**1. Introduction**

Chronic pain is a major health issue worldwide [1] that enacts considerable suffering [2,3]. Despite the limited effects of available drugs for the treatment of pain, chronic pain patients are often treated with opioids, which have a controversial role in chronic pain management. In fact, patient follow-ups and population studies reveal the low long-term analgesic efficacy of opioids that is accompanied by the development of tolerance, opioid induced hyperalgesia (OIH), adverse side-effects, addiction, and opioid-related

deaths [4,5]. New strategies to avoid the aversive effects of opioids while preserving their analgesic properties are therefore needed.

In this perspective, the serotonin 1 A receptor (5-HT1A) emerges as a promising candidate. 5-HT1A is an inhibitory presynaptic autoreceptor on serotonergic neurons and is also expressed postsynaptically in terminal regions innervated by serotonergic neurons [6]. In animal studies, 5-HT1A agonists have been reported to counteract opioid-induced hyperalgesia, opioid tolerance and to improve the analgesic potency of opioids while reducing their rewarding effects [6–8]. Contrary to opioids, a first order pronociceptive effect followed by an analgesic effect was documented for 5-HT1A agonists, suggesting opposing effects between opioids and 5-HT1A agonists [8]. Therefore, hypothetically, 5-HT1A/opioid interactions could be time-dependent with 5-HT1A antagonists initially enhancing opioid analgesia [9,10] and 5-HT1A agonists, having beneficial long-term effects when OIH has developed [6–8]. In agreement with this, a genetically inferred reduction of serotonergic signalling was associated with an increased analgesic response to the opioid drug fentanyl in healthy human subjects [11]. Furthermore, gene-to-gene interactions between the muopioid receptor (MOP) gene (*OPRM1*) and the serotonin transporter (*5-HTT*) or 5-HT1A (*HTR1A*) genes had antagonistic effects on endogenous descending pain modulation in healthy subjects and in fibromyalgia patients [12].

In the human brain, high densities of 5-HT1A [13] and MOP [14] have been reported in regions implicated in pain modulation, and high 5-HT1A binding potential was associated with more efficient endogenous pain inhibition [15]. Moreover, significant positive associations were found between the serotonin and the opioid systems in networks known to regulate pain and mood, including the cingulate cortex, thalamus, dorsolateral prefrontal cortex, amygdala, and the left parietal cortex [16]. The exact mechanisms responsible for the physiological, pain-related interactions between the opioid and the serotonergic signalling systems are not known [6]. One possible mechanism is the opioid-induced activation of 5-HT1A–expressing glial cells through the Toll-like receptor 4 [17], as activated glia has been implicated in the development of OIH and opioid tolerance [18,19]. In accordance with this reasoning, extensive cortical glia activation was documented in patients suffering from fibromyalgia [20], a chronic pain syndrome with aberrations in cerebral opioid signalling [21]. An additional explanation might be the co-localization of MOP and 5-HT1A receptors on the same neurons. In fact, Kishimoto et al. presented electrophysiological evidence of their co-localization on individual presynaptic GABAergic nerve terminals, and demonstrated that they synergistically inhibited GABA release in the periaqueductal gray (PAG), a structure that mediates opioid-based pain control [22]. In addition, the activation of GABA<sup>A</sup> receptors in PAG projecting neurons was shown to have a net pronociceptive effect [23]. Further support for interactions between MOP and 5-HT1A at the cellular level comes from a study showing that they can form functional heterodimers and that signalling of one receptor in the heterodimer is inhibited by the activation of the other [24]. We thus hypothesize that opioid induced heterodimerization of MOP and 5-HT1A inactivates the receptors, which then become unable to inhibit GABA release and promote pronociceptive pathways.

The primary aim of this study was to challenge this hypothesis by quantitatively characterizing interactions between the MOP and 5-HT1A receptors in live cells expressing near physiological levels of these receptors, and to assess the effects of commonly used non-peptide opioid drugs such as morphine, oxycodone, codeine, and fentanyl, on the extent of these interactions and their downstream effects. In particular, we have focused on intracellular Ca2+ levels and signalling dynamics, and on mitogen-activated protein kinases (MAPKs) p38 and the extracellular signal-regulated kinase (Erk1/2), both previously associated with the adverse effects of opioids [25,26].

#### **2. Results**

The effects of non-peptide opioids on the extent of interactions between the mu-opioid receptor fused with the enhanced Green Fluorescent Protein (MOP-eGFP) and the sero2. Results

tonin 1 A receptor fused with the red fluorescent protein Tomato (5-HT1A-Tomato) were examined in live cells using confocal laser scanning microscopy (CLSM) and fluorescence cross-correlation spectroscopy (FCCS). FCCS, a quantitative analytical method with singlemolecule sensitivity, is succinctly described in Section 4. Materials and Methods. Primary data, temporal autocorrelation curves (tACCs) and cross-correlation curves (tCCC) acquired using FCCS are shown in Figure 1. Determination of the so-called relative cross-correlation amplitude (RCCA) and its use to assess the apparent dissociation constant is described in Section 4. Materials and Methods and in the Supplementary Materials, Section S2. Calculation of the apparent dissociation constant and S3. Relative Cross-Correlation Amplitude (RCCA) increased upon opioid treatment. Verification by switching FCCS. More information can also be found in [27]. examined in live cells using confocal laser scanning microscopy (CLSM) and fluorescence cross-correlation spectroscopy (FCCS). FCCS, a quantitative analytical method with single-molecule sensitivity, is succinctly described in Section 4. Materials and Methods. Primary data, temporal autocorrelation curves (tACCs) and cross-correlation curves (tCCC) acquired using FCCS are shown in Figure 1. Determination of the so-called relative crosscorrelation amplitude (RCCA) and its use to assess the apparent dissociation constant is described in Section 4. Materials and Methods and in the Supplementary Materials, sections S2. Calculation of the apparent dissociation constant and S3. Relative Cross-Correlation Amplitude (RCCA) increased upon opioid treatment. Verification by switching FCCS. More information can also be found in [27].

The effects of non-peptide opioids on the extent of interactions between the mu-opioid receptor fused with the enhanced Green Fluorescent Protein (MOP-eGFP) and the serotonin 1 A receptor fused with the red fluorescent protein Tomato (5-HT1A-Tomato) were

Molecules 2022, 27, x FOR PEER REVIEW 3 of 21

Figure 1. fluorescence cross-correlation spectroscopy (FCCS). (A) Schematic presentation of the instrumental setup for dual colour CLSM imaging and FCCS. Incident laser light, 488 nm (blue) and 543 nm (green), is reflected by the main dichroic beam splitter (MDBS, 488/453/633) and focused by the objective into the sample. Fluorescence and scattered light are collected by the same objective and fluorescence is separated from the elastically scattered light by the MDBS. The fluorescence is spectrally separated by the secondary dichroic beam splitter (SDBS, 545) and further spectrally narrowed by emission filters (EF) before being recorded by avalanche photo diodes (APD) detectors. Magnified insert: Cross section through the observation volume element (OVE) in the radial (xy) plane in the sample. Fluctuations in fluorescence intensity are generated as fluorescently labelled molecules diffuse through the OVE (arrows). (B) CLSM image of a HEK293 cell genetically modified to stably express MOP-eGFP (green) and 5-HT1A-Tomato (red). Scale bar 10 µm. (C) Fluorescence intensity scan through a HEK293 cell expressing MOP-eGFP (green) and 5-HT1A-Tomato (red) in the axial (z-axis direction). The first peak in fluorescence intensity indicates the position of the basal (z = 0) and the second one the apical (z = 5 µm) plasma membrane of the same cell. Fluorescence intensity drops when the apical plasma membrane is crossed, as the OVE is now positioned in the surrounding cell culture medium. (D) Fluorescence intensity fluctuations recorded at the apical mem-**Figure 1.** Fluorescence cross-correlation spectroscopy (FCCS). (**A**) Schematic presentation of the instrumental setup for dual colour CLSM imaging and FCCS. Incident laser light, 488 nm (blue) and 543 nm (green), is reflected by the main dichroic beam splitter (MDBS, 488/453/633) and focused by the objective into the sample. Fluorescence and scattered light are collected by the same objective and fluorescence is separated from the elastically scattered light by the MDBS. The fluorescence is spectrally separated by the secondary dichroic beam splitter (SDBS, 545) and further spectrally narrowed by emission filters (EF) before being recorded by avalanche photo diodes (APD) detectors. *Magnified insert:* Cross section through the observation volume element (OVE) in the radial (*xy*) plane in the sample. Fluctuations in fluorescence intensity are generated as fluorescently labelled molecules diffuse through the OVE (arrows). (**B**) CLSM image of a HEK293 cell genetically modified to stably express MOP-eGFP (green) and 5-HT1A-Tomato (red). Scale bar 10 µm. (**C**) Fluorescence intensity scan through a HEK293 cell expressing MOP-eGFP (green) and 5-HT1A-Tomato (red) in the axial (z-axis direction). The first peak in fluorescence intensity indicates the position of the basal (*z* = 0) and the second one the apical (*z* = 5 µm) plasma membrane of the same cell. Fluorescence intensity drops when the apical plasma membrane is crossed, as the OVE is now positioned in the surrounding cell culture medium. (**D**) Fluorescence intensity fluctuations recorded at the apical membrane of a HEK293

brane of a HEK293 cell, originating from MOP-eGFP (green) and 5-HT1A-Tomato (red) lateral diffusion in the plasma membrane. (E) Representative auto- (green and red) and cross-correlation (black)

cell, originating from MOP-eGFP (green) and 5-HT1A-Tomato (red) lateral diffusion in the plasma membrane. (**E**) Representative auto- (green and red) and cross-correlation (black) curves recorded at the apical membrane of a HEK293 cell. (**F**) Cross-correlation curves recorded in live HEK293 cells expressing the positive (brown) and negative (champagne) control constructs. For detailed information see Section 4. Materials and Methods and Supplementary Materials, Section S1. Transfection, positive and negative control cells (Figures S1 and S2). Molecules 2022, 27, x FOR PEER REVIEW 4 of 21 live HEK293 cells expressing the positive (brown) and negative (champagne) control constructs. For detailed information see Section 4. Materials and Methods and Supplementary Materials, section S1. Transfection, positive and negative control cells (Figures S1 and S2).

#### *2.1. Non-Peptide Opioids Potentiate MOP and 5-HT1A Heterodimerization to a Different Extent* 2.1. Non-Peptide Opioids Potentiate MOP and 5-HT1A Heterodimerization to a Different Extent

CLSM imaging showed clear co-localization of both receptors, MOP-eGFP (green) and 5-HT1A-Tomato (red), in the plasma membrane in both the HEK293 (Figure 1B) and the PC12 (Figure 2A) cells. It also showed that treatment with non-peptide opioids did not cause the internalization of individual receptors or of heterodimer receptor complexes (Figure 2B). This contrasts with the effects of treatment with the opioid peptide DAMGO, which promoted MOP internalization, but not the internalization of the heterodimer MOPeGFP–5-HT1A-Tomato complex (Figure 2C). CLSM imaging showed clear co-localization of both receptors, MOP-eGFP (green) and 5-HT1A-Tomato (red), in the plasma membrane in both the HEK293 (Figure 1B) and the PC12 (Figure 2A) cells. It also showed that treatment with non-peptide opioids did not cause the internalization of individual receptors or of heterodimer receptor complexes (Figure 2B). This contrasts with the effects of treatment with the opioid peptide DAMGO, which promoted MOP internalization, but not the internalization of the heterodimer MOP-eGFP–5-HT1A-Tomato complex (Figure 2C).

Figure 2. Non-peptide opioids neither induce MOP-eGFP nor MOP-eGFP-5-HT1A-Tomato heterodimers internalization, whereas the opioid peptide DAMGO induces strong MOP-eGFP internalization. CLSM images of live PC12 cells stably expressing MOP-eGFP (green) and 5-HT1A-Tomato (red). (A) Cultured under standard conditions, without opioid treatment (control). (B) Treated for 18 h with 750 nM morphine. (C) Treated for 18 h with 500 nM DAMGO. Scale bar 10 µm. **Figure 2.** Non-peptide opioids neither induce MOP-eGFP nor MOP-eGFP-5-HT1A-Tomato heterodimers internalization, whereas the opioid peptide DAMGO induces strong MOP-eGFP internalization. CLSM images of live PC12 cells stably expressing MOP-eGFP (green) and 5-HT1A-Tomato (red). (**A**) Cultured under standard conditions, without opioid treatment (control). (**B**) Treated for 18 h with 750 nM morphine. (**C**) Treated for 18 h with 500 nM DAMGO. Scale bar 10 µm.

For FCCS analysis, data collected on cells expressing similar (within the experimental error of FCS measurements) receptor levels, NMOP = (27 ± 6) and N5-HT1A = (25 ± 3), were compared. At these expression levels, corresponding to concentrations: cMOP = (320 ± 70) nM and c5-HT1A = (300 ± 40) nM, FCCS analysis showed that MOP-eGFP and 5-HT1A-Tomato receptors not only co-localized in the plasma membrane, but also formed heterodimers, as evidenced by tCCCs (Figure 1E, black). FCCS showed that in untreated cells about 33% (RCCA = 0.33) of the 5-HT1A-Tomato receptors are bound in heterodimer complexes with MOP-eGFP (Figure 3A). Based on this, the apparent dissociation constant for a heterodimer receptor complex of MOP-eGFP–5-HT1A-Tomato with a 1:1 stoichiometry was estimated to be Kୢ ୟ୮୮ = (440 ± 70) nM. Moreover, FCCS showed that treatment with different concentrations of fentanyl in-For FCCS analysis, data collected on cells expressing similar (within the experimental error of FCS measurements) receptor levels, NMOP = (27 ± 6) and N5-HT1A = (25 ± 3), were compared. At these expression levels, corresponding to concentrations: cMOP = (320 ± 70) nM and c5-HT1A = (300 ± 40) nM, FCCS analysis showed that MOP-eGFP and 5-HT1A-Tomato receptors not only co-localized in the plasma membrane, but also formed heterodimers, as evidenced by tCCCs (Figure 1E, black). FCCS showed that in untreated cells about 33% (RCCA = 0.33) of the 5-HT1A-Tomato receptors are bound in heterodimer complexes with MOP-eGFP (Figure 3A). Based on this, the apparent dissociation constant for a heterodimer receptor complex of MOP-eGFP–5-HT1A-Tomato with a 1:1 stoichiometry was estimated to be Kapp d = (440 ± 70) nM.

creased the fraction of 5-HT1A-Tomato receptors in heterodimer complexes with MOPeGFP (Figure 3A,B). For fentanyl, the number of heterodimer receptor-receptor complexes increased in a dose dependent manner, as evident from the increase in RCCA from

ହ ୬ = 0.42 ± 0.09, which was not significantly different from the RCCA value

ହ ୬ = 0.49 ± 0.09 (P = 0.028) in cells

ହ ୬ = 0.62 ± 0.07 (P = 3.16 × 10−7) in cells treated

measured in untreated cells (P = 0.067), to RCCAୣ୬୲ୟ୬୷୪

treated with 500 nM fentanyl, and RCCAୣ୬୲ୟ୬୷୪

RCCAୣ୬୲ୟ୬୷୪

red line).

complexes and the known concentration of fentanyl, the effect of fentanyl on the extent of

Figure 3. Opioids differ in their potency to induce heterodimer formation between MOP-eGFP and 5-HT1A-Tomato in HEK293 cells. (A) Fentanyl induces a dose-dependent increase in MOP-eGFP and 5-HT1A-Tomato receptor heterodimer formation in the concentration range 0 < cFentanyl < 750 nM. For fentanyl concentrations ≥ 1 µM, the extent of heterodimer formation drops significantly. (B) Fentanyl dose response curve calculated from the experimentally obtained RCCA values in A and the known concentrations of fentanyl. (C) 18 h treatment with equimolar concentrations of different opioids, c = 750 nM induces in cells expressing the same levels of MOP-eGFP and 5-HT1A-Tomato different extent of receptor heterodimer formation. Relative cross-correlation amplitude (RCCA), defined as the limiting value, when the lag time τ → 0, of the amplitude of the cross-correlation curve relative to the amplitude of the green autocorrelation curve, yields the number of duallylabelled, i.e., heterodimer receptor complexes (Nrg) relative to the total number of the red labelled 5-HT1A-Tomato receptors (N<sup>r</sup> total = Nr + Nrg), where Nr is the number of unbound, single-labelled 5- HT1A-Tomato receptors, and Nrg is the number of double-labelled MOP-eGFP and 5-HT1A-Tomato complexes. To reduce the effect of noise and minimize the contribution of afterpulsing, the RCCAs values were calculated as an average value of five points, starting with the value at the lag time of 10 µs to the lag time of 50 µs. In the box-and-whisker plot, the solid line shows the mean value, the dashed line shows the median, box represents the standard deviation, and the whiskers give the 5- 95 percentiles. Statistical analysis: a Student's t-test was used to determine whether the difference between the mean values measured in untreated and treated cells, or in cells treated with different opioids, are significantly different from each other. The results are reported using a two-tailed Pvalue (P). The Benjamini–Hochberg method to control the false discovery rate (FDR) in sequential **Figure 3.** Opioids differ in their potency to induce heterodimer formation between MOP-eGFP and 5-HT1A-Tomato in HEK293 cells. (**A**) Fentanyl induces a dose-dependent increase in MOP-eGFP and 5-HT1A-Tomato receptor heterodimer formation in the concentration range 0 < cFentanyl < 750 nM. For fentanyl concentrations ≥ 1 µM, the extent of heterodimer formation drops significantly. (**B**) Fentanyl dose response curve calculated from the experimentally obtained RCCA values in A and the known concentrations of fentanyl. (**C**) 18 h treatment with equimolar concentrations of different opioids, c = 750 nM induces in cells expressing the same levels of MOP-eGFP and 5-HT1A-Tomato different extent of receptor heterodimer formation. Relative cross-correlation amplitude (RCCA), defined as the limiting value, when the lag time τ → 0, of the amplitude of the cross-correlation curve relative to the amplitude of the green autocorrelation curve, yields the number of dually-labelled, i.e., heterodimer receptor complexes (Nrg) relative to the total number of the red labelled 5-HT1A-Tomato receptors (Nr total = N<sup>r</sup> + Nrg), where Nr is the number of unbound, single-labelled 5-HT1A-Tomato receptors, and Nrg is the number of double-labelled MOP-eGFP and 5-HT1A-Tomato complexes. To reduce the effect of noise and minimize the contribution of afterpulsing, the RCCAs values were calculated as an average value of five points, starting with the value at the lag time of 10 µs to the lag time of 50 µs. In the box-and-whisker plot, the solid line shows the mean value, the dashed line shows the median, box represents the standard deviation, and the whiskers give the 5-95 percentiles. Statistical analysis: a Student's *t*-test was used to determine whether the difference between the mean values measured in untreated and treated cells, or in cells treated with different opioids, are significantly different from each other. The results are reported using a two-tailed *p*-value (*p*). The Benjamini–Hochberg method to control the false discovery rate (FDR) in sequential modified Bonferroni correction for multiple hypothesis testing showed that at an FDR value of 5%, *p* ≤ 0.012 was statistically significant.

modified Bonferroni correction for multiple hypothesis testing showed that at an FDR value of 5%, P 0.012 was statistically significant. By applying the standard mathematical formalism of ligand binding assays in the absence of competing reactions [28], and considering the concentration of heterodimer complexes as an dependent variable and the concentration of fentanyl as an independent variable, the concentration of fentanyl at which the number of heterodimer complexes would be doubled was determined to be (1.90 ± 0.05) µM. Unexpectedly, treatment with such high fentanyl concentrations showed a decrease, rather than the expected increase in the concentration of heterodimer complexes (Figure 3B, dashed red line) and the RCCA Moreover, FCCS showed that treatment with different concentrations of fentanyl increased the fraction of 5-HT1A-Tomato receptors in heterodimer complexes with MOPeGFP (Figure 3A,B). For fentanyl, the number of heterodimer receptor-receptor complexes increased in a dose dependent manner, as evident from the increase in RCCA from RCCA50 nM Fentanyl = 0.42 ± 0.09, which was not significantly different from the RCCA value measured in untreated cells (*p* = 0.067), to RCCA500 nM Fentanyl = 0.49 ± 0.09 (*p* = 0.028) in cells treated with 500 nM fentanyl, and RCCA750 nM Fentanyl = 0.62 <sup>±</sup> 0.07 (*<sup>p</sup>* = 3.16 <sup>×</sup> <sup>10</sup>−<sup>7</sup> ) in cells treated with 750 nM fentanyl. From the experimentally determined concentration of heterodimer complexes and the known concentration of fentanyl, the effect of fentanyl on the extent of MOP-eGFP and 5-HT1A-Tomato heterodimerization could be quantified (Figure 3B, solid red line).

decreased to 0.45 (SD = 0.11, P = 0.004). This suggested that other processes, such as receptor homodimer formation and/or higher-order receptor heterooligomer formation [29,30] and/or desensitization or feedback processes [30] may occur at high fentanyl concentrations. Finally, it may also happen that fentanyl at such high concentrations may be toxic to cells [31], but we have not observed any such indication. By applying the standard mathematical formalism of ligand binding assays in the absence of competing reactions [28], and considering the concentration of heterodimer complexes as an dependent variable and the concentration of fentanyl as an independent variable, the concentration of fentanyl at which the number of heterodimer complexes would be doubled was determined to be (1.90 ± 0.05) µM. Unexpectedly, treatment with such high fentanyl concentrations showed a decrease, rather than the expected increase in

ହ ୬ = 0.62 ±

ହ ୬ = 0.59 ± 0.07 (P =

0.07, which was significantly different from the RCCA value measured in untreated cells

(P = 3.16 × 10−7); RCCA୭୰୮୦୧୬ୣ

Due to this, concentrations higher than 750 nM were not investigated, and this con-

ହ ୬ = 0.47 ± 0.08 (P = 1.65 × 10−4); RCCAେ୭ୢୣ୧୬ୣ

the concentration of heterodimer complexes (Figure 3B, dashed red line) and the RCCA decreased to 0.45 (SD = 0.11, *p* = 0.004). This suggested that other processes, such as receptor homodimer formation and/or higher-order receptor heterooligomer formation [29,30] and/or desensitization or feedback processes [30] may occur at high fentanyl concentrations. Finally, it may also happen that fentanyl at such high concentrations may be toxic to cells [31], but we have not observed any such indication.

Due to this, concentrations higher than 750 nM were not investigated, and this concentration was selected in further studies to compare the effects of different opioids. FCCS showed that for treatment with 750 nM fentanyl, the RCCA was RCCA750 nM Fentanyl = 0.62 ± 0.07, which was significantly different from the RCCA value measured in untreated cells (*<sup>p</sup>* = 3.16 <sup>×</sup> <sup>10</sup>−<sup>7</sup> ); RCCA750 nM Morphine = 0.47 <sup>±</sup> 0.08 (*<sup>p</sup>* = 1.65 <sup>×</sup> <sup>10</sup>−<sup>4</sup> ); RCCA750 nM Codeine = 0.59 ± 0.07 (*<sup>p</sup>* = 5.25 <sup>×</sup> <sup>10</sup>−<sup>7</sup> ) and RCCA750 nM Oxycodone = 0.47 ± 0.09 (*p* = 0.0117) (Figure 3C). Moreover, the RCCA value measured for cells treated with fentanyl was significantly higher than that measured in cells treated with equimolar concentrations of morphine (*<sup>p</sup>* = 2.48 <sup>×</sup> <sup>10</sup>−<sup>4</sup> ) or oxycodone (*<sup>p</sup>* = 6.99 <sup>×</sup> <sup>10</sup>−<sup>4</sup> ), but not significantly higher than that for codeine (*p* = 0.24). The difference in RCCA values measured in cells treated with codeine was significantly higher than that in cells treated by morphine (*<sup>p</sup>* = 3.66 <sup>×</sup> <sup>10</sup>−<sup>3</sup> ). Based on these measurements and using Equation (6), the apparent dissociation constants for the MOPeGFP–5-HT1A-Tomato heterodimer complex in the presence of equimolar concentrations (750 nM) of different non-peptide opioids could be estimated: K app d, Fentanyl = (80 ± 70) nM, K app d,Morphine = (200 ± 70) nM, K app d, Codeine = (100 ± 70) nM and K app d, Oxycodone = (200 ± 70) nM. Likewise, the apparent heterodimer dissociation constants in the presence of different concentrations of fentanyl were determined to be: K app d, 50 nM Fentanyl = (260 ± 70) nM, K app d, 500 nM Fentanyl = (180 ± 70) nM, K app d, 750 nM Fentanyl = (80 ± 70) nM, and K app d, 1 *<sup>µ</sup>*M Fentanyl = (220 ± 70) nM.

#### *2.2. Non-Peptide Opioids Increase to a Different Extent the Brightness of eGFP and Tomato*

Prolonged treatment with non-peptide opioids increased eGFP brightness, as evident from the measured counts per second per molecule (CPM). In untreated cells, average eGFP brightness was CPMeGFP Untreated = (1.1 ± 0.3) kHz. In treated cells, eGFP brightness nearly doubled, showing statistically significant difference for all treatments: CPMeGFP Fentanyl = (1.9 <sup>±</sup> 0.7) kHz (*<sup>p</sup>* = 0.015), CPMeGFP Morphine = (2.0 <sup>±</sup> 0.5) kHz (*<sup>p</sup>* = 9.6 <sup>×</sup> <sup>10</sup>−<sup>3</sup> ), CPMeGFP Codeine = (1.9 <sup>±</sup> 0.5) kHz (*<sup>p</sup>* = 5.8 <sup>×</sup> <sup>10</sup>−<sup>3</sup> ), and CPMeGFP Oxycodone = (1.8 ± 0.7) kHz (*p* = 0.027). Interestingly, an increase in Tomato brightness was also observed in cells treated with 750 nM fentanyl or morphine, but not in cells treated with codeine or oxycodone. However, the increase in Tomato brightness was not as pronounced as for eGFP, and changed from CPMTomato Untreated = (0.8 <sup>±</sup> 0.2) kHz in untreated cells to: CPMTomato Fentanyl = (1.1 ± 0.3) kHz (*p* = 0.021) for treatment with 750 nM fentanyl; CPMTomato Morphine = (1.3 ± 0.3) kHz (*<sup>p</sup>* = 3.0 <sup>×</sup> <sup>10</sup>−<sup>3</sup> ) for treatment with 750 nM morphine, whereas it remained unchanged (within the limits of the experimental error) for treatment with 750 nM codeine, CPMTomato Codeine = (1.0 <sup>±</sup> 0.3) kHz (*<sup>p</sup>* = 0.20), or 750 nM oxycodone, CPMTomato Oxycodone = (0.9 ± 0.3) kHz (*p* = 0.12). While we do not know why the brightness of fluorescence reporters has changed following treatment with non-peptide opioids, two processes can independently and jointly cause such effects, receptor homodimerization and/or alteration of fluorescence lifetime due to environmental changes. However, to discern the contribution of one effect from the other, a stringent number and brightness analysis and fluorescence lifetime measurements would be needed. We reflect on this in more detail in Section 3.
