Inherent Fluorescence Demonstrates Immunotropic Properties for Novel Janus Kinase 3 Inhibitors

Abstract

There is a general question in small molecule pharmacology about how apparent compound concentrations in blood, plasma, and organs actually relate to actual amounts at the target site of a compound. In this study, we used inherently fluorescent JAK3 ligands and their macrolide conjugates to investigate the relationship between physical properties, apparent bulk concentration, and organ and subcellular distribution. In vitro uptake into immune cells suggested that much of the substance was associated with granules or organelles. Samples from murine pharmacokinetic studies were analyzed by both conventional mass spectrometry and cryofluorescence microscopy methods to show the distribution of a compound within organs and cells without artifacts of fixation. These observations confirm the uptake of granules observed in vitro. Data from macrolides carrying either a coumarin fluorophore or a JAK3 inhibitor were similar, suggesting that the distribution is directed by the properties of the larger macrolide. These data show a propensity for azalide macrolides to concentrate in the lung and gut epithelia and suggest that the plasma- or whole-blood-derived estimates of drug levels almost certainly underestimate concentrations of macrolides in the mucous membranes. Thus, their apparent efficacy at sub-bacteriostatic doses may reflect their higher levels in barrier layers.

1.

JAK3 is a member of the Janus kinase family that consists of JAK1–3 and tyrosine kinase 2 (TYK2). The JAKs are part of signal transduction cascades relevant to cell growth and function, e.g., for erythrocytes or immune cells.^1–3 JAK3 is predominantly expressed in leukocytes and some epithelia,⁴ which makes JAK3 a potential target for immunosuppressive and anti-inflammatory treatments that may be free of more general effects. As JAK3 is always coexpressed with JAK1 and they form heterodimers, the expectation is that a selective compound would modulate immune responses, but not abrogate immunity to pathogens.^5,6

Previously, we reported the synthesis, structure–activity relationship, and pharmacokinetic properties of covalent-reversibly binding, ATP competitive small molecule JAK3 inhibitors based on the highly selective and potent inhibitor FM-381 (1).⁷ Through variation of residues that are noncritical for binding to the kinase, a series of inhibitors with different tissue preferences were obtained (Figure 1).

Figure 1.

General structure of our series of JAK3 inhibitors. R₁ = 3–6 membered (hetero)alicyclic or aromatic ring; R₂ = short carbon chain or linker + macrolide carrier.

Next, we demonstrated their in vitro and in vivo efficacy in attenuating LPS-induced inflammatory reactions.⁸ We now report their structure-dependent tendency to accumulate in cells of the immune system, in particular, in the lysosomal compartment and within subsections of organ tissues, e.g., epithelial layers.

Lysosomes are cell organelles responsible for the degradation of macromolecules. The hydrolytic enzymes contained within lysosomes are most active under acidic conditions,⁹ and lysosomal pH, maintained by ATP-dependent proton pumps following activation, is around 4–5.¹⁰ Given the acidic milieu, we were particularly interested in the uptake of basic compounds, such as derivatives of azithromycin, into the lysosomal compartment.

Azithromycin (Figure 2) is a broad-spectrum macrolide antibiotic with unusual pharmacokinetic characteristics.¹¹ These pharmacokinetic properties are well-documented and include a long half-life, oral bioavailability, and preferential distribution into organ tissues. High concentrations can be found in the liver, lung, prostate, tonsil, spleen, and renal tissues even days after single-dose administration. On the other hand, plasma concentrations are comparably low. These levels are, however, stable due to slow redistribution from peripheral tissues. This slow release likely contributes to the postantibiotic effect exerted by the drug: even after the plasma concentration has fallen below the minimum inhibitory concentration, bacterial growth is still impaired for a considerable time.^12–14

Figure 2.

Structure of the macrolide antibiotic azithromycin. The two basic amino moieties and their pK_a values (corresponding acids) are highlighted.

Azithromycin accumulates strongly and rapidly in leukocytes. Concentrations 40-fold over the extracellular levels were observed in J774 cells. Up to 70% of this is reported to be found in lysosomes and only a smaller fraction remains in the cytosol.¹⁵ Both the uptake into the cytosol and the subsequent distribution into the lysosomal compartment are dependent on the local pH: azithromycin has two basic amines (pK_a 8.1 and 8.8)¹⁶ which are protonated under acidic conditions. Cationic molecules are hindered in their diffusion across cell membranes, limiting their uptake or outward diffusion, while the neutral, free base form can pass easily. On the other hand, once the compound enters the lysosomal environment (pH 4–5),¹⁰ it is protonated and subsequently trapped (Figure 3), leading to many-fold higher concentrations in comparison to the cytosol. In the case of dibasic molecules like azithromycin, the effect is even more potent compared to monobasic substances, as two monoprotonated and a bicationic form are in a dynamic equilibrium. Derendorf calculated a plausible accumulation ratio of up to 52.000, assuming a pH of 5.^17,18 The release of azithromycin from cells into extracellular space thus occurs in two distinct phases: A “fast” phase, where a large quantity of drug—the cytosolic portion—is released quickly and a prolonged, slower phase caused by redistribution from the acidic compartment(s) into the cytosol and then into extracellular space.^15,17

Figure 3.

Accumulation of basic compounds in the lysosomal compartment by “ion trapping.” Unprotonated compounds can freely pass through cellular and lysosomal membranes. In the acidic lysosomal compartment, equilibrium is shifted toward the protonated form, whose membrane passage is strongly hindered due to its charge. B = unprotonated compound/base; BH⁺ = protonated compound/base. Created with BioRender.com.

It has been demonstrated in animal models that azithromycin preferentially distributes to infected tissues by way of uptake into phagocytes which then migrate to the site of infection. There, the azalide is slowly released again, e.g., during lysis or degranulation processes. Even under leukopenic conditions, azithromycin concentrations are still significantly higher in sites of infection compared to normal tissues.^19,20

The advantageous pharmacokinetic properties of macrolides (particularly azalides), including their abundance in immune cells, can be transferred to other pharmacologically active compounds by conjugation. Carriers based on modified macrolide scaffolds have already been successfully linked to active “warheads” that retained their activity, while now also possessing improved distribution profiles.^7,8,21,22 This method is especially useful for anti-inflammatory drugs due to the aforementioned accumulation in acidic compartments and tissues. Furthermore, macrolides themselves are capable of nonspecific immune modulation, leading to potential synergistic effects.^21–25

The synthesis and characterization of JAK3 inhibitors conjugated to macrolide carriers improve pharmacokinetic properties while retaining adequate affinity to JAK3. Tissue concentrations of this class of conjugate compounds are substantially higher compared to their unlinked analogues, especially in the kidney, liver, and lungs. Plasma half-lives are also increased vs unconjugated compounds.⁷

In this report, we demonstrate the tendency of macrolides and macrolide-linked JAK3 inhibitors to accumulate in immune cells, especially in subcellular compartments, using either HPLC–MSMS methods or fluorescence microscopy. This series of compounds share a core pharmacophore with sufficiently strong inherent green fluorescence that can serve as its own reporter without modification. We employed the detection of fluorescence emissions at 525 ± 50 nm to directly observe their partition into cells.

Distribution of a selection of compounds was also investigated in vivo, where organs of treated animals were prepared to obtain cryo-preserved sections. These were then analyzed using the same microscopy methods to visualize their accumulation within the organ tissues following oral application. These observations allow the association between compound physical properties and apparent distribution to be assessed at both macro and subcellular levels.

2. Results and Discussion

2.1. UV/Vis Spectra

We recorded UV–vis absorption spectra of our library of JAK3 inhibitors to determine their absorption maxima (Figure 4) for fluorescence microscopy. Compounds were selected for high in vitro potency (e.g., 1), pharmacokinetic properties (e.g., 2, 4, and 5), or for being macrolide conjugates of other compounds (7, 8, 9, 10, 11, and 12) (Figure 5).⁷

Figure 4.

Absorption spectra of model compound 1 and several derivates at pH 7.4 (c = 50 μM). For clarity, only the relevant section at 250–600 nm is shown.

Figure 5.

Structures of the compounds used in this work.

Absorption spectra of the JAK3 inhibitor compounds and the positive control 13 were made at physiological pH 7.4 (Figure 4), 50 μM in phosphate-buffered saline. The majority of the λ_max values at pH 7.4 ranged between 370 and 390 nm. A noticeable difference in λ_max was observed for 3, which had absorption maxima at 270 and 315 nm (Figure 6).

Figure 6.

Comparison of the λ_max of 1 and its derivate 3. The Michael acceptor double bond was reduced, shortening the π system and shifting λ_max. For clarity, only the relevant section of 250–600 nm is shown.

3 is a derivative of 1 where the Michael acceptor double bond of the cyanoacrylamide moiety was reduced.⁷ This shortens the electron delocalization system and causes a shift of λ_max to lower wavelengths. 3 was thus employed as a negative control, as we expected neither intense fluorescence in the GFP channel (470 ± 40 nm), nor significant intracellular accumulation (lack of sufficiently basic amines) of this compound, nor covalent binding to the target.

Absorption spectra were also recorded at pH = 5.0 to determine whether lysosomal pH affects absorption properties. There was no effect of pH which simplified microscopy but ruled out the use of the compounds for spectral determination of local pH (e.g., 1, Figure 7). There were no major differences in λ_max values or absorption intensity (see the Supporting Information, Figure S1).

Figure 7.

Comparison of 1 absorption at cytosolic (green) and lysosomal (orange) pH. c = 50 μM.

Emission spectra of each compound were then recorded using a scanning fluorescence plate reader, using the compounds’ respective absorption λ_max values as excitation wavelengths. While the intensities varied between compounds, the highest emission levels were usually found between 500 and 600 nm, a range suitable for detection by the fluorescence microscope’s GFP filter (see the Supporting Information, Figure S2). This very high Stokes shift is an interesting feature of the compounds.

The absorption λ_max values of the JAK3 inhibitor compounds were in the excitation range of the blue 4′,6-diamidino-2-phenylindole (DAPI) channel (360 ± 40 nm), and the substances were detectable using 360 nm excitation in U937 cells (Figure 9).

Figure 9.

Visualization of 12 accumulating in U937 cells, seen through either the DAPI filter (left, ex. 360 ± 40 nm, det. 460 ± 50 nm) or the GFP filter (right, ex. 470 ± 40 nm, det. 525 ± 50 nm). c = 10 μM.

While the absorption maxima of the compounds were not ideal for the fluorescence microscope’s GFP channel (excitation range of 470 ± 40 nm), excitation in that range caused observable fluorescence in preliminary, cell-based tests using the macrolide-type compounds (data not shown). A selection of compounds was then screened for their emission spectra after excitation by wavelengths relevant to our microscopy experiments: excitation was performed at wavelengths corresponding to the mean excitation values of each of the used fluorescence filter cubes, as indicated by the manufacturer (DAPI filter: 360 nm; GFP filter: 470 nm; Texas Red filter: 560 nm). We observed potent emission in a suitable range by our test compounds when stimulated at 470 nm, confirming the feasibility of this fluorescence filter for our experiments (Figure S3).

Our reference compound, 13, has a λ_max of 438 nm at pH 7.4 (437 nm at pH = 5.0) and a sufficiently broad excitation range to be fluorescent in the GFP channel range (Figure 8). The very wide difference in excitation and emission wavelengths is potentially useful for other labeling purposes.

Figure 8.

Structure of the positive control compound 13 (left). Absorption spectra of 13 and its unlinked fluorophore 6 at pH 7.4 (right). For clarity, only the relevant section of 250–600 nm is shown.

Substance 13 consists of the fluorophore coumarin 343 (6) linked to an azithromycin scaffold over a short hydrocarbon chain as an amide. Its absorption spectrum was found to be nearly identical to that of the unconjugated precursor (Figure 8). We included 6 in our cell-based experiments as a negative control for 13, as it is ionic at physiological pH and thus highly polar, making passage across cellular membranes unlikely and decreasing the chance of visible intracellular accumulation.^26,27

2.2. Channel Tests

To assess the fluorescence of our compounds in a cellular environment, we first incubated U937 cells with the compounds without prior staining of nuclei or lysosomes. This way, a suitable filter set for the following experiments can be identified.

The fluorescent core scaffold of the JAK3 inhibitors led to visible signals in both the blue “DAPI” channel and the green “GFP” channel (Figure 9). The broad absorption range enables the compound to be detected in either channel, given sufficient exposure. As we stained the nuclei with Höchst dye, fluorescent in the DAPI channel, in the following experiments, we selected the GFP channel for detection of our compounds.

2.3 Effects of Macrolide Conjugation

Next, we investigated the influence of macrolide carriers on the intracellular distribution. To this end, we stained U937 cells with Höchst 33342 dye and the cell navigator lysosome staining kit (CNLSK) to visualize nuclei and lysosomes, respectively. Cells were then incubated with 10 μM of either 6 or its macrolide analogue 13, which has a nearly identical absorption/emission profile (see Figure 8 and Supporting Information, Figures S1 and S2). 6 did not accumulate in a sufficient concentration for visible green fluorescence, while 13 did (Figure 10).

Figure 10.

U937 cells incubated with 10 μM 6 (left) or its macrolide-linked analogue 13 (right). Exposure times for each channel were set identically for both compounds. Nuclei are stained blue and lysosomes red, while the compounds emit green fluorescence. Only 13 accumulates in sufficient concentrations to be visible in the GFP channel. Yellow color indicates the presence of 13 and CNLSK in the same compartments. The image suggests that there are 3 granule populations: predominantly green, predominantly yellow (both substances), and some largely red which are dominated by CNLSK. This difference suggests that there are macrolide binding sites that are not strictly pH-driven.

The macrolide compound 13 was observed throughout the cytosol, but especially high concentrations were localized in the lysosomal compartment, evidenced by the overlap of 13 and CNLSK staining (Figure 11).

Figure 11.

U937 cell incubated with 13 (c = 10 μM) and CNLSK, recorded in either the GFP channel (left), the TexasRed channel (middle), or an overlay of DAPI, GFP, and TexasRed channels (right).

The results clearly demonstrate how the conjugation of a drug warhead—or, in this case, an established fluorophore—to a basic carrier scaffold can affect its distribution and improve cellular and specifically lysosomal uptake.

2.4. Comparison of 1 with 2, 3, and Macrolide Analogues of 1

We then incubated cells of various types with our JAK3 inhibitor compounds. In the following paragraphs, representative examples are depicted (Figure 12); detailed information on all compounds and various cell types (U937 cells, PBMCs, HMC3, HBC, human whole blood) can be found in the Supporting Information repository. Negative control images for each cell type and filter channel are displayed in the Supporting Information (Figures S6–S10). A general trend, as predicted, was a lack of visible accumulation for the unconjugated compounds, e.g., model compound 1, while all macrolide-type compounds provided visible signals through the GFP filter, mainly localized in the putative lysosomal compartment. 2, while not linked to a carrier scaffold, possesses an additional basic function in the form of an N-methylpiperidine residue and is also concentrated in cells (Figure 12).

Figure 12.

Comparison of prototypic compound 1 (top left) to its derivatives 3 (top right), 2 (bottom left) and 7 (bottom right). At 10 μM, only the compounds with additional basic groups accumulate strongly enough in U937 cells to be detected by using a GFP filter set.

As expected, the dibasic macrolide scaffolds were effective in increasing concentration in immune cells, which was observed both for conjugate dye 13 and for macrolide-type JAK3 inhibitors. We attribute this mainly to the aforementioned trapping phenomenon (Figure 3). Green fluorescence was, generally, also detected for this compound class in immune cells other than U937 (e.g., HMC3 cells, PBMCs, HBC, and human whole blood; see the Supporting Information repository) to similar extents. Common components of these cells are the aforementioned acidic compartments, which could be the cause of the supposed accumulation and trapping mechanisms.

A similar phenomenon may be possible for granules based on heparin or chondroitin sulfate where the acidic groups could also preferentially bind dibasic macrolides. Among the nonmacrolide compounds, only 2 was visible in cells. Compared to the other unconjugated analogues, 2 has another basic amine in its N-methylpiperidine residue. The heterocycle is isolated from the delocalized π-electron system; therefore, its basicity is only marginally affected by the rest of the molecule. N-Methylpiperidine has a pK_a of 10.08 (conjugate acid).²⁸ If a similar value is assumed for the tertiary amine of 2 (and, by extension, its macrolide analogue 10), a major portion of the compound molecules will be protonated at lysosomal pH (about 6 orders of magnitude vs unprotonated).

The JAK3 inhibitors’ tricyclic core scaffold itself is basic as well; however, pK_a/pK_b values have not been experimentally determined. Given the lack of observable intracellular accumulation for the nonmacrolide compounds in the majority of experiments, we assume the extent of basicity is not sufficient to lead to local concentrations which would be visibly fluorescent under the conditions we used. The pK_a of 7-azaindole is reported as 4.59 (conjugate acid),^29,30 already making it a weaker base than azithromycin and its derivates, the electron withdrawing mesomeric effects of the amide and nitrile moieties likely further decrease the basicity of the ring nitrogen.

2.5. Quantification of Cell Uptake

As not all compounds were expected to concentrate in cells to be visible by fluorescence microscopy—and also to quantify the uptake—we tested the distribution of our compounds into cells in vitro by incubation with human peripheral leukocytes for 2 h. Uptake was evaluated by calculating the ratio between the concentration of the compound measured in a cell pellet and the concentration of the supernatant (Table 1). We used freshly taken human peripheral leukocytes (“buffy coat”) for the uptake experiments. While this limits the amount of replicates (due to donor-specific variations of blood composition and time of sampling restricting a comparison between experiments), it is more clinically relevant than using a singular, defined cell line.

Table 1. Uptake of Test Compounds into Human Peripheral Leukocytes^a.

compound	ratio
1	4.3 ± 1.3
2	7.1 ± 1.1
3	5.2 ± 2.4
4	4.0 ± 1.3
5	5.0 ± 2.3
6	3.6 ± 0.2
7	114 ± 54
8	44 ± 3.5
9	23 ± 1.3
10	13 ± 2.2
11	42 ± 11
12	11 ± 2.1
13	17 ± 0.9

^aHuman peripheral leukocytes were incubated with the test compounds for 2 h at 37 °C in a shaking incubator. Then, samples were centrifuged, and the respective compound concentrations in the supernatant and cell pellets were quantified by HPLC-MS (n = 2).

As expected, the macrolide conjugate compounds (7–13) were found to be in substantially higher concentrations within the cell pellets compared to the cell-free supernatant. The increased uptake vs the unconjugated analogues was especially notable when comparing 1 to its conjugate derivates 7, 8, and 9 (pellet to supernatant ratio increased from 4.3 ± 1.3 to 114 ± 54, 44 ± 3.5, and 23 ± 1.3, respectively) and the control compound 13 to the free fluorophore 6 (pellet to supernatant ratio increased from 3.6 ± 0.2 to 16.9 ± 0.9). For 7, an unexpectedly high deviation was observed, including earlier time points (Figure S5) and previous experiments. It appears to be a property of a particular substance. Nevertheless, the concentrations were considerably higher than those of the other compounds and were consistent with fast and high accumulation in the cell pellets.

Among the unconjugated compounds, the strongest intracellular accumulation was observed for the N-methylpiperidine-substituted 2, most likely due to the additional basic residue. Data for different incubation times (30 s, 30 min, and 120 min, respectively) are depicted in the Supporting Information (Figure S5). The concentration ratios obtained from the earlier time points suggest that uptake into leukocytes takes place quickly and the concentrations remain relatively constant over time.

Quantification of uptake at 10 μM (data not shown) provided lower concentration ratios, suggesting that the uptake may be saturated at higher compound concentrations. Large quantities of the basic compounds may also influence the lysosomal pH (or membrane function) strongly enough to affect the overall uptake.

These data are in line with the findings from the in vitro microscopy experiments, where fluorescence was observed for compounds taken up in a concentrative manner. It is important to note that while this method shows the difference between intra- and extracellular concentrations of compounds, it does not differentiate between lysosomal and cytosolic localization, as cell pellets were analyzed as a whole. Furthermore, only “intact” molecules were able to be quantified. In other words, the data do not take into account a loss of compound due to cellular metabolism. In vitro metabolism for the compound class in murine liver microsomes has been reported. Typical metabolic transformations were observed at structural elements that are common to the compound class as a whole, e.g., the core scaffold or amide moiety.^7,31

2.6. In Vivo Distribution Studies

To translate our findings to in vivo conditions, C57BL/6 mice were treated p.o. with 24 μmol/kg 13. Dose, vehicle, strain, and termination time points were established in a previous pharmacokinetics study (data not shown). Compound concentrations of 13, 2, 7, and 9 in murine tissues were deemed sufficiently high for the following histology experiments, so the aforementioned study parameters were used again.

Tail plasma was collected to observe plasma levels of 13 over time, and organs were taken 4 h after treatment to obtain cryosections for analysis by fluorescence microscopy. Organs were also analyzed by HPLC–MSMS (Figure 13).

Figure 13.

Plasma (A) and organ (4 h post-treatment) (B) concentrations of 13 in C57BL/6 mice after administration of 24 μmol/kg p.o. (n = 3). Values are displayed as mean ± 1 standard deviation. For (B) concentrations are depicted on a logarithmic scale.

As expected, 13 displayed pharmacokinetic properties typical of macrolides. High concentrations were detected in most organs, especially in liver (20 ± 2 μM) and ileum (34 ± 4 μM) samples. The compound was also found in plasma samples taken over the course of the study, with an average of 297 ± 53 nM remaining even 4 h after treatment. These data are consistent with our previous PK studies on macrolide conjugates, where similar distribution patterns were observed for the conjugated JAK3 inhibitors.⁷ Compound concentrations were about 30-fold lower in the colon compared with the ileum. While this could indicate enrichment in ileum epithelium after uptake from the luminal space, it could also be a result of redistribution into ileal tissues from the blood or simply reflect the effects of the timing of intestinal passage.

Compound-derived fluorescence was observed in organs obtained from this study, especially for lung, ileum, kidney, and liver tissues (Figure 14 and Supporting Information repository). As with the in vitro experiments, the accumulation of 13 was clearly visible in the GFP channel. Distinct distribution patterns were apparent for each organ. For example, in lung tissue, the compound displayed a preference for epithelial cells (Figure 15, left). In the ileum, fluorescence was most intense at the surface of the villi (Figure 15, right). In addition, cell pellets obtained by centrifugation of tail blood were analyzed by fluorescence microscopy (see the Supporting Information repository). As with the in vitro experiments, green fluorescence in the GFP filter channel was observed in cells whose morphology was typical of immune cells.

Figure 14.

Fluorescence micrographs of cryosections of organ sections from mice treated with 24 μmol/kg of 13, taken 4 h after administration. From top left, clockwise: lung, ileum, liver, and kidney. The compound emits green fluorescence, while nuclei have been stained with DAPI. Images were digitally merged from a series of 20× images which allows detailed enlargement.

Figure 15.

Fluorescence microscopy of a lung (left) and an ileum (right) sample from a mouse treated with 24 μmol/kg 13 (green), 4 h after application. Nuclei were stained with DAPI (blue). Images were obtained through a 20x lens. In lung tissue, the compound appears to distribute preferentially into the epithelial cell layers. In the ileum, strong accumulation on the surface of the villi is apparent.

Given the detection of 13 in vivo, we conducted a similar study with three JAK3 inhibitors: The macrolide conjugates 7 and 9 as well as the N-methylpiperidine substituted 2. Organs were taken 2 h after treatment.

Plasma and organ concentrations after oral administration were consistent with data obtained in previous studies (Figures 16 and 17).⁷ Of the three compounds, 9 reached the highest plasma levels, which were consistently around 1 μM over the course of the experiment. 7, another macrolide conjugate, was found in concentrations about 1 order of magnitude lower; as we discussed before, this is likely a result of fast distribution into peripheral organs.⁷ The unstable methyl ester 2 was barely measurable in plasma samples regardless of sampling time, reflecting its poor stability versus that of unspecific esterases. However, the resulting free acid metabolite is likely still pharmacologically active, as the affected residue is not critical for interactions with JAK3.^7,32 As before, cell pellets obtained by the centrifugation of tail blood were analyzed. We observed GFP filter range fluorescence in non-erythrocyte cells of all three treatment groups, but not of the vehicle group, indicating in vivo accumulation of the compounds in immune cells (see Supporting Information repository).

Figure 16.

Plasma concentrations of JAK3 inhibitors 2, 7, and 9 in C57BL/6 mice after administration of 24 μmol/kg p.o. (n = 3). Values are displayed as mean ± 1 standard deviation.

Figure 17.

Concentrations of JAK3 inhibitors 2, 7, and 9 in organs of C57BL/6 mice 2 h after administration of 24 μmol/kg p.o. (n = 3). Values are displayed as mean ± 1 standard deviation.

As expected, the JAK3 inhibitors accumulated in most peripheral organs, especially 9, which reached one-to-two-digit micromolar concentrations in heart muscle, lung, liver, ileum, and particularly in the kidney and spleen. The N-methylpiperidine substituted 2 reached bulk brain levels of 1.2 μM. Ileum levels were lower than in earlier studies, where, for example, 7 accumulated to around 19 μM, even 8 h post-administration.⁷ Other data are consistent with previous data as well as our observations from fluorescence microscopy (see the next paragraph and Supporting Information).

Compound 2 fluorescence was detectable in heart muscle, brain, ileum, liver, lung, and spleen (Figures 18–20). The observed patterns of distribution thus are in line with the data obtained from HPLC–MSMS analysis. Within the ileum, the compound was especially present in villi (Figures 18–20). Unlike the macrolides 7 and 9, distribution of the unconjugated 2 in the lung was diffuse and less focused on epithelia.

Figure 18.

Fluorescence micrographs of cryosections of brain, heart, and ileum from a mouse treated with 24 μmol/kg p.o. of 2. The left pictures show the overlay of blue DAPI and green compound fluorescence, the right images are green fluorescence alone. Images were digitally merged from a series of 20× images which allows detailed enlargement.

Figure 20.

Fluorescence image of an ileum section from a mouse treated with 24 μmol/kg p.o. of 2. The compound emits green fluorescence, while nuclei have been stained with DAPI. Images were merged from singular frames obtained with a 4× lens.

Figure 19.

Fluorescence micrographs of cryosections from kidney, lung, and spleen from a mouse treated with 24 μmol/kg p.o. of 2. The left pictures show the overlay of blue DAPI and green compound fluorescence, the right images are green fluorescence alone. Images were digitally merged from a series of 20× images which allows detailed enlargement.

The macrolide conjugate 7 demonstrated intense fluorescence in heart muscle, liver, lung, and kidney (Figures 21 and 22). Intensities were lower in the ileum and spleen yet still readily detectable. In lung samples, the fluorescence was most intense around blood vessels or bronchioles, indicating a preferential distribution into these tissues.

Figure 21.

Fluorescence microscopy of a heart, ileum, and kidney from a mouse treated with 24 μmol/kg p.o. of 7. The left pictures show the overlay of blue DAPI and green compound fluorescence, the right images are green fluorescence alone. Images were digitally merged from a series of 20× images which allows detailed enlargement.

Figure 22.

Fluorescence micrographs of a liver, lung, and spleen from a mouse treated with 24 μmol/kg p.o. of 7. The left pictures show the overlay of blue DAPI and green compound fluorescence, the right images are green fluorescence alone. Images were digitally merged from a series of 20× images which allows detailed enlargement.

The other macrolide we tested, 9, displayed a similar distribution as 7, accumulating in most organs apart from the brain (Figures 23 and 24). The amount of 9 in the ileum was lower than expected from the PK data, but fluorescence was more intense in the kidneys. Interestingly, only the outer areas of the liver, distal to the portal vein, were stained green.

Figure 23.

Fluorescence microscopy of a heart, ileum, and kidney from a mouse treated with 24 μmol/kg p.o. of 9. The left pictures show the overlay of blue DAPI and green compound fluorescence, the right images are green fluorescence alone. Images were digitally merged from a series of 20× images which allows detailed enlargement.

Figure 24.

Fluorescence micrographs of cryosections of liver, lung, and spleen from a mouse treated with 24 μmol/kg p.o. of 9. The left pictures show the overlay of blue DAPI and green compound fluorescence, the right images are green fluorescence alone. Images were digitally merged from a series of 20× images which allows detailed enlargement.

3. Discussion

In this report, we set out to address various questions of practical pharmacology using two example classes of substance: JAK3 inhibitors and macrolides. We used fluorescence to visualize the tissue and subcellular location of the substances and LC-MSMS to quantify them in tissue and plasma. This approach has many advantages including the fact that it is technically simple and that the signal is associated with an intact compound. With radioactivity, or PET, it is often unclear what signal is a metabolite and what is the parent substance. Radiolabels are often incorporated in the last step in potentially labile positions, meaning that some radioactivity is not from the compound being investigated.

The JAK3 inhibitors introduced here have interesting properties for such studies. The first is that they are inherently fluorescent even as optimized ligands. This means that therapeutic candidates from this class can be their own labels. The second is that the substances have a very broad Stokes shift, which gives them practical advantages as fluorophores, albeit at the expense of relatively low quantum yield. In this study, we took advantage of these inherent properties to examine why certain compounds and conjugates appear to be very effective JAK3 inhibitors in vivo. These conjugates were azalide macrolides. The very same compounds then provided the chance to observe macrolide distribution in vivo.

The questions that we addressed using this system included.

• 1Where do azalide macrolides distribute in tissue?
• 2Is this distribution influenced by specific properties of the macrolide or pendant structures?
• 3Do effective compounds share properties of distribution?
• 4Do JAK3 ligands distribute in a characteristic way?
• 5Do small molecules distribute homogeneously?
• 6Are there differences in organs, layers, or tissue structures in terms of drug disposition? What properties or structure aspects influence these effects?

3.1. Where Do Azalides Distribute In Vivo?

We examined the azalide distribution using two different pendant fluorophores, a pharmacologically inert coumarin (13), and an active JAK3 covalent ligand (7, 9) bound by various linking groups. Both classes had similar properties of uptake and distribution in terms of bulk kinetics. In lung, 13 was more obviously associated with epithelial cell layers whereas 7 and 9 had higher relative fluorescence in parenchyma vs epithelium. Given that pure JAK3 ligand 2 was mainly in parenchyma, these data suggest that the JAK3 ligand may have influenced binding beyond the epithelium.

In the kidney, the substances were all found in the cortex and not in the medulla. 13 appeared more concentrated in the outer cortex than other substances. In the ileum, the substances varied in the amount in basement membrane vs villi. 13 appeared most concentrated in villi, while the other macrolides were similar between villi and basement membrane.

The common element of high epithelial partition for macrolides is of practical interest for their mode of action. Macrolides are often considered to be unusual in that they are active at plasma levels below the MIC for target bacteria. These data suggest that the very high partition to epithelium could, in part, explain retention of therapeutic activity at doses that give sub-MIC plasma levels.³³

3.2. Is This Distribution Influenced by Specific Properties of the Macrolide or Pendant Structures?

We examined 3 macrolide structures: 7, 9 and 13. Each differed in absolute distribution, but the trends in relative tissue disposition were broadly similar. In terms of in vitro uptake, 7 had the highest degree of concentration (Table 1) in primary cells, however, it generally had the lowest plasma and tissue levels in vivo. 13 was intermediate between 7 and 9 in terms of organ and plasma levels. 9 had the highest plasma and tissue levels of these 3 examples. 9 differs from 7 and 13 in that it has a longer linker with an additional amide; however, it is not clear why that should improve in vivo partition or stability. These data show that despite the dominant properties of the macrolide carrier, there is variability in the stability and partition due to small structural differences. This means that when using macrolides as carriers, variation in linkage will yield different degrees of uptake, while distribution patterns remain broadly similar.

3.3. Do Effective Compounds Distribute Similarly?

Of the unconjugated JAK3 inhibitors, substance 2 provided clear fluorescence in cells in uptake studies. We compared its distribution to that of 7, 9, and 13. 2 differed most profoundly in brain levels, where it was strongly associated with the meninges and parts of the cortex. However, 2 was similar to conjugates 7 and 9 in terms of its partition to lung parenchyma. 2 also partitioned to intracellular granules in a manner similar to that of 7, 9, and 13. These data suggest that amines in the structure aid retention in cells and are associated with improved in vivo pharmacological effects for this target. Compounds with neutral side chains were not sufficiently present in cells to be observed directly.

3.4. Do JAK3 Ligands Distribute in a Characteristic Way?

The JAK3 ligands used here are highly selective and covalent binders.^7,8,31 In this regard, they may have served to label JAK3.

3.5. Are Small Molecules Always Homogeneously Distributed?

In this study, we treated animals p.o. and were interested in distribution following uptake from the gut. In the kidney, ileum, and spleen, substance disposition appeared uniform, albeit with specific tissue layers retaining more labels (e.g., kidney cortex). In contrast, sections of the liver in particular but also the heart and lung were much less homogeneous. The latter organs are further from the point of absorption in the circulatory system, and their homogeneity may be explained by the fact that peripheral blood levels may be more uniform in that part of the circulation. In the organs nearer to the point of resorption (presumably upper intestine), the liver, heart, and lungs in that order, the lack of apparent homogeneity needs some explanation. We investigated the potential for artifacts during sectioning, staining, and drying. However, such artifacts should have influenced all materials and controls suggested retention of substance at the site of origin at the point of sampling.

In particular, liver sections showed zones of substance, either on the edges of tissue or around particular vessels. While we cannot explain these observations mechanistically, they could be explained by temporal variation in the perfusion of parts of the liver. Thus, one explanation may be that portal blood may be variably distributed, with various parts of the liver being more or less perfused over time. Alternatively, uptake via lymph and chylomicrons may influence which parts of the liver are most exposed.

Similar patterns were observed in the lung, with certain lobes more labeled than others. The idea of variable perfusion of lung lobes is highly counterintuitive, and we will continue to investigate this.

3.6. Are There Differences in Organs, Layers, or Tissue Structures in Terms of Drug Disposition? What Properties or Structural Aspects Influence These Effects?

For macrolides, there appear to be very clear effects of organ subanatomy on the final distribution. We noted that macrolides tended to exhibit higher fluorescence in the epithelial layers and kidney cortex. Variation between the intestine basement membrane and villi suggests that the degree of epithelial to parenchymal distribution can be modified through optimization of physical properties. In a general sense, the properties optimal for organ penetration and retention appear to include the presence of amines. The effects appear to merit more detailed study in multiple pharmacophores, because they may be useful means to promote the activity of small molecules.

More generally, these data demonstrate how the introduction of basic moieties to a drug can greatly increase uptake into lysosome-bearing cells, e.g., leukocytes, as well as certain organ tissues, e.g., epithelial layers. This was achieved either by using compounds with simple amino substituents (2) or by conjugation to a macrocyclic carrier scaffold. Both approaches led to strong intracellular accumulation.

We then investigated methods to increase cell permeability and uptake of an established fluorophore, 6—without impacting its fluorescence range or intensity. By conjugation to an azithromycin-derived carrier, accumulation in immune cells was increased. The resulting compound, 13, displayed a similar fluorescence spectrum and can now, among other uses, be employed to stain acidic intracellular compartments. The compound was also used in an in vivo study to determine its distribution within organs, where it was most present in the epithelium and endothelium.

In general, uptake into immune cells other than U937 (e.g., HBC, PBMCs, HMC3 cells, and immune cells in human whole blood) was similar, which we attribute to the presence of acidic compartments in all of these cell types. Among these observations, several compounds (e.g., 2 and 7–12) were found to visibly accumulate in HMC3 cells (see the Supporting Information repository). Of particular interest was substance 2, which unlike other analogues from the series has significant uptake to the brain in murine models which was previously shown by LC–MS.⁷ Examination of the in vivo fluorescence data shows that the substance has general tissue partition and is particularly present in the meninges and cortex in the mouse. This is of interest because we are investigating the role of immune modulators in CNS disease and the question of whether modulation of JAK3 is relevant to CNS resident immune cells, or whether it is purely a peripheral target. We are now investigating the distribution to brain organs to evaluate these phenomena.

Taken together, our findings elucidate how conjugation of a pharmacophore to a basic carrier scaffold increases the accumulation into immune cells and organ tissues. These data provide a useful impression of how small molecules behave in vivo and how their distribution varies in organs. They also show that specific cell layers may be more exposed than others. In particular, these data serve to report on the distribution of macrolides in vivo and their tendency to reach higher levels in epithelial cell layers, e.g., those of the lung and gut mucosa. This may explain in part their benefits in their anti-infective uses. More generally, the “fluorescent pharmacology” described here provides a data set to help better understand the short-term partition properties of drugs in vivo.

4. Materials and Methods

4.1. Compounds and Compound Synthesis

The fluorophore conjugate 13 and the JAK3 inhibitors 1–5 and 7–12 were synthesized according to previously described procedures (see also Supporting Information).^7,32

Coumarin 343 (6) was obtained from Fisher Scientific.

4.2. Cellular Assays

Human full blood and human peripheral blood leukocytes extracted from human full blood were obtained from the ZKT Tübingen GmbH (center for transfusion medicine) under licenses ZKT-FoPro202106-2305-01 and ZKT-FoPro202012-2211T (Otfried-Müller-Straße 4/1, 72076 Tübingen).

PBMCs were extracted from human peripheral blood leukocytes by addition of 1/3 volume of Ficoll–Paque PLUS (density 1.077 g/mL) using a syringe. After centrifugation at 1000 rpm for 15 min, the resulting cell layer between plasma and the Ficoll phase was aspirated carefully. The aspirated cell layer was then diluted 1:10 in RPMI medium and centrifuged again at 400 rpm, and the supernatant was discarded. The cell pellets were diluted and centrifuged again multiple times until the supernatant was clear; then pellets were resuspended in medium to a concentration of 50–70 million cells/mL.

U937 and HMC3 cells were obtained from ATTC.

4.3. Determination of λ_max

Absorption spectra of compounds were determined by using an Implen Nanophotometer NP80. Compounds were dissolved either in phosphate-buffered saline (pH 7.4) containing 0.5% DMSO, or citrate buffer (pH 5.0) and 0.5% DMSO at 50 μM concentrations in a 10 mm Hellma Analytics QS High Precision Cell cuvette. Spectra (200–900 nm) were measured at ambient temperature (see the Supporting Information for spectra of each compound).

4.4. Emission Spectra

Emission spectra were recorded on a SpectraMax Gemini XS Microplate Reader with SoftMax Pro software, using a Sarstedt ELISA plate white High Bind. F plates. Compounds (c = 50 μM) were dissolved in phosphate-buffered saline (pH 7.4) containing 0.5% DMSO. Excitation wavelengths for each compound were identical to the respective λ_max values obtained from absorption spectroscopy (see above) or to the mean excitation wavelengths for each of the used Keyence fluorescence microscopy filter cubes as indicated by the manufacturer (DAPI filter: 360 nm, GFP filter: 470 nm, TexasRed filter: 560 nm). Emission spectra were recorded in steps of 5 nm in the range of 350–750 nm (cutoff: 420 nm, except for 3, 6, and 13). The cutoff for data shown in Figure S3 was 40 nm above the respective excitation wavelength. Spectra are represented as the average of three measurements per compound.

4.5. Fluorescence Microscopy

To visualize cell nuclei and lysosomes, samples were incubated for 15 min with Hoechst 33342 solution and the CNLSK, respectively (1:1000 dilution for both). Cells were then incubated with compounds at ambient temperature for 30 or 2 h to a final concentration of 10 μM and 0.1% DMSO. A 20 μL amount of the resulting suspensions was then pipetted and subsequently gently distributed on microscope slides (Roth). Nuclei were stained by Hoechst 33342 to appear in blue (OP-87762 BZX-Filter DAPI, excitation wavelength 360/40 nm, absorption wavelength 460/50 nm), lysosomes stained by CNLSK to appear in red (OP-87765 BZX-Filter Texas Red, excitation wavelength 560/40 nm, absorption wavelength 630/75 nm), and cells with high concentrations of the fluorescent JAK3 inhibitors appear in green (OP-87763 BZX-Filter GFP, excitation wavelength 470/40 nm, absorption wavelength 525/50 nm). Images were acquired with a Keyence BZ-810 fluorescence microscope (Keyence, Tokyo, Japan) using a 100X/a.45 oil-immersion objective lens, using the same exposure and intensity for each sample. Then, haze reduction was applied to the captured images with the corresponding software (ImageJ).

4.6. Uptake with Human Buffy Coat Cells

HBC cells were used at a concentration of 5 × 10⁶ cells/mL and incubated with the inhibitor compounds at a concentration of 1 μM for different time points (30 s, 30 and 120 min).³⁴ Incubation was done in duplicate for each time point and compound.

After incubation at 37 °C, tubes were centrifuged for 5 min at 400 g, and then supernatants were transferred into separate tubes. To both supernatants and pellets, acetonitrile and internal standard (5 nM terbuthylazine and 1 nM sulfentrazone) were added. Measurements of samples were performed by reverse-phase HPLC on an Agilent series system using the 1260 HiP Degasser, the 1260 BinPump, the 1260 HiP ALS, and the 1290 Thermostat, using Agilent C18 2.7 μm columns.

For detection, the system was connected to an AB SCIEX API 4000 MS using a TurboSpray ion source (settings: source voltage 45 V, ion spray voltage 4.5 kV, temperature 300 °C, gas flow 5 L/min).

4.7. Experimental Animals

All animal experiments were carried out in accordance with German law (application number SYN 06/20/35/9185.81-7). Mice were purchased from Janvier Laboratories and maintained in our dedicated specific-pathogen-free animal facility. Mice were kept for at least 7 days after arrival for acclimatization. Animals had ad libitum access to mouse chow and tap water. Body weights were approximately 20 g per mouse with each mouse being weighed right before treatment to ensure equal doses for all animals.

4.8. 13 Pharmacokinetics and Distribution Study

Seven-to-eight week-old C57BL/6 male mice were treated orally with either vehicle (0.5% citric acid, 5 mL/kg) or 24 μmol/kg of 13 (n = 3). Blood was taken from the tail vein at predetermined time points. Four hours after treatment, animals were sacrificed by CO₂ asphyxiation, and organs were taken for subsequent analysis by LC–MS or for histological preparation, respectively.

4.9. 2, 7, and 9 Pharmacokinetics and Distribution Study

Seven-to-eight week-old C57BL/6 male mice were treated orally with either vehicle (5% DMSO in 0.5% citric acid, 5 mL/kg) or 24 μmol/kg of test compounds (n = 3). Blood was taken from the tail vein at predetermined time points. Two hours after treatment, animals were sacrificed by CO₂ asphyxiation and organs were taken for subsequent analysis by LC–MS or for histological preparation, respectively.

4.10. Collection of Analytic Samples

Blood was collected in heparinized tubes and centrifuged for 8 min at 8000 rpm and 4 °C. The supernatant was used to determine plasma concentrations. Both plasma and organ samples for analytics were immediately stored at −25 °C until workup. The cell pellets obtained by the centrifugation step were stored at 4 °C until analysis by microscopy.

4.11. Sample Workup and Analysis by HPLC-MS

Plasma samples were diluted with ACN, homogenized in a FastPrep FP-120 instrument (6 m/s for 40 s), and then centrifuged at 14,000 rpm for 7 min at 4 °C. Organ samples were treated with 1 μL of proteinase K solution (0.5 mg/mL in 20 mM phosphate buffer) per mg organ weight and then worked up analogously to the plasma samples. For the homogenization step, glass beads and sand were added to the tubes to ensure thorough homogenization. Bile samples were diluted with water, homogenized, and then further diluted with ACN plus internal standards, followed by homogenization and centrifugation.³⁵

Compound concentrations were measured using reversed-phase HPLC with MS detection. The procedure used a mobile phase comprised of 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B). Method: 10% B for 1 min, to 100% B in 4 min, 100% B for 2 min, to 10% B in 1 min, 10% B for 2 min, stop time 10 min, flow rate 500 μL/min, injection volume 6 μL. Using a thermostat, a constant column temperature of 45 °C was maintained, while the samples were kept at 6 °C.

4.12. Collection of Histological Samples

Organ samples for histological analysis were then incubated for 4 h in 4% paraformaldehyde solution at 4 °C. Then, the organs were transferred to tubes containing 15% sucrose in PBS and incubated for another 8 h. Lastly, they were transferred to 30% sucrose in PBS and incubated for 24 h.

4.13. Workup of Histological Samples

After the incubation steps, organs were embedded in the Tissue Tek OCT compound at −20 °C. Cutting was performed using a Microtome-386040-HM-500_OM (MICROM International) to obtain slices of 5 μm thickness. Immediately after the collection of slices, they were fixed on SuperFrost microscopy slides using a ROTI-Mound FluorCare DAPI glue. Prepared slides were stored at 4 °C and protected from light until analysis via microscopy was conducted.

4.14. Analysis of Blood and Histological Samples

Analysis of blood pellets and organ slides was performed on a Keyence-BZ-X810 fluorescence microscope (see above). Exposure time was adjusted until no autofluorescence of cells was observed in vehicle samples (for vehicle/negative control organ images, see Supporting Information, Figures S11–S18), and then the same parameters were used to obtain images of treatment group samples.

Glossary

Abbreviations

ACN

acetonitrile

CNLSK

cell navigator lysosomal staining kit

DAPI

4′,6-diamidino-2-phenylindole

DMSO

dimethyl sulfoxide

GFP

green fluorescent protein

granulocyte

monocyte colony-stimulating factor

HBC

human buffy coat (peripheral leukocytes)

HMC3

human microglial cells

HPLC

high-performance liquid chromatography

JAK

Janus kinase

LPS

lipopolysaccharides

MS

mass spectrometry

PBMC

peripheral blood mononucleated cells

PBS

phosphate-buffered saline

TYK2

tyrosine kinase 2

UV/vis

ultraviolet/visible spectrum of light

Data Availability Statement

The microscopy data underlying this study (additional images acquired by fluorescence microscopy, DOI: 10.17632/y7crpxfvcf.1) are openly available in Mendeley Data at: https://data.mendeley.com/datasets/y7crpxfvcf/1.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.3c00119.

• Synthesis and analysis of 11–13, absorption and emission spectra of test compounds, crystallographic data, X-ray crystal structure of intermediate 17, time course of uptake into HBC and negative control images (PDF)

• SMILES spreadsheet uptake and fluorescence (CSV)

Author Contributions

J.L. and M.M. contributed equally to the work. J.L. was responsible for the design, synthesis, and characterization of the featured test compounds, planning, coordination, and interpretation of in vitro and in vivo studies, and writing. M.M. was responsible for microscopy experiments and image processing, preparation of histology samples, and the planning and performance of the HBC uptake experiment. S.S. carried out the design, synthesis, and analytical characterization of 13. D.S. carried out the X-ray crystallography of 17. M.B., F.M., and S.L. were responsible for general guidance and supervision of the project.

The authors declare the following competing financial interest(s): J.L., M.M., S.S., F.M., and M.B. are employees of Synovo GmbH, a pharmaceutical company that has an interest in the development of this class of compounds.