Transporters and Drug Secretion into Human Breastmilk

Abstract

Introduction:

Medication use is highly prevalent in breastfeeding persons, posing potential risks for drug exposure to nursing infants. Transporters in the lactating mammary gland carry pharmacological and toxicological significance, as they can mediate the active transfer of drugs and nutrients into breastmilk.

Areas Covered:

In this narrative review, we searched and compiled current knowledge on the transport of drugs in the human mammary gland from literature indexed in PubMed (current as of October 25, 2024), and clinical evidence demonstrating active transport of drugs into milk is provided. In vitro and in vivo models of the mammary gland are outlined in brief and known drug transporters at the blood-milk barrier and their potential relevance to drug concentrations in milk are described in detail.

Expert Opinion:

Although clinical data show that membrane transporters mediate the transfer of multiple drugs into breastmilk, our ability to predict milk concentrations for these drugs is limited. Improving our understanding of the transporter biology and pharmacology in the mammary gland is crucial for developing models to predict drug concentrations in human milk, which will support clinicians and lactating individuals in making rational decisions to balance the benefits of breastfeeding and the risks of drug exposure to infants.

1. Introduction

Exclusive breastfeeding is currently recommended for the first six months of life with continued breastfeeding along with appropriate complementary foods up to 2 years or more by the American Academy of Pediatrics to support infant immune development and long-term health [1]. According to the US Centers for Disease Control and Prevention, 83% of infants born in the US in 2019 received breastmilk after delivery (either exclusively or in combination with formula), and 56% of infants continued receiving some breastmilk at 6 months of age [2]. While there is ample evidence that breastmilk is beneficial for both maternal and infant health [3–6], medication use is exceedingly common during lactation, with over 90% of breastfeeding people in the US taking at least one medication [7]. Medications used by lactating individuals can be transferred into breastmilk to a significant extent, leading to unintentional infant exposure and, in some cases, adverse health outcomes for the infants. Although an overwhelming majority of people take medications while breastfeeding, there is a dearth of clinical evidence to support the safety of most drugs for breastfed infants.

Pregnant and breastfeeding individuals have historically been excluded from clinical studies during drug development due to safety concerns and risk of legal liabilities, which limits the ability of patients and clinicians to make evidence-based decisions about possible benefits and risks of medications during pregnancy and lactation [8]. Clinical pharmacokinetic studies in lactating individuals are challenging and logistically problematic when considering every marketed drug, which necessitates the use of mechanism-based prediction methods for assessing drug passage into breastmilk. One historical approach is the prediction of the milk-to-plasma (M/P) AUC ratios of drugs during breastfeeding [9]. For drugs that enter the milk by passive diffusion, the M/P ratio can be adequately predicted based on partition theories using both the physicochemical properties of drugs and the physiological characteristics of breastmilk. However, many drugs and nutrients are known to be actively secreted into milk by transporters expressed in mammary epithelial cells (MECs) [10–13]. For these drugs, no methods currently exist to accurately predict their transfer into human milk.

The purpose of this narrative review is to summarize the current understanding of drug transport mechanisms in the human mammary gland, with a particular focus on carrier-mediated active transport and transporters expressed in human MECs, and to identify knowledge gaps in our understanding of carrier-mediated drug transport into breastmilk. We compiled literature indexed in PubMed (current as of October 25, 2024) using search terms that described: 1) mechanisms of mammary gland transport in humans, animals, or vitro models; 2) the function(s) of specific membrane transporters identified in the mammary gland; and 3) clinical studies or case reports examining active drug transfer into breastmilk. Articles were excluded if they were not in English, or if full text was not available. While much of the literature in this review was published in the last 10–15 years, older publications were included if literature on the topic is very limited, or if they represented seminal work that shaped our understanding of mammary gland transport. We conclude by providing our perspectives and recommendations for future research to enhance our understanding of drug transport mechanisms in MECs and to support the development of predictive methods for active drug transfer into human milk.

2. Mechanisms of Drug Transfer into Human Milk

The adult human mammary gland is comprised of a network of ducts that branch and differentiate into terminal ductal lobules during pregnancy and lactation [14,15]. Estrogen and progesterone concentrations increase during pregnancy stimulating branching of lactiferous ducts and alveologenesis [14,16]. In late pregnancy and lactation, MEC (mammary epithelial cell) gene expression of hormone receptors for estrogen, progesterone, and prolactin increase [17]. Both progesterone and prolactin stimulate terminal differentiation of alveolar cells into milk-producing MECs, also referred to as lactocytes, luminal epithelial cells, or alveolar secretory cells [16]. MECs form a single layer of epithelium lining the apical lumen of alveoli and are responsible for producing and secreting milk into alveoli. The apical membrane of MECs face the lumen where milk is secreted and the basolateral membrane is perfused by maternal blood.

MECs possess multiple transport mechanisms that tightly control milk components (Fig. 1). Milk contains a plethora of nutrients for infant development, including carbohydrates, lipids, proteins, fat- and water-soluble vitamins, minerals and trace elements [18,19]. In addition, milk contains immunoglobulins, hormones, and growth factors that provide immune protection and promote nutrient absorption in the infant [20,21]. The concentrations of nutrients and bioactive compounds are highly regulated by MECs and change as the infant grows [19,22]. Most drugs are thought to enter breastmilk by passive diffusion from capillaries in the mammary gland across the plasma membranes of MECs; however, drugs may enter breastmilk through multiple mechanisms based on their pharmacokinetics, physiochemical properties, and interaction with specific transport proteins in the MECs as described below [23]. Drugs and nutrients may also share the same pathway(s) to enter breastmilk, which may potentially carry implications for drug-nutrient interactions.

Figure 1. Transport pathways at the blood-milk barrier.

Drugs and nutrients may cross into the breastmilk via passive diffusion, protein-mediated active transport, lipid co-transport, transcytosis, or exocytosis depending on their physiochemical properties, interactions with transport proteins in the MEC, as well as binding to or partitioning into milk constituents. Note that lipid co-transport is a proposed mechanism that has not been observed experimentally, and lipophilic compounds may passively diffuse into milk and subsequently be dissolved in lipid droplets or may be packaged into lipid droplets prior to secretion into milk (or a combination of these mechanisms). In early lactogenesis, the blood-milk barrier has fewer tight junctions between MECs, allowing larger molecules such as antibodies to pass between MECs via paracellular diffusion in late pregnancy and shortly after childbirth. Created in BioRender. Beers, J. (2025) https://BioRender.com/d20t078

2.1. Passive diffusion

Passive diffusion across the MEC plasma membrane is the major route of transfer for most small molecule drugs into breastmilk [24]. In this model, unbound, unionized drug molecules move across MEC membranes from maternal plasma into breastmilk driven by the concentration gradient. According to the pH partition theory, passive diffusion is affected by the drug’s lipophilicity (LogP), molecular weight, degree of ionization, and protein binding. Small, highly lipid-soluble drugs (less than approximately 200–500 Da [25]) can readily diffuse across the MEC membrane and partition into the lipid-rich milk. As the milk has a slightly more acidic pH (~7.1) compared to plasma (7.4), weakly basic drugs will be ionized (hence trapped) to a greater extent in the milk [26,27]. Lastly, since only the free (unbound) fraction of the drug is available for diffusion, drugs highly bound to plasma proteins in the maternal blood only have a small portion of the drug transfer into milk [26].

The passage of drugs into breastmilk is often characterized using the milk-to-plasma ratio (M/P ratio), which is calculated as a ratio of the drug AUC (area under the curve) in milk vs. plasma. While some studies estimated the this ratio using single time point drug concentrations simultaneously measured in milk and maternal plasma, this approach may not provide an accurate M/P ratio as drug levels in maternal plasma and breastmilk do not necessarily rise and fall in parallel [28]. Rasmussen first applied the pH partition theory to predict the free concentrations of drugs in milk in 1958 [29,30]. While this model predicted the ratio of unbound drug in milk and plasma, it failed to predict the total M/P ratio for many drugs. Fleishaker et al. later incorporated drug binding to plasma and milk proteins, which allowed for more accurate predictions of the M/P ratio [31,32]. This traditional model for predicting M/P ratios relies on the free drug hypothesis combined with the assumption that passive diffusion is the only mechanism for drug transfer into milk. The drug’s pKa and fraction unbound in milk and plasma are used to determine the unbound, unionized fraction of drug available to diffuse through the MEC membrane into milk [24,31]. Based on its physiochemical properties, the predicted M/P ratio of a given drug may vary considerably from 1. This model adequately predicts the M/P ratio for many passively diffused drugs, including caffeine [33,34], diazepam [35–38], labetalol [39,40], metronidazole [11,39,41,42], and pseudoephedrine [43,44], among many others. However, this model fails to account for active transport in the mammary gland, which has been found to influence the concentrations of transporter substrates and leads to greatly underpredicted M/P ratios [45].

2.2. Active transport

Uptake and efflux membrane transporters are known to facilitate the movement of electrolytes, vitamins, and other small molecules between MECs and breastmilk [46,47]. They play a crucial role in secreting essential nutrients for infant development. However, some drug molecules can also be recognized as substrates by MEC transporters, leading to their active transport into human milk. Carrier-mediated active transport can result in drug concentrations in human milk exceeding those in maternal blood, which increases the infant’s exposure – and consequently the risk of toxicity – to medications taken by the lactating individual. For instance, clinical studies previously showed that atenolol, a beta-blocker used to treat hypertension, is actively excreted into breastmilk with an observed M/P ratio much higher than predicted (see Table 1) [48]. A case report of infant toxicity (cyanosis, bradycardia, and hypothermia) related to atenolol exposure through breastmilk has been reported [49].

Table 1. Examples of Drugs Secreted into Breastmilk.

Putative transport mechanisms are proposed based on prior literature reports, known mechanism(s) of transport in the mammary gland or other tissues, and reported discrepancies between predicted and observed M/P ratios. For a list of predicted M/P ratios for commonly prescribed drugs, we refer interested readers to references [11,39,51,192].

Drug	Species	Milk:Plasma Ratio	Putative Transport Mechanism	Reference(s)
Abacavir	Humans	1.03 (concentration ratio at matched timepoints)	Possibly active transport; substrate of BCRP and ENTs	[23,193,194]
Acyclovir	Humans	2.25 3.24 (average concentration ratio over 48 h)	Active transport; substrate of OCT1 and BCRP	[10,195–197]
Atenolol	Humans	3.6 (concentration ratio at Cmax, milk†) 4.5 (AUC ratio) 4.9–5.7 (AUC ratio) 1.1–3.1 (AUC ratio)	Unknown; substrate of OCT1 and OATPs	[48,198–203]
Azithromycin	Humans	2.49	Active transport; substrate of P-gp and OATPs	[23,159,204]
Bupropion	Humans	2.51–8.58 (concentration ratios)	Active transport (BCRP)	[11,205]
Cannabidiol	Humans	2.6 (Css ratio)	Unknown; possibly passive diffusion, lipid co-transport and/or active transport (BCRP)	[76,78]
Chloroquine	humans	1.96–4.26 (AUC ratio)	Active transport (BCRP)	[206]
Cimetidine	humans, rabbits, rats, mice	Human: 5.77 (AUC ratio), 1.7 (AUC ratio) Rabbit: 1.03–1.08 (Css ratio‡) Rat: 24.6–31.9 (Css ratio); 29.3 (milk:serum Css ratio) Mouse: 13.8 (concentration ratio 30 min after dose)	Active transport; substrate of OCT1 and BCRP	[10,12,45,50,117,120,207]
Ciprofloxacin	Mice	Humans: 1.6–2.14 (concentration ratios) Mice: 3.08 (concentration ratio)	Active transport (BCRP)	[208,209,11]
Delta-9-tetrahydrocannabinol	Humans	8.4 (Css ratio); 7.0 (Css ratio); 6.1 (concentration ratio collected at matched timepoints)	Unknown; possibly passive diffusion, lipid co-transport and/or active transport (BCRP)	[76,78,210,211]
Doxorubicin	Humans	9.9 (AUC ratio)	Active transport; substrate of OCT1, P-gp, BCRP, MRP1, and OATP1A2	[168,212]
Emtricitabine	Humans	1.77 (AUC ratio) 3.01 (AUC ratio) 3.92 (concentration ratio at matched timepoints)	Unknown; substrate of ENTs and CNTs	[193,194,213,214]
Flecainide	Humans	2.3 (concentration ratio 24h after last dose) 2.9 (concentration ratio 48h after last dose) 1.57–2.18 (Css ratio)	Active transport; substrate of P-gp	[11,215–218]
Lamivudine	Humans	3.34 (median concentration ratio)	Active transport (BCRP); substrate of CNTs and ENTs	[23,194,219]
Nadolol	Humans	4.6 (Milk:serum Css ratio)	Possibly active transport; substrate of OCT1 and P-gp	[11,220,221]
Nifedipine	Humans	0.73 (median Css ratio)	Possibly active transport; substrate of BCRP	[23,182,222,223]
Nitrofurantoin	humans, mice, rats	Humans: 6.21 (milk:serum AUC ratio) Mice: 45.7 (concentration ratio 30 min after dose) Rats: 30.15 (milk:serum Css ratio); 23 (AUC ratio)	Active transport (BCRP)	[10–12,88,91,96]
Ofloxacin	Humans	1.19 (AUC ratio) 1.28 (mean milk:serum concentration ratio collected at matched time points over 24h)	Possibly active transport; substrate of BCRP	[11,192,208]
Procainamide	Humans	3.18 (AUC ratio)	Possibly active transport; substrate of OCT1	[11,192,224,225]
Riboflavin	Humans, mice	Mice: 25 (Css ratio in mice fed chow supplemented with riboflavin)	Active transport (RVFT2 and BCRP)	[116]
Sumatriptan	Humans	4.9 (AUC ratio)	Possibly active transport; substrate of OCT1 and P-gp	[226–228]
Tacrolimus	Humans	2.89 (AUC ratio)	Active transport; substrate of P-gp	[23,158]
Tenofovir alafenamide	Humans	4.09 (median concentration ratio)	Active transport (BCRP)	[23,193]
Topotecan	Mice	6.8 (concentration ratio 30 min after dose)	Active transport (BCRP)	[117,229]
Venlafaxine	Humans	3.26–3.99 (AUC ratio)	Possibly active transport; substrate of P-gp	[230,231]
Zidovudine	Humans	1.11–1.78 (AUC ratio) 3.21 (median concentration ratio)	Possibly active transport; substrate of BCRP, CNTs, and ENTs	[11,194,232]
Zonisamide	Humans	0.93 (average concentration ratio collected at matched timepoints)	Possibly active transport; substrate of BCRP	[11,233]

^†C_{max, milk} = maximal drug concentration in milk;

^‡C_ss = drug concentration in plasma, serum, or milk at steady state.

Generally speaking, an observed M/P ratio more than 2-fold greater than that predicted by the pH partition theory suggests an active transport mechanism may be involved in drug transfer [28]. One of the first reported instances of active transport into breastmilk was described in lactating rats given cimetidine, where steady state milk-to-serum ratios were ~6 times higher than values predicted by the traditional diffusional model [45]. This finding was later demonstrated in lactating individuals, with observed cimetidine milk-to-serum ratios ~5.5-fold higher than predicted [50]. In Table 1, we compiled a list of drugs with in vivo evidence of active secretion into breastmilk. It is important to note that while the observed M/P ratios for humans and rodents were similar in this case, species differences in milk composition and transporter expression may contribute to differences in the M/P ratio for the same drugs [51] (see Section 3).

Despite these clinical examples suggesting carrier-mediated active transport of drugs into human milk, the underlying molecular mechanisms involved in MEC transport of a specific drug are less clear. While we also proposed putative transport mechanisms for the drugs listed in Table 1, based on available in vivo studies using transporter knockout mice and/or in vitro transporter substrate data, additional studies are required to clearly define the specific roles and contributions of various MEC transporters (see section 2.2) in drug secretion into human milk.

2.3. Paracellular diffusion

Paracellular diffusion plays a brief but important role in nutrient transport during the early lactation stage. Colostrum and early milk have high concentrations of antimicrobial and immunostimulatory proteins such as lactoferrin, lysozyme, cytokines, and secretory IgA [21,52]. In the colostral phase (days 0–5 after childbirth), paracellular diffusion through gaps between MECs is a common mechanism of diffusion for larger nutrients and immunoglobulins [28,53,54]. In addition to early lactogenesis, gaps between MECs are also present during involution and inflammation [55]. Leukocytes may enter milk through paracellular gaps in response to inflammatory signals [47,55]. Approximately 1 week after delivery, tight junctions form between MECs, allowing only small molecules to pass through [28,53,56]. Nguyen et al. found that paracellular diffusion of radiolabeled sucrose (molecular weight ~342 Da) occurs across the mammary epithelium of pregnant mice but is almost completely inhibited in lactation [56]. Tight junction formation is thought to be regulated by the decline in progesterone concentrations that occur after childbirth combined with an increase in glucocorticoids and prolactin [56]. During this transition, concentrations of most whey proteins and immunoglobulins decrease while lipid and carbohydrate concentrations increase [21,54,57].

2.4. Exocytosis

Exocytosis is used to transport ions and endogenous nutrients including calcium, lactose, phosphate, and citrate into milk [47]. The MEC Golgi complex synthesizes proteins and lactose, which are packaged alongside casein micelles and endocytosed serum lipids in vesicles derived from the Golgi membrane [47,58,59]. These secretory vesicles are then transported to the apical membrane and their contents are released after fusion of the vesicular and apical membranes [47,55]. This transport pathway has mainly been described with regards to intracellular proteins, lactose, and ions, and the potential role of exocytosis in secreting drugs into milk remains unknown. Exocytosis may be relevant for mineral drugs and supplements (e.g. lithium, calcium, etc.); however, this area remains unexplored.

2.5. Transcytosis

Large molecules such as immunoglobulins, serum albumin, and endocrine hormones are transported in vesicles from surrounding cells and serum through transcytosis [19,47,60]. In addition to being synthesized within MECs, the circulating iron-binding protein transferrin and the hormone prolactin bind to the transferrin receptor (TfR) and the prolactin receptor (PRLR), respectively, located on the basolateral MEC membrane [60]. They are then internalized and shuttled through the cell in endosomes, where they engage in cross-talk with secretory vesicles from the MEC Golgi complex before release into the alveolar lumen through a process that is not completely understood [60–62]. Immunoglobulins are thought to enter MECs through interaction with Fc receptors in a manner similar to other epithelial tissues, and intravenously administered immunoglobulins have been identified in both colostrum and milk [63]. This may carry implications for antibody-based therapeutics; however, the transfer of these drugs into breastmilk has not been adequately studied [64,65]. Available clinical examples of IgG-based therapeutics studied in lactation were recently reviewed by Guinn et al. [64]. Limited evidence from case reports and cohort studies suggest that IgG-based monoclonal antibody secretion into milk is minimal [66,67]. A case report measuring rituximab concentrations in a single patient measured a milk-to-serum ratio of 1/240 approximately seven days after infusion of rituximab [67]. Clowse et al. reported that milk concentrations of certolizumab pegol were negligible in a cohort of 19 breastfeeding individuals receiving therapeutic doses of the drug [66]. The authors hypothesized that the lack of an Fc region in the certolizumab antibody may have contributed to limited milk absorption [66].

2.6. Lipid co-transport

Milk lipids are secreted intact through an apocrine mechanism that differs from serum lipid exocytosis [18,68]. Triglycerides that make up the core of lipid droplets are synthesized in the endoplasmic reticulum of MECs and trafficked to the apical membrane [68]. Lipid droplets are then secreted as milk fat globules enveloped by budded portions of the MEC apical plasma membrane [18,19,69]. Lipophilic drugs can passively diffuse across the MEC cells into the milk (see Section 2.1) and then partition into the lipids in the milk. An alternative lipid co-transport pathway has also been postulated [23]. In this model, a lipophilic drug enters the MEC by passive diffusion, dissolves into the lipid droplets in the MECs, and is subsequently co-transported into milk when the lipid droplets are secreted. However, there is no experimental evidence to support such a mechanism, and the observation of higher drug concentrations in milk with higher fat content does not constitute proof of either a diffusion or a co-transport mechanism. It is also possible that drug transfer may occur through a combination of these mechanisms.

The milk lipid content increases as the breast empties during breastfeeding, with hindmilk containing 2–3 times the amount of fat compared to foremilk [70]. Concentrations of lipophilic compounds may therefore be markedly higher in hindmilk. This phenomenon has been demonstrated with duloxetine [23,71,72] and mirtazapine [23,73]. Relatively few clinical studies have compared drug concentrations in foremilk vs. hindmilk or examined the underlying mechanism(s) of transport for lipophilic drugs. Cannabinoids, for example, are highly lipophilic compounds that are known to accumulate in the adipose tissue of chronic cannabis users [74,75]. Both delta-9-tetrahydrocannabinol (THC) and cannabidiol (CBD), two major pharmacologically active constituents of cannabis, are known to rapidly accumulate in breastmilk [76,77]. These drugs are also known substrates of the efflux transporter BCRP [78]. Cannabinoid milk concentrations have not been compared in foremilk and hindmilk, and it is currently unknown whether these drugs may enter milk through passive diffusion and/or lipid co-transport, efflux by BCRP, or a combination of these mechanisms.

3. Experimental Models of Mammary Gland Transport

To date, few clinical studies have been performed to describe drug passage into breastmilk due to ethical and practical challenges. Most human M/P ratios have been calculated from case studies that included a small number of lactating individuals prescribed one or more drugs. For this reason, evidence derived from experimental in vitro and animal models remains crucial in understanding mechanisms of lactational transfer for drugs. Species differences and limitations of in vitro to in vivo extrapolation must be considered, however, when comparing these non-clinical models to humans. For a more in-depth discussion of non-human models of lactation, the authors refer interested readers to a recent review by Nauwelaerts et al. [79].

3.1. Cell Lines

Drug transfer in lactation has been studied using both human and animal derived cell lines. Human breast tissue-derived cells, including the Michigan Cancer Foundation 7 (MCF7), MCF12A, and MCF10A cell lines, have been used to model the transport of vitamin C, hormones, carnitine, and the environmental toxin β-N-methylamino-alanine (BMAA) into breastmilk [80–83]. MCF7 is a breast cancer-derived cell line that expresses both estrogen and progesterone receptors, proliferates in response to estradiol exposure, and is very commonly used as a model for breast cancer research [84,85]. While there is significant overlap in transporter expression with healthy MECs, several xenobiotic transporters have not been identified in MCF7 cells, including but not limited to P-gp, OCTs, CNTs, and PEPTs [79]. Importantly, tight junction formation in MCF7 cells appears to be highly variable and unstable [86], further limiting its utility for predicting drug transfer into breastmilk.

The MCF12A cell line is non-tumorigenic. It was originally cultured from excised fibrocystic tissue removed during the reduction mammoplasty of a 63 year old woman [85]. Unlike MCF7 cells, MCF12A cells express both OCT1 and OCT3 at the mRNA level; however, it does not appear to express estrogen or progesterone receptors, and receptor expression is not inducible by estrogenic hormones [82,85]. OCTN1 and OCTN2 protein expression and activity have been described in MCF12A cells using the substrates carnitine and tetraethylammonium (TEA) [82]. MCF10A cells are an immortalized non-tumor origin cell line used as an in vitro model for the normal breast [79]. These cells similarly do not express hormone receptors, and their utility as a model for mammary epithelia and drug secretion was recently questioned as they do not form proper tight junctions [86] or exhibit a normal mammary epithelial phenotype in 2D or 3D culture [87].

The murine mammary epithelial cell lines HC11 and CIT3 have similarly been used to model the transport of nitrofurantoin, dipyridamole, nucleosides, mitoxantrone, digoxin, and BMAA into breastmilk [88–91]. In addition to expressing CSN2 (the gene encoding β-casein), differentiated HC11 cells in culture express transcripts for murine P-gp, BCRP, MRP1, OATP1A5, OCTN1, and OCT1 [92]. In differentiated HC11 cells, protein expression of murine P-gp was decreased and BCRP expression was increased compared to undifferentiated cells [93]. When bovine BME-UV mammary epithelial cells were cultured in the same study, they were found to only express bovine P-gp and MRP1 following differentiation [92]. While drug transporter expression in CIT3 cells appears to be less well-studied, Athavale et al. generated an in vitro model of the mammary gland using CIT3 cells cultured in transwells to estimate the M/P ratios of several passively diffused drugs [94]. However, pronounced species differences in lactational physiology, milk composition, and transporter expression limit the utility of these cell lines in modeling xenobiotic transfer into human milk.

In addition, although its phenotype is not representative of the mammary gland, the Madin-Darby Canine Kidney-II (MDCK-II) cell line transfected with human BCRP has also been used in several studies to help verify the role of BCRP in transporting its substrates into milk [11,95,96].

3.2. Induced Pluripotent Stem Cells (iPSCs)

In response to pregnancy- and lactation-related hormones, the mammary gland may generate its own population of induced pluripotent stem cells (iPSCs) that are capable of differentiating into MECs and myoepithelial cells in vitro [97]. This has been demonstrated using explanted mammary tissue from mice [97,98]. Three-dimensional mammary organoids have also been developed from fibroblast- and blood-derived human iPSCs [99]. While iPSCs have been used to model the mammary gland in breast cancer research, they have not been applied to understanding drug transport into breastmilk to the authors’ knowledge.

3.3. Primary Cells

Multiple studies have cultured primary MECs derived from both animals and humans. This has been performed using cells isolated directly from breast tissue [86,100–103], and from cells shed into breastmilk during lactation [17,104–109]. Relatively few of these studies, however, have examined drug transport using primary MECs. A recent study by La Mantia et al. cultured healthy human MECs (HMECs) in transwells and conducted functional transporter studies using probe substrates for P-gp, MRP2/3, BCRP, and OAT transporters [86]. Though mRNA was detected for each of these transporters in cultured HMECs, there was no measurable transporter activity for any of the substrates tested [86]. Nevertheless, HMECs formed a functional barrier with tight junctions in transwells that may prove useful in predicting the milk concentrations of passively diffused drugs [86].

3.4. Organoids

Three dimensional mammary organoids have been developed for cancer and lactation research using iPSCs and primary patient-derived MECs (reviewed in [110]). Many mammary organoid models consist of MECs suspended in a matrix of collagen, Matrigel, or basement membrane, allowing cells to self-assemble into ductal and alveolar structures mimicking breast tissue [110,111]. Subsequent exposure to media containing prolactin, insulin, and hydrocortisone has been shown to trigger production of lipid droplets and milk proteins in mammary organoids [99,112,113]. Whether these models are capable of recapitulating drug transport in lactation remains to be studied.

3.5. Animal Studies

Many in vivo animal studies have been conducted using mice [51,93,95,96,114–118], rats [12,119–121], rabbits [32,34,38,45], cows [117,122,123], and goats [123] to model drug transfer into milk. A comparison of lactational anatomy, physiology, and milk composition across common model organisms was recently reviewed by Ventrella et al. [124]. While useful for mechanistic studies comparing drug transport between wild-type and knockout animals, it is important to note that physiological differences between species may lead to very different estimations of the M/P ratio. Ito et al. reported that while the unbound M/P ratios of mice and humans were similar for a panel of 27 drugs, the total M/P ratios were approximately 2-fold higher on average in mice compared to humans [51]. This difference was attributed mainly to the higher protein and lipid content in mouse milk, leading to increased binding to milk proteins and partitioning to fat in mice [51]. In addition, the expression and activity of membrane transporters may influence the M/P ratio across different species. For instance, although the OCT and BCRP substrate cimetidine has been shown to accumulate in the milk of mice, rats, and humans, cimetidine concentrations appear to be governed by passive diffusion in rabbits [45,50,120].

3.6. Human Studies

Most of our knowledge of M/P ratios for drugs in humans come from case studies or small cohort studies. Due to limited sample sizes and variability in the time from childbirth for most study cohorts, it is difficult to determine how drug passage into breastmilk changes from colostrum to mature milk, though this is suspected due to changes in human milk composition and observed changes in transporter gene expression during lactation in animals [46,125]. Eyal et al. conducted a prospective, longitudinal cohort study that measured the pharmacokinetics of atenolol in plasma and breast milk at 2–4 weeks (n=32), 3–4 months (n=22), and 6–8 months postpartum (n=17) [48]. Atenolol breastmilk concentrations and subsequent infant exposure decreased over time, with relative infant doses (expressed as a percentage of the parent’s weight-adjusted dose) decreasing from 14.6% at 2–4 weeks to 5.9% at 6–8 months [48]. The average M/P ratio similarly decreased from 5.7 to 4.9 from the first to the last study visit [48]. While the exact cause remains unclear, this shift likely reflects a change in lactational physiology resulting in reduced mammary clearance over time. The authors hypothesized that decreased atenolol mammary clearance may be associated with longitudinal changes in OATP activity in mammary epithelial cells [48].

4. Transporters in the human mammary gland

As stated earlier, carrier-mediated transport can lead to greater drug accumulation in human breastmilk. Consequently, transporters involved in this process are of significant clinical and toxicological importance, as they can influence infant drug exposure and potential toxicity. In humans, there are two superfamilies of transporters: the ATP-binding cassette transporters (ABC), and the solute carrier transporters (SLC). Referred to as efflux transporters, the ABC transporters utilize ATP hydrolysis to actively transport their substrates out of cells. Although frequently referred to as uptake transporters, the SLC transporters have diverse transport modes. They can operate as facilitative transporters or secondary, tertiary active transporters through coupling to a co-transported substrate (e.g. Na⁺, H⁺) or counter ions.

Numerous membrane ABC and SLC transporters have been identified at the mRNA and protein level in human MECs (see Fig. 2 and Table 2). Transporters localized to the basolateral and apical membranes of MECs regulate the milk concentrations of ions, vitamins, bile acids, xenobiotics, and other small molecules as they are transported between plasma and breastmilk. Polymerase chain reaction (PCR) evidence shows changes in transporter gene expression during lactation as MECs transition into milk-producing cells. Gene expression of PEPT2, CNT1, and ENT3 is induced from nondetectable levels in lactation, and transcripts of OCT1, CNT3, and SVCT1 are increased more than 2-fold [46]. A recent transcriptomic study found increased expression of BCRP in lactating vs. non-lactating human MECs [17], and a murine in vitro mammary gland model similarly showed upregulation of mBcrp in differentiated vs. undifferentiated HC11 cells [93]. In contrast, transcript levels of P-gp, OCT3, MRP1, OATP4A1, and PEPT1 are all downregulated by greater than 2-fold in lactation [46,93].

Figure 2. Mammary gland anatomy and transporters at the blood-milk barrier.

The localization of transporters shown in color (BCRP, P-gp, PEPT2, and MRP4) has been demonstrated from imaging studies. For transporters shown in gray, membrane localization is inferred based on localization in other tissues, and further work is needed to confirm their location in MECs. Additional transporters that have been identified with unknown localization are shown in Table 2. Created in BioRender. Beers, J. (2025) https://BioRender.com/w91e923

Table 2. Mammary Gland Transporter Expression.

The localization within MECs has been proposed for MDR3, MRP1, MRP2, OCT1, OCTN1, OCTN2, CNT1, CNT3, and OATP1A2, which are consistently localized at either the apical or basolateral membrane in other human tissues; however, this has not been confirmed experimentally in MECs.

Transporter	Gene	Function	Localization in MECs	Species	mRNA or protein?	Detection method(s)	Reference(s)
P-gp	ABCB1	ATP-dependent efflux transporter	Apical membrane	Humans, mice	mRNA, protein	PCR (humans, mice), in-cell western assay (mice)	[10,46,92,93]
MDR3	ABCB4	ATP-dependent efflux transporter	Unknown (likely apical based on other tissues)	Humans (MCF12A cell line, MECs)	mRNA	PCR	[10,161]
MRP1	ABCC1	ATP-dependent efflux transporter	Unknown (likely basolateral based on other tissues)	Humans, mice	mRNA	PCR	[46,92]
MRP2	ABCC2	ATP-dependent efflux transporter	Unknown (likely apical based on other tissues)	Humans	mRNA	PCR	[46]
MRP4	ABCC4 (humans), Abcc4 (mice, rats)	ATP-dependent efflux transporter	Basolateral membrane of non-lactating women and lactating mice	Humans, mice, rats	mRNA, protein	PCR (mice, rats), Western blot (rats), IHC¥ (humans, mice), immunofluorescence (mice)	[118]
MRP5	ABCC5	ATP-dependent efflux transporter	Unknown	Humans	mRNA	PCR	[46]
BCRP	ABCG2	ATP-dependent efflux transporter	Apical membrane	Humans, mice, cows	mRNA (mice, cows) protein (mice, human milk fat globule membrane)	PCR (mice, cows), Western blot (mice), IHC (mice), immunofluorescence (mice), proteomics (humans)	[92,117,118,140]
PEPT1	SLC15A1	Proton-coupled secondary active peptide uptake transporter	Unknown	Humans	mRNA	PCR	[46]
PEPT2	SLC15A2	Proton-coupled secondary active peptide uptake transporter	Apical membrane	Humans, rats	mRNA, protein	PCR, IHC	[46,171]
OCT1	SLC22A1	Organic cation uptake transporter	Unknown (likely basolateral based on other tissues)	Humans, mice	mRNA	PCR	[46,92,125]
OCT3	SLC22A3	Organic cation uptake transporter	Unknown	Humans	mRNA	PCR	[46]
OCTN1	SLC22A4	Organic cation uptake transporter	Unknown (likely apical based on other tissues and IHC)	Humans, mice, cows, rats	mRNA (humans, mice, cows, rats) protein (mice, rats)	PCR, IHC	[46,92,115,153]
OCTN2	SLC22A5	Organic cation uptake transporter	Unknown (likely apical based on other tissues)	Humans, mice, rats	mRNA (humans, mice, rats) protein (mice, rats)	PCR, IHC	[46,115]
SVCT2 (formerly NCBT1)	SLC23A2 (formerly SLC23A1)	Na+-coupled secondary active vitamin C uptake transporter	Unknown	Humans	mRNA	PCR	[46]
CNT1	SLC28A1	Na+-coupled secondary active nucleoside uptake transporter	Unknown (likely apical based on other tissues)	Humans	mRNA	PCR	[46]
CNT3	SLC28A3	Na+-coupled secondary active nucleoside uptake transporter	Unknown (likely apical based on other tissues)	Humans	mRNA	PCR	[46]
ENT1	SLC29A1	Facilitative nucleoside uptake transporter	Unknown	Humans	mRNA	PCR	[46]
ENT3	SLC29A3	Facilitative nucleoside uptake transporter	Unknown	Humans	mRNA	PCR	[46]
RFVT2	SLC52A2 (humans), mSlc52a2 (mice)	Na+-independent riboflavin uptake transporter	Basolateral and apical membrane in transfected cell lines; unknown in MECs	Humans, mice	mRNA, protein	PCR (humans, mice), IHC (mice)	[175]
OATP1A2	SLCO1A2 (humans), mOatp1a5 (mice), bOatp1a2 (cows)	Secondary active uptake transporter	Unknown (likely apical based on other tissues)	Humans, mice, cows	mRNA	PCR	[46,92]
OATP2B1	SLCO2B1	Organic anion uptake transporter	plasma membrane; exact localization unknown	Humans	mRNA, protein	PCR, IHC, immunofluorescence	[46,167]
OATP3A1	SLCO3A1	Organic anion uptake transporter	plasma membrane; exact localization unknown	Humans	mRNA, protein	PCR, IHC, immunofluorescence	[46,167]
OATP4A1	SLCO4A1	Organic anion uptake transporter	Unknown	Humans	mRNA	PCR	[46]
OATP5A1	SLCO5A1	Organic anion uptake transporter	plasma membrane; exact localization unknown	Humans	mRNA, protein	PCR, IHC, immunofluorescence	[167]

^¥IHC = immunohistochemistry.

Changes in transporter expression may be stimulated by activation of nuclear receptors via pregnancy- and lactation-related hormones. Concomitant release of estrogen and progesterone has been shown to increase BCRP expression in placental BeWo cells through activation of the estrogen (ER) and progesterone (PR) receptors [126,127], and PR activation in pregnancy increases expression of P-gp in the endometrium [127,128]. Similarly, both of the nuclear receptors pregnane X receptor (PXR) and constitutive androstane receptor (CAR) are modulated by estradiol and progesterone [129,130]. Activation of PXR and CAR by xenobiotics upregulates the expression of efflux transporters BCRP, P-gp, and MRP2–3 in the brain, liver, and intestine [127,131,132]. While PXR is known to be expressed in both healthy and neoplastic breast tissue [133], little information exists on CAR expression in the lactating mammary gland. The Human Protein Atlas reports negligible CAR expression in breast tissue, though CAR is present at low levels in some breast cancer-derived cell lines [134,135]. Xu et al. recently reported that the pesticide contaminant 1, 4-bis [2-(3, 5-dichloropyridyloxy)] benzene (TCPOBOP) inhibited lipid accumulation and accelerated fibrosis in mammary adipose tissue through CAR activation; however, TCPOBOP also activates PXR [136]. In addition, expression of the nuclear hormone vitamin D receptor (VDR) is upregulated in healthy lactating mammary epithelial cells, likely in response to the lactogenic hormones insulin, cortisol, and prolactin [137]. Activation of VDR by calcitriol (the active form of vitamin D) has also been demonstrated to upregulate both P-gp and MRP2, and led to increased P-gp-mediated efflux of digoxin in the brains and kidneys of mice [138,139]. Vitamin D is excreted into breastmilk, though levels are considered insufficient to meet an infant’s nutritional needs and supplementation is recommended for exclusively breastfed infants [21].

Below, we describe the most significant xenobiotic transporters expressed in human MECs, starting with BCRP--the most prominent and well-studied transporter influencing drug concentrations in breastmilk.

4.1. BCRP

The most well-studied transporter in lactation is BCRP, which is encoded by the ABCG2 gene. ABCG2 and BCRP protein expression is induced during lactation in the mammary glands of humans, mice, and cows, and imaging studies show localization to the MEC apical membrane [92,117,118]. A recent proteomic study of human milk fat globules identified BCRP in globule membranes, which are formed through tethering of milk lipid droplets to the MEC apical plasma membrane [140]. BCRP expression is also upregulated following cell differentiation in several in vitro models of the mammary gland, including murine HC11 cells and bovine BME-UV cells [92,93].

Located at the apical membrane, BCRP mediates active efflux of substrates from MEC cells into the alveolar lumen. Multiple BCRP substrates have been identified in breastmilk, and clinically observed M/P ratios suggest an active transport mechanism that cannot be accounted for using traditional M/P calculation methods (see Table 1). To illustrate the impact of BCRP on drug distribution into milk, Ito et al. compared observed M/P ratios from the literature (calculated based on AUC) to those predicted by the pH partition theory for 166 drugs [11]. Of the 16 known BCRP substrates identified, 13 had M/P ratios more than 1.5 times higher than their predicted values. The correlation between observed and predicted M/P values for BCRP substrates was improved after correcting for BCRP transport using flux ratios obtained from in vitro assays with BCRP-expressing MDCK-II cells (r² = 0.89 with correction for BCRP-mediated transport vs. r² = 0.09 with pH partition theory alone) [11].

BCRP transports multiple xenobiotics into breastmilk, and substrate milk concentrations are altered in the absence or inhibition of BCRP. The B vitamin riboflavin is a substrate of BCRP and an essential milk nutrient needed for cellular maintenance and infant growth [141]. In lactating Abcg2^−/− mice fed a diet supplemented with riboflavin, levels of riboflavin and its metabolite flavin mononucleotide were significantly decreased in milk, whereas plasma riboflavin concentrations were increased compared to the wild type [116]. Similarly, milk concentrations and M/P ratios of the BCRP substrates topotecan and cimetidine were both significantly decreased in lactating Abcg2^−/− mice compared to wild-type mice [117]. Milk concentrations of topotecan were also reduced in wild-type mice following administration of the BCRP inhibitor elacridar, suggesting the potential for BCRP-mediated drug interactions in breastmilk [117]. Elacridar administration also led to intracellular accumulation of the BCRP substrate mitoxantrone in undifferentiated and differentiated HC11 cells [93].

BCRP and other ABC transporters may also mediate the concentrations of bile acids in the mammary gland. Blazquez et al. reported that bile acid concentrations in milk and serum were lower in Abcg2^−/− mice compared to wild-type mice. Following administration of taurocholic acid, serum levels of bile acids were increased to a similar degree in both groups of mice; however, milk bile acids increased to a lesser extent in Abcg2^−/− mice, suggesting that BCRP may secrete bile acids into milk [118].

4.2. OCTN1/2

OCTN1 and OCTN2 are ubiquitously expressed, polyspecific, cation uptake transporters that are members of the SLC22 family [142]. OCTN2 is primarily responsible for the uptake of carnitine from the bloodstream into cells of the lung, kidney, liver, heart, and muscle, among others [142]. Carnitine is an essential nutrient for fatty acid metabolism and required dietary supplement for healthy infant development [115,143]. In contrast, carnitine is a weak substrate of OCTN1 [142]. Though the exact role of OCTN1 is not well understood, it has been found to transport the antioxidant ergothioneine and the neurotransmitter acetylcholine [142,144,145]. Xenobiotic substrates of OCTN1 include oxaliplatin [146], pregabalin [147], and metformin [148], among many others [149]. Substrates of OCTN2 include but are not limited to etoposide [150], entecavir [151], and imatinib [152].

Transcripts of both OCTN1 and OCTN2 have been identified in human and murine MECs, though the effect of lactation on their expression is disputed. Alcorn et al. reported approximately 6-fold upregulation of OCTN1 and 4-fold downregulation of OCTN2 in human lactating MECs [46]. In lactating murine HC11 cells, mOctn1 expression was downregulated in gestation and lactation compared to cells from virgin NMRI mice (also known as the Naval Medical Research Institute mouse strain) [92]. To the contrary, Lamhonwah et al. found increased protein expression of mOctn1 and mOctn2 in pregnant and lactating C3H mice (a popular type of inbred mouse strain) [115]. Immunostaining revealed that both proteins appear to co-localize with intracellular fat globules, which travel to the apical membrane via transcytosis and appear on the membranes of fat globules in the alveolar lumen [115]. Fifteen days after ceasing lactation, protein expression of mOctn1 and mOctn2 appeared markedly decreased by immunostaining [115].

In the mammary glands of Sprague-Dawley rats, OCTN2 mRNA and protein levels were higher early in lactation (day 4) compared to later in lactation (day 10), while OCTN1 mRNA and protein levels remained relatively stable [153]. Immunostaining experiments performed on day 4 showed OCTN1 was mainly localized to the apical MEC membrane with slight staining of the basolateral membrane. While OCTN2 was found at the alveolar membrane of the mammary gland, staining was unfortunately inconclusive with regards to MEC localization [153]. An approximately 2-fold increase in both carnitine milk concentrations and the milk/serum ratio was observed on lactation day 4 compared to day 10. When the OCTN2 substrate cefepime was administered, carnitine milk concentrations decreased by 56% on day 4 but had no effect at day 10. Likewise, the cefepime milk/serum ratio was higher at day 4 compared to day 10, further suggesting a time-dependent change in transporter expression [153]. This work demonstrates the potential for drug-nutrient interactions mediated by OCTN2 in a manner similar to BCRP.

4.3. MRPs

The multidrug resistance-associated proteins (MRPs) are a family of ABC efflux transporters capable of transporting a wide range of compounds, though they are typically known for transporting bile acids in addition to glutathione and glucuronide conjugates. Of all the MRPs, MRP1, MRP2, MRP4, and MRP5 have been identified at least on the mRNA level in lactating human MECs [46,118]. The Mrp3 gene has been detected in the mammary gland of lactating rats, but not mice [118].

MRP1 is ubiquitously expressed and transports leukotrienes, prostaglandins, bile acids, glutathione, drugs, and hormone conjugates [154]. MRP2 is an apical efflux transporter that mainly transports the conjugated forms of drugs and bile salts [155]. MRP4 is an efflux transporter for bile acids and multiple drugs that has tissue-specific membrane localization [156]. MRP5 is known to efflux cGMP and cAMP throughout the body [157]. MRP1 and MRP2 gene expression appears to be downregulated in lactation, whereas MRP5 is upregulated compared to non-lactating human tissue [46].

Blazquez et al. sought to understand the impact of cholestasis on bile acid concentrations in breastmilk and related ABC transporter expression in rodent models of the mammary gland [118]. They found that Mrp4 was expressed on the basolateral membrane in MECs of lactating mice and in human MECs derived from non-lactating breast tissue [118]. Bcrp was also identified on the apical membrane of lactating murine MECs. Following bile duct ligation in rats, mRNA and protein levels of Mrp4 and Bcrp were increased, and bile acid concentrations in serum and milk were increased 22-fold and 15-fold, respectively. This shift in bile acids was accompanied by a decrease in the milk/serum ratio of bile acids from 0.29 in control rats to 0.18 in rats with bile duct ligation [118]. These results suggest that MRP4 may play a protective role in regulating bile acid concentrations in milk via reabsorption to prevent infant cholemia. It is currently unknown whether MRP4 or other membrane transporters may reabsorb drugs from breastmilk.

4.4. MDR1 (P-gp) and MDR3

Unlike BCRP, P-gp is downregulated during lactation, with ABCB1 transcript levels decreased up to 52-fold compared to non-lactating human MECs [46]. Nevertheless, P-gp appears to regulate the transfer of multiple substrates into breastmilk, as demonstrated by the predicted vs. observed M/P ratios of tacrolimus and azithromycin [158,159]. P-gp is expressed in human MECs, as well as the mammary epithelial cell line MCF12A, murine HC11 cells, and bovine BME-UV cells [10,92,93]. The exact mechanism for P-gp downregulation in lactation is unknown, though signaling from proinflammatory cytokines has been proposed to play a role [10,160].

Transcripts of the closely related transporter MDR3 have been identified in MCF12A cells, an epithelial cell line derived from healthy non-lactating human mammary tissue [10]. MDR3 protein has also been identified in human MECs in vitro [161]. MDR3 is commonly known for its role in the apical efflux of phospholipids from hepatocytes into bile canaliculi, though it shares some xenobiotic substrates with P-gp [162]. Lactation-related hormones appear to downregulate its expression [10].

4.5. OCT1/3

Transcripts of the organic cation uptake transporters OCT1 and OCT3 have been identified in human mammary gland tissue and the mammary epithelial cell line MCF12A [46,82,163]. OCT1 uptake activity has also been reported in MCF12A cells in vitro using the substrate tetraethyl-ammonium [163]. In addition to many cationic drugs, OCT1 and OCT3 mediate the transport of acetylcholine, hormones, neurotransmitters, and thiamine (vitamin B1) [164]. The membrane localization of OCT1 and 3 in MECs has not been reported. However, OCT1 is expressed on the basolateral membrane of hepatocytes and transfected MDCK cells [165]; hence, it is likely expressed on the basolateral membrane of MECs (Fig. 2).

During lactation, OCT1 gene expression increases in both human and murine MECs, whereas OCT3 expression decreases in human tissue [46,92,125]. In pregnant mice, mOct1 gene expression initially increases in late gestation, peaks in lactation, and decreases following weaning in the murine mammary gland [92,125]. Lactation stage-dependent changes in OCT1 and other transporters may carry implications for a fluctuating drug-nutrient interaction risk occurring for some substrates throughout lactation.

Lactating Sprague-Dawley rats also were found to express transcripts for rOCT1 and rOCT3, but not rOCT2, in mammary gland tissue [12]. Observed M/P ratios following steady-state infusions of the OCT substrates cimetidine, nitrofurantoin, and probenecid were vastly higher than predicted values in rats, suggesting that OCT1 and OCT3 regulate drug concentrations at the blood-milk barrier [12]. In a murine Oct1/2^−/− model, the M/P ratio of thiamine decreased 28-fold compared to wild-type mice, likely due to a lack of Oct1-mediated uptake of thiamine into the MECs in the knockout mice [166].

4.6. OATPs

Organic anion co-transporting polypeptides mediate the uptake of drugs, bile acids, steroid hormones, bilirubin, and signaling peptides [81,132]. Genes for OATP1A2, 2B1, 3A1, 4A1, and 5A1 have been identified in the human mammary gland by PCR [46,92,167]. Alcorn et al. reported that transcripts for OATP1A2 and OATP2B1 are moderately increased in lactation, while OATP3A1 and OATP4A1 transcripts are decreased compared to non-lactating MECs [46]. In contrast, a more recent PCR study found that mOatp1a5 (an analogue of human OATP1A2) decreases beginning in gestation and continuing throughout lactation and involution in mice [92]. In non-lactating human mammary tissue, OATP2B1, OATP3A1, and OATP5A1 proteins are present on the plasma membranes of MECs in the alveoli and lactiferous ducts, though localization on the basolateral vs. apical membrane remains uncertain [167]. While the presence of these OATPs has been described, functional studies have not been performed with MECs or animal models of the mammary gland to the authors’ knowledge. It is currently unknown how or whether these transporters mediate the transfer of endogenous substrates into milk. Furthermore, while multiple OATP xenobiotic substrates have been identified in breastmilk, the roles of these transporters in contributing to the observed M/P ratio have not been described. Active transport has been proposed as a possible mechanism for the secretion of the OATP substrates azithromycin, methotrexate, atenolol, and doxorubicin into breastmilk (see Table 1) [39,48,159,168].

4.7. PEPT1/2

Though both PEPT1 and PEPT2 genes have been identified in lactating MECs, few studies have been performed to investigate their function [46]. PEPT1 is important for dietary uptake of di- and tripeptides from the lumen of the small intestine [169]. While PEPT2 also transports di- and tripeptides, it is located in the kidney, brain, and lung epithelia [169,170]. Located in the apical membrane of the renal proximal tubule cells, PEPT2 is responsible for the reabsorption of peptides from the glomerular filtrate. PEPT2 appears to have inducible gene expression in lactation, as expression is negligible in non-lactating MECs [46]. Interestingly, PEPT2 protein is localized to the apical membrane of MECs in the alveoli and lactiferous ducts of humans and rats [171]. Several xenobiotic PEPT2 substrates, including amoxicillin, cephalosporins, and captopril, have been identified at low concentrations in breastmilk [171]. Ito et al. hypothesized that PEPT2 may reabsorb the substrates captopril, cephalexin, and cefadroxil, which were found to have much lower observed M/P ratios than those predicted by passive diffusion [11]. It is currently unknown whether PEPT2 significantly affects the concentrations of these drugs in milk, or whether these drugs interfere with PEPT2-mediated nutrient transport.

4.8. SVCT2

SVCT2 is a sodium/vitamin C cotransporter that mediates the uptake of vitamin C in the brain, eye, bone marrow, heart, lung, adrenal glands, intestine, and skeletal muscle [172]. Vitamin C is an essential nutrient and antioxidant that is secreted into breastmilk, and milk concentrations increase with dietary intake [173]. While oral vitamin C supplementation produces mild increases in milk concentrations for well-nourished individuals (1.2-fold increase), concentrations of vitamin C increased 3-fold in the milk of poorly nourished individuals, suggesting that vitamin C concentrations are regulated via a transport mechanism in MECs [173]. SVCT2 gene expression is increased approximately 2-fold in lactating vs. non-lactating MECs [46]. SVCT2-mediated vitamin C uptake is inhibited by steroid hormones, flavonoids, and the NSAIDs diclofenac and indomethacin [174]. Although the localization of SVCT2 is not known in lactating MECs, in breast cancer cells SVCT2 is localized mainly to the nucleus and mitochondria with lower expression on the plasma membrane [80]. The potential involvement of SVCT2 in transporting drugs remains unknown.

4.9. RFVT2

RFVT2, encoded by SLC52A2, is a ubiquitously expressed membrane transporter responsible for the uptake of vitamin B2, or riboflavin. Murine RFVT2 mRNA and protein are upregulated in the MECs of lactating mice [175]. In transfected MDCK-II cells, RFVT2 was localized to both the apical and basolateral membranes; however, the exact membrane localization has not been confirmed in human or animal mammary gland tissue [175]. It appears that BCRP and RFVT2 may work together to maintain riboflavin milk content, with RFVT2 mediating MEC uptake on the basolateral membrane and BCRP (and possibly RFVT2) regulating riboflavin levels in milk at the apical membrane.

4.10. CNT1/3 and ENT1/3

Little is known about the roles of constitutive nucleoside transporters (CNTs) and equilibrative nucleoside transporters (ENTs) at the blood-milk barrier. As their names suggest, both transporter families are responsible for the uptake of nucleosides and some nucleobases, which serve as the building blocks of nucleic acids, coenzymes, and signaling molecules for the breastfed infant [176]. CNT1, CNT3, ENT1, and ENT3 transcripts have all been identified in human MECs [46]. Gene expression is upregulated for all of these transporters in lactation with the exception of ENT1, which appears to be mildly downregulated [46]. Multiple anti-infective and anti-cancer nucleoside analogues are substrates and/or inhibitors of CNT and ENT transporters, including abacavir, entecavir, emtricitabine, ribavirin, gemcitabine, 5’-fluorouracil, cytarabine, and zidovudine [177–181] (note that abacavir, emtricitabine, and zidovudine are listed in Table 1; there is insufficient data on the excretion of the other drugs listed here in breastmilk). It is currently unknown whether these CNTs and ENTs play a significant role in affecting the concentrations of drugs in milk.

5. Conclusion

A multitude of SLC and ABC transporters are expressed on the membrane of MECs where they may work in concert to regulate the concentrations of nutrients and drugs in breastmilk. For many drugs, knowledge of ion trapping, lipophilicity, and protein binding properties may be sufficient to reasonably predict breastmilk concentrations and infant drug exposure. In some instances, however, xenobiotics can be actively transported into milk by transporters in MECs, resulting in increased infant exposure to drugs, potentially posing a safety risk for this vulnerable population. Medications taken by lactating individuals may also competitively inhibit transporters involved in nutrient secretion, which in turn can impact milk composition [12]. This historically overlooked phenomenon may result in reduced concentrations of essential dietary vitamins and endogenous compounds needed for developing infants. While the influence of BCRP on drug and milk nutrient composition has been fairly well described, the localization, function, and drug interaction potential of many transporters have yet to be fully characterized in human MECs. This review illustrates the need for further functional studies to understand the significance of these transporters at the blood-milk barrier to better understand the mechanisms affecting active drug and nutrient transport into human milk. Such insights are crucial for developing methods to predict drug concentrations in human milk and to support clinicians and lactating individuals in making rational decisions to balance the benefits of breastfeeding and the risks of drug exposure to infants.

6. Expert Opinion: Knowledge Gaps and Recommendations for Future Research

Apart from BCRP, little information exists on the roles of most membrane transporters identified in the human mammary gland, including OATPs, OCTs, most MRPs, P-gp, MDR3, ENTs, CNTs, SCVT2, RFVT2, and PEPTs. To gain a mechanistic understanding of carrier-mediated drug transfer into the breastmilk, it is imperative to identify and characterize the transporters in MECs that influence drug transfer into human breastmilk. Approaches such as quantifying the absolute protein expression of these transporters and determining their membrane localization in human MECs in various lactational stages would provide molecular and biochemical clues to infer their potential roles in drug transport across the blood-milk barrier. Pharmacokinetic studies in lactating knockout animal models would be invaluable in ascertaining the in vivo role of a specific transporter in drug transport into milk, although species differences should be considered in translating to humans. The development of in vitro cell models that replicate human lactational physiology and transporter expression would be crucial for gaining a mechanistic understanding of carried-mediated, or non-mediated, transport pathways in MECs. Importantly, quantitative analysis of the drug transport characteristics in these models may allow accurate prediction of M/P ratios through in vitro to in vivo extrapolation (IVIVE). In addition to mechanistic and translational studies in non-clinical models, several other knowledge gaps should be addressed.

To date, very few studies have sought to determine whether pharmacogenetic differences in transporter expression and/or activity impact the concentrations of key nutrients and drugs in breastmilk. BCRP is known to have a common c.421C>A single nucleotide polymorphism in the ABCG2 gene, which is found in around 10% of Caucasians and Brazilians, 1–4% of African Americans, and up to 36% of Japanese and Chinese people [182–184]. The c.421C>A polymorphism results in a Q141K amino acid change (Gln141Lys) and has been associated with a decrease in BCRP protein expression and activity [185]. Surprisingly, a recent study reported that the c.421C>A is significantly associated with a higher M/P ratio of nifedipine [182]. Clearly, additional studies are needed to resolve this apparently paradoxical observation and to determine whether pharmacogenetic variations in BCRP could influence drug concentrations in breastmilk. To our knowledge, the impact of pharmacogenetic differences in other transporters has not been evaluated with regards to breastmilk drug concentrations.

Epigenetic regulation is known to influence transporter expression and activity in other tissues; however, this has yet to be fully examined in relation to breastmilk nutrient composition. Individual factors that may influence transporter expression and activity in the mammary gland include but are not limited to age, disease state, diet, BMI, alcohol intake, and smoking, among others. Environmental exposure to endocrine disrupting compounds such as bisphenol A has been associated with altered mammary gland development and impaired lactation in animals, and may lead to an increased risk of breast cancer [186,187]. The pesticide degradation product DDE is excreted into breastmilk, and serum DDE levels have been associated with shortened lactation in humans [186]. It is currently unknown how exposure to these compounds in early life and mammary gland development influence the expression of MEC transporters and subsequent nutrient levels in milk.

Physiologically based pharmacokinetic (PBPK) models are physiologically relevant mathematical models used to predict the absorption, distribution, metabolism, and excretion of drugs in humans. Several lactation PBPK models that incorporate mammary gland physiology, as well as drug ionization and partitioning, have been developed to predict milk concentrations of drugs transported into milk through passive diffusion [188,189]. Nauwelaerts et al. recently described a PBPK model that reasonably predicts the M/P ratios and relative infant doses of a physiochemically diverse set of ten medications that have been identified in human milk [190]. The authors noted that single-dose exposure of the OCT substrate metformin in milk was underpredicted, possibly owing to active transport. They recommended further model refinement to better predict the elimination of transporter substrates using an in vitro model of the blood-milk barrier [79,190]. These findings highlight the current need for basic studies that characterize the kinetics of drug transport across the blood-milk barrier, which are needed to build accurate predictive models. As our understanding of the expression, localization, and function of various transporters in MECs advances, it will be possible to integrate carrier-mediated transport into the general lactation PBPK models. Using drug-specific information derived from in vitro systems, such as transporter-expressing cell lines or vesicles, the milk secretion of drugs with carrier-mediated transport can be predicted through IVIVE and PBPK modeling. Once validated, these models can predict milk concentrations and exposure levels of drugs that undergo similar carrier-mediated transport pathways in MECs.

Little clinical evidence of transporter-mediated excretion in breastmilk exists beyond case reports or underpowered prospective studies. Despite the need for clinical research, practical challenges make recruitment and retention of pregnant and lactating persons difficult as research subjects navigate the major life event of caring for an infant. In addition, individuals may face multiple societal, policy, health, and personal barriers to continued breastfeeding, making participant retention difficult [191].

In the near future, improved predictive models that combine conventional M/P ratio predictions with knowledge of MEC membrane transporters may be designed to estimate breastmilk drug concentrations and the impact of drugs on milk nutrients. Such models, once validated with clinical data, would help to support evidence-based decisions to continue, dose adjust, or discontinue a drug during lactation, and would help bridge the knowledge gap on medication use in breastfeeding.

Article Highlights:

• Traditional models for predicting milk concentrations of drugs rely on the free drug hypothesis and the assumption that most drugs passively diffuse into milk; however, this model falls short of predicting milk concentrations for drugs with active transfer into milk.

• Drugs and nutrients may be transferred to milk via passive diffusion, active transport, lipid co-transport, transcytosis, exocytosis, or paracellular transport depending on their physiochemical properties and specific interaction with transport proteins in the mammary epithelial cells.

• Many ATP-binding cassette (ABC) and solute carrier (SLC) transporters have been identified in the human mammary epithelial gland and could actively influence the concentrations of xenobiotics at the blood-milk barrier.

• While the impact of the breast cancer resistance protein (BCRP) has been demonstrated, the localization, function, and drug interaction potential of many transporters have yet to be fully characterized in human mammary epithelial cells.

• A comprehensive and quantitative understanding of transporters in mammary epithelial cells could help predict active milk drug transfer as well as risk of drug-nutrient interaction at the blood-milk interface.