Functional Heterogeneity of Protein Kinase A Activation in Multipotent Stromal Cells

Multipotent stromal cells (MSC) demonstrate remarkable functional heterogeneity; however, its molecular mechanisms remain largely obscure. In this study, we explored MSC response to hormones, which activate Gs-protein / cyclic AMP (cAMP) / protein kinase A (PKA) dependent signaling, at the single cell level using genetically encoded biosensor PKA-Spark. For the first time, we demonstrated that about half of cultured MSCs are not able to activate the cAMP/PKA pathway, possibly due to the limited availability of adenylyl cyclases. Using this approach, we showed that MSC subpopulations responding to various hormones largely overlapped, and the share of responding cells did not exceed 40%. Using clonal analysis, we showed that signaling heterogeneity of MSC could be formed de novo within 2 weeks.

Previously, using single-cell signaling analysis of Ca 2+ -dependent intracellular signaling in living cells, we demonstrated that MSCs functional heterogeneity could be due to different ability to initiate G q /calcium-dependent cascades in response to hormones, such as noradrenaline and angiotensin II [7,10]. Moreover, Ca 2+ -dependent functional activity of MSC is regulated by cyclic AMP (cAMP)/protein kinase A (PKA)-dependent signaling activated by β-adrenoceptors [15]. cAMP/PKA signaling pathway also regulates pivotal functions of MSC, such as adipogenic differentiation [16] or regulation of resident stem cells [17].   Interestingly, noradrenaline induced fast dissolving of fluorescent droplets in some previously stimulated cells, which means predominantly activation of Gi-associated α2-adrenoceptors in these particular cells (Supplementary Figure S7). That also means that long-term preservation of the fluorescent droplets after washing the hormone corresponds to the real dynamics of PKA-dependent phosphorylation but is not a biosensor-dependent artifact. Evaluating the share of responding cells (Figure 2b), we observed that it does not exceed 40% for any of hormones tested (mean values are 37.0% for adenosine; 34.7% for noradrenaline; 25.2% for dopamine; 16.6% for serotonin; and 4.2% for histamine). The share of the cells responding to adenosine and noradrenaline was close to the number of MSCs responding to forskolin, i.e., which were capable of PKA activation. The close proximity of the numbers of responding cells to the observed maximum may indicate that distinct hormones activate cAMP/PKA pathway in the same cell. To find out whether the same cell responds to different cAMP-activating hormones, we added different hormones sequentially. Due to the biosensor peculiarities, we could not return cells to the basal state by simply washing off added hormones (see Methods section for details). Therefore, we intensively washed the cells before adding the next hormone but did not wait for returning the biosensor to the basal state. As Figure 2c shows, nearly 35% of MSC responded to adenosine, and adding of noradrenaline led to additional activation of only a few percent of the cells. At the same time, a different order of hormonal stimulation, for example, adenosine after dopamine, as shown in Figure 2d, or dopamine after noradrenaline, or adenosine after noradrenaline, and so on, showed the same results: the second hormone activated the same cells as the previous one with only a small addition. We did not see additivity in cellular responses and the overall share of responding cells was consistent with the observation that only about 40% of MSCs are capable of cAMP/PKA activation.

MSC Demonstrate Heterogeneous Response to the Hormonal Stimuli
Interestingly, noradrenaline induced fast dissolving of fluorescent droplets in some previously stimulated cells, which means predominantly activation of G i -associated α2-adrenoceptors in these particular cells (Supplementary Figure S7). That also means that long-term preservation of the fluorescent droplets after washing the hormone corresponds to the real dynamics of PKA-dependent phosphorylation but is not a biosensor-dependent artifact.
More detailed examination of cAMP-mediated responses of MSC showed some patterns in cellular responses to individual hormones. Adenosine, dopamine, and noradrenaline induced strong and transitory responses, whereas histamine and serotonin activated long but weak cAMP-dependent responses (see Figure 3 and Supplementary Figures S2-S6). Interestingly, some particular cells activated PKA for a long time even in response to adenosine, noradrenaline, and, in one case, dopamine (see Figure 3b It is clear that long responses predominantly have much lower amplitude compared to transitory signals. Thus, MSCs respond to adenosine, dopamine, histamine, noradrenaline, and serotonin, but the amplitude and temporal dynamics of the signal vary between different hormones, as well as between the responses of different cells to a particular hormone.

MSC Functional Heterogeneity Is Highly Dynamic
The lack of sensitivity to cAMP-dependent signals in some MSCs can be either a dynamic property of the entire population or stable characteristic of some groups of the cells. To find out whether MSC are able to form functional heterogeneity with respect to cAMP-dependent signaling, we obtained single cell-derived clones of MSC expressing PKA-Spark. We used immortalized adipose-derived MSC to get higher efficiency of clonal growth. Cells were analyzed two weeks after sorting 1 cell per well. The entire plate with grown MSC clones is shown in Supplementary Figure S8. The cells from one clone showed similar phenotype regarding size, morphology, and PKA-Spark expression level (see Figure 4). At the same time, forskolin stimulation led to PKA-Spark response in 43.6 ± 7.5% of the cells of clones, which is highly consistent with the data obtained in the total population. Thus, functional heterogeneity of MSC seems to be acquired during cooperative growth of the cells, even in the case of genetically identical cells from one clone.
(section (b) in all figures) show the results presented in a different fashion, where it is possible to compare signal amplitudes. It is clear that long responses predominantly have much lower amplitude compared to transitory signals. Thus, MSCs respond to adenosine, dopamine, histamine, noradrenaline, and serotonin, but the amplitude and temporal dynamics of the signal vary between different hormones, as well as between the responses of different cells to a particular hormone.

Discussion
MSCs carry out a huge number of functions in the tissues they locate, from differentiation into several mesenchymal directions to regulation of a number of physiological functions, such as angiogenesis, neurogenesis, modulation of immune response, and regulation of tissue specific stem cells differentiation. It is hard to imagine that every cell of the MSC population performs all these functions. An idea of heterogeneity of the MSC population is rather old, but attempts to link the particular cell' phenotype with its function or sorting of the MSCs to functionally homogeneous sub-populations were often without effect. Several attempts, however, were successful. Mendez-Ferrer and colleagues described nestin-positive subpopulation of bone marrow derived

Discussion
MSCs carry out a huge number of functions in the tissues they locate, from differentiation into several mesenchymal directions to regulation of a number of physiological functions, such as angiogenesis, neurogenesis, modulation of immune response, and regulation of tissue specific stem cells differentiation. It is hard to imagine that every cell of the MSC population performs all these functions. An idea of heterogeneity of the MSC population is rather old, but attempts to link the particular cell' phenotype with its function or sorting of the MSCs to functionally homogeneous sub-populations were often without effect. Several attempts, however, were successful. Mendez-Ferrer and colleagues described nestin-positive subpopulation of bone marrow derived MSC, which uniquely transmits regulatory signals from sympathetic nervous system to hematopoietic stem cells [17]. Recently, using the separation of the MSC population based on the hormonal sensitivity of individual cells and their signaling properties, we identified a subpopulation of MSC, which differentiates into adipocytes faster and more effectively than other MSC, but, at the same time, they are not committed preadipocytes [7]. These cells differ from other MSC by co-expression of angiotensin II receptors of types 1 and 2, which can heterodimerize and change signaling properties of MSC and their response to differentiation stimuli [20]. Responsivity to the hormonal stimuli and the type of response define functional activity of MSC; thus, it is a sensitive and effective approach to study functional heterogeneity of MSC. cAMP-dependent signaling is activated by a number of critical regulators of MSC functional activity, such as noradrenaline or adenosine. Moreover, as we recently showed, activation of cAMP/PKA-dependent signaling leads to heterologous sensitization of MSC and potentiation of α1A-adrenoceptor/Ca 2+ -dependent response [15].
In this work, we demonstrated, for the first time, heterogeneity of hormone-activated cAMP-dependent responses of MSC at the single cell level. We showed that MSC responded to the hormones heterogeneously, and different hormones activated PKA with different time dynamics and amplitude. Moreover, individual cells may respond to the same hormone differently. This variable signaling may play a crucial role in the ability of cells to react to the complex hormonal regulation in the tissue. cAMP-dependent signaling lacks the unparalleled specificity and networking of signaling pathways associated with receptor Tyr-kinases, as cAMP elevation mainly affects PKA. However, PKA specificity may be regulated by temporal profile of PKA activation and through its interaction with A-kinase anchoring proteins (AKAPs), which is mediated by diversity of AKAP isoforms [21]. Thus, responsive cells are capable of sensing more than one hormone at the time and forming a wide variety of cellular answers, depending on external hormonal stimuli. These findings on type and format of cAMP-dependent response of particular cells may be associated with the functional activity of tissue, but this requires further research.
Signaling heterogeneity of PKA-dependent response is due to the lack of adenylyl cyclase in some MSCs. This is an extremely unexpected result indicating a possible new level of the regulation of cAMP/PKA signaling by controlling the availability of adenylyl cyclase. Direct analysis of adenylyl cyclase expression at the single cell level is complicated by the abundance of adenylyl cyclase isoforms in mammalian cells [22]. Although methods of multiplex protein analysis at the single cell level have already appeared, these approaches are still laborious and low-accessible. Another possible approach to the analysis of protein expression in both multiplex and single cell modes, is single cell RNAseq. To partially testify our findings about heterologous activity of adenylyl cyclases, we analyzed openly available results of single cell RNAseq of human and mouse adipose derived MSCs [3,4,19]. We found near 60% of the cells that lacked mRNA of all isoforms of adenylyl cyclase (Adcy1 to Adcy10 genes in mice and ADCY1 to ADCY10 genes in humans). Although mRNA level does not necessary reflects the expression of corresponding protein, especially in case of their detection at the single cell level, we suggest that cells lacking adenylyl cyclase mRNA may have a reduced level of these proteins; thus, they do not respond to cAMP/PKA-mediated hormonal stimuli.
There were some reports about heterogeneous expression and responses of adenylyl cyclases in clonally derived cells [23]. Due to the lack of reliable probe for registration of cAMP production or PKA activity at single cell level, the authors obtained single cell derived clones to compare cell-to-cell heterogeneity indirectly. Here, we showed that PKA-mediated responses are varied even between the cells of one clone. PKA-associated signal heterogeneity can be formed during the proliferation of single-cell derived clones, which means that this heterogeneity is a dynamic process. Thus, it can indicate that MSCs are capable of self-organization and specialization via cell-to-cell interactions (probably both contact and paracrine). This phenomenon may determine the ability of MSC to form cell niches in tissues where cells can form a complex system composed of cells primarily attributing to the reception of stimuli and cells dedicated to driving tissue cell renewal via proliferation and differentiation.
PKA signaling heterogeneity forming in cell clones can be viewed as a manifestation of «gene expression noise» [24]. Previously, a similar phenomenon was described in regard to stem cell marker Sca-1 [25]. It is known that progenitor cells demonstrate stochasticity in gene expression, which results from fluctuations in transcription and translation [26]. Moreover, this stochasticity plays an important role in maintaining the multipotency of stem cells and decreases as cells become more and more lineage-restricted [27].

RNA Isolation and PCR Detection of Receptors
The expression of receptors coupled with intracellular PKA-signaling was evaluated using RT-PCR. Total RNA was isolated from cells using RNeasy Mini Kit (Qiagen, Germantown, MD, USA) according to the user manual. 1 µg of total RNA was used for reverse transcription using MMLV Reverse transcription kit (Evrogen, Moscow, Russia). Resulting cDNA was used as a template for PCR with specific primers pairs (Evrogen, Moscow, Russia; see Table 1 for details). All specified procedures were carried out according to the manufacturer's instructions. PCR products were then analyzed by electrophoresis in the agarose gel (2% agarose, visualization with EtBr), and product length was estimated using GeneRuler 50 bp DNA Ladder (Thermo Fisher, Logan, UT, USA).

Lentiviral Transduction of MSC
To deliver the cAMP sensor to MSCs, it was cloned from pcDNA3-PKA-SPARK (#106920, Addgene, Watertown, MA, USA) into the LeGO-iG2 lentiviral transfer vector (#27341, Addgene, Watertown, MA, USA) using restriction endonucleases MluI and EcoRI. Lentiviral particles (LVPs) were assembled in HEK293T cells using the standard PEI transfection protocol [28]. The conditioned medium containing LVPs was collected 48-72 h post-transfection and separated from cell debris by centrifugation at 4000× g 4 • C for 15 min. In order to increase the efficiency of transduction, protamine sulfate (50 µg/mL) was added to the medium containing LVPs. An immortalized MSC line (ASC52telo) or MSC of early (<3) passages were used for transduction. Cells cultured on 24-well plates were incubated in the medium containing LVPs and subjected to centrifugation at 800× g RT for 1.5 h to assist virus entry. After centrifugation, the medium was changed to standard culture medium and cells were incubated for~8 h. The transduction procedure was repeated 2 times. Between transductions, the virus stock was stored at +4 • C.

Registration of PKA Activation with PKA-Spark
To registrate PKA activation with PKA-Spark, we grew transduced cells in 24-or 48-well plates at low densities to prevent cell-to-cell communications during imaging of signaling. Before the experiment the growth medium was changed to To optimize the detection of local PKA activation using PKA-Spark biosensor, we used the HeLa-Kyoto cell line. These cells were transfected with PKA-Spark encoding plasmid using FuGENE HD transfection reagent (Promega, Logan, UT, USA) according to manufacturer's instruction. Twenty-four hours after transfection, we stimulated these cells by serotonin and registered the formation of fluorescent aggregates in the cytoplasm, which corresponded to the activation of G s /AC/PKA-dependent signaling. Importantly, only cells with high PKA-Spark expression level demonstrated the formation of aggregates in response to the hormone ( Figure 5). The response of other cells was possible to register only 1-2 days later, when PKA-Spark expression increased sufficiently. Therefore, only cells with enough level of biosensor expression should be analyzed. For further analysis of MSC responses, we first identified the threshold level of biosensor expression, which allows the detection of the formation of fluorescent aggregates in the cytoplasm, and only cells with PKA-Spark expression exceeding established threshold were taken into analysis. To optimize the quantification of fluorescent aggregates formation, we compared 3 alternative approaches: 1-SPARK signal was quantified as the sum of fluorescent droplets' pixel intensity divided by sum of cells' pixel intensity [18]; 2-signal was quantified as changes in standard deviation (SD) of the picture intensity; and 3-SPARK signal was calculated as the sum of the area of all fluorescent droplets after background subtraction using a "rolling ball" tool (ImageJ software, Bethesda, MD, USA) (Figure 5b-e). We found out that the latter approach, a calculation of total area of fluorescent aggregates, shows the minimal level of signal in the absence of stimulation (Figure 5d) and the highest contrast (Figure 5e). Therefore, for further analysis, we used this approach to quantify MSC responses. Images were processed using ImageJ software. We also analyzed the signal reversibility and found that PKA-Spark demonstrates a very slow dynamics of the fluorescent droplets disassembly, and the droplets partially remain 30 min after washing off the hormone. Thus, the repetitive hormonal stimulation of the same cells which is commonly used during calcium signaling registration [10] is not possible for this detection method.

Immunofluorescent Analysis of PKA Expression in MSCs
Cells were fixed in 4% paraformaldehyde for 10 min. After several washes by phosphate buffer saline (PBS) (PanEco, Moscow, Russia), cells were incubated in 0.1% bovine serum albumin (BSA) containing 10% normal donkey serum to block non-specific binding of antibodies. This was followed by incubation with specific primary antibodies against PKA (PA5-70360, Thermo Fisher Scientific, Logan, UT, dilution 1:100), for 1 h and subsequent extensive washing in PBS. Then, cells were incubated with Alexa488-conjugated donkey anti-rabbit (Molecular Probes, Logan, UT, USA). Cell nuclei were counterstained with DAPI (Sigma-Aldrich, St. Louis, MO, USA) and mounted in Aqua Poly/Mount (Polysciences Inc, Warrington, PA, USA). For negative controls, rabbit non-specific IgGs were used in appropriate concentration. Images were obtained using the confocal microscope LSM 780 and ZEN2010 software (Zeiss, Oberkochen, Germany).

scRNAseq Data Processing
To analyze patterns of different isoforms of adenylyl cyclases expression at the single cell level, we used freely available data of single cell RNAseq analysis of adipose-derived MSC. Raw data were downloaded from databases ArrayExpress (EBI) иGEO (NCBI, SRA): mouse-derived MSC (E-MTAB-6677) [3], (SRP145475) [4], and human-derived MSC (SRP148833) [19]. Downloaded datasets were processed using Cellranger 3.0 software with standard parameters, and analysis of cloupe files was performed using Loop Browser 4.0 software (10x Genomics, San Francisco, CA, USA). The number of the cells lack of adenylyl cyclases mRNA was counted using 'Filter' tool in Loop Browser 4.0 and filtering on adenylyl cyclase isoforms genes Adcy1 (ADCY1 in case of human MSC population analysis) to Adcy10 (ADCY10 in case of human MSC).

MSC Sorting and Clonal Analysis
To obtain single cell derived clones of PKA-Spark expressing cells, hTERT-MSC 6 days after lentiviral transduction were re-suspended in BSA buffer (DPBS, 0.1% BSA) and the 10% of the brightest cells were sorted in single-cell mode separately into 96-well tissue culture plates. Cells were sorted based on fluorescence in 488-530/30 channel on a BD FACSAria III sorter. Doublets were excluded based on side scatter height and width. Then, single clones were cultured in AdvanceSTEM full growth medium for two weeks. The medium was changed to the fresh every 3-4 days.

Statistics
The share of responding cells was calculated as the ratio of the number of cells that formed fluorescent droplets in response to the hormonal stimulation, to the number of all cells in which brightness exceeds the established threshold level of biosensor expression.

Conclusions
This study demonstrates that cAMP-activating hormones could exert their cAMP-dependent effects only in about half of MSCs. At the same time if a particular MSC activates cAMP production in response to one cAMP-activating hormone, it would respond to another one as well. Surprisingly, the heterogeneity of hormone-activated cAMP-dependent responses results from various availability of adenylyl cyclases. Thus, our data indicate that despite generally accepted redundancy and ubiquitous expression of adenylyl cyclase isoforms, the presence of this enzyme in particular MSC determines their ability to respond to hormones by activation of PKA/cAMP signaling. Interestingly, this signaling heterogeneity observed in total population re-establishes in single-cell derived clones. This may underline heterogeneous functional responses of MSC to hormones. Molecular mechanisms of establishing of MSC heterogeneity caused by the regulation of adenylyl cyclase expression as well as functional consequence of such heterogeneity need to be further elucidated.