ser 71 phosphorylation regulates both cell proliferation and chemotherapy-induced apoptosis but in different fashions. Dissection of the functional influences of subcellular localization further indicated that while tumor inhibition by Pfn1 is driven largely by its nuclear activities, apoptosis-sensitizing effect depends on its cytoplasmic localization. Thus, by identifying and characterizing a previously unknown inhibitory phosphorylation event for actin-binding of Pfn1, we provided further mechanistic insights into its multifaceted tumor-inhibitory activities which are regulated both by ligand-binding and subcellular localization.

Cell Culture
All cell lines were purchased from ATCC with the exception of MDA-MB-231 cells stably expressing a tri-modal reporter fusion used to inject NOD/SCID mice as previously described (Diamond et al., 2015). All cell lines were authenticated and tested for mycoplasma within 3 months prior to the experiments. MDA-MB-231, MCF-7, and BT-549 were grown in RPM1 1,640 containing 5 or 10% fetal bovine serum (FBS) with gentamicin and supplements (50 µg/mL gentamycin, 1mM sodium pyruvate, 10 mM HEPES and glucose to 4.5 g/L). HEK293T cells were grown in high glucose DMEM supplemented with 5 or 10% fetal bovine serum and 50 µg/mL gentamicin. Transient transfection was performed using Fugene HD or Lipofectamine 2000. Lentiviruses were generated using HEK293T cells as previously described (Diamond et al., 2015).

Pull-Down Assays
To study Pfn1 interaction with actin and VASP, HEK293T cells grown in 6-well dishes were transfected with Myc-tagged or HAtagged Pfn1 constructs (WT and mutants), lysed by RIPA buffer, and subjected to immunoprecipitation using antibodies against Myc or HA tags as described previously (Shao et al., 2008b;Diamond et al., 2015). To affinity purify endogenous Pfn1, parental MDA-MB-231 cells were lysed and bound to PLPconjugated agarose beads as described previously (Shao et al., 2008b;Diamond et al., 2015).

In Vitro Drug Treatment
For drug treatment in 2D cultures, MDA-MB-231 or BT-549 stable cells were seeded at 1,000 cells per well in 96-well plates or 500 cells per well in 24-well plates, and treated on the next day with vehicle or paclitaxel in quadruplicate wells. Viable cells were quantified 5-7 days later by the Alamar Blue assay (Diamond et al., 2015;Zhu et al., 2021). Briefly, they were incubated with 100ul (for 96-well) or 500ul (for 24-well) growth media containing 44 µM resazurin for 2-4 h at 37 • C, and the fluorescence intensity of resorufin (converted product) in the media was measured at 540λ Ex/590λ Em on a fluorescence plate reader (Tecan Infinite M200). Relative drug effects were calculated as the percentage of live cells in drug vs. vehicle wells.

Mouse Xenografts
The animal experiment was carried out in strict accordance with the guidelines recommended for care and use of laboratory animals by the National Institutes of Health. The Animal Studies Committee at Washington University (St. Louis, MO, United States) approved all animal protocols. Five-week old female NOD/SCID and NU/NU mice were purchased from Charles River and kept under standard institutional care. Experimental details for orthotopically inoculating MDA-MB-231 stable cells were previously described (Diamond et al., 2015;Zhu et al., 2021). For paclitaxel dosing, when tumors in the NU/NU mice reached an average of ∼70 mm 3 , they were randomly divided and treated with 0.9% sodium chloride or paclitaxel (10 mg/kg) by weekly intraperitoneal injection for two consecutive weeks. Clinical grade paclitaxel (6 mg/ml) was purchased from Siteman Cancer Center pharmacy at Washington University School of Medicine. Primary tumors were measured by Caliper on a weekly (for NOD/SCID) and semiweekly (NU/NU) basis until tumor resection and euthanasia of the mice.

Statistical Analysis
Unpaired two-tailed student t-test was used to determine the statistical significance of the differences in cell growth rates, tumor sizes and weights between control and experimental groups. All statistical analyses were performed using GraphPad Prism 7.0. P-values < 0.05 were considered significant.

Ser 71 Phosphorylation Abolishes Actin-Binding of Pfn1
Hypothesizing that actin-binding of Pfn1 can be inhibited by phosphorylation, we selected five candidate serine/threonine residues at the actin-binding region of Pfn1 guided by the actin/Pfn1 co-crystal structure ( Figure 1A; Ferron et al., 2007). Each residue was mutated to alanine (S/A) or aspartate (S/D) to prevent or mimic phosphorylation. The resulting Myctagged Pfn1 mutants were transfected into HEK293T cells and immunoprecipitated using an anti-Myc antibody followed by Western blot to determine their actin-binding. All mutants were successfully expressed except T89D which was likely due to a destabilizing effect on Pfn1 structure and its consequent degradation. Out of the five candidate residues, Ser 71 appears to be a bona fide phosphorylation site as preventing (S71A) and mimicking (S71D) phosphorylation had opposite effects on actin-binding of Pfn1. While Pfn1(S71A) binds more actin than Pfn1(WT), Pfn1(S71D) completely fails to bind actin ( Figure 1B). Introducing the S71A and S71D mutations into an HA-tagged Pfn1 showed the same effects on actin-binding as for Myc-Pfn1, but neither mutation had detectable effect on Pfn1 interaction with vasodilator-stimulated phosphoprotein (VASP), a wellknown PLP-containing ligand of Pfn1 (Reinhard et al., 1995;Kang et al., 1997;Ferron et al., 2007; Figure 1C).
A search at PhosphoSitePlus revealed that Ser 71 phosphorylation of Pfn1 has been detected in several unbiased mass spectrometry datasets (Klammer et al., 2012;Mertins et al., 2013Mertins et al., , 2016. To confirm this, we developed a polyclonal antibody FIGURE 1 | Mimicking Ser 71 phosphorylation by the S71D mutation abolishes actin-binding of Pfn1. (A) X-ray diffraction structure of human Pfn1 (right, cyan) bound with actin (left, purple) (Protein Data Bank ID 2PAV) (Ferron et al., 2007). Ser 71 and four other serine and threonine residues in the actin-binding region of Pfn1 are displayed in the sticks mode. (B) HEK293T cells were transfected with Myc-tagged wild type or mutant Pfn1 containing alanine (A) or aspartate (D) substitutions on the five candidate serine or threonine residues. Lysates were subjected to immunoprecipitation using a Myc-tag antibody, and Western blot was used to detect Myc-Pfn1 and co-precipitated actin. (C) HEK293T cells were transfected with wild type or mutant forms of HA-tagged Pfn1 containing S71A or S71D, which were subsequently immunoprecipitated by an anti-HA tag antibody. Lysates were subjected to immunoprecipitation, and samples were blotted for actin, VASP, and HA-tag (to visualize HA-Pfn1).
FIGURE 2 | Pfn1 is phosphorylated at Ser 71 in cultured cells. (A) Equal amounts of recombinant 6His-Pfn1 (WT, S71A, or S71) purified from bacteria was stained by Coomassie blue or analyzed by Western blot using the pSer 71 -Pfn1 antibody F5675. (B) HA-tagged Pfn1 (WT, S71A, or S71D) was transfected into HEK293 cells and immunoprecipitated using an anti-HA antibody. Samples were analyzed by Western blot using the pSer 71 -Pfn1 antibody (F5675) or the anti-HA antibody. Densitometry was used to quantify the intensity of bands detected by F5675 and expressed as relative values to HA-Pfn1(WT). (C) Endogenous Pfn1 was purified from MDA-MB-231 cell lysates using poly-L-proline (PLP)-conjugated agarose beads, and subjected to Western blot analysis using the pSer 71 -Pfn1 antibody (F5675) or a pan anti-Pfn1 antibody. Input was also blotted with the pSer 71 -Pfn1 antibody (F5675) for comparison. named F5675 using an antigenic Pfn1 peptide harboring pSer 71 . Using recombinant 6His-Pfn1 purified from bacteria (where serines/threonines are largely unphosphorylated), we observed by Western blot little detection of Pfn1(WT) and the phospho-resistant Pfn1(S71A) but robust detection of the phosphomimetic Pfn1(S71D) by F5675 (Figure 2A). Though chemically different from phospho-serines, phosphomimetic amino acids such as aspartate (D) or glutamate (E) have been reported by us and others to react with phospho-antibodies raised against different proteins (Shao et al., 2008b;Shu et al., 2008). This thus supports the phospho-specificity of the pSer 71 -Pfn1 antibody. Consistent with this, HA-Pfn1 immunoprecipitated from HEK293T cells was positively detected by the pSer 71 -Pfn1 antibody, and the signal was significantly reduced by the S71A mutation while enhanced by S71D ( Figure 2B). However, we could not readily detect endogenous Pfn1 within whole cell lysates using the pSer 71 -Pfn1 antibody because it recognizes many other proteins due to its non-specific, polyclonal nature ( Figure 2C, input, second lane). To overcome this, we used PLP-conjugated agarose beads to purify and enrich endogenous Pfn1 from the triple-negative breast cancer MDA-MB-231 cell line. This resulted in clear detection of endogenous Pfn1 by the pSer 71 -Pfn1 antibody (Figure 2C, first lane), as confirmed by the side-by-side blotting with a pan Pfn1 antibody (third lane). Thus, our data suggest that Pfn1 is phosphorylated on Ser 71 in mammalian cells.

Ser 71 Is Phosphorylated by Protein Kinase A
By comparing Pfn1 and Pfn2 (a functionally related isoform) (Haarer and Brown, 1990;Witke, 2004;Jockusch et al., 2007;Birbach, 2008) sequences from various species, we find that Ser 71 is evolutionarily conserved in vertebrates (a chemically similar threonine is present in zebrafish Pfn1) ( Figure 3A). Ser 71 is invariably preceded at −2 position by a basic residue (most commonly Arg), suggesting that it could be phosphorylated by the AGC kinase family (Pearce et al., 2010). A closer examination of the sequence surrounding Ser 71 and comparing it with the phosphorylation consensus sites of common AGC kinases at Human Protein Kinase Knowledgebase revealed that cAMPdependent protein kinase [also known as protein kinase A (PKA)] is a possibility. The preferred amino acids at +1 and +2 positions for PKA substrates are hydrophobic (F, I, L, and V) while those for +5 and +6 positions are leucine. The Ser 71 -containing motif in Pfn1 meets all these requirements in particular a perfect match at +5/+6 positions. To test whether PKA is indeed the kinase for Ser 71 , we co-transfected HEK293T cells with GFP or GFP-tagged alpha catalytic subunit of PKA with HA-Pfn1, and examined the level of Ser 71 phosphorylation by Western blot after anti-HA immunoprecipitation. While equal amount of HA-Pfn1 was pulled down, co-transfection with GPKA significantly increased its pSer 71 level detected by the F5675 antibody ( Figure 3B). We next treated HEK293T cells co-transfected with HA-Pfn1 and GPKA with DMSO or PKA-specific inhibitor H89, and performed the same anti-HA immunoprecipitation and Western blot analyses. H89 significantly decreased pSer 71 -Pfn1 level without affecting the total level of HA-Pfn1 ( Figure 3C). To confirm that Ser 71 , as opposed to other PKA phosphorylation sites in Pfn1, is detected by the F5675 antibody, we performed the same co-transfection experiment using either wild type or S71A mutant form of HA-Pfn1 with GFP or GPKA. We observed no GPKA-induced increase in the detection of HA-Pfn1(S71A) by the pSer 71 -Pfn1 antibody in contrast to HA-Pfn1(WT) (Figure 3D). As a separate control, mutating Ser 137 to alanine (S137A) did not affect the ability of GPKA to increase the detection of HA-Pfn1 by the F5675 antibody ( Figure 3E). Taken together, our data suggest that Ser 71 is the site of PKA-dependent phosphorylation of Pfn1.

Antitumor Activity of Pfn1 Requires Reversible Ser 71 Phosphorylation and Dephosphorylation
Pfn1 overexpression inhibits the growth of various cancer cell lines in vitro and in vivo (Janke et al., 2000;Wittenmayer et al., 2004;Wu et al., 2006;Zou et al., 2007Zou et al., , 2010Das et al., 2009;Yao et al., 2014). We have previously demonstrated that the antitumor effect of Pfn1 requires its PLP-binding ability which is inhibited by Ser 137 phosphorylation (Diamond et al., 2015). To determine the effect of Ser 71 phosphorylation, we virally expressed untagged wild type Pfn1 and its S71A or S71D mutants in the MDA-MB-231 breast cancer cells at levels 2-3folds over endogenous Pfn1 (Figure 4A). GUS, a reporter gene encoding bacterial β-glucuronidase, was expressed as the negative control as previously described (Diamond et al., 2015). We first compared the cell proliferation rates in vitro by Alamar blue assay. While Pfn1(WT) showed anti-proliferative effect as observed previously, both Pfn1(S71A) and Pfn1(S71D) mutants were inactive (Figure 4B). We next determined the in vivo effects by injecting the same stable cells orthotopically in the mammary fat pads of female NOD/SCID mice as previously described (Diamond et al., 2015). Caliper measurements showed that both S71A and S71D mutations abolish the antitumor effect of Pfn1 and caused an additional increase in tumor growth compared to the GUS control ( Figure 4C). End-point tumor weights confirmed these effects ( Figure 4D). Thus, unlike the toggling effect of Ser 137 phosphorylation, tumor inhibition by Pfn1 appears to require reversible Ser 71 phosphorylation and dephosphorylation.

Ser 71 Phosphorylation Inhibits the Apoptosis-Sensitizing Activity of Pfn1 in Response to Paclitaxel
In addition to suppressing proliferation, Pfn1 also sensitizes cancer cells to apoptosis induced by cytotoxic agents (Zou et al., 2010;Yao et al., 2013;. When transfecting MDA-MB-231 cells using lipofectamine (which is cytotoxic), we consistently observed higher number of surviving cells expressing Pfn1(S71D) than those expressing Pfn1(WT) and Pfn1(S71A) (Supplementary Figure 1A). A similar effect was also observed in the transfected MCF-7 cells (Supplementary Figure 1B). Hypothesizing that Ser 71 phosphorylation may be a pro-survival event by inhibiting the pro-apoptotic activity of Pfn1, we treated the stable MDA-MB-231 cells with paclitaxel, a commonly used chemotherapy agent which was reported to cause apoptosis in breast cancer cells more effectively upon Pfn1 overexpression . Indeed, we detected significantly decreased viability of Pfn1(WT)-expressing cells compared to the control cells. While Pfn1(S71A) showed a similar drug-sensitizing effect as Pfn1(WT), Pfn1(S71D) was completely inactive (Figure 5A). Western blot for cleaved caspase-7 confirmed the pro-apoptotic effect of Pfn1(WT) and Pfn1(S71A) but not Pfn1(S71D) upon paclitaxel treatment ( Figure 5B). Similar results were also observed in MDA-MB-231 stable cells treated with doxorubicin and staurosporine (Supplementary Figure 1C,D), two other cytotoxic agents whose apoptosisinducing abilities can be augmented by Pfn1 overexpression (Yao et al., 2013;. To confirm the toggling effect of Ser 71 phosphorylation of Pfn1 on chemotherapy-induced apoptosis in vivo, we inoculated the MDA-MB-231 stable cells expressing Pfn1(S71A) vs. Pfn1(S71D) in the mammary fat pads of female nude mice. Tumors formed by both cell lines grew at similar rates as observed in the NOD/SCID mice. When the average tumor volumes in both groups reached ∼70 mm 3 , mice were randomly divided into two subgroups (n = 5 per group) which were treated with vehicle or paclitaxel (10 mg/kg, weekly intraperitoneal injection) for 2 weeks. Semiweekly Caliper measurement showed an obvious tumor-regressing effect of paclitaxel in the Pfn1(S71A) group, with the tumor volume difference between the vehicle and paclitaxel groups being statistically significant at the last two time points (Figure 5C). In contrast, tumors in the Pfn1(S71D) group responded much less to paclitaxel and no statistically significant difference in tumor volumes between vehicle and drug groups was detected. P-values were based on two-tailed unpaired t-test (versus GUS control) at day 42. (D) End-point tumor weights. Data are the mean ± SEM. P-values were based on two-tailed unpaired t-test (versus GUS control). *p < 0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001. GUS and Pfn1(WT) tumor values were published in our prior study (Diamond et al., 2015) as they were the same controls for Pfn1(S137A vs. S137D) (published) and Pfn1(S71A vs. S71D) (shown here, unpublished) in the same mouse xenograft experiment.
Western blot using tumor lysates showed significantly higher levels of cleaved caspase-7 in the paclitaxel-treated Pfn1(S71A) tumors than the Pfn1(S71D) tumors ( Figure 5D). Interestingly, baseline cleaved caspase-7 levels in untreated tumors, though much lower than in paclitaxel-treated tumors, were also higher in the Pfn1(S71A) vs. Pfn1(S71D) tumors. Collectively, these data suggest that Ser 71 phosphorylation, by blocking actinbinding of Pfn1, abolishes the apoptosis-sensitizing activity of Pfn1 particularly in response to chemotherapy agents such as paclitaxel.

Apoptotic Sensitization by Pfn1 Rrequires Actin-Binding and Cytoplasmic Localization
The ability of Pfn1 to bind actin is crucial for its nuclear export by exportin-6 (Stuven et al., 2003). Since Ser 71 phosphorylation disrupts actin-binding of Pfn1, we examined its effect on Pfn1 subcellular localization. Upon transfection into HEK293T cells, YFP-tagged Pfn1(WT) and Pfn1(S71A) were predominantly localized in the cytoplasm. However, YFP-Pfn1(S71D) was diffusely present within cytoplasm and nucleus (Figure 2A). We observed the same phenotype using lentivirus-infected MDA-MB-231 cells ( Figure 6A). The opposite effects of S71A vs. S71D mutations on cell survival upon paclitaxel treatment were observed in the context of YFP-Pfn1 similar to untagged Pfn1 (Supplementary Figure 2B). Intrigued by the effects of Ser 71 phosphorylation on both Pfn1 subcellular localization and chemo-sensitizing activity, we tested whether these two phenotypes are causally linked. To do that, we forced YFP-Pfn1 expression either in the cytoplasm or nucleus by tagging it with a nuclear export sequence (NES, recognized by exportin-1) or nuclear localization sequence (NLS) as previously described (Diamond et al., 2015;Zhu et al., 2021). We virally introduced YFP, YFP-NES-Pfn1, or YFP-NLS-Pfn1 into MDA-MB-231 cells and determined their relative responses to paclitaxel treatment. Compared to YFP control cells, YFP-NES-Pfn1 cells showed higher sensitivity to paclitaxel treatment while YFP-NLS-Pfn1 cells showed less ( Figure 6B). These opposite effects on cellular survival by YFP-NES-Pfn1 and YFP-NLS-Pfn1 were also observed in BT-549 cells, another triple-negative breast cancer cell line (Figure 6B). Data are mean ± SEM of four technical replicates in one experiment. Similar results were observed in three independent experiments. P-values were based on two-tailed unpaired t-test (versus GUS control). (B) Stable MDA-MB-231 cells were treated with DMSO or 5 nM paclitaxel for 48 h, followed by Western blot analysis for full-length or cleaved caspase-7 or tubulin. Cleaved caspase-7 levels were quantified by densitometry and normalized by the levels in GUS control cells in DMSO or paclitaxel groups. (C) MDA-MB-231 stable cells expressing Pfn1(S71A) or Pfn1(S71D) were injected bilaterally into the 4th mammary fat pads of female nude mice (10 mice per group). Tumor-bearing mice were evenly divided and treated with vehicle (0.9% sodium chloride) or paclitaxel (10 mg/kg) by intraperitoneal injection for 2 weeks. Tumor volumes were measured semiweekly by Caliper. Data are mean ± SEM. P-values were based on two-tailed unpaired t-test between vehicle and paclitaxel treatment groups at individual time points. *p < 0.05; **p < 0.01. (D) Randomly selected tumors harvested from vehicle or paclitaxel-treated mice in (C) were blotted for full-length or cleaved caspase-7 and GAPDH. Each tumor was from a different mouse.
The pro-apoptotic activity of Pfn1 has been causally linked to its ability to increase p53 and IκBα (negative regulator of NFκB) levels in breast cancer cell lines (Yao et al., 2013;. We found that MDA-MB-231 stable cells expressing untagged Pfn1(S71D) contain significantly lower levels of IκBα than those expressing Pfn1(S71A) (Figure 6E). The difference in IκBα levels between MDA-MB-231 cells expressing NES-Pfn1(S71A) and NES-Pfn1(S71D) can also be seen but to a lesser extent ( Figure 6E). Interestingly, p53 levels did not differ significantly between Pfn1(S71A) and Pfn1(S71D) cells regardless of whether Pfn1 is untagged or NES-tagged ( Figure 6E). Collectively, our data suggest that cytoplasmic Pfn1 promotes apoptosis in response to cytotoxic treatments and such an activity is abolished by Ser 71 phosphorylation via abolishing actin-binding of Pfn1 as well as causing its nuclear retention, both of which contribute to IκBα destabilization and increased NFκB signaling. (F) Hypothetical model for the pro-apoptotic activity of cytoplasmic Pfn1 which requires actin-binding and is abolished by PKA-dependent Ser 71 phosphorylation. Ser 71 may also be phosphorylated by PKA in the nucleus. Actin-free pSer 71 -Pfn1, due to its small size, can freely diffuse in and out of nucleus. However, nuclear pSer 71 -Pfn1 cannot be actively transported into the cytoplasm by exportin-6 due to its loss of actin-binding.

DISCUSSION
Pfn1 was the first actin-binding protein identified more than four decades ago (Carlsson et al., 1977), yet it remains uncertain to this day whether its actin-binding ability undergoes negative regulation by post-translational modifications. In this paper, through candidate mutagenesis and validation using public proteomic data and a custom-made phospho-specific antibody, we provided evidence that the evolutionarily conserved Ser 71 is a bona fide phosphorylation site in Pfn1 selectively inhibiting its actin-binding with little effect on its PLP-binding. This is consistent with the fact that Ser 71 is located within the actinbinding site of Pfn1 and distal to its PLP-binding pocket (Ferron et al., 2007). Sequence analysis combined with genetic and pharmacological testing suggested that PKA may be one of the kinases phosphorylating Ser 71 . Interestingly, it was recently suggested that PKA can also phosphorylate Ser 137 of Pfn1 (Gau et al., 2019), a shared target site for PKC and ROCK which specifically inhibits Pfn1 interaction with PLPs (Singh et al., 1996;Sathish et al., 2004;Shao et al., 2008a). Nonetheless, based on coimmunoprecipitation data using HA-Pfn1-transfected HEK293T cells, the stoichiometry of PKA-mediated Ser 71 phosphorylation of total cellular pool of Pfn1 appeared low, as overexpression of the alpha catalytic subunit of PKA did not cause detectable reduction in the levels of associated actin (data not shown), despite increasing Ser 71 phosphorylation levels. This could be due to several possible reasons. First, PKA-mediated Ser 71 phosphorylation may occur to a small fraction of total cellular Pfn1 either transiently or at specific subcellular locations. Given the high cellular abundance of Pfn1 (>50 µM in most tissues) (Witke, 2004), the net effect on steady-state actin-binding would not be detectable by co-IP using whole cell lysates. Second, PKA activity is regulated by complex mechanisms including inhibitory binding of the regulatory subunits as well as tethering to specific subcellular locations via the diverse A-kinase-anchoring proteins (AKAP) (Francis and Corbin, 1994;Harada et al., 1999). Therefore, it is possible that overexpressing the catalytic subunit of PKA alone is insufficient to achieve optimal Pfn1 phosphorylation. In either case, more molecular details regarding the newly identified PKA/pSer 71 -Pfn1 axis and the possible involvement of other kinases remain to be determined in the future.
Despite being an essential actin-binding protein, Pfn1 simultaneously functions as a non-classical tumor suppressor across different malignancies including breast (Roy and Jacobson, 2004;Wittenmayer et al., 2004;Zou et al., 2007Zou et al., , 2009Zou et al., , 2010Bae et al., 2009Bae et al., , 2010Das et al., 2009;Diamond et al., 2015), pancreatic (Yao et al., 2014), and liver (Wu et al., 2006) cancers. Its ability to inhibit cell cycle progression and tumor cell growth has been demonstrated by many in vitro and in vivo studies. Our recent work suggested that tumor growth inhibition by Pfn1 is mediated at least in part by its nuclear function in repressing SEC-dependent transcription of pro-cancer genes including c-MYC (Diamond et al., 2015;Zhu et al., 2021). Such an activity can be toggled off and on by Ser 137 phosphorylation and dephosphorylation (mimicked by S137D and S137A) which, respectively, blocks and enables Pfn1 binding to PLPs that are present in the SEC component ENL (Diamond et al., 2015;Zhu et al., 2021). In this paper, by analyzing the effects of pSer 71 -resistant and mimetic mutants (S71A and S71D) in MDA-MB-231 cells, we made the unexpected observation that both cannot inhibit tumor growth. This indicates that, rather than functioning as an on/off switch, reversible phosphorylation and dephosphorylation at Ser 71 , causing dynamic actin dissociation and rebinding, are required for Pfn1 to inhibit tumor growth. Given that actin-binding promotes nuclear export of Pfn1 (Stuven et al., 2003), this raises an interesting possibility that Ser 71 phosphorylation may occur in a transient and tightly controlled manner to nuclear Pfn1 to allow its escape from nuclear export, which is subsequently followed by Ser 71 dephosphorylation to render actin-rebinding by Pfn1 for its functional engagement in transcriptional repression. Though speculative at the moment, this theory is consistent with our current knowledge regarding Pfn1 functions and can be tested in the future.
In addition to suppressing tumor growth, Pfn1 also sensitizes cancer cells to drug-induced apoptosis. This has been best demonstrated in breast cancer cells treated with cytotoxic agents many of which are common cancer chemotherapies including paclitaxel (Zou et al., 2010;Yao et al., 2013;. Although detailed molecular mechanisms underlying apoptosissensitization by Pfn1 remain unknown, such an activity has been linked to p53 and NFκB signaling. It was found that Pfn1 and p53 co-exist in the same complex in breast cancer cells, and Pfn1 overexpression increases total p53 protein levels as well as its redistribution to cytoplasm and mitochondria (Yao et al., 2013; where p53 can drive intrinsic apoptotic pathway in a transcription-independent fashion (Marchenko et al., 2000;Chipuk et al., 2004). In support of the functional link between Pfn1 and p53, apoptosissensitizing effect of Pfn1 was greatly reduced in p53-null and knockdown cells (Yao et al., 2013;. In addition to enhancing p53 activity, Pfn1 was also found to decrease NFκB signaling by preventing cytotoxin-induced IκBα phosphorylation and degradation and consequently preventing p65 nuclear translocation and transcription of pro-survival genes . Although these studies implicate the involvement of cytoplasmic Pfn1 in apoptotic sensitization, direct evidence was unavailable. In this study, by tagging Pfn1 with NES or NLS, we showed that apoptotic sensitization by Pfn1 indeed requires its cytoplasmic localization. Interestingly, such an activity of Pfn1 can be switched off and on by pSer 71 -Pfn1 mimicking and preventing mutants both in vitro and in vivo, indicating an essential role of actin-binding. Our data suggested that pSer 71 -Pfn1 could be a novel predictive biomarker for cancer chemotherapy response. They are also consistent with the well-known pro-survival effects of PKA, which have been mechanistically linked to activating phosphorylation of the cAMP-response element binding protein (CREB) in the nucleus (Wilson et al., 1996;De Cesare and Sassone-Corsi, 2000;Mayr and Montminy, 2001;Naqvi et al., 2014) and deactivating phosphorylation of the pro-apoptotic protein Bad in the cytoplasm (Harada et al., 1999;Lizcano et al., 2000). Thus, Pfn1 may be a previously unknown downstream effector of PKA function in apoptotic inhibition through Ser 71 phosphorylation. Given that Ser 71 phosphorylation prevents nuclear Pfn1 export, we speculate that it may inhibit the apoptosis-sensitizing effect of Pfn1 via a two-pronged mechanism by abolishing its actinbinding and reducing its cytoplasmic levels ( Figure 6F). Taken together, our data in this paper demonstrated that Ser 71 is a bona fide phosphorylation site of Pfn1 capable of inhibiting its actinbinding ability, preventing its nuclear export, and influencing its tumor-inhibitory functions.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by The Animal Studies Committee at Washington University in St. Louis.