*Aggregatibacter actinomycetemcomitans* **LtxA Hijacks Endocytic Tra**ffi**cking Pathways in Human Lymphocytes**

**Edward T Lally 1, Kathleen Boesze-Battaglia 1, Anuradha Dhingra 1, Nestor M Gomez 1, Jinery Lora 1, Claire H Mitchell 1, Alexander Giannakakis 1, Syed A Fahim 1, Roland Benz <sup>2</sup> and Nataliya Balashova 1,\***


Received: 17 December 2019; Accepted: 16 January 2020; Published: 21 January 2020

**Abstract:** Leukotoxin (LtxA), from oral pathogen *Aggregatibacter actinomycetemcomitans*, is a secreted membrane-damaging protein. LtxA is internalized by β2 integrin LFA-1 (CD11a/CD18)-expressing leukocytes and ultimately causes cell death; however, toxin localization in the host cell is poorly understood and these studies fill this void. We investigated LtxA trafficking using multi-fluor confocal imaging, flow cytometry and Rab5a knockdown in human T lymphocyte Jurkat cells. Planar lipid bilayers were used to characterize LtxA pore-forming activity at different pHs. Our results demonstrate that the LtxA/LFA-1 complex gains access to the cytosol of Jurkat cells without evidence of plasma membrane damage, utilizing dynamin-dependent and presumably clathrin-independent mechanisms. Upon internalization, LtxA follows the LFA-1 endocytic trafficking pathways, as identified by co-localization experiments with endosomal and lysosomal markers (Rab5, Rab11A, Rab7, and Lamp1) and CD11a. Knockdown of Rab5a resulted in the loss of susceptibility of Jurkat cells to LtxA cytotoxicity, suggesting that late events of LtxA endocytic trafficking are required for toxicity. Toxin trafficking via the degradative endocytic pathway may culminate in the delivery of the protein to lysosomes or its accumulation in Rab11A-dependent recycling endosomes. The ability of LtxA to form pores at acidic pH may result in permeabilization of the endosomal and lysosomal membranes.

**Keywords:** *Aggregatibacter actinomycetemcomitans*; RTX toxin; localized aggressive periodontitis; LFA-1; leukotoxin (LtxA); endocytosis

#### **1. Introduction**

The RTX (Repeats in ToXin) toxins are membrane-damaging proteins secreted by some Gram-negative bacteria [1]. The organisms producing these proteins are important human and animal pathogens, implicating the toxins' role in the bacterial virulence. RTX toxins' common features are using the type I secretion system as a mode of export across the bacterial envelope, employing an uncleaved C-terminal recognition signal [2–4], and the characteristic nonapeptide glycine- and aspartate-rich repeat binding of Ca2<sup>+</sup> ions [5,6]. The toxins are modified with fatty acid moieties attached to internal lysine residues, which is an unusual characteristic for bacterial proteins [7–10]. RTX toxins can be divided into three groups: (i) broadly cytolytic RTX hemolysins, (ii) species-specific RTX leukotoxins, and (iii) large, multifunctional, autoprocessing RTX toxins (MARTX) [1]. RTX

leukotoxins exhibit a narrow cell type and species specificity due to cell-specific binding through protein receptors of the β<sup>2</sup> integrin family [1]. The β<sup>2</sup> integrins are expressed on the surface of leukocytes and share a common β<sup>2</sup> subunit, CD18, which is combined with either one of the unique α chains, α<sup>L</sup> (CD11a), α<sup>M</sup> (CD11b), α<sup>X</sup> (CD11c), or α<sup>D</sup> (CD11d) [11].

*Aggregatibacter actinomycetemcomitans* (*Aa*), a facultative anaerobe and common inhabitant of the human aerodigestive tract, causes localized aggressive periodontitis (LAP) [12]. LAP is a rapidly progressing periodontal disease that results in loss of tooth attachment and alveolar bone destruction in adolescents. If left untreated in teenagers, the infection will lead to the loss of the permanent first molars and central incisors [13]. Recent data indicate that *Aa* plays a role in the early stages of the disease. Specific *Aa* virulence factors can trigger the disease by suppressing the host response, which will allow for the overgrowth of *Aa* and other "toxic" bacteria in the local environment [12]. The pivotal virulence factor of *Aa* is an RTX leukotoxin, LtxA, that kills both human innate and adaptive immune cells [14]. *Aa* isolated from LAP patients predominantly belongs to a single clone, JP2 [15], which is characterized by increased LtxA production, implicating a role for LtxA in disease development [16]. Analysis of a primary LtxA sequence consisting of 1055 amino acids predicts four LtxA domains [17]. The hydrophobic domain encompasses residues 1–420 and incorporates cholesterol recognition amino acid consensus (CRAC) [18]. CRAC motif mediates LtxA binding to cholesterol and is essential for LtxA association with the plasma membrane of human T lymphocytes and monocytes [18,19]. The central domain (residues 421–730) contains two internal lysine residues (K562 and K687) that are the sites of post-translational acylation, required for LtxA activation [9]. The repeat domain (residues 731–900) contains the typical repeated amino acid sequence of the RTX family with the C-terminal domain (residues 901–1055), and is believed to play a role in secretion [17].

Recent findings suggest that recirculating and resident memory T cells in gingival tissue play an important role in the maintenance of periodontal homeostasis [20]. In an experimental rat periodontal disease model, antigen-specific CD4 T lymphocytes were required for bone resorption [21]. Hence, the investigation of the LtxA effect on T lymphocytes is important for our understanding of how *Aa* causes periodontal disease. In our previous studies, the Jurkat cell line, subclone Jn.9, served as a model to study LtxA interaction with T lymphocytes' cell membrane. Jn.9 cells express cell-surface LFA-1 and are susceptible to LtxA-induced toxicity [22]. LtxA toxicity requires the presence of the β<sup>2</sup> integrin LFA-1(LFA-1, CD11a/CD18 or αL/β2) and cholesterol on the surface of Jn.9 cells [18,23,24]. LFA-1 is a native ligand for intercellular adhesion molecule (ICAM-1) located on vascular endothelial cells [25]. In immunocytes, LFA-1/ICAM-1 binding is one of the molecular mechanisms for leukocyte adhesion and migration to the site of infection [26]. LFA-1 is constantly endocytosed and then rapidly recycled back to the plasma membrane through vesicular transport [27,28] using the "long-loop" of recycling involving GTPase Rab11A-positive endosomes [29]. Also, LFA-1 activity is regulated by the ability of these receptors to switch between active and inactive conformations [25].

In the proposed mechanism of LtxA interaction with the Jn.9 cells membrane, the initial binding of the toxin with the membrane elevates cytosolic Ca2<sup>+</sup> independent of the toxin binding to LFA-1. Ca2<sup>+</sup> elevation involves the activation of calpain and talin cleavage, and the subsequent clustering of LFA-1 in lipid rafts on the membrane [24]. LtxA binds to the extracellular domains of LFA-1 subunits, CD11a and CD18. The toxin then transverses the cell membrane, binds to the cytoplasmic tails of LFA-1, and causes activation of LFA-1 [23]. Following from the results of the liposomal study, LtxA adopts a U-shaped conformation in the membrane, with the N- and C-terminal domains residing outside of the membrane [30].

After binding to the LFA-1 subunits, LtxA is quickly internalized into the cytosol, where it is found in vesicular structures [23]. Since LtxA binds to LFA-1, there is a possibility that LtxA could be using an integrin endocytic trafficking pathway to gain access to the target cell cytosol. The mechanism of LtxA uptake and the pathway of intracellular toxin trafficking has not been investigated. Here, we examined the components of the cytosol of LtxA-treated cells for co-localization of the toxin and the CD11a subunit of LFA-1 with different organelle markers. LtxA association with endosomal and

lysosomal markers suggests a receptor-mediated endocytic process that may culminate in the delivery of the toxin to lysosomes. Additionally, the toxin can be redirected to the plasma membrane due to the LFA-1 receptor Rab11A-mediated recycling. This study provides new insight into convergent mechanisms of LFA-1 and LtxA trafficking, and the ability of LtxA to function in acidic environments.

#### **2. Results**

#### *2.1. LtxA Does Not Damage Host Cell Membrane When Entering the Cell*

The membrane damaging properties of LtxA have been documented [31,32]. Therefore, the first question we asked was whether the initial steps of LtxA interaction with the cells result in the plasma membrane damage. The green-fluorescent impermeable nucleic acid stain YO-PRO®-1 is used to detect early membrane damage as it permeates cells immediately after membrane destabilization [33]. Propidium iodide (PI) is used to identify late cell death in which the integrity of the plasma and nuclear membranes significantly decreases, allowing PI to penetrate the membranes and intercalate into nucleic acids [34]. To study the effect on membrane permeability, we first incubated Jn.9 cells with 20 nM LtxA at different times over a 10 h time period, and then flow cytometry analysis of LtxA-treated cells was performed to determine YO-PRO®-1 and PI internalization. The YO-PRO®-1 membrane permeabilization assay showed no evidence of plasma membrane damage in LtxA-treated Jn.9 cells at least within first 3 h of treatment (Figure 1A,B). However, our flow cytometry data demonstrated that 20 nM LtxA-DY488 became internalized with Jn.9 cells 30 min after the toxin was added and internalization steadily increased over time (Figure 1A,B). Lymphocytes are known to be moderately susceptible to LtxA and are killed by apoptosis [35]. The staining of Jn.9 cells with PI was observed after 10 h of treatment with LtxA, suggesting that cells are in the late apoptotic stage (Figure 1B). Hence, our data indicate that LtxA is quickly internalized by Jn.9 cells but the toxin does not rupture the plasma membrane when it enters the host cells.

**Figure 1.** Damage to the plasma membrane in Jn.9 cells by LtxA. Flow cytometry analysis was used to detect YO-PRO®-1, PI and LtxA internalization with Jn.9 cells over time. Cells (1 <sup>×</sup> <sup>10</sup>6) were incubated with 0.1 μM YO-PRO®-1/ 1.5 μM PI alone or after treatment with 20 nM LtxA at indicated times at 37 ◦C. Another set of cells was treated with 20 nM LtxA-DY488 at different times. The extracellular fluorescence of the cells was quenched with 0.025% trypan blue [23] and the intracellular fluorescence was determined. (**A**). Uptake of YO-PRO®-1 (black) and internalization of LtxA-DY488 (red) at various times, presented as mean channel fluorescence (MCF). The data shown are the results of three independent experiments. Error bars indicate ± SEM, \* *p* ≤ 0.05. (**B**). Top and Middle: Flow cytometry histograms showing YO-PRO®-1 and PI dyes uptake by LtxA-treated cells (red line) vs. the dyes uptake by untreated Jn.9 cells (black line) at different times. Bottom: Flow cytometry histograms showing LtxA-DY488 internalized with Jn.9 cells. The data shown are representative of three independent experiments.

#### *2.2. LtxA Uptake is Diminished by Dynamin Inhibitors*

We hypothesized that LtxA could get access to the cytosol of Jn.9 cells through endocytic uptake. The exposure of cells to cold temperatures is used for the nonspecific inhibition of endocytosis [36]. Therefore, we treated Jn.9 cells with fluorescent-labeled LtxA in ice-cold medium and analyzed the toxin entry to cells by flow cytometry and confocal microscopy. We demonstrated that cold temperatures affected the binding of and slowed down internalization of the toxin with Jn.9 cells (Figure 2A, Figure S1). We then set up an experiment employing a set of chemical and pharmacological inhibitors of endocytosis (Table 1) to define the mechanism of the toxin uptake by the cells. The effect of dynamin- and clathrin-mediated endocytosis inhibitors on transferrin uptake is well established [37,38]. To confirm the efficiency and select the inhibitors' concentrations for our study, the internalization of transferrin conjugated to Alexa Fluor®555 and LtxA-DY650 by Jn.9 cells was followed using confocal microscopy Figure S2. At 10 μM Dynasore, 10 μM Dynole 34-2 and 5 μM Pitstop 2, the inhibitory effect on transferrin uptake was observed. First, we wanted to identify whether GTPase dynamin activity is essential for LtxA uptake. The LtxA-DY650 internalization in the presence of dynamin inhibitor 10 μM Dynasore was evaluated by live confocal imaging (Figure S3) and flow cytometry analysis (Figure 2C). To confirm our results, we performed flow cytometry analysis in cells treated with another dynamin inhibitor, 10 μM Dynole 34-2, which blocks the GTPase activity of dynamin [37,39]. Fluorescent-labeled toxin internalization was significantly reduced in cells pre-treated with dynamin-inhibitors. Cells pretreated with 10 μM Dynasore and Dynole 34-2 for 20 min internalized much less LtxA (Figure 2C). However, the inactive control for Dynole 34-2, Dynole 31-2, did not inhibit the toxin internalization. The inhibitors affecting the clathrin-mediated endocytic pathway, such as potassium-depleted medium [40] and 5 μM Pitstop 2 [41], did not change the efficiency of LtxA internalization with Jn.9 cells (Figure 2B,C). Collectively, these data suggest that LtxA internalization with Jn.9 cells is dynamin-dependent and predominantly clathrin-independent. Next, we evaluated the toxicity of LtxA on Jn.9 cells containing the above inhibitors. However, we identified some toxic effect of endocytosis inhibitors on Jn.9 cells. After 18 h of treatment with the 2 nM toxin, we observed some decrease in LtxA toxicity on Jn.9 cells pretreated with 10 μM Dynasore and 10 μM Dynole 34-2 (Figure S4), suggesting that these compounds could attenuate LtxA intoxication.


**Table 1.** Chemical inhibition of LtxA uptake.

\* To measure LtxA internalization inhibition, Jn.9 cells were preincubated with 5–10 μM inhibitors for 20 min in serum free medium. The 0.5–1 mM chemical stocks were prepared in dimethyl sulfoxide (DMSO) and were added in the 1 μl volume to 1 ml of cells. No adverse effect of DMSO alone on Jn.9 cells was observed.

**Figure 2.** Effect of endocytosis inhibitors on LtxA internalization. (**A**). Flow cytometry analysis of LtxA internalization with Jn.9 cells at different temperatures. The cells were treated with LtxA-DY488 for 30 min on ice or 37 ◦C. In a set of cells, the total cell-associated fluorescence was measured by flow cytometry analysis (shown in red). In another set of cells, the extracellular fluorescence was quenched (0.025% trypan blue) [42] and intracellular fluorescence (red peak) was determined by flow cytometry analysis. (**B**). Flow cytometry analysis of LtxA internalization with Jn.9 cells in K+-free buffer. Jn.9 cells (1 <sup>×</sup> 106) were incubated in K<sup>+</sup>-containing (top) or K+-free buffer (bottom), and then 20 nM LtxA-DY488 was added for 30 min. Flow cytometry analysis to determine the amount of internalized toxin (red peak) was performed as described in Figure 2A. (**C**). Flow cytometry analysis of LtxA-DY488 internalization with Jn.9 cells pretreated with chemical inhibitors. Jn.9 cells (1 <sup>×</sup> <sup>10</sup>6) were preincubated with 5–10 <sup>μ</sup><sup>M</sup> endocytosis inhibitors for 20 min in serum free medium, and then were treated with 20 nM LtxA-DY488 for 30 min at 37 ◦C. The extracellular fluorescence was quenched 0.025% trypan blue, and intracellular fluorescence (red peak) was determined by flow cytometry analysis.

#### *2.3. LtxA and CD11a Are Found in Early and Recycling Endosomes*

In our imaging studies, LtxA was found in vesicular structures after entry into Jn.9 cells [23]. The co-distribution of LtxA and LFA-1 heterodimer components on the surface of target cell membranes indicates that LtxA could intrude into the cytosol as individual LtxA molecules or as part of an LtxA/LFA-1 complex. In order to characterize LtxA-containing endocytic vesicles, Jn.9 cells were treated with fluorescent-labeled LtxA for 30 min and were used to perform immunocytochemistry experiments with endocytic pathway markers, including GTPase Rab5 and Rab11A. Our imaging studies demonstrated the abundant colocalization of LtxA, early endosome membrane protein Rab5 and CD11a, suggesting toxin uptake through receptor-mediated endocytosis. Figure 3 and Figure S5

show confocal images of Jn.9 cells with a co-localization of LtxA, CD11a, and Rab5 after treatment of the cells with LtxA-DY650 for 30 min at 37 ◦C.

**Figure 3.** LtxA localization in early endosomes of Jn.9 cells. The cells were treated with 20 nM LtxA for 30 min at 37 ◦C. (**A**). 3D confocal images showing the distribution of LtxA, CD11a and Rab5. LtxA is pseudo colored in red and CD11a is in red or green (pseudo colored), as indicated on images. The 3D images were reconstructed from seventeen confocal planes using Nikon Elements AR 4.30.01 software. Bounding box dimensions are: width 14.19 μm; height 17.30 μm; depth 5.20 μm. (**B**). Localization of LtxA-DY650 is shown in cyan, CD11a recognized with mouse Alexa Fluor™ 594 clone HI111 is shown in red, and Rab5, recognized by rabbit anti-Rab5 antibody followed by staining with anti-rabbit IgG Alexa Fluor®488, is shown in green. The nucleus was stained with Hoechst dye and is shown in blue. (**C**). Top: Merged image "B" showing colocalization of LtxA DY650 (cyan), CD11a (red) and Rab5 (green). Bottom: Intensity profiles for LtxADY650 (cyan), CD11a (red) and Rab5 (green) across the line depicted in the image above. The degree of overlap in the LtxA-containing area was estimated with the Pearson's correlation coefficient in the LtxA-containing area of 0.78 for LtxA and Rab5, 0.68 for LtxA and CD11a, and of 0.88 for CD11a and Rab5. Representative cells are shown. Additional data are shown in Figure S5.

LFA-1 is exocytosed via GTPase Rab11A-mediated recycling [43] a process that involves trafficking through the perinuclear recycling compartment (PNRC), before reaching the plasma membrane. We found co-localization of CD11a and LtxA with Rab11A, a marker of recycling endosomes in approximately 1/3 of LtxA-containing spots (Figure 4 and Figure S6). The interaction of CD11a and LtxA with Rab11A in recycling suggests that after entering the early endosome a significant amount of LtxA is redirected back to the membrane in the LFA-1 recycling turnover. Alternatively, the release of

LtxA into PNRC can provide access to the nuclear membrane for LtxA. Indeed, in our imaging studies we often observed the toxin surrounding nuclei (Figure S7).

**Figure 4.** LtxA localization in Rab11A-positive endosomes of Jn.9 cells. The cells were treated with 20 nM LtxA for 30 min at 37 ◦C. (**A**). 3D confocal images showing the distribution of LtxA, CD11a and Rab11A. LtxA is pseudo colored in red and CD11a is in red or green (pseudo colored), as indicated on images. The 3D images were reconstructed from seventeen confocal planes using Nikon Elements AR 4.30.01 software. Bounding box dimensions are: width 22.58 μm; height 16.47 μm; depth 5.20 μm. (**B**). Localization of LtxA-DY650 is shown in cyan, CD11a recognized with mouse Alexa Fluor™ 594 clone HI111 is shown in red, and Rab11A, recognized by rabbit anti-Rab11A antibody followed by staining with anti-rabbit IgG Alexa Fluor®488, is shown in green. The nucleus was stained with Hoechst dye and is shown in blue. (**C**). Top: Merged image "B" showing co-localization of LtxA DY650 (cyan), CD11a (red) and Rab11A (green). Bottom: Intensity profiles for LtxADY650 (cyan), CD11a (red) and Rab11A (green) across the line depicted in the image above. The degree of overlap in the LtxA-containing area was estimated with the Pearson's correlation coefficient of 0.75 for LtxA and Rab11A, 0.72 for LtxA and CD11a, and of 0.76 for CD11a and Rab11A. Representative cells are shown. Additional data are shown in Figure S6.

#### *2.4. LtxA and CD11a Are Found in Late Endosomes and Lysosomes*

Atlater timepoints, Jn.9 cells treatedwith fluorescent-labeled LtxAwere usedinimmunocytochemistry experiments with the endocytic pathway markers GTPases Rab7 and Lamp1. After 1 h of treatment with LtxA-DY650, LtxA was associated with the late endosome membrane protein Rab7 (Figure 5

and Supplemental Figure S8). Colocalization was detected in approximately 1/10 of LtxA-containing spots. Colocalization of LtxA with lysosomal marker Lamp1 after 2 h of treatment with LtxA-DY650 indicated that the toxin trafficking culminates in its delivery to the lysosomes, where LtxA was found separated from CD11a (Figure 6 and Figure S9).

**Figure 5.** LtxA localization in Rab7-positive endosomes of Jn.9 cells. The cells were treated with 20 nM LtxA for 1 h at 37 ◦C. (**A**). 3D confocal images showing the distribution of LtxA, CD11a and Rab7. LtxA is pseudo colored in red and CD11a is in red or green (pseudo colored), as indicated on images. The 3D images were reconstructed from seventeen confocal planes using Nikon Elements AR 4.30.01 software. Bounding box dimensions are: width 22.17 μm; height 19.37 μm; depth 5.20 μm. (**B**). Localization of LtxA-DY650 is shown in cyan, CD11a recognized with mouse Alexa Fluor™ 594 clone HI111 is shown in red, and Rab7, recognized by rabbit anti-Rab7 antibody followed by staining with anti-rabbit IgG Alexa Fluor®488, is shown in green. The nucleus was stained with Hoechst dye and is shown in blue. (**C**). Top: Merged image "B" showing colocalization of LtxA DY650 (cyan), CD11a (red) and Rab7 (green). Bottom: Intensity profiles for LtxADY650 (cyan), CD11a (red) and Rab7 (green) across the line depicted in the image above. The degree of overlap in the LtxA-containing area was estimated with the Pearson's correlation coefficient of 0.88 for LtxA and Rab7, 0.03 for LtxA and CD11a, and of 0.13 for CD11a and Rab7. Representative cells are shown. Additional data are shown in Supplementary data Figure S8.

**Figure 6.** LtxA localization in lysosomes of Jn.9 cells. The cells were treated with 20 nM LtxA for 2 h at 37 ◦C. (**A**). 3D confocal images showing the distribution of LtxA, CD11a and Lamp1. LtxA is pseudo colored in red and CD11a is in red or green (pseudo colored), as indicated on images. The 3D images were reconstructed from seventeen confocal planes using Nikon Elements AR 4.30.01 software. Bounding box dimensions are: width 25.17 μm; height 19.37 μm; depth 5.20 μm. (**B**). Localization of LtxA-DY650 is shown in cyan, CD11a recognized with mouse Alexa Fluor™ 594 clone HI111 is shown in red, and Lamp1, recognized by rabbit anti-Lamp1 antibody followed by staining with anti-rabbit IgG Alexa Fluor®488, is shown in green. The nucleus was stained with Hoechst dye and is shown in blue. (**C**). Top: Merged image "B" showing colocalization of LtxA DY650 (cyan), CD11a (red) and Lamp1 (green). Bottom: Intensity profiles for LtxADY650 (cyan), CD11a (red) and Lamp1 (green) across the line depicted in the image above. The degree of overlap in the LtxA-containing area was estimated with the Pearson's correlation coefficient of 0.72 for LtxA and Lamp1, 0.15 for LtxA and CD11a, and of 0.11 for CD11a and Lamp1. Representative cells are shown. Additional data are shown in Supplementary data Figure S9.

#### *2.5. Rab5 siRNA Knockdown limits LtxA Toxicity*

Irrespective of routes of internalization, endocytic cargoes are trafficked to early endosomes, where Rab5 GTPases is the key player in subsequent trafficking events [44]. We investigated the impact of Rab5a downregulation on LtxA uptake and toxicity on cells (Figure 7). Western blot analysis 24 h after transfection with Rab5a siRNA confirmed that Rab5a was significantly downregulated (≥ 90%) in Jn.9 cells compare to scrambled siRNA transfected cells. When transfected cells were treated with 20 nM LtxA for 18 h, the toxic effect of the toxin on Rab5a downregulated cells was 30% less than on control cells (Figure 7A). Internalization of LtxA was analyzed by flow cytometry after 30 min of treatment with 20 nM LtxA-DY488. No significant variations in the amount of internal fluorescence were detected in cells transfected with Rab5a siRNA (mean channel fluorescence (MCF) 28.2 ± 0.6) and

cells using scrambled siRNA (MCF 29.8 ± 0.6) (Figure 7B). Our results suggest that the abolishment of Rab5a function does not affect LtxA internalization, but affects cytotoxicity.

**Figure 7.** Modulation of Rab5a function in Jn.9 cells. A. Jn.9 cells (1 <sup>×</sup> <sup>10</sup><sup>6</sup> cells) were transfected with siRNA control (SCR) or with siRNA against Rab5a and collected 24 h post-transfection for Rab5a expression analysis by Western blotting. The cell viability testing was performed by trypan blue assay after 18 h of treatment with 20 nM LtxA. (**A**). representative expression of Rab5a protein was shown for 24 h of siRNA treatment. The Rab5a protein expression (inset) was analyzed in extracts obtained from <sup>1</sup> <sup>×</sup> 106 Jn.9 cells by Western blot, <sup>β</sup>-actin served as a loading control. Error bars indicate <sup>±</sup>SEM, \* *p* ≤ 0.05 compared with siRNA SCR-treated cells. The experiment was performed three independent times. (**B**). Jn.9 cells (1 <sup>×</sup> 106 cells) were transfected with siRNA control (SCR) or with siRNA against Rab5a, then were collected 24 h post-transfection and treated with 20 nM LtxA-DY488 for 30 min at 37 ◦C. The extracellular fluorescence of the cells was quenched (0.025% trypan blue) [42,45] and intracellular cell fluorescence (red peak) was determined by flow cytometry analysis. No residual fluorescence was detected in 0.1% Triton X-100 permeabilized cells after the trypan blue treatment. Untreated cells (black) served as a negative control. Representative flow cytometry histograms are shown.

#### *2.6. LtxA Causes Lysosomal Damage in Jn.9 Cells*

We detected LtxA in Jn.9 lysosomes, and therefore we wanted to see whether LtxA was able to cause lysosomal damage in the cells. We have probed the effect of LtxA on lysosomal integrity in the cells using lysosomal dye, LysoTracker®Green DND-26. We followed changes in lysosomal properties of the cells after the addition of 20 nM LtxA to the cells by live cell confocal imaging. No changes in LysoTracker staining intensity were detected within the first 90 min of treatment and about 15% decrease in the intensity was identified in Jn.9 cells after 2 h of treatment (Figure 8A), which may indicate lysosomal damage due to lysosomal membrane permeabilization or lysosome alkalization. In order to assess if LtxA causes the rupture of lysosomes, we identified the intensity of Lamp1 staining in the presence and absence of toxin. Similar Lamp1 staining intensity in both conditions would indicate that this is likely because of a rise in pH in intact lysosomes. A decrease in Lamp1 intensity would suggest that change in lysosomal pH is due to lysosomal rupture (Figure 8B).

**Figure 8.** Lysosomal damage by LtxA. Jn.9 cells (1 <sup>×</sup> <sup>10</sup>6) were incubated in presence or absence of 20 nM LtxA for 2 h at 37 ◦C. (**A**). LtxA-treated and untreated cells were stained with 100 nM LysoTracker® Green DND-26 for 15 min at 37 ◦C. The LysoTracker® Green DND-26 intensity was evaluated by flow cytometry analysis. Representative flow cytometry histograms are shown on the left. The fluorescence of LtxA treated cells is shown in red, and untreated cells in black. The mean channel fluorescence (MCF) of LysoTracker® Green DND-26 in LtxA-treated vs. untreated cells is presented on the right. (**B**). Average fluorescence intensity of Lamp1 staining in Jn.9 cells was evaluated by confocal microscopy. The intensities of Lamp1 staining in 39 LtxA-treated cells and 44 untreated were analyzed and shown on the right. Error bars indicate ± SEM of three independent experiments. \* *p* ≤ 0.05.

#### *2.7. LtxA is Active in Lipid Bilayer Membranes at a Low pH*

Pore formation by LtxA was studied in detail at a neutral pH [32,46]. In lipid bilayer membranes formed by asolectin, LtxA forms cation-selective channels with a single-channel conductance of approximately 1.2 nS in 1 M KCl (pH 6.0) [46]. Since LtxA is found in endocytic vesicles, we asked whether LtxA is also able to form ion-permeable channels and damage membranes at an acidic pH. To address this, we performed lipid bilayer experiments with wildtype LtxA at different pH-values ranging from pH 3.5 to pH 10.0. LtxA formed ion-permeable channels in 1 M KCl solutions under all these conditions (pH 3.5, 4.7, 7.5, 8.5 and 10.0). However, because the membranes became very fragile at very low and very high pH values (3.7 and 10.0) it was not possible to record too many single-channel events under these conditions. At the other pH values, the membranes were rather stable, and a sufficient number of single-channel events could be recorded in the experiments. Figure 9 shows a single channel recording of LtxA in 1 M KCl, 10 mM MES-KOH, pH 4.7. The channel had a somewhat reduced lifetime at this pH as compared with a neutral pH [46]. Figure 9B,C shows a histogram obtained from 47 LtxA channels recorded under these conditions. A fit of the histogram with a Gaussian function yielded an average single-channel conductance of 1.1 ± 0.3 nS, somewhat smaller than that at pH 6.0 (G = 1.2 ± 0.3 nS) [46]. Again, we found that the single-channel distribution was quite broad, similar to the conditions at pH 6.0 (Figure 9B,C, Table 2).

**Figure 9.** Pore-forming activity of LtxA in asolectin/*n*-decane membranes at different pH values. (**A**). Single-channel recording of LtxA in an asolectin/n-decane membrane at pH 4.7. Current recording of an asolectin/n-decane membrane, performed in the presence of 10 nM LtxA added to the cis-side of the membrane. The aqueous phase contained 1 M KCl, 10 mM MES-KOH, pH 4.7. The applied membrane potential was 20 mV at the cis-side (indicated by an arrow), at 20 ◦C. (**B**). Histogram of the probability P (G) of an occurrence of a given conductivity unit observed for LtxA with membranes formed of 1% asolectin dissolved in *n*-decane in a salt solution at pH 4.7. The histogram was calculated by dividing the number of fluctuations with a given conductance unit by the total number of conductance fluctuations. The average conductance was 1.1 ± 0.31 nS for 47 conductance steps derived from nine individual membranes. The value was calculated from a Gaussian distribution of all conductance fluctuations (solid line). The aqueous phase contained 1 M KCl, 10 mM MES-KOH, pH 4.7 and 10 nM LtxA; the applied membrane potential was 20 mV at 20 ◦C. (**C**). Histogram of the probability P(G) for the occurrence of a given conductivity unit observed for LtxA with membranes formed of 1% asolectin dissolved in n-decane in a salt solution at pH 6.0. The average conductance was 1.20 ± 0.31 nS for 95 conductance steps derived from 17 individual membranes. The aqueous phase contained 1 M KCl, 10 mM MES, pH 6.0 and about 10 nM LtxA; the applied membrane potential was 20 mV at 20 ◦C.

We also studied the effect of high pH values on channel formation, mediated by LtxA. Ion-permeable channels were also observed at these conditions. The average single channel conductances at pH 7.5, 8.5 and 10 are shown in Table 2. The influence of the aqueous pH was rather small on the conductance of the LtxA channel, despite a possible shift of the selectivity of the LtxA channel from being slightly cation selective at pH 6.0 to a higher selectivity for potassium ions over chloride.

No residual fluorescence was detected in 0.1% Triton X-100 permeabilized cells after the trypan blue treatment. Untreated cells (blue or black) served as a negative control. Representative flow cytometry histograms are shown.


**Table 2.** Influence of the aqueous pH on the conductance of channels formed by LtxA.

\* The LtxA conductance (G ± variance/SD) in each 1 M KCl solution was either taken from Gaussian distributions (see Figure 9) or directly from the statistics of single-channel data (n number of single events). To analyze the conductance in each case, n channels were reconstituted in asolectin/n-decane membranes at 20 mV voltage at 20 ◦C. The number of events analyzed at pH 3.7 and 10 was low due to the instability of the lipid bilayers at extreme pH values.

#### **3. Discussion**

Leukocytes need to rapidly move from blood vessels to tissues upon inflammation or infection. A crucial mechanism regulating this process is the subcellular trafficking of adhesion molecules, primarily integrins [47]. Integrins undergo constant endo/exocytic turnover, necessary for the dynamic regulation of cell adhesion. Bacterial toxins have developed a number of schemes to cross the membrane in order to enter the cell. LtxA evolved the strategy to target specifically β<sup>2</sup> integrin LFA-1 on leukocytes' surface [22]. This binding is required for toxin internalization [23].

We here report that LtxA is delivered to the cytosol of Jn.9 cells through endocytic trafficking. Historically, endocytic pathways are classified as either clathrin-dependent or clathrin-independent. The large GTPase dynamin [48] is hypothesized to be directly involved in closing off endocytic vesicles from the plasma membrane. The key players in the formation of clathrin-coated vesicles are dynamin [48] and adaptor proteins [49]. The studies with *Mannheimia haemolytica* LktA, another RTX leukotoxin, show that LktA is internalized in a *dynamin*-*2* and *clathrin*-*dependent* manner [50]. The following LktA-trafficking events involve the toxin binding to the mitochondria and interaction with cyclophilin D, a mitochondrial chaperone protein, in bovine lymphoblastoid cells [51].

Our data indicate that LtxA enters Jn.9 cells using a clathrin-independent mechanism (or predominantly uses this pathway). Our results correlate with the finding that LFA-1 is internalized through a clathrin-independent, cholesterol-dependent pathway and this process is essential for cell migration [52]. In this scenario, non-clathrin-coated lipid raft microdomains form 50–100 nm flask-shaped vesicles in the plasma membrane regions rich in lipid rafts [53]. Lipid-raft dependent endocytosis was shown to be dynamin-dependent [54] and may involve caveolae formation. Thus, we hypothesize that LtxA/LFA-1 is endocytosed through caveolae-mediated endocytosis.

Bacterial toxins often piggyback existing endocytic trafficking pathways [55,56] to deliver active proteins to subcellular targets. The small GTPases Rab are essential regulators of intracellular membrane trafficking and exist in an inactive GDP-bound form and an active GTP-bound form [57]. The co-localization experiments with Rab5, Rab7, Lamp1 revealed that LtxA can follow the degradation pathway process that culminates in the delivery of the toxin to lysosomes. Rab5 localizes to early endosomes where it is involved in the recruitment of Rab7 and the maturation of these compartments to late endosomes [58]. Impaired Rab5a function affects endo- and exocytosis rates and, conversely, Rab5 overexpression increases the release efficacy [59]. Therefore, the termination of Rab5 function blocks the movement of proteins downstream of the endocytic pathway. Downregulation of Rab5a decreased LtxA toxicity, suggesting that further toxin trafficking is required for intoxication by LtxA. LFA-1 was suggested to undergo endocytic recycling through the long-Rab11A-dependent pathway with a transitional step at PNRC [29]. Here we, for the first time, demonstrated CD11a localization in Rab11A-containing endocytic vesicles. Extensive colocalization of CD11a and Rab11A was found in Jn.9 cells which were not treated with LtxA (Figure S10). While some LtxA follows LFA-1 in its

recycling turnover, a portion of LtxA is separated from LFA-1 and the toxin proceeds to late endosomes and lysosomes. A proposed model of LtxA trafficking in lymphocytes is shown in Figure 10.

**Figure 10.** Proposed mechanism of LtxA entry and trafficking in human lymphocytes. LtxA binds to cholesterol and LFA-1 on the surface of Jn.9 cell. LtxA/LFA-1 complex internalization is dynamin-dependent. Internalized LtxA/LFA-1 complexes are quickly transported to early endosomes. The small GTPase Rab5 regulates membrane binding and fusion in the early endocytic pathway. The interruption of Rab5 expression in Jn.9 cells results in the abolishment of the LtxA activity. LFA-1 undergoes endocytic recycling through the long-Rab11A-dependent pathway with a transitional step at PNRC [29]. While some LtxA follows LFA-1 in its recycling turnover, a portion of LtxA is separated from LFA-1 and the toxin proceeds to late endosomes and lysosomes. The ability of LtxA to damage lipid membranes at a low pH may cause endocytic vesicles and lysosomal rupture and release of the toxin to the cytosol.

Interaction between integrins and their β-integrin ligands typically leads to enhanced cell survival and several immunological changes [60,61]. Our experiment using cell impermeable dye, YO-PRO®-1, serves to demonstrate that LtxA gains access to the Jn.9 cell cytosol without evidence of plasma membrane damage. Our study and others suggest that LtxA could accumulate in the lysosomes and alter lysosomal pH [62,63]. Damage to lysosomes by LtxA in human and rat monocytes cells [62,64] and in human erythroleukemia cells [62] was reported. In our previous study, treatment with 100 ng/mL LtxA led to cytosol acidification in K562 cells expressing LFA-1, presumably due to the leakage of lysosomal content, as was identified using a pH-sensitive indicator pHrodo®. This process correlated with the disappearance of lysosomes in the cytosol, as determined by both acridine orange and LysoTracker®Red DND-99 staining. Similarly, using LysoTracker®Red DND-99 dye, lysosomal damage was detected in malignant monocytes (THP-1 cells) as early as 15 min after treatment with LtxA, and reached 70% after 2 h of treatment (unpublished data). In these cells, LtxA was shown to localize to the lysosome where it induces active cathepsin D release [64]. Here, we demonstrate that

LtxA causes changes in lysosomal pH in T lymphocytes, however, to a leser extent. As the Lysotracker dye is sensitive to luminal pH, the decrease in the dye staining could have resulted from either a rise in lysosomal pH or a decrease in the number of lysosomes. To distinguish between these possibilities, cells were stained for lysosomal marker Lamp1. The clear decrease in Lamp1 staining, combined with the decrease in the Lysotracker signal, forms a strong argument that the toxin decreased the number of lysosomes. As the transcription factor EB (TFEB) feedback systems try to increase lysosomal biogenesis [65], the most likely explanation for this is that the toxin has ruptured the lysosomes and overridden the TFEB pathways. The pore-forming properties of LtxA are well established [32,46]. Therefore, we propose that LtxA can cause permeabilization of the lysosomal membrane, and possibly other intracellular organelles after the toxin is released from lysosomes.

LtxA was reported to cause different cellular responses leading to cell death in LFA-1-expressing cells. Kelk et al. reported that LtxA lyses healthy monocytes by the activation of inflammatory caspase 1 and causes release of IL-1b and IL-18. In contrast to myeloid cells, LtxA uses a "slow mode" of lymphocyte killing. The killing of malignant lymphocytes requires Fas receptors and caspase 8 in both T and B lymphocytes [66]. In B lymphocytes (JY cells), LtxA caused loss of the mitochondrial membrane potential, cytochrome c release, reactive oxygen species release, and activation of caspases 3,7,9 [24]. One possible explanation for the cell death mechanism induced by LtxA is the degree of lysosomal damage caused by the toxins in the cell. The extent of lysosomal rupture will determine morphological outcomes following lysosomal membrane permeabilization. Extensive lysosomal injury may lead to necrotic cell death, while less substantial damage to lysosomes may instigate several apoptotic pathways, which can be attenuated by the inhibition of lysosomal cathepsins [66–69].

The planar lipid bilayer assay is a highly sensitive method that allows the characterization of the membrane damaging activity of RTX-toxins in different physical conditions [70]. A current model proposes that RTX-toxins form cation-selective channels with a diameter of 0.6–2.6 nm in artificial membranes formed of lipid mixtures such as the asolectin/n-decane membrane [46]. It was demonstrated that the membrane-damaging activity of LtxA in artificial bilayers did not require the presence of the receptor [71]. In the endocytic pathway, subsequent acidification may initiate proteolysis and conformational changes, resulting in the ability of toxins and viruses to cross the endocytic vesicle membrane, since drugs that interfere with the endosomal pH are able to block the infection [72,73]. In this study, we used this method to observe and compare the pore formation of LtxA at different pH. We demonstrated that LtxA is functional in acidic pH found in endocytic vesicles and lysosomes, which may result in their damage. RTX toxins are intrinsically disordered proteins, therefore changes in pH may affect their secondary structure and consequently change their activity [74]. Further investigation is required to improve our understanding of the intracellular events leading to LtxA-induced cytolysis.

In conclusion, our results show that LtxA enters the cytosol of Jurkat cells without evidence of plasma membrane damage, utilizing receptor-mediated endocytic mechanisms. In our studies, colocolization between LtxA/CD11a was demonstrated on the plasma membrane [23] and in the early steps of LtxA endocytic trafficking. Our results suggest that LtxA can accompany LFA-1 in its recycling pathway; however, the toxin molecules can apparently dissociate from the receptor in an acidic environment of endocytic vesicles and independently follow the degradative pathway. LtxA delivery to the terminal point of this route results in the lysosomal membrane rupture.

#### **4. Materials and Methods**

#### *4.1. Antibodies and Chemicals*

The following primary antibodies were used; CD11a Alexa Fluor™ 594 clone HI111 (Biolegend, San Diego, CA), rabbit polyclonal anti-Rab5, anti-Rab11A, anti-Rab7, or anti-Lamp1 antibody (Abcam, Cambridge, UK), anti-beta-actin antibody (AnaSpec, Fremont, CA) (1:1000), and anti-LtxA monoclonal antibody 107A3A3 [75] in hybridoma supernatants (1:10 dilution). The following secondary antibodies

were used: goat anti-rabbit IgG Alexa Fluor®488 (1:1000); horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (Fc) or (HRP)-goat anti-rabbit (Pierce, Rockford, IL) (1:10,000). Transferrin labeled with Alexa Fluor®555 was from Invitrogen (Waltham, MA, USA). Dynamin inhibitor Dynole 34-2 and its inactive control, Dynole 31-2, were purchased from SigmaAldrich (St. Louis, MO), Dynasore and Pitstop 2 (Abcam, Cambridge, UK). The inhibitors were used in the following concentrations: 10 μM Dynole 34-2; 10 μM Dynole 31-2; 10 μM Dynasore; 5 μM Pitstop 2.

#### *4.2. Cell Culture*

Jn.9, a subclone of Jurkat cells [76] was utilized in this study. The cells were cultivated in RPMI 1640 medium containing 10% FBS, 0.1 mM MEM non-essential amino acids, 1x MEM vitamin solution, and 2 mM L-glutamine, and 0.5 μg/mL gentamicin at 37 ◦C under 5% CO2.

#### *4.3. LtxA Purification and Labeling*

*Aa* strain JP2 [77] was grown on solid AAGM medium [78] for 48 h at 37 ◦C in 10% CO2 atmosphere. The colony was then inoculated in 1.5 L of liquid AAGM medium and the culture was incubated for 18 h. The toxin was purified from cell culture supernatants as described previously [79]. Purified LtxA was labeled with DyLight™ 650 (LtxA-DY650) or DyLight™ 488 (LtxA-DY488) using DyLight™ Amine-Reactive dyes (Pierce). The toxin was purified after labeling using a Zeba™ Spin Desalting column (40 K MWCO, Thermo Fisher™ Scientific), according to previously published protocol [23].

#### *4.4. Immunofluorescence*

For LtxA trafficking studies, 1 <sup>×</sup> 106 of Jn.9 cells were incubated with 20 nM LtxA-DY650 for 15 min to 2 h at 37 ◦C in the growth medium. The cells were then washed with PBS, fixed with 2% paraformaldehyde for 10 min, washed twice with PBS, and permeabilized with 0.2% Triton X-100 for 20 min. The cells were subsequently blocked with 4% BSA for 30 min at 37 ◦C, incubated with primary antibody for 18 h at 4 ◦C, washed, and incubated with secondary antibody conjugated to Alexa Fluor 488 for 1 h at 37 ◦C. The nuclei were stained with 1 μg/ml Hoechst 33342 (Molecular Probes™, Eugene, OR) for 15 min at 37 ◦C. Samples were mounted in Cytoseal mounting medium (Electron Microscopy Sciences, Hatfield, PA) and images captured with a Nikon A1R laser scanning confocal microscope (Nikon Instruments Inc., Melville, NY ) with a PLAN APO VC 60 × water (NA 1.2) objective at 18 ◦C. Data were analyzed using Nikon Elements AR 4.30.01 software. For co-distribution analyses, the Pearson's' coefficient of 0.55 was used as a cut off and was identified using circular ROIs. A Z-stack series consisting of seventeen individual planes 0.33 μm apart were assembled in 3D animations using Nikon Elements AR 4.30.01 software. Maximum intensity projection and standard LUT adjustment were used for the images' presentation.

For live imaging of LtxA uptake, Jn.9 cells were washed with in the serum free medium and were placed to attach for 20 min in ibiTreat 60 μ-dishes (Ibidi, Madison, WI) coated with poly-l-lysine (Sigma St. Louis, MO, USA). After floating cells were removed, the attached cells were pretreated with specific inhibitor, if necessary, and 20 nM LtxA-DY650 or 1 μM transferrin labeled with Alexa Fluor®555 was added. The cells were examined using a Nikon A1R laser scanning confocal microscope with a 60× water objective at different intervals of treatment at 37 ◦C.

#### *4.5. Inhibitors*

Chemicals stocks were prepared in DMSO and were added in the 1 μl volume to 1 ml of cells. To measure LtxA internalization inhibition, Jn.9 cells (1 <sup>×</sup> 106 cells) were pre-incubated with 5–10 <sup>μ</sup><sup>M</sup> inhibitors for 20 min in the serum free medium at 37 ◦C and then 20 nM LtxA-488 was added for 30 min. The effect of intracellular K+-depletion was evaluated using previously published protocol [80]. Pelleted Jn.9 cells (1 <sup>×</sup> <sup>10</sup><sup>6</sup> cells) were incubated in 2 ml hypotonic medium (RPMI/water, 1:1) for 5 min, followed by incubation in isotonic K+-free buffer (50 mM Hepes and 100 mM NaCl at pH 7.4) for 40 min at 37 ◦C, and then 20 nM LtxA-488 was added for 30 min. The toxin internalization assay was

performed as described in the "*Flow cytometry*" section. Live imaging of LtxA uptake was performed as described in the "*Immunofluorescence*" section. For cytotoxicity evaluation, the cells were treated with 2 nM LtxA and cell viability was evaluated as described in the "*Cytotoxicity assay*" section. The cells treated with specific inhibitors alone served as a control

#### *4.6. Flow Cytometry*

YO-PRO®-1 and PI internalization was investigated using Membrane permeability/dead cell apoptosis kit (Invitrogen, Carlsbad, CA) according to the manufacture's protocol. Jn.9 cells (1 <sup>×</sup> 106) were treated with 20 nM LtxA at 37 ◦C in Jn.9 culture medium at indicated times, washed with PBS and then treated with 0.1 μM YO-PRO®-1 or 1.5 μM PI, followed by flow cytometry analysis. To detect internalized LtxA, Jn.9 cells (1 <sup>×</sup> 10<sup>6</sup> cells) were incubated with 20 nM LtxA-DY488 for the specified time on ice or at 37 ◦C in Jn.9 culture medium, washed with PBS, and total cell-associated fluorescence was analyzed. To quench the extracellular fluorescence, LtxA-DY488-treated cells were incubated with 0.025% trypan blue (Sigma, St. Louis, MO) for 20 min as described previously [42,45]. To quench the intracellular fluorescence cells were permeabilized using 0.1% Triton X-100 (SigmaArdrich, St. Louis, MO) for 10 min and then subjected to 0.025% trypan blue treatment. Fluorescence was measured using a BD LSR II flow cytometer (BD Biosciences). Ten thousand events were recorded per sample, and MCF values were determined using WinList™7.0 software (Verity Software House). No residual fluorescence was detected in 0.1% Triton X-100 permeabilized cells after the trypan blue treatment. Samples that were not treated with LtxA or LtxA-DY488 served as a control.

For lysosomal staining analysis, Jn.9 cells were incubated with 20 nM LtxA for 2 h at 37 ◦C in the growth medium. Then 100 nM LysoTracker® Green DND-26 (Life Technologies, Carlsbad, CA, USA) was added to LtxA-treated and control cells for 15 min.

#### *4.7. Protein Analyses*

The protein concentration was determined by absorption at 280 nm on A1 NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA). Proteins were resolved on 4% to 20% SDS-PAGE and visualized by staining with GelCode blue stain reagent (Pierce, Rockford, IL). The Western blot analysis was performed as described previously [70].

#### *4.8. siRNA*

The validated Silencer®Select siRNA targeting human Rab5a (ID s11678) and Silencer® Select Negative Control #2 siRNA (catalog# 4390846) were synthesized by Life Technology (Carlsbad, CA, USA). Jn.9 cells were transfected with lipofectamine 2000 (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instructions. For each transfection, 5 μl of the 20 μM siRNA stocks were added to 400 μl of Jn.9 cells grown to 90% confluency. Rab5a levels in 1 x 106 Jn.9 cells were confirmed by Western blot analysis 24 h after transfection. β-actin served as a loading control.

#### *4.9. Cytotoxicity Assay*

For toxicity tests, 2–20 nM LtxA was added to 1 x 10<sup>6</sup> Jn.9 cells in growth medium and incubated for 18 h at 37 ◦C. The cell membrane permeability was determined with trypan blue assay using Vi-cell Cell Viability Analyzer (Beckman Coulter, Miami, FL). All reactions were run in duplicate; the assay was performed three independent times. Untreated cells were used as controls.

#### *4.10. Planar Lipid Bilayers*

Lipid bilayer measurements have been described previously in detail [81]. In short, A Teflon chamber, containing two 5 mL compartments connected by a small circular hole with a surface area of about 0.4 mm2, were filled with 1 M KCl, 10 mM MES, pH 6.0. Black lipid bilayer membranes were created by painting onto the hole solutions of 1% (w/v) asolectin (phospholipids from soybean, Sigma-Aldrich) in *n*-decane. The temperature was maintained at 20 ◦C during all experiments. The current across the membrane was measured with a pair of Ag/AgCl electrodes with salt bridges switched in series with a voltage source and current amplifier Keithley 427 (Keithley Instruments, INC. Cleveland, OH). The amplified signal was recorded by a strip chart recorder (Rikadenki Electronics GmbH, Freiburg, Germany).

#### *4.11. Statistical Analysis*

The statistical analyses were performed using either Student's test or one-way analysis of variance using SigmaPlot® (Systat Software, Inc. Chicago, IL, USA). The following statistical criteria were applied: *p* < 0.001, *p* < 0.05, and *p* < 0.01.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-0817/9/2/74/s1, Figure S1: "Confocal imaging of LtxA and CD11a in Jn.9 cells"; Figure S2: "Effect of inhibitors on transferrin uptake by Jn.9 cells"; Figure S3: "Effect of 10 μM Dynasore on LtxA uptake by Jn.9 cells."; Figure S4 "LtxA toxicity on Jn.9 cells"; Figure S5 "Localization of LtxA in early endosomes of Jn.9 cells"; Figure S6: "Localization of LtxA in Rab11A-positive endosomes of Jn.9 cells."; Figure S7: "Localization of LtxA around nuclear membrane of Jn.9 cells"; Figure S8: "Images of late endosomes in Jn.9 cells"; Figure S9: "Images of lysosomes in Jn.9 cells". Figure S10: "Confocal imaging of CD11a trafficking in Jn.9 cells".

**Author Contributions:** Conceptualization, E.T.L. and N.B.; data curation, N.M.G. and N.B.; formal analysis, N.M.G., A.D., R.B., J.L. and N.B.; funding acquisition, E.T.L. and N.B.; investigation, N.M.G., A.G., S.A.F. and R.B.; methodology, K.B.-B., C.H.M. and R.B.; project administration, N.B.; software, A.D.; supervision, N.B.; writing—review and editing, N.B., K.B.-B., and C.H.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the United States National Institute of Health grants R01DE009517 (ETL and NB), R01DE022465 (KBB).

**Acknowledgments:** The authors thank Juan Reyes-Reveles (JRRDesign Inc.) for technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **The Cell-Cycle Regulatory Protein p21CIP1**/**WAF1 Is Required for Cytolethal Distending Toxin (Cdt)-Induced Apoptosis**

#### **Bruce J. Shenker 1,\*, Lisa M. Walker 1, Ali Zekavat 1, Robert H. Weiss <sup>2</sup> and Kathleen Boesze-Battaglia <sup>3</sup>**


Received: 14 November 2019; Accepted: 28 December 2019; Published: 2 January 2020

**Abstract:** The *Aggregatibacter actinomycetemcomitans* cytolethal distending toxin (Cdt) induces lymphocytes to undergo cell-cycle arrest and apoptosis; toxicity is dependent upon the active Cdt subunit, CdtB.We now demonstrate that p21CIP1/WAF1 is critical to Cdt-induced apoptosis. Cdt induces increases in the levels of p21CIP1/WAF1 in lymphoid cell lines, Jurkat and MyLa, and in primary human lymphocytes. These increases were dependent upon CdtB's ability to function as a phosphatidylinositol (PI) 3,4,5-triphosphate (PIP3) phosphatase. It is noteworthy that Cdt-induced increases in the levels of p21CIP1/WAF1 were accompanied by a significant decline in the levels of phosphorylated p21CIP1/WAF1. The significance of Cdt-induced p21CIP1/WAF1 increase was assessed by preventing these changes with a two-pronged approach; pre-incubation with the novel p21CIP1/WAF1 inhibitor, UC2288, and development of a p21CIP1/WAF1-deficient cell line (Jurkatp21<sup>−</sup>) using clustered regularly interspaced short palindromic repeats (CRISPR)/cas9 gene editing. UC2288 blocked toxin-induced increases in p21CIP1/WAF1, and JurkatWT cells treated with this inhibitor exhibited reduced susceptibility to Cdt-induced apoptosis. Likewise, Jurkatp21<sup>−</sup> cells failed to undergo toxin-induced apoptosis. The linkage between Cdt, p21CIP1/WAF1, and apoptosis was further established by demonstrating that Cdt-induced increases in levels of the pro-apoptotic proteins Bid, Bax, and Bak were dependent upon p21CIP1/WAF1 as these changes were not observed in Jurkatp21<sup>−</sup> cells. Finally, we determined that the p21CIP1/WAF1 increases were dependent upon toxin-induced increases in the level and activity of the chaperone heat shock protein (HSP) 90. We propose that p21CIP1/WAF1 plays a key pro-apoptotic role in mediating Cdt-induced toxicity.

**Keywords:** *Aggregatibacter actinomycetemcomitans*; cytolethal distending toxin; lymphocytes; apoptosis; virulence

#### **1. Introduction**

The cytolethal distending toxin (Cdt) is a putative virulence factor that is produced by a wide range of human pathogens capable of colonizing mucocutaneous tissue, resulting in disease characterized by persistent infection and inflammation (reviewed in References [1,2]). In general, Cdts are heterotrimeric complexes encoded by an operon of three genes designated *cdtA*, *cdtB*, and *cdtC* which encode three polypeptides: CdtA, CdtB, and CdtC with molecular masses of 23–30, 28–32, and 19–20 kDa, respectively [3–13]. Analyses of subunit structure and function indicate that the heterotrimeric

holotoxin functions as an AB2 toxin; the cell binding unit (B) is responsible for toxin association with the cell surface and is composed of subunits CdtA and CdtC. These subunits deliver the active subunit (A), CdtB, to intracellular compartments. Cdt binding and CdtB internalization are both dependent upon toxin binding to target cell cholesterol in the context of cholesterol-rich membrane microdomains (reviewed in Reference [14]).

Cdt B internalization leads to irreversible cell-cycle arrest and eventually apoptotic cell death. These toxic effects were originally attributable to CdtB's ability to function as a DNase, thereby causing DNA damage which in turn leads to G2/M arrest and death [9,15–23]. Over the past several years, our studies suggested an alternative paradigm to account for *Aggregatibacter actinomycetemcomitans* Cdt-mediated toxicity which is based upon a novel molecular mode of action for CdtB. In this regard, we demonstrated that, in addition to exhibiting DNase activity, CdtB is a potent lipid phosphatase capable of converting the signaling lipid phosphatidylinositol (PI)-3,4,5-triphosphate (PIP3) to PI-3,4-diphosphate [24–28]. Moreover, our investigations demonstrated that the ability of CdtB to function as a PIP3 phosphatase enables this toxin subunit to intoxicate cells via blockade of the PI-3K signaling pathway. Indeed, we demonstrated that the toxic effects of Cdt on lymphocytes, macrophages, and mast cells results in PI-3K signaling blockade characterized by decreases in PIP3, leading to concomitant reductions in the phosphorylation status of downstream targets: Akt and GSK3β. Additionally, we demonstrated that the induction of both G2/M arrest and apoptosis is dependent upon CdtB-mediated PI-3K blockade.

In order to more accurately define the molecular mechanisms that link CdtB-mediated PI-3K blockade with G2/M arrest and apoptosis, we investigated the role of the cyclin-dependent kinase inhibitor known as CDK-interacting protein 1 (Cip1) and wild-type p53-activated fragment 1 (WAF1) (p21CIP1/WAF1). P21CIP1/WAF1 was originally identified as a negative regulator of the cell cycle, as well as a tumor suppressor. However, recent studies demonstrated additional functions for p21CIP1/WAF1 that are associated with regulation of a number of cellular processes including cell differentiation, migration, senescence, and apoptosis [29–33]. Thus, it is not surprising that several investigators demonstrated an association between p21CIP1/WAF1 expression and exposure to Cdt [16,34–37]. It should be noted, however, that these studies did not provide any information as to whether the p21CIP1/WAF1 levels were mechanistically linked to and/or required for Cdt toxicity. In this study, we investigated the relationship between lymphocyte exposure to *A. actinomycetemcomitans* Cdt, altered p21CIP1/WAF1 levels, and induction of toxicity. We now report that Cdt-treated human lymphocytes exhibit dose-dependent increases in levels of p21CIP1/WAF1 and the chaperone HSP90 within 4–16 h of exposure to the toxin. To study the biologic consequence of these increases, we employed a two-pronged approach to modify the ability of Cdt to alter expression of p21CIP1/WAF1: gene editing and pharmacologic intervention. Additionally, these interventions were assessed for their ability to alter cell susceptibility to Cdt toxicity. Our results indicate a requisite role for p21CIP1/WAF1 in Cdt-induced apoptosis.

#### **2. Results**

#### *2.1. Cdt Induces Elevations in Lymphocyte Levels of p21CIP1*/*WAF1*

Cdt derived from *A. actinomycetemcomitans*, *Haemophilus ducreyi*, and *Helicobacter hepaticus* were shown to induce increases in p21CIP1/WAF1 within 24–48 h in several cell lines including fibroblasts, lymphocytes, enterocytes, and hepatocytes [16,34–38]. Likewise, we now demonstrate that *A. actinomyetemcomitans* Cdt induces increases in p21CIP1/WAF1 levels in Jurkat cells in a time- and dose-dependent manner. Jurkat cells were treated with varying amounts of Cdt (0–400 pg/mL) for 4, 8, and 16 h and then analyzed by Western blot to assess total p21CIP1/WAF1 levels (Figure 1A,B). Analysis indicates that a small, but consistent, increase in p21CIP1/WAF1 was detected within 4 h in cells exposed to the highest concentration of Cdt (400 pg/mL). Following an 8-h exposure, significant increases of nine- and 18-fold were observed in cells exposed to 100 and 400 pg/mL Cdt, respectively. After a 16-h exposure, the relative levels of p21CIP1/WAF1 remained elevated; cells treated with 25, 100, and 400 pg/mL Cdt exhibited three-, six-, and seven-fold increases, respectively.

**Figure 1.** Effect of cytolethal distending toxin (Cdt) on p21CIP1/WAF1 levels in Jurkat cells. Jurkat cells were exposed to 0–400 pg/mL Cdt for 4, 8, and 16 h; cells were then harvested, and extracts were fractionated by SDS-PAGE and analyzed by Western blot for the presence of p21CIP1/WAF1. Panel (**A**) shows representative Western blots of p21CIP1/WAF1 for cells treated with each dose of toxin at 4, 8, and 16 h; glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as a gel loading control. Panel (**B**) shows the results from multiple blots which were analyzed by digital densitometry; the value of relative intensity obtained from digital densitometry is presented as the mean ± standard error of the mean (SEM) of four experiments. Panel (**C**) shows the relative levels of pp21CIP1/WAF1 in cells treated with 50 pg/mL Cdt or 0.5 μM GSK690693 for 16 h expressed as a percentage observed in control (medium only) cells. A representative blot is shown along with compiled results from three experiments: results are expressed as the percentage (mean <sup>±</sup> SEM) of pp21CIP1/WAF1 observed in control cells; \* indicates statistical significance (*p* < 0.05) when compared to untreated cells.

To verify that the effects of Cdt on p21CIP1/WAF1 levels were not unique to the Jurkat cell line, the cutaneous T-cell lymphoma line, MyLa, was also assessed for altered p21CIP1/WAF1 levels when exposed to the same doses of Cdt. As shown in Figure 2A,B, MyLa cells treated with 0–400 pg/mL Cdt for 16 h exhibited significant increases in p21CIP1/WAF1levels: 4.5- (100 pg/mL Cdt) and 5.7-fold (400 pg/mL Cdt) over control levels. In addition to lymphoid cell lines, p21CIP1/WAF1 levels were assessed in primary human lymphocytes (HPBMCs). As shown in Figure 2A,B, exposure to 25 pg/mL Cdt resulted in a detectable but not statistically significant increase in p21CIP1/WAF1. Significant increases

in p21CIP1/WAF1 levels were observed in the presence of 100 and 400 pg/mL Cdt leading to 361% <sup>±</sup> 82% and 673% ± 185% over that observed in untreated control cells.

**Figure 2.** Effect of Cdt on p21CIP1/WAF1 levels in primary human lymphocytes (HPBMCs) and MyLa cells. HPBMC and MyLa cells were treated with 0–400 pg/mL Cdt for 16 h. The cells were harvested, and extracts were fractionated by SDS-PAGE and analyzed by Western blot for the presence of p21CIP1/WAF1. Panel (**A**) shows representative Western blots, and panel (**B**) shows results from four experiments assessed by digital densitometry; results (intensity) are expressed as the mean <sup>±</sup> SEM. Panel (**C**) shows the effect of 50 pg/mL of Cdt containing wild-type CdtB (CdtBWT) or CdtB containing mutations (CdtBR117A, CdtBR144A, and CdtBA163R) on p21CIP1/WAF1 levels in MyLa cells. A representative Western blot is shown, as well as results from four experiments that were analyzed by digital densitometry; results are expressed as a percentage of p21CIP1/WAF1 levels observed in control (untreated) cells; \* indicates statistical significance (*p* < 0.05) when compared to untreated cells.

As noted earlier, we previously demonstrated that Cdt toxicity in lymphocytes (both primary and cell lines), macrophages, and mast cells is dependent upon PIP3 phosphatase activity exhibited by the active Cdt subunit, CdtB [24–28]. We generated and previously reported on the characterization of the enzymatic and toxic activities of Cdts containing targeted mutations within the CdtB subunit [24]. In each instance, we observed that the retention of lipid phosphatase activity, and not DNase activity, was a requisite for both Cdt-induced cell-cycle arrest and apoptosis in lymphocytes, as well as the induction of pro-inflammatory responses in macrophages. Therefore, we next determined the requirement for PIP3 phosphatase activity in Cdt-induced increases in the levels of p21CIP1/WAF1. Cdt containing CdtB mutant proteins that were previously described were employed [24]: CdtBA163R retains lipid phosphatase activity, lacks DNase activity, and is toxic; CdtBR144A exhibits low lipid phosphatase activity, exhibits increased DNase activity and is not toxic; CdtBR117A exhibits low lipid

phosphatase activity, retains DNase activity, and is not toxic. As shown in Figure 2C, elevations in p21CIP1/WAF1 levels were only observed when MyLa cells were treated with toxin containing the active wild-type subunit, CdtBWT (2.5-fold increase), or the CdtB mutant, CdtBA163R (3.2-fold increase) [24]. Cells exposed to CdtB mutant proteins that we previously demonstrated to be deficient in phosphatase activity and lack toxicity (CdtBR144A and CdtBR117A) did not exhibit significant changes in levels of p21CIP1/WAF1: 1.2-fold and 1.7 fold, respectively. It is noteworthy that we previously reported that *A. actinomycetemcomitans* Cdt-induced cell-cycle arrest and apoptosis in human lymphocytes does not involve activation of the DNA damage response (DDR) [24,39]. These observations were confirmed as we now demonstrate that Cdt containing CdtBWT, as well as the other CdtB mutant proteins employed in this study, does not induce phosphorylation of the histone H2AX, a commonly employed indicator of DDR activation due to DNA damage (Figure S1, Supplementary Materials).

The biologic activity of p21CIP1/WAF1 is governed by post-translational modifications such as phosphorylation; therefore, we also assessed Cdt-treated Jurkat cells for changes in phosphorylation status (T145; (pp21CIP1/WAF1)) (Figure 1C). Baseline levels of p21CIP1/WAF1 were very low, but detectable; as described above, these levels increased at 8–16 h in the presence of Cdt (see above). In contrast, the relative amount present as phosphorylated p21CIP1/WAF1 (pp21CIP1/WAF1) significantly declined in the presence of Cdt. Cells treated with 50 pg/mL Cdt for 16 h exhibited a reduction in the amount of pp21CIP1/WAF1 to 20.6% <sup>±</sup> 3.4% of the amount observed in untreated control cells. Interestingly, the PI-3K signaling pathway exhibits cross-talk with other regulatory pathways; specifically, p21CIP1/WAF1 was shown to be a downstream target of activated Akt (pAkt) [40]. Furthermore, we showed that Cdt-treated cells exhibit reduced levels of pAkt (reduced kinase activity) [24]. These observations, along with our current findings that toxin-treated cells also exhibit reduced levels of pp21CIP1/WAF1, are consistent with the proposed molecular mode of action for CdtB which involves PIP3 phosphatase activity leading to PI-3K signaling blockade. To further support our findings and provide additional "proof-of-principle" evidence for a relationship between pAkt and p21CIP1/WAF1, we employed GSK690693, an Akt inhibitor [41]. Jurkat cells treated for 16 h with 0.5 μM GSK690693, exhibited a reduction in pp21CIP1/WAF1 to 0.7% <sup>±</sup> 0.3% of that observed with untreated cells (Figure 1C).

#### *2.2. Blockade of Cdt-Induced Increases in p21CIP1*/*WAF1 Results in Reduced Jurkat Cell Susceptibility to Cdt Toxicity*

We next extended our investigation to address the biological significance of Cdt-induced increases in p21CIP1/WAF1 by utilizing a two-pronged approach: pharmacologic modulation and altered expression using gene editing. The novel inhibitor of p21CIP1/WAF1, UC2288, was demonstrated to reduce p21CIP1/WAF1 levels [42]; therefore, we firstly employed UC2288 as a potential inhibitor of Cdt toxicity by virtue of its ability to block increases in p21CIP1/WAF1. As shown in Figure 3A, pre-treatment of JurkatWT cells with UC2288 (2.5–10 μM) resulted in reduced Cdt-induced apoptosis in the presence of the highest concentration of the inhibitor (10 μM); interestingly, this is the same concentration reported to be effective in other studies [42]. Untreated control cells exhibited 7.7% ± 1.1% terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labeling (TUNEL)-positive cells and, in the presence of 50 pg/mL Cdt, the percentage of TUNEL-positive cells increased to 35.0% ± 1.9%. Pre-treatment of cells with UC2288 reduced the percentage of TUNEL-positive cells to 19.7% ± 1.4%. Two inactive analogues of UC2288 were employed: UC1770 and UC1472; neither of these altered Cdt-induced apoptosis, as the percentage of TUNEL-positive cell was not reduced below 31% when cells were treated with equivalent concentrations of these analogues. The ability of UC2288 to block Cdt-induced increases in p21CIP1/WAF1 levels was also confirmed (Figure 3B). Results are expressed as a percentage of p21CIP1/WAF1 levels observed in cells exposed to toxin alone (100%); the latter represents almost a five-fold increase over control cells (medium only). Pre-treatment with 10 μM UC2288 resulted in a significant reduction of p21CIP1/WAF1 levels to 24.7% of that observed with cells treated with toxin alone.

**Figure 3.** Effect of UC2288 on Cdt-induced apoptosis and on p21CIP1/WAF1 levels. JurkatWT cells were pre-incubated with 0–10 μM UC2288 or its inactive analogues UC1770 and UC1472 for 30 min. Cdt (50 pg/mL) was added to the cell cultures, and the cells were harvested 24 h later and analyzed for apoptosis using the TUNEL assay as described (panel (**A**)). Results are expressed as a percentage of apoptotic cells versus drug concentration; the mean ± SEM for four experiments is plotted. Panel (**B**) shows the effect of UC2288 on Cdt-induced p21CIP1/WAF1 levels at 16 h. A representative blot is shown along with results from four experiments. Western blots were analyzed by digital densitometry; the data are expressed as a percentage of the p21CIP1/WAF1 levels observed in cells treated with Cdt alone; \* indicates statistical significance (*p* < 0.05) when compared to untreated cells.

To confirm our observations with UC2288 and to better understand the biological significance of Cdt-induced elevations in p21CIP1/WAF1 levels, we utilized CRISPR/Cas9 gene editing to establish a stable Jurkat cell line deficient in p21CIP1/WAF1 expression (Jurkatp21−). As shown in Figure 4A, Jurkatp21<sup>−</sup> cells were unable to express p21CIP1/WAF1 when challenged with either Cdt or etoposide for 16 h; in contrast, JurkatWT cells exhibited clear elevations in this protein when challenged with either agent under identical conditions. As noted above, we established that Cdt-induced toxicity is dependent on the CdtB subunit's ability to function as a PIP3 phosphatase, thereby mediating blockade of the PI-3K signaling pathway [26,27]. Therefore, we next verified that Jurkatp21<sup>−</sup> cells remained susceptible to toxin-induced signaling blockade. Specifically, cells were assessed after 2 and 4 h of exposure to Cdt for changes in the phosphorylation status of two downstream PI-3K signaling targets: Akt and GSK3β. JurkatWT and Jurkatp21<sup>−</sup> cells were treated with 50 pg/mL Cdt and then analyzed by Western blot for the presence of Akt, pAkt, GSK3β, and pGSK3β at 0, 2, and 4 h. Figure 4B shows representative immunoblots indicating that the levels of both pAkt and pGSK3β were reduced at 2 and 4 h in both JurkatWT and Jurkatp21−; in contrast, the total amount of these proteins (Akt and GSK3β) remained unchanged. In previous studies, we reported that decreases in pAkt and pGSK3β

within JurkatWT cells were statistically significant with levels of pAkt reduced to 48.7% <sup>±</sup> 9.3% (2 h) and 45.5% ± 11.6% (4 h) relative to untreated control cells; likewise, pGSK3β levels were reduced to 55.7% ± 6.0% and 47.6% ± 6.1% of control values at 2 and 4 h, respectively [24]. Figure 4C shows similar compiled results from multiple experiments for Jurkatp21<sup>−</sup> cells. Akt and GSK3β levels remained relatively constant for the 4-h period. In contrast, pAkt levels were reduced to 56.4% ± 19% (2 h) and to 34.8% ± 16.7% (4 h); pGSK3β levels were reduced to 59.7% ± 22.7% (2 h) and 49.5% ± 21.7% (4 h).

**Figure 4.** Analysis of Jurkatp21<sup>−</sup> cells. CRISPR/Cas9 gene editing was employed to produce a Jurkatp21<sup>−</sup> cell line. Panel (**A**) shows the results of p21CIP1/WAF1 analysis by Western blot of both JurkatWT and Jurkatp21<sup>−</sup> cells at 16 h following exposure to either Cdt (50 pg/mL) or etoposide (Eto; 50 μM); GAPDH is shown as a gel loading control. Panels (**B**,**C**) show the effect of Cdt on PI-3K signaling targets. JurkatWT and Jurkatp21<sup>−</sup> cells were treated with medium alone (0) or with Cdt (50 pg/mL) for 2 or 4 h and then analyzed by Western blot for the expression of Akt, pAkt (S473), GSK3β, and pGSK3β (S9), as well as actin which served as a loading control. Representative blots are shown in panel (**B**), and the results of three experiments for Jurkatp21<sup>−</sup> cells are shown in panel (**C**). Data are plotted as the percentage of protein expressed in untreated control cells; the mean ± SEM is shown and \* indicates statistical significance (*p* < 0.05) when compared to untreated cells. Panel (**D**) shows the effect of Cdt on apoptosis (TUNEL-positive) in JurkatWT and Jurkatp21<sup>−</sup> cells. The percentage of apoptotic cells was determined at 24 h and is plotted versus Cdt concentration; the mean ± SEM is shown for three experiments; \* indicates statistical significance (*p* < 0.01) when compared to untreated cells.

JurkatWT and Jurkatp21<sup>−</sup> cells were next assessed and compared for their susceptibility to Cdt-induced apoptosis using the TUNEL assay (Figure 4D) following a 24-h treatment with toxin. Consistent with our previous findings, JurkatWT cells exhibit dose-dependent apoptosis; the percentages of TUNEL-positive were 4.2% ± 0.9% (0 Cdt), 28.3% ± 1.6% (25 pg/mL Cdt), 47.7% ± 1.7% (100 pg/mL Cdt), and 63.5% <sup>±</sup> 0.9% (400 pg/mL Cdt). In contrast, Jurkatp21<sup>−</sup> cells were resistant to Cdt-induced apoptosis; cells incubated with 0–400 pg/mL toxin exhibited 5.7% ± 1.7%, 8.9% ± 2.6%, 10.5% ± 3.0%, and 12.6% <sup>±</sup> 3.0% apoptotic cells. It is noteworthy that Jurkatp21<sup>−</sup> cells retained the capacity to undergo apoptotic cell death as they remained sensitive to paclitaxel (Figure S2, Supplementary Materials).

Previously, we demonstrated that Cdt-induced apoptosis involves the intrinsic apoptotic pathway and, in particular, development of the mitochondrial permeability transition (MPT) [39,43]. Therefore, we assessed and compared JurkatWT and Jurkatp21<sup>−</sup> for expression of pro-apoptotic members of the Bcl-2 protein family in response to Cdt. Treatment of JurkatWT cells with 50 pg/mL Cdt resulted in significant increases in both Bid and Bax at 8 hrs (Figure 5A); Bid levels increased by 498.3% ± 290.5% relative to control cells and Bax levels increased by 637.0% ± 274.7%. A consistent, but not statistically significant, increase was observed for Bak to 160% ± 38.0%. In comparison, Cdt failed to induce increases in the levels of any of the three pro-apoptotic proteins in Jurkatp21<sup>−</sup> cells; relative to untreated cells, Bid, Bax, and Bak expression levels were 73.2% ± 7.5%, 94.2% ± 10.5%, and 80.4% ± 20.4%, respectively.

**Figure 5.** Effect of Cdt on the expression of pro-apoptotic Bcl-2 family members, as well as on the ΔΨm. Panel (**A**): JurkatWT and Jurkatp21<sup>−</sup> cells were incubated with medium or 50 pg/mL Cdt for 8 h. Cells were then analyzed by Western blot for Bid, Bax, Bak, and GAPDH (loading control). Representative blots are shown on top for Bid, Bax, Bak, and GAPDH for JurkatWT cells exposed to medium and Cdt-treated, as well as for Jurkatp21<sup>−</sup> cells exposed to medium and Cdt-treated. Western blots were analyzed by digital densitometry; results represent the levels of protein expressed as a percentage of that observed in respective control cells. The mean ± SEM for five experiments is shown; \* indicates statistical significance (*p* < 0.05) when compared to untreated cells. Panels (**B**–**I**) show the effect of Cdt (0–400 pg/mL) on the ΔΨm in both JurkatWT (panels (**B**–**E**)) and Jurkatp21<sup>−</sup> cells (panels (**F**–**I**)). ΔΨm was determined using DIOC6(3), and the percentage of cells exhibiting a reduction the membrane potential was determined using the analytical gate indicated; the numbers represent the mean ± SEM of three experiments, each performed in duplicate.

Cdt was also assessed for its ability to induce the MPT in both JurkatWT and Jurkatp21<sup>−</sup> cells by measuring a decline in the mitochondrial transmembrane potential (ΔΨm) using the fluorochrone DIOC6(3) [39,43]. As shown in Figure 5B–E, JurkatWT cells treated with 0–400 pg/mL Cdt exhibited a dose-dependent increase in the percentage of cells exhibiting a decline in the ΔΨm, with 9% in control cells versus 34.4%, 41.5%, and 45.5% in cells exposed to 100, 200, and 400 pg/mL Cdt. In contrast, Jurkatp21<sup>−</sup> did not exhibit a significant change in the ΔΨm; 15.3% of control cells exhibited reduced ΔΨm, and Cdt treatment resulted in a slight, but not statistically significant increase to 21.2%, 23.1%, and 23% in the presence of 100, 200, and 400 pg/mL Cdt, respectively (Figure 5 F–I).

#### *2.3. Cdt-Induced Increases in p21CIP1*/*WAF1 Levels and Induction of Apoptosis Are Blocked by the Chaperone HSP90*

The intracellular levels of p21CIP1/WAF1 are controlled transcriptionally, as well as post-translationally, by the proteasome. Experiments were carried out to account for the mode of Cdt-induced increases in p21CIP1/WAF1. We firstly addressed the effect of Cdt on p21CIP1/WAF1 messenger RNA (mRNA) levels in Jurkat cells. As shown in Figure 6A, p21CIP1/WAF1 mRNA levels were marginally elevated at high toxin doses within 4 h and declined in the presence of the same doses of Cdt at 16 h; however, these changes were not statistically significant. It should also be noted that *A. actinomycetemcomitans* Cdt-induced increases in lymphocyte p21CIP1/WAF1 levels were previously shown to be p53-independent [35]. These findings were confirmed as we demonstrate that Cdt-induced p21CIP1/WAF1 increases were observed in the presence of the p53 inhibitor, pifithrin-α (Figure S3, Supplementary Materials); this inhibitor also failed to block Cdt-induced apoptosis. Similar results were observed with Molt-4 cells (data not shown). On the post-translational side, the chaperone HSP90 was shown to stabilize p21CIP1/WAF1 and prevent its degradation by the proteasome [44–46]. Jurkat cells were assessed for changes in the levels of HSP90 by Western blot following 4 and 8 h of exposure to Cdt (Figure 6B,C). Untreated cells exhibited detectable levels of HSP90 at both time points. Exposure of Jurkat cells to Cdt for 4 h resulted in 1.5- (25 pg/mL), 2.0- (100 pg/mL), and 1.9-fold (400 pg/mL) increases in HSP90 levels. Treatment with the same doses of Cdt for 8 h resulted in increases of 1.6-, 1.5-, and 3.6- fold. The dependence of Cdt-induced increases in HSP90 on CdtB-associated PIP3 phosphatase activity was also assessed (Figure 6D). Jurkat cells treated with Cdt containing the active CdtB subunits CdtBWT and CdtBA163R exhibited increased HSP90 levels of 5.1-fold and 4.6-fold over levels observed in control cells, respectively. The HSP90 levels in cells treated with the PIP3 phosphatase-inactive CdtB units, CdtBR117A and CdtBR144A, did not change.

The relationship between HSP90 and Cdt-induced increases in p21CIP1/WAF1 levels, as well as the induction of apoptosis, was demonstrated by employing geldanamycin (GM), an inhibitor of HSP90 ATPase activity [47]. As shown in Figure 6E, GM was assessed for its ability to inhibit HSP90 and in turn Cdt-induced apoptosis. Jurkat cells were pre-treated with medium or GM (0–5 μM) for 1 h, followed by the addition of medium or Cdt (50 pg/mL). Cells were assessed for apoptosis (TUNEL-positive) 24 h later (Figure 6E). Cells treated with medium exhibited 6.2% ± 1.8% TUNEL-positive cells; in comparison, cells treated with 50 pg/mL Cdt exhibited 63.7% ± 8.8% TUNEL-positive cells. Pre-treatment of cells with GM followed by the addition of toxin resulted in a dose-dependent reduction in apoptosis. In the presence of 1.25, 2.5, and 5 μM GM, the percentage of TUNEL-positive cells was reduced to 21.6% ± 0.7%, 16.4% ± 1.7%, and 8.7% ± 2.3%. It should be noted that GM alone exhibited low levels of toxicity (net 5–10% over untreated cells).

In addition to evaluating the effects of GM on Cdt-induced apoptosis, Jurkat cells were also assessed for p21CIP1/WAF1 levels by Western blot (Figure 6F). Untreated Jurkat cells contained marginally detectable levels of p21CIP1/WAF1; cells treated with diluent (DMSO) or GM (2.5 μM) exhibited slight increases in p21CIP1/WAF1 to 1.2 <sup>±</sup> 1.0 and 1.1 <sup>±</sup> 0.98, respectively. Cells exposed to Cdt alone exhibited a significant increase in p21CIP1/WAF1 levels to 2.4 <sup>±</sup> 1.0; pretreatment of cells with 2.5 <sup>μ</sup>M GM reduced p21CIP1/WAF1 to near baseline levels (0.56 <sup>±</sup> 0.19).

**Figure 6.** Cdt-induced increases in p21CIP1/WAF1 and induction of apoptosis are dependent upon HSP90. Panel (**A**) shows the effects of Cdt on p21CIP1/WAF1 messenger RNA (mRNA) by RT-PCR. Jurkat cells were incubated with 50 pg/mL Cdt for 4 and 16 h. RNA extraction, complementary DNA (cDNA) synthesis, RT-PCR, and changes in mRNA were calculated as previously described [28]. Results are the mean of three experiments plotted as a percentage of mRNA levels observed in control cells (time 0). Panels (**B**,**C**) show the results of experiments that assess the effect of Cdt on HSP90 levels. Jurkat cells were treated with Cdt (0–400 pg/mL) for 4 and 8 h, and the cell extracts were assessed by Western blot. Panel (**B**) shows a representative Western blot with GAPDH as a loading control. Panel (**C**) shows the results of four experiments; Western blots were analyzed by digital densitometry, and the results are plotted as the mean intensity ± SEM. Panel (**D**) shows Western blot analysis of the effect of 50 pg/mL Cdt containing CdtBWT or CdtB containing mutations (CdtBR117A, CdtBR144A, and CdtBA163R) on HSP90 levels in Jurkat cells. A representative Western blot is shown, as well as results from three experiments that were analyzed by digital densitometry; results are expressed as the mean intensity ± SEM; \* indicates statistical significance (*p* < 0.05) when compared to untreated cells. The effects of geldanamycin D (GM) on Cdt-induced apoptosis are shown in panel (**E**). Jurkat cells were incubated with 0–5 μM GM for 60 min, followed by the addition of medium and 50 pg/mL Cdt, before being analyzed for apoptosis (TUNEL assay) 24 h later. The percentage of TUNEL-positive cells is plotted as the mean ± SEM of four experiments. Panel (**F**) shows the effects of GM on Cdt-induced increases in Jurkat cell p21CIP1/WAF1 level. Jurkat cells were treated with medium or Cdt (50 pg/mL) in the presence of DMSO (vehicle) or 2.5 μM GM for 16 h. Cells were analyzed by Western blot and digital densitometry for the levels of p21CIP1/WAF1. A representative blot and the results from three experiments is shown; results are plotted as the mean ± SEM of three experiments; \* indicates statistical significance (*p* < 0.05) when compared to untreated cells.

#### **3. Discussion**

#### *3.1. Functional Significance of Cdt-induced Increases in the Levels of p21CIP1*/*WAF1*

We now report that *A. actinomycetemcomitans* Cdt induces intracellular increases of p21CIP1/WAF1 in human lymphocyte cell lines, as well as primary HPBMCs. These observations are consistent with those of other investigators who also demonstrated similar increases in response to Cdt derived from either *A. actinomycetemcomitans* or *Haemophilus ducreyi* [16,18,35,36]. P21CIP1/WAF1 increases were observed in a murine B-cell hybridoma cell line, fibroblasts, the Hep-2 carcinoma cell line, and a gingival squamous cell carcinoma cell line, Ca9-22. These investigators did not demonstrate the biological significance of the observed Cdt-dependent changes in cellular levels of p21CIP1/WAF1. As noted earlier, increases in p21CIP1/WAF1 levels are typically observed in response to cell stress or following DNA damage. The binding of p21CIP1/WAF1 to cyclins and their CDK binding partner (e.g., CDK1 or CDK2) results in inhibition of the kinase complex, resulting in cell-cycle arrest (reviewed in References [29,31–33]). It is generally accepted that p21CIP1/WAF1 regulates both cell-cycle arrest, i.e., checkpoint activation, and apoptosis. Collectively, the actions of p21CIP1/WAF1 promote cell survival and provide time for cells to undergo DNA repair before completing the cell cycle. However, it is becoming increasingly clear that the role of p21CIP1/WAF1 in regulating apoptosis is more complex than originally thought [30,48]. For example, several investigators demonstrated that pro-apoptotic agents induce increases in p21CIP1/WAF1 and, furthermore, over-expression of p21CIP1/WAF1 enhances apoptosis [49–52]. Likewise, cells and animals deficient in p21CIP1/WAF1 expression exhibit reduced susceptibility to apoptotic cell death [53,54].

In the current study, we employed a combination of gene editing (CRISPR/cas9) and pharmacologic intervention to assess the functional consequence of Cdt-induced increases in p21CIP1/WAF1. Collectively, our observations indicate that Cdt-induced apoptosis is dependent upon increased levels of p21CIP1/WAF1. Jurkatp21<sup>−</sup> cells failed to exhibit apoptotic death following treatment with Cdt. Nonetheless, Jurkatp21<sup>−</sup> cells retained their susceptibility to other pro-apoptotic agents such as paclitaxel. Additionally, we employed the novel p21CIP1/WAF1 inhibitor UC2288 whose structure is based upon sorafenib, another inhibitor of p21CIP1/WAF1. UC2288 is a new-generation inhibitor that exhibits greater selectivity with higher potency to attenuate p21CIP1/WAF1 levels [41,55]. Similar to our findings with Jurkatp21<sup>−</sup> cells, JurkatWT cells treated with Cdt in the presence of 10 μM UC2288 also failed to exhibit increased levels of p21CIP1/WAF1; moreover, the drug suppressed Cdt-induced apoptosis. It should also be noted that the effective dose of UC2288 used in this study was identical to that employed in other studies [41,55].

Our results demonstrate a pivotal role for p21CIP1/WAF1 in Cdt-induced apoptosis (Figure 7); these observations encompass a non-conventional role for p21CIP1/WAF1 as it proposes a pro-apoptotic function when increases in this regulatory protein are commonly thought to, at least initially, serve a pro-survival role. It is well established that the pleiotropic effects of p21CIP1/WAF1 are regulated by post-transcriptional modification. For example, phosphorylation of p21CIP1/WAF1 is known to alter its function which includes a shift in the balance between its pro- and anti-apoptotic effects [29,31,56,57]. In this regard, p21CIP1/WAF1 was shown to serve as a downstream target for phosphorylation (pp21CIP1/WAF1 (T-145)) by pAkt, the active form of Akt [58,59]. Consistent with these findings is our observation that Cdt-induced increases in p21CIP1/WAF1 were not only associated with the induction of apoptosis but also accompanied by a substantial shift in the distribution of pp21CIP1/WAF1 levels relative to untreated cells. It is noteworthy that, in addition to inducing p21CIP1/WAF1 increases, we previously established that treatment of lymphocytes with *A. actinomycetemcomitans* Cdt results in blockade of the PI-3K signaling cascade and, furthermore, Cdt toxicity is dependent upon perturbation of these early signaling events. Cdt-mediated PI-3K blockade is the result of potent lipid phosphatase activity, specifically PIP3 phosphatase activity, associated with the active Cdt subunit, CdtB [26]. Cells exposed to Cdt exhibit PI-3K signaling blockade within 2 h; this is characterized by PIP3 depletion and reduced Akt activity [24,27,28]. It is important to note that Jurkatp21<sup>−</sup> cells retained their susceptibility to Cdt-induced PI-3K signaling blockade but were unable to undergo apoptosis due to their inability to

produce p21CIP1/WAF1. We propose that depletion of pAkt in Cdt-treated cells limits phosphorylation of p21CIP1/WAF1 (Figure 7); this relationship is further supported by our finding that the Akt inhibitor, GSK690693, also depletes JurkatWT cells of pp21CIP1/WAF1. Collectively, these findings support a critical pro-apoptotic role for non-phosphorylated p21CIP1/WAF1 in mediating Cdt-induced cell death.

**Figure 7.** Overview of the proposed mechanism via which Cdt induces p21CIP1/WAF1-dependent apoptosis. Based upon the data presented in this paper, the proposed pathway of Cdt-induced p21CIP1/WAF1-dependent apoptosis is shown with red arrows. The dotted line shows the proposed relationship between Akt and the putative downstream target, p21CIP1/WAF1.

Finally, the link between increased p21CIP1/WAF1 levels and the onset of apoptosis was established in experiments in which we assessed the ability of Cdt to alter the expression of pro-apoptotic members of the Bcl-2 family. In previous studies, we, and others, established that Cdt-induced apoptosis involves development of the mitochondrial permeability transition state characterized by a decrease in the transmembrane potential and an increase in production of reactive oxygen species [43,60,61]; these events were followed by activation of the caspase cascade. Of particular relevance to our current study were our previous findings that over-expression of Bcl-2 blocked Cdt-induced apoptosis [39]. In this context, the pro-apoptotic requirement for p21CIP1/WAF1 in Cdt-induced apoptosis is further supported by our current observation that the toxin induces p21CIP1/WAF1-dependent upregulation of the pro-apoptotic proteins Bid, Bax, and Bak. P21CIP1/WAF1-dependency was demonstrated by the observation that the levels of these proteins were not altered in Cdt-treated Jurkatp21<sup>−</sup> cells. Moreover, Jurkatp21<sup>−</sup> cells also failed to exhibit changes in the ΔΨm as we observed with Cdt-susceptible cells. It is noteworthy that Gogada et al. [62] demonstrated that p21CIP1/WAF1 played a critical role in circumin-induced apoptosis by altering mitochondrial permeability, thereby facilitating the release of cytochrome c.

#### *3.2. Cdt Toxicity Requires HSP90-Dependent Increases in the Levels of p21CIP1*/*WAF1*

Eukaryotic cell-cycle progression and cell survival are regulated at multiple checkpoints involving a number of critical regulatory proteins. One such regulatory protein is p21CIP1/WAF1 which functions as a cyclin-dependent kinase inhibitor [29]. Elevated levels of p21CIP1/WAF1 are critical to its function; typically, p21CIP1/WAF1 is expressed at low levels under normal growth conditions as an unstable protein with a short half-life. It is well established that p21CIP1/WAF1 can be upregulated via a number of p53-dependent and -independent mechanisms (reviewed in Reference [31]). Likewise, Cdt-induced p21CIP1/WAF1 increases were reported to be both p53-dependent and -independent (reviewed in Reference [1]); these findings led others to propose that the role of p53 with respect to Cdt is dependent upon both the specific target cell and the source of toxin. Of particular relevance to this study, Sato et al. [35] demonstrated that *A. actinomycetemcomitans* Cdt-treated lymphocytes exhibit p53-independent increases in p21CIP1/WAF1. We confirmed these observations by demonstrating that *A. actinomycetemcomitans* Cdt-treated lymphocytes exhibited increases in p21CIP1/WAF1 and apoptosis when pre-exposed to the p53 inhibitor pifithrin-α.

Stress and/or DNA damage are known to induce increases in p21CIP1/WAF1 levels as a result of elevated transcription, RNA stability and/or decreased proteasomal degradation (reviewed in References [29,33]). Cdt-induced increases in p21CIP1/WAF1 were not accompanied by significant changes in mRNA levels. These findings led us to explore the possibility that the observed increases were instead due to protein stabilization and reduced proteasomal degradation. It is in this context that we considered the role of the heat shock protein HSP90 in both Cdt-induced increases in p21CIP1/WAF1 and toxicity. As noted above, p21CIP1/WAF1 is typically expressed at low levels as an unstable protein with a short half-life in normally growing cells. HSP90 is known to function as a chaperone that is critical to stabilizing proteins involved in several cellular processes [46]. In particular, HSP90 was shown to be recruited to p21CIP1/WAF1 where it binds via the WAF1/CIP1 stabilizing protein 39 (WISp39). The complex of HSP90, WISp39, and p21CIP1/WAF1 was reported to stabilize and protect the cell-cycle inhibitor from proteasomal degradation [44,45].

We extended these observations to Cdt-induced stress in lymphocytes by firstly assessing toxin-treated Jurkat cells for changes in both WISp39 and HSP90. Jurkat cells constitutively express WISp39, and we observed that exposure to Cdt did not result in further increases (data not shown). In contrast, Cdt-treated cells exhibited significant increases in HSP90 levels within 4 h, prior to the peak elevation in p21CIP1/WAF1. It should also be noted that the ability of CdtB to alter HSP90 levels, as demonstrated for p21CIP1/WAF1, was found to be dependent upon PIP3 phosphatase activity, since toxin-containing CdtB subunit mutants that lacked phosphatase activity were unable to induce increases in HSP90 levels. To further investigate the relationship between HSP90, p21CIP1/WAF1, and apoptosis, we employed the HSP90 inhibitor, GM. Specifically, our observations demonstrate that cells pre-treated with GM prior to exposure to Cdt prevented toxin induces increases in p21CIP1/WAF1; moreover, these cells exhibited decreased susceptibility to toxin-induced apoptosis, demonstrating a linkage between HSP90, changes in p21CIP1/WAF1 levels, and toxicity.

#### **4. Conclusions**

In summary, this study advances our understanding of the molecular events leading to *A. actinomycetemcomitans* Cdt-mediated toxicity in lymphocytes. We previously demonstrated that Cdt toxicity is dependent upon PI-3K signaling blockade in lymphocytes, mast cells, and macrophages [2,24–26,28,63,64]. A key to impairment of this signaling pathway is the ability of the active Cdt subunit, CdtB, to function as a PIP3 phosphatase. Our current observations utilized both pharmacologic and gene editing approaches to demonstrate that toxin-induced apoptosis is also dependent upon increased levels of p21CIP1/WAF1. Furthermore, toxin-induced increases in this critical regulatory protein are dependent upon HSP90. Moreover, the timeline for these and previous observations suggest a sequence in which the earliest events involve: Cdt binding to membrane cholesterol via CdtC and CdtB, internalization of CdtB, and depletion of PIP3, leading to a concomitant decrease in pAkt (loss of activity), increased expression/activation of HSP90, increased levels of p21CIP1/WAF1, and increased levels of pro-apoptotic Bcl-2 family proteins [28,65]. We propose that the ability of Cdt to impair lymphocyte proliferation and promote cell death therefore compromises the host response to Cdt-producing organisms. Our observations are of particular significance, as Cdts are produced by not only *A. actinomycetemcomtans*, but also by over 30 γ- and ε- Proteobacteria [1,2]. Therefore, we further propose that the action of this putative virulence factor contributes to the pathogenesis of a range of diseases leading to persistent infection by Cdt-producing pathogens. Our current findings not only contribute to a greater understanding of the molecular events critical to Cdt toxicity, but also provide avenues for developing novel approaches to attenuating the immunoinhibitory effects of the toxin.

#### **5. Materials and Methods**

#### *5.1. Reagents and Antibodies*

We previously reported on the construction and expression of the plasmids which contain the *cdt* genes for the holotoxin (pUCAacdtABChis), as well as those constructs containing CdtB mutations [10]. The histidine-tagged holotoxin was isolated by nickel affinity chromatography as previously described [66]. All antibodies were obtained from commercial sources as indicated. UC2288 was provided by RH Weiss), GM was purchased from Thermofisher Scientific (Waltham, MA, USA), and GSK690962 was purchased from Cayman Chemical (Ann Arbor, MI, USA).

#### *5.2. Culture Conditions and CRISPR*/*cas9-Mediated Genome Editing*

Two human lymphoid cell lines were employed in these studies: the T-cell leukemia cell line Jurkat (E6-1) and the cutaneous T-cell lymphoma cell line, MyLa2059. Cells were maintained as previously described [10]. JurkatWT cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 supplemented with fetal bovine serum (FBS) (10%), glutamine (2 mM), 4-(2-hydroxyethyl)-1 piperazineethanesulfonic acid (HEPES) (10 mM), penicillin (100 U/mL), and streptomycin (100 μg/mL). MyLa2059 cells were maintained in the same medium containing 20% FBS. Human peripheral blood mononuclear cells (HPBMCs) were prepared and incubated as described previously [67]; blood was obtained using an Institutional Review Board-approved protocol, and all donors provided written consent.

To generate p21CIP1/WAF1-deficient Jurkat cells (Jurkatp21<sup>−</sup>), we utilized CRISPR/cas9 technology (Santa Cruz Biotechnology; Santa Cruz, CA, USA) as previously described [68]. Cells were transfected (Amaxa Nucleofector system; Lonza, Basel) with a pool of three plasmids, each encoding the Cas9 nuclease and a p21CIP1/WAF1-specific 20-nt guide RNA (gRNA). Cells were co-transfected with a pool of three plasmids each containing a homology-directed DNA repair (HDR) template; this corresponded to sites generated by the p21CIP1/WAF1 CRISPR/cas9 knockout plasmids. HDR plasmids insert the puromycin resistance gene that facilitates selection of stable knockout cells. Cells were firstly incubated for a five-day period following transfection and then for an additional seven days in incubation in puromycin (5 μg/mL). Limiting dilution of surviving cells was utilized to clone cells; clones were expanded and assessed by Western blot analysis for both the presence of p21CIP1/WAF1 their ability to increase p21CIP1/WAF1 levels in response to Cdt or etoposide. Clones determined to be deficient in p21CIP1/WAF1 were then cloned a second time using limiting dilution (Figure 4A). Jurkatp21<sup>−</sup> cell lines were maintained in medium containing puromycin (1 μg/mL); experiments were conducted in medium without puromycin.

#### *5.3. Assessment of Apoptosis*

JurkatWT and Jurkatp21<sup>−</sup> cells were challenged with Cdt or medium (control) for 24 h, and apoptosis was assessed by measuring DNA fragmentation (In Situ Cell Death Detection Kit; Sigma Aldrich, St. Louis, MI, USA) [43]. Briefly, after 24 h of incubation, the cells were re-suspended in freshly prepared 4% formaldehyde and permeabilized with 0.1% Triton X-100 for 2 min at 4 ◦C; then, they were washed and stained with a solution containing FITC-labeled nucleotide and terminal deoxynucleotidyl transferase (TdT). Flow cytometry was employed to measure FITC fluorescence with a laser at 488 nm to excite the fluorochrome; fluorescence emission was measured through a 530/30-nm bandpass filter.

To measure development of the permeability phase transition, Jurkat cells were incubated for 18 h in medium alone or containing Cdt under conditions described above. Changes in the transmembrane potential (ΔΨm) were determined using 4 nM 3,3 -dihexyloxacarbocyanine (DIOC6(3); Thermofisher) [39,43]. Cells were stained for 15 min (37 ◦C) with the fluorochrome and fluorescence measured following excitation with a laser at 488 nm (250 mW), and emission was monitored through a 530/30-nm bandpass filter; at least 10,000 cells were analyzed per sample.

#### *5.4. Western Blot Analysis*

Cells were incubated with and without Cdt as described above; following the indicated incubation period, cells were solubilized in 20 mM Tris-HCl buffer (pH7.5) containing 150 mM NaCl, 1 mM EDTA, 1% NP-40, 1% sodium deoxycholate, and a protease inhibitor cocktail (ThermoFisher Scientific; Waltham, MA, USA). Samples (30 μg) were fractionated on 12% SDS-PAGE and then blotted onto PVDF membranes; the membranes were blocked with BLOTTO and then incubated with one of the following primary antibodies (Cell Signaling Technology; Danvers, MA, USA) for 18 h at 4 ◦C [12]: anti-Akt, anti-pAkt (S473), anti-GSK3β, anti-pGSK3β (S9), or anti-GAPDH; anti-p21CIP1/WAF1, anti-pp21CIP1/WAF1, and anti-HSP90 antibodies were also employed (Abcam; Cambridge, MA, USA). The membranes were incubated with goat anti-rabbit immunoglobulin conjugated to horseradish peroxidase (Southern Biotech Technology; Birmingham, AL, USA) after they were blocked and washed. The Western blots were developed using chemiluminescence and analyzed by digital densitometry (Li-Cor Biosciences; Lincoln, NE, USA) as previously described [25].

#### *5.5. Statistical Analysis*

The mean ± standard error of the mean was calculated for replicate experiments. Significance was determined using a Student's *t*-test using SigmaPlot Software (Systat; San Jose, CA); a *p*-value of less than 0.05 was considered to be statistically significant.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-0817/9/1/38/s1: Figure S1: Effect of Cdt containing CdtBWT or one of the CdtB mutants on phosphorylation of H2AX; Figure S2: Susceptibility of Jurkatp21<sup>−</sup> cells paclitaxel; Figure S3: Effect of pifithrin-α (PFT) on Cdt-induced increases in p21CIP1/WAF1 and apoptosis.

**Author Contributions:** Conceptualization, B.J.S., K.B.-B., and R.H.W.; methodology, B.J.S., L.M.W., A.Z., and K.B.-B.; validation, L.M.W. and A.Z.; formal analysis, B.J.S., L.M.W., A.Z., and K.B.-B.; investigation, L.M.W. and A.Z.; resources, B.J.S., R.H.W., and K.B.-B.; data curation, L.M.W. and A.Z.; writing—original draft preparation, B.J.S.; writing—review and editing, R.H.W. and K.B.-B.; supervision, B.J.S. and K.B.-B.; project administration, B.J.S. and K.B.-B.; funding acquisition, B.J.S. and K.B.-B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Institutes of Health grants DE0006014 and DE023071.

**Acknowledgments:** The authors wish to acknowledge the expertise and assistance of the Flow Cytometry Core Facility at the University of Pennsylvania School of Dental Medicine.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
